Comparative study between nadolol and 5-isosorbide mononitrate vs. endoscopic band ligation plus sclerotherapy in the prevention of variceal rebleeding in cirrhotic patients: a randomized controlled trial


Dr D. Kravetz, Division of Gastroenterology, VA San Diego Healthcare System, 3350 La Jolla Village Drive 9111D, San Diego, CA 92161, USA.


Background  After variceal bleeding, cirrhotic patients should receive secondary prophylaxis.

Aim  To compare nadolol plus 5-isosorbide mononitrate (5-ISMN) with endoscopic band ligation. The end points were rebleeding, treatment failure and death.

Methods  One hundred and nine cirrhotic patients with a recent variceal bleeding were randomized: nadolol plus 5-ISMN in 57 patients and endoscopic band ligation in 52 patients.

Results  The mean follow-up was 17 and 19 months in nadolol plus 5-ISMN and endoscopic band ligation groups, respectively. No differences were observed between groups in upper rebleeding (47% vs. 46%), variceal rebleeding (40% vs. 36%), failure (32% vs. 22%), major complications (7% vs. 13.5%) and death (19% vs. 20%), respectively. The actuarial probability of remaining free of rebleeding, failure and deaths were similar in both groups. Time to rebleeding shows that endoscopic band ligation patients had an early rebleed, with a median of 0.5 month (95% CI: 0.0–4.2) compared with patients from nadolol plus 5-ISMN, 7.6 months (95% CI: 2.9–12.3, P < 0.013). Multivariate analysis indicated that outcome-specific predictive factor(s) for rebleeding was Child A vs. B + C (P < 0.01); for failure was Child A vs. B + C (P < 0.02); and for death ascites (P < 0.01) and rebleeding (P < 0.02).

Conclusion  This trial suggests no superiority of endoscopic band ligation over nadolol plus 5-ISMN mononitrate for the prevention of rebleeding in cirrhotic patients.


Variceal rebleeding is a common event that develops frequently after acute variceal bleeding in patients with cirrhosis. The incidence can be as high as 30–50% within the first days of the acute episode and 60–80% at 1 year.1–3 Therefore, after controlling acute variceal bleeding, prevention of a rebleeding episode would be mandatory.4 Both non-selective beta-blockade and endoscopic treatment have been shown to reduce variceal rebleeding and mortality compared with untreated controls.5–7 Meta-analysis of controlled trials demonstrated that the rate of rebleeding is decreased by 40% and overall survival is improved by 20% with propranolol.7 However, this treatment presents some problems, such as a high number of non-responders and contraindications for its use. Non-selective beta-blockers decrease portal pressure and blood flow in the porto-collateral system.8 Although some decrease in portal pressure is observed in 60–70% of patients receiving propranolol, a reduction of ≥20% or <12 mmHg9, 10 is achieved in only 10–30% and 10–15% respectively.11, 12 A vasodilator, 5-isosorbide mononitrate (5-ISMN), has been shown to decrease portal pressure in cirrhotic patients. These beneficial homodynamic effects of nitrates are due to a reduction of the intrahepatic resistance, probably through a relaxation of the smooth muscle and myofibroblasts.13–15 Two recent controlled clinical trials found the combination of beta-blockers with 5 ISMN to be superior to sclerotherapy16 and beta-blockade alone.17 Endoscopic band ligation (EBL) has also shown to decrease the frequency of rebleeding, complications and death compared to sclerotherapy and should be the preferred endoscopic treatment.18–20 In addition, complete eradication of varices is obtained in a shorter period of time with banding than sclerotherapy. Recently, Lo et al.21 demonstrated that adding small volume sclerotherapy at the end of variceal band ligation was superior to band ligation alone in terms of rebleeding and variceal recurrence. Until now three studies have been published comparing the efficacy of beta-blocker plus 5-ISMN with EBL in the treatment of oesophageal variceal rebleeding. These three studies show different results in relation to variceal rebleeding: one showed that pharmacological therapy was better;22 another showed a benefit of EBL23 and the third did not find any significant differences between treatment groups.24 Therefore, the aim of the present study was to compare the clinical efficacy between the two ‘most’ effective treatments, nadolol (N) + 5-ISMN vs. EBL with one or two sessions of small amount of sclerotherapy, at the end of banding treatment23 for the prevention of variceal rebleeding in cirrhotic patients.

This trial was designed on the basis of a superiority hypothesis (i.e. that endoscopic therapy decreases the 2-year risk of rebleeding from 35% (with drugs) to 10%).

Material and methods

Patients selection

This randomized controlled clinical trial was conducted in two centres: Hospital de Gastroenterologia Bonorino Udaondo (Coordinating Center) and Hospital Nacional Prof. A. Posadas. From August 1998 to December 2002, 192 consecutive cirrhotic patients with a history of variceal bleeding were evaluated at the two hospitals. The inclusion criteria were: (i) cirrhosis diagnosed by clinical history, laboratory, ultrasonography and liver biopsy when possible; (ii) the index variceal bleeding episode, demonstrated by emergency endoscopy, in the prior 3 months without any other evidence of bleeding within this time; and (iii) informed consent signed by the patient. Exclusion criteria were: (i) portal vein thrombosis; (ii) fundal gastric varices; (iii) any malignant tumour; (iv) more than one endoscopic treatment after the control of acute variceal bleeding; (v) creatinine ≥1.6 mg/%; (vi) contraindication to receive beta-blockers; (viii) bacterial infection and/or encephalopathy; and (ix) inability to attend follow-up visits. In compliance with the exclusion criteria, 83 patients had to be excluded (Figure 1).

Figure 1.

 Trial profile.

The patients were classified following the Child–Pugh classification17 on the basis of data collected at randomization. In addition to the general laboratory tests, all the patients were tested for the presence of hepatitis B (HBsAg), C (HCV) and human immunodeficiency virus.

Design of the study

Randomization was carried out utilizing consecutively numbered opaque-sealed envelopes containing the treatment assignment. Randomization of 109 (Figure 1) patients was determined by means of a table of random numbers.

Group N + 5-ISMN: 57 patients received the pharmacological treatment.

Group EBL: 52 patients were treated with EBL and one or two low-volume sclerotherapy session at the end of ligation.21

The human investigational review committees of the participating hospitals approved this protocol.

Pharmacological treatment

Patients allocated to pharmacological treatment started with a dose of nadolol 40 mg/day in a single daily administration. Every other day, the dose was titrated up to achieve a 25% decrease in resting heart rate or until 55 bpm. Then 5-ISMN was added starting with 10 mg/b.i.d., which was increased to 40 mg/b.i.d. unless hypotension (systolic ≤90 mmHg) or severe headache occurred. Patients with side effects related to beta-blockers that required withdrawal of the treatment were considered treatment failure. Compliance was assessed at each follow-up visit by careful questioning of the patient, his/her relatives and by measurement of heart rate. Patients who failed to attend more than two follow-up visits, and/or showed no decrease in heart rate more than once, were classified as non-compliant.

Endoscopic procedures

After light sedation with intravenous diazepam as needed, an endoscope (Fujinon UGI F7, Buenos Aires, Argentina) was introduced and since initially the single band device (Bard Interventional Products, Tewksbury, MA, USA) was used, the aid of the overtube was necessary. Later, a commercial multibanding ligation device became the standard instrument. Variceal ligation was performed beginning at the most distal variceal column and proceeding to the next proximal varix; no more than 10 bands were placed per session. The same treatment was repeated after 2 weeks until variceal eradication was obtained. The 2-week interval was set to allow the healing of banding ulcers; however, this schedule was modified only if large ulcers or oesophageal stenosis was found. Thereafter, one or two sclerotherapy sessions were given to the residual varices. No more than 8 cm3 of polidocanol 2% injection was performed per session.21


In the pharmacological group, after the medication dose was reached the patients were followed every 3 months. In the EBL group, after eradication of the varices the patients were followed by endoscopy at 1-, 3- and 6-month intervals and every 6 months thereafter and were censored at the time of the last visit. If varices recurred after eradication, patients received the same treatment.

Patients who failed to show up for a period of at least 6 months were counted as lost to follow-up. During the treatment EBL patients received sucralfate 1 g/q.i.d. Patients presenting with ulcers or stenosis after 1 month were treated with 40 mg/day of omeprazole during 8 weeks instead of sucralfate. Phone and letter recalled patients who did not attend follow-ups. The study was stopped at 6 months of follow-up of the last patient included.

End points

The primary end points of this investigation were gastrointestinal (GI) and variceal rebleeding. Secondary end points were survival time and complications.


Variceal eradication

Variceal eradication was considered when varices disappeared or reduced to a grade I and/or was not possible to be aspirated into the ligation chamber.

Gastrointestinal rebleeding was defined as any episode of haematemesis, melaena or both that occurred during the follow-up period.

Variceal rebleeding

Variceal rebleeding was considered if an upper endoscopy performed within 24 hours of a haematemesis and/or melaena showed: (i) presence of active variceal bleeding; (ii) clot over a varix that could not be removed by washing; (iii) presence of white nipple over a varix; (iv) oesophageal varices and fresh blood in the stomach without any other identified bleeding lesions. All the rebleeding patients were treated with sclerotherapy. Severe variceal bleeding was considered if at least two of the following criteria were observed: (i) admission to the hospital; (ii) transfusion of one unit of blood per hour during ≥3/h; (iii) decrease of ≥3 points in haemoglobin from baseline; (iv) systolic blood pressure ≤100 mmHg or a postural change of ≤20 mmHg and/or HR ≥100/min; (v) haematocrit drop ≥5% in relation to the last value.

Treatment failure

Treatment failure was considered when (i) two or more episodes of severe variceal bleeding occurred;16 and (ii) the presence of an episode of acute variceal bleeding that could not be controlled by pharmacological, endoscopic and/or balloon therapy and required surgical treatment. These patients were treated according to the protocol prevailing at our units. Patients who developed major complications related to the treatment were withdrawn from the trial and censored as treatment failure.


Oesophageal ulcers was only considered a complication if it produced a bleeding episode or a ulcer which did not disappear after 1 month of sucralfate administration and omeprazole had to be added to the treatment. Oesophageal stenosis was considered if persistent dysphagia was present and reduction in oesophageal calibre was observed during endoscopy. Other complications were defined as any abnormal physical signs, symptoms or laboratory changes occurring during the study and considered to be related to the treatments under investigation.

Follow-up was defined from the date of randomization until the date of treatment failure or death. Statistical analyses were performed using the last visit date or death as the end point for patients without rebleeding. In patients who remained alive at the end of the study, the survival analysis was performed at the time of last visit.

Statistical analysis

A sample size of at least 51 patients in each group was required to achieve a decrease of variceal rebleeding of 25% by EBL with an α and β errors of 0.05 and 0.2 respectively. This difference was assumed because at the time of the trial design only one study was published comparing N + 5-ISMN vs. sclerotherapy that show the actuarial risk for recurrent bleeding at 2 years was 35%.16 At the same time for EBL plus sclerotherapy the actuarial risk of rebleeding at 2 years was approximately 10%.21 Values are given as mean ± S.E. The statistical analysis was performed according to an intention-to-treat strategy. Student's t-test was used to compare mean differences between groups. Qualitative data were analysed by chi-squared or Fisher's exact test. Actuarial probabilities were calculated by the Kaplan–Meier method, and differences between groups were gauged with the log-rank test. The Cox proportional hazards model was used to identify the variables, predicting time to rebleeding, failure, survival, variceal eradication and recurrence. All the Cox regression analyses included the following variables: treatment, previous treatment and bleeding episodes, age, sex, size of oesophageal varices, aetiology, number of sessions, number of bands, Child–Pugh class and score, alanine aminotransferase, platelet count, blood urea nitrogen and creatinine. Two-sided P-values of ≤0.05 were considered to be statistically significant.


Both groups at the time of inclusion were similar with regard to: demographics and clinical and laboratory data. The time elapsed from the index bleeding to randomization were also similar between groups, through day 15, 49% and 50% of patients in the N = 5-ISMN and EBL groups were randomized, and 25% and 27% of patients were randomized between days 16 and 30. (Table 1). Only one patient did not take nadolol because of a massive variceal bleeding and death after randomization. Eight patients did not tolerate the 5-ISMN for arterial hypotension and were treated with nadolol alone.

Table 1.   Demographic characteristics of patients treated with nadolol plus isosorbide mononitrate (N + 5-ISMN) or with endoscopic ligation at randomization
 N + 5-ISMN (n = 57)EBL (n = 52)
  1. Values are given as mean ± S.E.

  2. EBL, endoscopic band ligation.

  3. n = number of episodes.

  4. † Time elapsed between bleeding index and inclusion in the study.

  5. ‡ This patients were not censored for treatment failure and mortality rate.

  6. § For variceal eradication.

Age (years)51 ± 1053 ± 10
M/F (n)37/2035/17
Aetiology (%)
 Alcohol30 (53)24 (46)
 Alcohol + virus C 7 (12) 6 (11.5)
 Virus C and/or B 8 (14) 7 (13.5)
 Others12 (21)15 (29)
Child class (%)
Child score7 ± 1.97 ± 1.6
Ascites (%)3736
Haemoglobin (g/L)10.8 ± 2.19.9 ± 1.7
Bilirubin (mg/dL)1.9 ± 1.82.1 ± 2.9
Albumin (g/dL)3.0 ± 0.72.8 ± 0.6
Prothrombin (%)64 ± 18.967 ± 18.8
Creatinine (mg/dL)0.98 ± 0.90.76 ± 0.18
Time of inclusion (days)†26 ± 2425.4 ± 22
Follow-up (months)
Lost of follow-up (n)‡49
Nadolol mg/day88 ± 68
5-ISMN mg/day57.7 ± 27
Number EBL session§3.4 ± 1.2

Among the patients in EBL, varices were eradicated in 36 patients (69%) after a mean of 3 ± 2 sessions and recurred in 16 (42%) during follow-up. In 16 patients (31%), the varices could not be eradicated due to rebleeding and death in five, rebleeding and failure in five, liver failure and death in one, and loss to follow-up in five.


During follow-up a non-significant difference was observed in GI rebleeding between treatment groups, 27 (47.4%) patients in the N + 5-ISMN group and 24 (46.2%) in EBL group (P = 0.8; RR 0.97; 95% CI 0.7–1.7). In addition, no significant differences were observed when the cause of GI rebleeding, the total number of rebleeding and the number of rebleeding per patient were analysed (Table 2). The probability of variceal rebleeding was 27 (47.4%) patients in the N + 5-ISMN group and 24 (46.2%) in EBL group (P = 0.6; RR 0.9; 95% CI 0.6–1.7). The cumulative probability to be free of GI rebleeding was similar between groups (Figure 2a). The probability of being free of variceal rebleeding at 1, 2 and 3 years was similar in N + 5-ISMN (57%, 44% and 32% respectively) and in EBL (64%, 61% and 51% respectively, log-rank test P = 0.96) (Figure 2b).

Table 2.   Rebleeding rates and sources in both treatment groups
 N + 5-ISMN n = 57 (%)EBL n = 52 (%)
  1. GI, gastrointestinal; N + 5-ISMN, nadolol plus 5-isosorbide mononitrate; EBL, endoscopic band ligation.

  2. Values are given in absolute numbers (%). Non-significant differences were observed between groups.

GI rebleeding27 (47.4)24 (46.2)
 Oesophageal varices21 (78)17 (71)
 pH gastropathy1 (4)2 (8)
 Oesophageal ulcers3 (8)
 Others2 (7)2 (16)
 Undetermined3 (11)
Rebleeding episodes4535
Variceal rebleeding episodes3526
Rebleeding episodes/patients0.630.51
Figure 2.

 Probability to be free of gastrointestinal rebleeding (a) and variceal rebleeding (b) in patients treated with nadolol plus 5-isosorbide mononitrate (—) and endoscopic band ligation (- - - -). No statistical significant difference was observed between groups.

We also performed a statistical analysis of the patients included until day 15, between 16 and 30 days and more than 30 days of the index bleeding episode. No statistical significant differences were observed in relation to: GI rebleeding, variceal rebleeding, treatment failure and mortality between groups. In addition, the actuarial probability of survival (Kaplan–Meir) was analysed for these periods and no statistical significant differences were observed between groups (Breslow test P = 0.61).

To analyse if GI rebleeding rate might have been different if all patients lost to follow-up were available for analyses, we performed the following statistics. Considering the nine patients in the EBL group lost to follow-up as GI rebleeders plus those who actually rebled (total 33 patients) vs. the patients who rebled in the pharmacological group (27 patients) the result showed no difference (P = 0.39). Vice versa, considering the four patients in the pharmacological group lost to follow-up as GI rebleeders plus those who actually rebled (total 31 patients) vs. the patients who rebled in the EBL group (24 patients) the result showed no difference (P = 0.09).

When time to rebleeding was analysed, we observed that EBL patients had an early GI rebleeding time, with a median of 0.5 month (95% CI: 0.0–4.2) compared to patients from the N + 5-ISMN with a median time of 7.6 months (95% CI: 2.9–12.3, P < 0.013). A total of 40 patients in both groups had ascites at inclusion, and 60% of them had an episode of GI rebleeding, this was significantly higher than the GI rebleeding rate observed in patients without ascites (39%; P < 0.03; RR 2.3; 95% CI 1.2–5.2). The univariate analysis demonstrated that the factors associated with GI rebleeding were Child–Pugh grade A vs. B + C and ascites at inclusion. However, Cox logistic regression multivariate analysis shows that only Child–Pugh grade A vs. B + C independently predicted GI rebleeding (Table 3).

Table 3.   Multivariate analyses
Outcome variableExplanatory variablesOdds ratio (95% CI)P
RebleedingChild A vs. B + C2.3 (1.2–4.5)0.01
FailureChild A vs. B + C3.2 (1.2–8.8)0.022
MortalityAscites at inclusion3.4 (1.3–9.1)0.015
Rebleeding3.3 (1.2–9.4)0.026

Treatment failure

Our results did not show any significant differences for treatment failure between the groups. Treatment failure was observed in 18 (32%) patients in N + 5-ISMN and 12 patients (22%) in EBL. The cumulative probability to be free of treatment failure at 1, 2 and 3 years were also similar (74%, 56% and 42% in N + 5-ISMN vs. 75%, 71% and 53% in EBL; log-rank test P = 0.62). Multivariate analyses demonstrated that Child–Pugh grade have an independent prognostic factor (Table 3).

When treatment failure was analysed in relation to the inclusion time no significant difference was observed between groups in relation to treatment failure as well as the probability of being free of failure at 2 years.


Complications that required suspension of the treatment and/or admission to the hospital were not significantly different in N + 5-ISMN (7%) than in EBL (13.5%) (P = 0.2; RR 2; CI 0.6–10). Only one patient in each group had to be withdrawn from treatment. Minor complications that did not require any modification of the treatment were also similar in both groups [22 patients in N + 5-ISMN (39%) vs. 16 patients (31%) in EBL] (Table 4).

Table 4.   Complications associated with pharmacological treatment and band ligation
  1. N + 5-ISMN, nadolol plus 5-isosorbide mononitrate; EBL, endoscopic band ligation; SBP, spontaneous bacterial peritonitis.

 N + 5-ISMN (n = 57) EBL (n = 52) 
  Bleeding ulcers3
 Total4 (7%)Total7 (13.5%)
 Hypotension8Transitory dysphagia10
 Weakness4Chest pain1
 Sexual dysfunction1Fundal varices1
 Total22 (39%)Total16 (31%)


Mortality rate as well as bleeding-related deaths was similar in both groups. A total of 11 patients (19.3%) died in N + 5-ISMN and 10 patients in EBL (19.2%) (P = 0.9; RR 1.0; 95% CI 0.4–2.1), of whom 82% and 60% (ns) respectively were variceal rebleeding-related deaths (Table 5). The cumulative probability of survival at 1, 2 and 3 years were similar in both groups of patients (83%, 69% and 69% for N + 5-ISMN and 84%, 80% and 64% for EBL; log-rank test P = 0.94) (Figure 3).

Table 5.   Cause of deaths
 N + 5-ISMN n = 57 (%)EBL n = 52 (%)
  1. No statistical significant difference was observed between groups. Values are absolute numbers and (%).

  2. N + 5-ISMN, nadolol plus 5-isosorbide mononitrate; EBL, endoscopic band ligation.

Total mortality11 (19.3)10 (19.2)
Variceal bleeding9 (81)6 (60)
Liver failure12
Figure 3.

 Probability of survival in patients treated with nadolol plus 5-isosorbide mononitrate (—) and endoscopic band ligation (—-). Kaplan–Meier curve calculated with all the death patients. No statistical significant difference was observed between groups.

A survival analysis was performed between the group of patients included at different times and no significant difference was observed in relation to survival rate as well as the cumulative probability of survival at 2 years (Figure 4). The univariate analysis showed that the factors associated with death were ascites at inclusion (HR 6.33; 95% CI 2.2–18.1; P < 0.001), Child A vs. B + C (HR 4.35; 95% CI 1.19–15.9; P < 0.014), GI rebleeding (HR 4.4; 95% CI 1.6–14.4; P < 0.03) and variceal rebleeding (HR 4.3; 95% CI 1.6–11.8; P < 0.007). However, multivariate analysis demonstrated that only ascites at inclusion and GI rebleeding independently predicted death (Table 3). The probability to be free of variceal rebleeding, treatment failure and survival in Child–Pugh A patients vs. the Child–Pugh B + C patients was investigated. These probabilities were significantly higher in Child A patients than in B + C patients (Table 6).

Figure 4.

 Probability of survival in patients included at different periods of time from the acute variceal bleeding episode. No statistical significant difference was observed between groups.

Table 6.   Comparison between all the patients Child–Pugh A vs. B/C
 1 year (%)2 years (%)3 years (%)
  1. * Long rank test P = 0.0024.

  2. † Long rank test P = 0.0033.

  3. ‡ Long rank test P = 0.0049.

Probability to be free of variceal rebleeding
Probability to be free of treatment failure
Probability to survival


In the present study we compared the efficacy of N + 5-ISMN in the secondary prophylaxis of oesophageal variceal bleeding, with EBL plus one or two sessions of small sclerotherapy injection. The results demonstrated that no significant differences were observed in relation to the rebleeding rate, causes of rebleeding, complications and mortality rate between the two treatment groups. These results are consistent with a recent meta-analysis published by Mela et al.25 with three similar studies.22–24 These three studies present some differences in the protocol design, but similar comparison between N + 5-ISMN and EBL, with the present clinical trial. When the rebleeding was analysed they showed differences. In one of these trials the N + 5-ISMN was superior,22 another trial demonstrated a benefit of EBL23 and the third study did not find any significant differences between both groups.24 The different results obtained in these trials are not completely clear and are also difficult to explain.26, 27 They could be related to patient selection, approach to the pharmacological treatment, such as giving both drugs to all the patients22, 23 or only adding 5-ISMN to the patients who did not respond to beta-blockade.24 Possible explanations, difficult to demonstrate, include differences in the study centre and local expertise of treatments26 and/or the unpredictable bias committed when the studies cannot be performed in a double-blinded design. In addition, it is very difficult to understand how the combination of all these factors could influence the results of any given treatment. Besides confirming the results of the recent meta-analysis, the present study evaluates the efficacy of similar treatments in a country with a different healthcare system. In that regard it is important to clarify that we included a high percentage of patients where their index bleeding were treated in the participant's hospitals, however, other patients were transferred after the acute variceal bleeding was controlled.27 This difference in the healthcare system could be in part responsible for the time elapses between the index bleeding and the randomization time as well as the higher loss of follow-up between our study and the other previous studies.

However, it is important to comment that most of the patients (75%) were included during the first month after the index bleeding, and no significant differences were observed between the groups included until day 15, between days 16 and 30 and more than 30 days of the index bleeding in relation to upper GI rebleeding, variceal rebleeding and treatment failure. Furthermore, the actuarial probability of variceal rebleeding, treatment failure at 2 years did not show any significant difference between these groups. It is possible that the patients lost to follow-up might have influenced the results observed in the rebleeding, however, the best- and worst-case scenario analysis performed did not show any statistical differences, suggesting that the patients lost did not influence our results. Another important difference in the rebleeding rate observed in patients included in the pharmacological group could be related to the doses of beta-blockers and 5-ISMN administered in the four studies. The present study in comparison with the one of Villanueva et al.22 presented a higher rebleeding rate (47.3% vs. 33% respectively), even though the patients received a very similar dose of nadolol (88 mg/day vs. 96 mg/day respectively) and 5-ISMN (58 mg/day vs. 66 mg/day respectively). On the contrary, the higher rebleeding rate observed by Lo et al. (57%)23 in the N + 5-ISMN may be more understandable. In that study, the patients were treated with much lower doses of nadolol (48 mg/day) and 5-ISMN (30 mg/day). The rebleeding rate observed by Patch et al.24 is difficult to compare with the rest of the trials because in this study only 41% of the patients in the pharmacological group received both drugs.

When the results of EBL were analysed the rebleeding rate observed by us and the ones from Villanueva et al.22 and Patch et al.24 were similar (46%, 49% and 53% respectively). These results are different from the one reported by Lo et al.23 in which the rebleeding rate in that group of patients was much lower (38%). The reason for this difference is not clear, however, it could partially be due to the fact that the latter study took place in a unit specialized in endoscopy, where a better efficacy and/or lower complication rate in comparison with other studies have been previously published.26, 28 The only significant difference observed in the present study was the median of the rebleeding time observed between treatment groups that were significantly shorter in the EBL than the N + 5-ISMN. Several reasons for this difference are possible. One is that the patients treated with endoscopic banding were still at risk of bleeding until the varices were eradicated;29, 30 another is that some patients may have presented a rebleeding secondary to oesophageal ulcers. As recently published, this early rebleeding time could also be in part related to the increase in portal pressure during the first 48 h after EBL.31 By contrast, it has been reported that N + 5-ISMN treatment can produce a rapid decrease in portal pressure of ≥20% of basal values in 50% of the patients.15 Furthermore, this difference in the rebleeding time suggests that may be, the secondary prophylaxis should be started with N + 5-ISMN treatment because, theoretically, this will protect against rebleeding in the period before variceal obliteration, even though the final decision is to treat the patients with EBL. The addition of sclerotherapy at the end of EBL was performed following the results published by Lo et al.21 In that study after variceal eradication with EBL patients receiving one or two sessions of small volume sclerotherapy show a very low (8%) rebleeding rate and oesophageal varices recurrence. However, the rebleeding observed in the present study by EBL, despite performing the same endoscopic techniques, suggests that the addition of low-volume sclerotherapy is not recommended. Furthermore, a recent meta-analysis of seven randomized controlled trials including 453 patients showed no significant difference in variceal rebleeding (OR 1.12; 95% CI 0.69–1.81) and mortality (OR 1.1; 95% CI 0.70–1.74) rates. The number of treatment sessions required to achieve complete eradication of the varices was similar for the two groups (7–12 sessions). In addition, the combination-therapy group experienced a higher incidence of oesophageal stricture (P < 0.001).32

The incidence of complications observed in this study was similar in both treatment groups and the incidence and types of complications were similar to those reported in previous trials22–24 with the exception of the oesophageal perforation produced at the beginning of the study when the overtube was utilized. The mortality rate observed between the trials of Lo et al.23 and Villanueva is similar and lower than the one observed by Patch et al.24 in which a higher number of Child–Pugh C patients were present. In the present study, the mortality rate is among the lowest ones, probably reflecting the small proportion of Child–Pugh C patients included, especially in the endoscopic group. Another possible cause could be related to the inclusion time of the different studies. In the present study, the mean inclusion time was higher (4 weeks) than the other studies (1 week), and it is known that variceal rebleeding episodes and deaths are higher during the first 4–6 weeks after the index bleeding.2, 24 In addition, the loss of follow-up in the present study probably may also contribute in part to the lower mortality rate observed. However, the analysis performed between the groups of patients included at different periods of time did not show any significant difference in the survival rate and the probability of survival at 2 years.

It is important to point out that when we compared the efficacy of both treatments between Child class A vs. Child B/C patients both treatments showed a significant difference in favour of the Child A. This result suggests that N + 5-ISMN and EBL treatments may be better for Child–Pugh A patients and, another treatment/s should be explored to treat Child–Pugh B/C patients.

In conclusion, the results of this trial do no suggest a superiority of EBL over N + 5-ISMN for the prevention of rebleeding, treatment failure and mortality in cirrhotic patients. The difference observed in relation to the rebleeding time as well as the efficacy between Child–Pugh classes should be demonstrated in future clinical trials.


The authors thank Mrs Roya Morady Kravetz for reviewing the English version of the manuscript. This study was supported in part by a grant from Fundacion Argentina para el Estudio de las Enfermedades del Higado (FUNDHIG).