After initial control of acute variceal bleeding, patients have up to a 70% risk of rebleeding and a mortality of 20%-35%.1, 2 The importance of preventing rebleeding in patients surviving an episode of acute variceal hemorrhage is well recognized and a lot of studies have been performed.2–4 In the recent decade, endoscopic variceal ligation (EVL) has been shown to be safer and more effective than endoscopic injection sclerotherapy (EIS) and becomes the endoscopic treatment of choice in the management of bleeding esophageal varices.5–6 On the other hand, the addition of isosorbide-5-mononitrate (ISMN) to propranolol enhanced the reduction of portal pressure and effectiveness in the prevention of variceal rebleeding.7-8 Up to now, four controlled studies have compared the effectiveness of EVL with a combination of beta-blocker and ISMN but with divergent results.9-12 Previous reports generally observed for only a mean of 1 to 2 years. Thus, we extended our study to compare the long-term effectiveness and survival of EVL with nadolol and ISMN (N+I) in the prevention of rebleeding from esophageal varices.
Both medical therapy and endoscopic variceal ligation (EVL) have proven to be comparable in the prevention of variceal rebleeding. However, the long-term results are still lacking. Our previous study enrolled 121 patients with history of esophageal variceal bleeding and randomized to receive EVL (EVL group, 60 patients) or drug therapy, nadolol plus isosorbide-5-mononitrate (N+I) (N+I group, 61 patients) to prevent variceal rebleeding. The EVL group received ligation regularly until variceal obliteration. The N+I group received N+I during the study period. Patients were followed for up to 8 years. After a median follow-up of 82 months, recurrent upper gastrointestinal bleeding developed in 28 patients (47%) in the EVL group and 49 patients (80%) in the N+I group (P = 0.001). Recurrent bleeding from esophageal varices occurred in 18 patients (30%) in the EVL group and 39 patients (64%) in the N+I group. The actuarial probability of rebleeding from esophageal varices was lower in the EVL group (P = 0.001). A total of 42 patients of the EVL group and 30 patients of the N+I group died (P = 0.013). The multivariate Cox analysis indicated that age, serum albumin, presence of encephalopathy, and treatment were the factors predictive of mortality. Conclusion: Our long-term follow-up study showed that combination of N+I therapy was inferior to banding ligation in the reduction of variceal rebleeding, but with enhanced survival. (HEPATOLOGY 2008;48:580–587.)
Patients and Methods
Between December 1997 and February 2001, patients presenting with hematemesis or tarry stool admitted to our hospital with bleeding source proven to arise from esophageal varices were considered for enrollment in the study. Inclusion criteria, exclusion criteria, method of randomization, and standard therapies of acute variceal bleeding have been described in detail.10 Briefly, our study enrolled primarily patients with cirrhosis presenting with acute esophageal variceal hemorrhage without hepatocellular carcinoma (HCC) or other debilitating comorbidities. The flow chart of enrolled patients is shown in Fig. 1.
Eligible patients were randomized into two groups: the EVL group and the N+I group. Patients in the EVL group were treated with repeated EVL. Patients in the N+I group were treated with N+I. All the surviving patients of the previous study except those lost to follow-up agreed to continue the follow-up study. All the patients signed an informed consent. Additional informed consents were obtained for patients receiving repeated elective sessions of EVL, emergency EVL, or other rescue therapies. Our study was approved by our hospital Ethics Committee.
Technique of EVL and Methods of Medications.
Our technique of EVL has been described elsewhere.10 Briefly, the Saeed Six-Shooter (Wilson-Cook Medical, Winston-Salem, NC) attached to a video endoscope (Olympus XQ 230) was utilized. The interval between two sessions of EVL was 3 to 4 weeks until variceal obliteration was achieved. Variceal obliteration was defined as complete disappearance of esophageal varices or when the sizes of esophageal varices were too small to be ligated.13 After variceal obliteration, patients were scheduled to receive endoscopic surveillance every 6 months in the first year and annually thereafter. Recurrent esophageal varices were treated with repeated EVL until obliteration. Patients who refused to receive regular elective EVL to achieve variceal obliteration or refused to receive regular endoscopic surveillance after variceal obliteration were regarded as noncompliant.
Among the N+I group, nadolol 40 mg once daily (E.R. Squibb, Sons, Inc.) was given initially and then adjusted according to the dosage that reduced resting pulse rate up to 25% or 55 beats per minute. Subsequently, oral ISMN 20 mg once per day was administered concomitantly. The dose was increased up to 20 mg twice a day. The compliance was assessed by reduction of pulse rate and/or by quantifying the number of tablets consumed. In case of side effects, nadolol was continued in those who could not tolerate ISMN, and ISMN was continued in patients whose side effects were related to nadolol. Patients were asked to visit the clinic monthly to get drugs as well as for measurement of blood pressure and pulse rate. Patients who refused to take drugs for portal hypertension or refused to come to the scheduled clinic >3 times were regarded as noncompliant.
At the time of enrollment, the severity of liver disease was classified according to the Child-Turcotte Pugh classification. The size of esophageal varices was classified according to Beppu's criteria. Clinical evaluations were performed monthly. Liver function tests, complete blood cells counts, renal function tests, serum alpha-fetoprotein measurements, and ultrasonographic examination of liver were performed every 3 months in both treatment groups. Development of ascites, hepatic encephalopathy, spontaneous bacterial peritonitis (SBP), and HCC were treated and recorded. Patients in both groups were advised to abstain from alcohol as much as possible. If patients had hepatitis B virus–related decompensation, the antiviral drug lamivudine (100 mg per day) was administered. Rebleeding from esophageal varices was defined as the presence of hematemesis and/or melena; the bleeding site was identified to be from esophageal varices by emergency endoscopy. Episodes of gastrointestinal bleeding from sources other than esophageal varices were also recorded. Only those who had a drop in hemoglobin and needed blood transfusion of two or more units were considered to have rebleeding. When rebleeding from esophageal varices was encountered, emergency EVL was performed within 12 hours of hospitalization in both treatment groups. Among the N+I group, nadolol and ISMN were discontinued temporarily until vital signs became stable for 48 hours. Repeated sessions of EVL were performed in the EVL group until the varices were obliterated once again. Patients in both groups with bleeding from gastric varices were treated with histoacryl injection. Patients with two or more rebleeding episodes received alternative rescue therapy such as cross-over, transjugular intrahepatic portosystemic stent shunt, EIS with 3% sotradecol as sclerosant, or a devascularization operation at the discretion of doctors in charge. Blood was transfused to keep hemodynamics stable when a patient rebled and hemoglobin concentration was less than 9 g/dL. The requirements of blood transfusion were recorded during hospitalization for patients with rebleeding during the study period. Complications were evaluated in detail and recorded. The primary end points were rebleeding and death. The causes of all mortalities were searched for in detail. If patients died in other hospital, the chart records were reviewed. Patients who were noncompliant or lost to follow-up at scheduled visits in both groups were contact by a study nurse by either telephone or mail.
The data are expressed as mean ± standard deviation. Statistical analysis was based on an intention-to-treat principle. Patients lost to follow-up were censored according to the day of their last visit. Quantitative variables were compared according to Student t test, and qualitative variables were compared using the chi-square test and Fisher's exact test when appropriate. For survival analysis, the final status of patients was traced by all means if possible. Kaplan-Meier estimation was applied to examine the time to a first occurrence of recurrent bleeding from gastroesophageal varices and the time to death. The log rank test was used to examine the variation of rebleeding episodes and survival. Cox's regression analysis was used to detect possible prognostic variables other than treatment modality on rebleeding rate and survival. All P values were two-tailed. P value < 0.05 was considered significant. Analyses were performed using SPSS software.
The EVL group comprised 60 patients and the N+I group comprised 61 patients. The baseline characteristics of both groups are shown in Table 1. Both groups were comparable in age, sex, etiologies of cirrhosis, and severity of liver disease. A total of seven alcoholic patients in the EVL group and six alcoholic patients in the N+I group continued to drink. Every patient was followed for at least 5.5 years or until death. The mean follow-up period was 82 months (range, 10 days to 8 years and 2 months) in the EVL group and 81 months (15 days to 8 years and 4 months) in the N+I group. A total of four patients in EVL group and five patients in N+I group were lost to follow up after a mean follow-up of 15 months (range, 1 month to 50 months). A total of nine patients (four patients refused elective EVL, five patients refused regular surveillance) in EVL group and four patients in the N+I group did not comply with the protocol of this study.
|Characteristic||EVL Group (n = 60), (%)||N+I group (n = 61), (%)||P|
|Age (years)||52 ± 12||51 ± 13||0.44|
|Causes of cirrhosis|
|Alcoholism||16 (27)||22 (36)||0.33|
|Hepatitis B||18 (30)||20 (33)||0.87|
|Hepatitis C||14 (23)||12 (20)||0.85|
|Hepatitis B+C||7 (12)||4 (6)||0.65|
|Cryptogenic||5 (8)||3 (5)||0.70|
|Albumin (gm/dL)||3.5 ± 1.5||3.1 ± 0.7||0.22|
|Bilirubin (mg/dL)||2.5 ± 2.1||2.9 ± 4.1||0.23|
|GPT (U/dL)||70 ± 107||58 ± 80||0.08|
|Creatinine (mg/dL)||1.0 ± 0.5||1.0 ± 0.4||0.88|
|Ascites present||39 (65)||36 (59)||0.14|
|Prothrombin time (second)||13.6 ± 4.2||13.2 ± 3.8||0.21|
|Encephalopathy||4 (7)||2 (3)||0.36|
|A||13 (22)||13 (21)|
|B||35 (58)||35 (57)|
|C||12 (20)||13 (21)|
|Child-Turcotte-Pugh score||8.0 ± 1.5||7.8 ± 1.6||0.90|
|Previous bleeding||14 (23)||22 (31)||0.18|
|Active bleeding at initial session||7 (12)||8 (13)||0.97|
|Control of index bleeding|
|Hemoglobin (gm/dL)||9.1 ± 2.2||9.0 ± 2.5||0.93|
|Size of varices: F1/F2/F3||0/41/19||1/30/30||0.25|
|Red color signs on varices mild/moderate/severe||23/32/5||26/26/9||0.72|
Among EVL group, 40 (67%) patients achieved variceal obliteration after a mean of 3.3 ± 1.1 sessions of EVL. The duration from the start of treatment to obliteration was 62 ± 13 days. Variceal recurrence occurred in 26 patients (65%). A mean of 2.1 ± 0.3 (range, 1 to 7) sessions of EVL were required to achieve obliteration of recurrent varices.
Among the N+I group, the mean dose of nadolol was 48 ± 10 mg a day (range, 20 to 120 mg). The mean dose of ISMN was 30 ± 6 mg per day (range, 10 to 40 mg). The mean pulse rates and blood pressure obtained in patients treated with medical therapy were 65 ± 7 and 115 ± 12 mmHg, respectively. A total of two patients did not take nadolol and three patients did not take ISMN because of side effects.
The results are shown in Table 2. Upper gastrointestinal tract rebleeding from all sources occurred in 28 patients (47%) of the EVL group, and 49 patients (80%) of the N+I group (P = 0.001, Fig. 2). When considering only rebleeding from esophageal varices, 18 patients (30%) rebled in the EVL group and 39 patients (64%) in the N+I group. Only four patients in the N+I group were noncompliant measured by pill counts. The variceal rebleeding rate was 63% (36/57) in compliant patients and 75% (3/4) in noncompliant patients (P = 0.54). The actuarial probability of rebleeding from esophageal varices was significantly lower in the EVL group than in the N+I group (P = 0.001; Fig. 3). The incidence of bleeding from gastric varices was slightly higher in the N+I group. The total number of variceal rebleeding episodes was higher in the N+I group than in the EVL group (70 versus 27 episodes). The number of rebleeding episodes per patient was significantly higher in the N+I group (P < 0.05). The mean total blood requirement per patient for rebleeding episodes was slightly lower in the EVL group than in the N+I group (5.3 ± 7.6 versus 6.8 ± 8.2 units, respectively) (P = 0.25). Cox regression analysis revealed that serum bilirubin level (hazards ratio, 1.1; 95% confidence interval, 1.01-1.15) and treatment were predictive factors of rebleeding (N+I versus EVL; hazards ratio, 2.48; 95% confidence interval, 1.42-4.3). A total of 10 (17%) patients in the EVL group and 21(34%) patients in the N+I group experienced two or more rebleeding episodes from esophageal varices (P < 0.05). Rescue therapies for patients with two or more rebleeding episodes are shown in Table 3. One patient in each group received orthotopic liver transplantation. Three patients in the EVL group and two patients in the N+I group did not receive further treatment.
|EVL group (n = 60), (%)||N+I group (n = 61), (%)||P|
|Patients with rebleeding||28 (47)||49 (80)||0.001|
|Sites of rebleeding*|
|Esophageal varices||18 (30)||39 (64)||0.001|
|Gastric varices||3 (5)||9 (15)|
|Esophageal ulcer||2 (3)||4 (7)|
|Peptic ulcer||11 (18)||10 (16)|
|Congestive gastropathy||3 (3)||3 (5)|
|Number of variceal rebleeding episodes||27||70||<0.05|
|Blood transfusions (units)||5.3 ± 7.6||6.8 ± 8.2||0.25|
|EVL group||N+I group|
|Nadolol + ISMN||2||0|
|EVL + nadolol +ISMN||2||6|
Complications developed in 14 patients (23%) in the EVL group and 15 patients (24%) in the N+I group (no significant difference). Most of the treatment-related complications were mild in the two groups. The most frequent complication in the EVL group was transient dysphagia (12%) and retrosternal pain (10%), whereas the most frequent complications in the N+I group were dizziness (13%). Ligation-induced ulcer bleeding occurred in two patients (3%). There were no mortalities resulting from complications in either group.
During the long-term follow-up, ascites occurred in 39 patients (65%) in the EVL group and 44 patients (72%) in the N+I group. SBP developed in 11 patients in each group. Hepatic encephalopathy was noted in 17 patients in the EVL group and 19 patients in the N+I group. HCC was detected in 25 patients (42%) in the EVL group and 19 patients (31%) in the N+I group. Five patients in the EVL group and seven patients in the N+I group had ever received lamivudine treatment for hepatic decompensation. No significant differences in these complications existed between the two treatment groups.
There were 42 deaths in the EVL group and 30 deaths in the N+I group. The causes of death are shown in Table 4. Eight-year survival probability was significantly higher in the N+I group than in the EVL group (49% versus 30%; P = 0.013; hazards ratio, 0.55; 95% confidence interval, 0.346-0.889). The Kaplan-Meier plot of the survival curve is shown in Fig. 4. The majority of both treatment groups died of hepatic failure or variceal bleeding. Two patients in the EVL group died in traffic accidents. Both were male, alcoholics, and died at 7 months and 2 years after enrollment, respectively. There appeared to be no causal relationship between treatment and the occurrence of the traffic accidents. The number of patients who died of uncontrollable variceal hemorrhage was similar between both treatment groups. The causes of three mortalities could not be exactly identified due to death at home or at a home-care center. If mortalities unrelated to chronic liver disease, such as cerebral vascular accident, traffic accident, myocardial infarction, lymphoma, gastric carcinoma, and uncertain causes were censored at the time of an excluded endpoint, the difference in survival between both treatment groups was still significant (P = 0.046; Breslow test; Fig. 5; hazards ratio, 0.59; 95% confidence interval, 0.372-0.910).Univariate analysis revealed that age, serum albumin level, presence of ascites or encephalopathy during index bleeding, and treatment modality were predictive factors of mortality. Cox regression analysis showed that age, serum albumin level, encephalopathy, and treatment modality were predictive factors of mortality (Table 5).
|EVL Group (n = 60)||N+I Group (n = 61)|
|Esophageal variceal bleeding||6||8|
|Gastric variceal bleeding||1||0|
|Cerebral vascular accident||1||1|
|Spontaneous bacterial peritonitis||3||2|
|Variable||Univariate Analysis||Multivariate Analysis|
|Hazards Ratio||95% CI||P value||Hazards Ratio||95% CI||P value|
It is generally anticipated that prevention of variceal rebleeding is the first step to achieve the goal of enhancing survival in patients who have a history of esophageal variceal hemorrhage. Previous studies comparing EIS with beta-blockers showed that EIS is more effective than propranolol in preventing variceal rebleeding, but the survival rates were similar between the two therapies.14 Due to lower rebleeding rates, EVL is emerging as the endoscopic treatment of choice to prevent variceal rebleeding. However, most studies, except two articles, suggested that the survival rates were similar between patients treated with EIS and EVL.5, 15, 16 It appears that the effectiveness of reduction in variceal rebleeding does not necessarily translate into improved survival.
The use of medical therapy for portal hypertension, obviating the invasiveness of endoscopic therapy, is probably more acceptable to patients with cirrhosis. The combination of beta-blockers and ISMN has been shown to be superior to beta-blockers alone in the reduction of variceal rebleeding.8 Either EVL or a combination of beta-blockers and ISMN is currently considered to be suitable for prevention of variceal rebleeding.2 Up to now, four controlled studies have compared the effectiveness of EVL with combination of beta-blocker and ISMN but with divergent results.9-12 Villaneuva et al.9 showed that beta-blockers combined with ISMN was superior to EVL, whereas ours showed that EVL was superior to medical therapy in the reduction of variceal rebleeding.10 All these studies observed for a short duration, and the survival benefit was not demonstrable.
We extended the observation period up to 8 years; to our knowledge, this could possibly be the longest follow-up period for a controlled study of EVL and the pharmacological prevention of variceal rebleeding. Our study demonstrated that EVL was definitely better than combination drug therapy in terms of prevention of rebleeding from esophageal varices. The number of our patients who were able to tolerate being placed on the combination of beta-blockers and nitrates appeared to be higher than the rates of other studies.8, 10 This may be related to the lower dosage of nitrates used in our study and the fact that patients with contraindications to beta-blockers were excluded from our trial. The cumulative variceal rebleeding rate was 64% in our patients treated with a combination of nadolol and ISMN. This rate was higher than previous studies. This may be ascribed to a longer term of observation and lower dose of ISMN used in this trial. A higher proportion of our patients with ascites who were classified as Child-Pugh C may also partly explain the higher rebleeding rate.
However, the higher rebleeding rate in the patients treated with combination drug therapy did not fare worse. Conversely, the survival in patients treated with combination drug therapy was significantly better than those treated with repeated EVL. The 8-year survival probability was 49% in the N+I group and 30% in the EVL group. The majority of patients in both groups died of hepatic failure and variceal bleeding. Much of the observed difference in the survival curves occurred in the first 2 years. When compared with other studies,9, 11, the difference is probably not just due to longer follow-up. Our data suggests that the effective obliteration of varices with reduced variceal rebleeding achieved by EVL contributed limited value to prolongation of survival. Some studies suggested that the responders, defined by a reduction of the hepatic venous pressure gradient to <12 mmHg, or by >20% of baseline, is associated with a marked reduction in the long-term risk of developing complications of portal hypertension and with better survival.17, 18 Our study did not measure hemodynamic response in patients receiving nadolol and ISMN. However, routine hemodynamic measurement of hepatic venous pressure gradient is not applicable in clinical practice. The other complications of portal hypertension, such as ascites, SBP, and hepatic encephalopathy occurred at a similar rate between both treatment groups. Development of HCC was more frequently encountered in patients treated with EVL than in patients receiving drug therapy; however, the difference was not significant. The study from Patch et al.11 also showed a higher frequency of HCC development in patients treated with EVL than with drug therapy (6/51 patients versus 1/51 patients, respectively). Whether this was coincidence or an effect of treatment is not known. Though our patients in the N+I group rebled from esophageal varices more frequently, the incidence of dying of variceal hemorrhage was not much more than that in the EVL group. This should be ascribed to the effectiveness of using vasoconstrictors and EVL in the management of our patients with acute variceal hemorrhage; that is, 97% hemostasis of active variceal bleeding.19 There were two patients who died in traffic accidents in the EVL group. However, this does not appear to explain the discrepant results. Previous studies showed that the complications of EIS may offset its efficacy and result in similar survival between patients treated with EIS and with propranolol.14 The frequency and severity of complications were similar between the two treatment groups. This result suggests that the complications of repeated EVL were mild just like previous studies.15 Thus, the higher mortality rate in the EVL group could not be ascribed to treatment-related complications. A slightly higher incidence of noncompliance was noted in the EVL group, which may contribute partly to the higher mortality. It is likely that drugs like beta-blockers are more tolerable than the procedure of EVL. Moreover, medical therapy may have other beneficial effects in addition to reduction of the risk of variceal hemorrhage. In contrast, increased portal pressure after repeated EVL may negatively impact survival.20 As shown in the multivariate analysis, age, serum albumin level, presence of hepatic encephalopathy, and treatment modality were predictive factors of mortality. It is difficult to change aging and poor liver reserve of patients with portal hypertension; however, a treatment modality for prevention of variceal rebleeding indeed may influence the outcome.
Our study is similar to the prophylactic shunt operation or EIS, which has been noted to be effective in the prophylaxis of first variceal bleeding but with increased morbidity and mortality.21 If mortalities unrelated to chronic liver disease were excluded, the difference in survival was still significant. Our patients did not receive prophylaxis for SBP. Though patients receiving medical therapy had a higher variceal rebleeding rate and the need for rescue therapy, the number of patients dying of variceal bleeding was not significantly higher than that in the EVL group. This could be ascribed to the highly effective hemostasis achieved by emergency EVL. Similar to the prophylactic EIS carried out in the United States, the causes of increased mortality in our patients receiving repeated EVL to prevent rebleeding were not actually known. We could only presume that the use of nadolol and ISMN to prevent variceal rebleed is more physiological than repeated EVL, the latter being aimed at merely focal points instead of systemic effects induced by the former.
In conclusion, our study showed that repeated EVL is more effective than a combination of nadolol and ISMN in the prevention of variceal rebleeding. However, the survival was significantly higher in those treated with combination drug therapy. It may be recommended that combination drug therapy is preferable to EVL as the first choice of treatment for preventing variceal rebleeding. Further studies are needed to confirm our findings.