Endoscopic Sclerotherapy as Compared with Endoscopic Ligation for Bleeding Esophageal Varices

Selection of Patients

Patients with esophageal varices that were bleeding actively or had recently bled were eligible for this study if they were 18 years of age or older and there was documentation that their varices were caused by cirrhosis. Patients were excluded from the study if they had a contraindication to endoscopy (e.g., esophageal stricture), previous endoscopic or operative treatment for esophageal varices, gastric fundal varices, intercurrent illness with death expected within 12 months, or symptoms of esophageal dysfunction. Previous treatment with balloon tamponade, vasopressin, or beta-adrenergic-antagonist agents did not disqualify patients, but the use of the latter was not permitted during the trial.

Patients who were actively bleeding were resuscitated, underwent endoscopy, were randomized, and were treated within 12 hours of admission. Those in whom bleeding had ceased spontaneously were treated at the earliest opportunity, usually within 24 hours of the apparent cessation of bleeding.

Informed Consent and Randomization

The study design was approved by the human investigational review boards of all the institutions, and informed consent was obtained from all the patients or their next of kin. Treatment assignments were made in blocks of 10 patients, with use of computergenerated random numbers. Randomization was done after endoscopy confirmed that the patient met none of the criteria for exclusion and that varices were the source of hemorrhage. Variceal bleeding was defined as bleeding from an esophageal varix visible at the time of endoscopy; the presence of a blood clot over an esophageal varix, with no other endoscopically observed source of bleeding; or the presence of large esophageal varices, blood in the stomach, and no other bleeding lesion.

Technique of Endoscopic Sclerotherapy and Endoscopic Ligation

All the investigators had previous experience in the performance of sclerotherapy, and each had performed at least 10 previous treatments using endoscopic ligation. Pentax FG-34, Olympus Q, GIF-1T, GIF-2T, GIF-V10, or GIF-1TV10 endoscopes were used. A sedative agent (midazolam, diazepam, or meperidine) was given before endoscopy at the discretion of the investigator. At each endoscopic session the size of the varices was graded from 1 to 4. Variceal size was estimated or compared with an object of known size, as follows: grade 1, <3 mm; grade 2, 4 to 6 mm; grade 3, 7 to 10 mm; grade 4, >10 mm or large enough to fill the esophageal lumen completely. The initial treatment was given in the hospital, but most subsequent treatments were provided on an outpatient basis.

Endoscopic Sclerotherapy

The sclerosant used was 3.0 percent sodium tetradecyl sulfate (Elkins-Sinn, Cherry Hill, N.J.) diluted with saline (with or without methylene blue) to a 1.0 percent solution. The varices were injected with 25-gauge disposable needles. The injections were intravariceal, were begun at or near the gastroesophageal junction, and delivered up to 2 ml of sclerosant at each site. No more than 20 ml of sclerosant was injected during each session of elective treatment. Treatment was confined to the distal 7 cm of the esophagus and to the proximal 1 to 2 cm of the stomach. In patients who were bleeding actively, the bleeding varix was injected first, if it was identifiable, to control hemorrhage. The injections were then continued as in the case of patients not bleeding actively. Balloon tamponade was not routinely used after either sclerotherapy or ligation.

Endoscopic Ligation

Endoscopic ligation was performed with an endoscopic ligating device and overtube (Bard Interventional Products, Tewksbury, Mass.). The endoscopic overtube consisted of a plastic sheath 20 mm in diameter and 25 cm long that was inserted into the proximal esophagus before ligation was performed. The overtube remained in the esophagus throughout the procedure and facilitated the repeated removal and reinsertion of the endoscope, necessary for multiple ligations to be effected. The details of the technique have been described elsewhere.¹⁴ In brief, varices were ligated individually with a single elastic O ring, starting at or just below the gastroesophageal junction and continuing cephalad to 7 cm above that junction. All individual varices were ligated at least once per treatment, and larger varices were often ligated twice at separate points, one caudad and one cephalad. No more than eight individual ligations were performed during sessions of elective treatment. In patients bleeding actively, ligations were performed at or around the site of bleeding, to control hemorrhage. Additional varices were then ligated as in patients without bleeding.

Follow-up Treatment

The supportive care and monitoring provided during follow-up were similar in both groups. The physicians providing care were not blinded to the treatment given. Control of variceal bleeding was defined as the absence of upper gastrointestinal hemorrhage (i.e., stable hematocrit and vital signs and the absence of hematemesis) for eight hours after treatment. Recurrent bleeding was defined as a subsequent upper gastrointestinal hemorrhage that resulted in unscheduled endoscopy, a need for blood transfusion, or exsanguination. Repeat endoscopy with sclerotherapy or ligation was performed as needed for recurrences of bleeding and at intervals of 5 to 21 days until all varices in the distal esophagus were eradicated —i.e., until all varices were reduced to grade 1 or distal esophageal veins were no longer present. After varices were eradicated, patients were followed by repeat endoscopic examinations at three-month intervals to detect and treat any varices that recurred.

Treatment failure was defined as the occurrence of any of the following: continued bleeding not controlled by two endoscopic treatments during any hospitalization in which four or more units of blood were required after the last endoscopic procedure, three or more episodes of rebleeding (requiring transfusion), complications that contraindicated endoscopic therapy, a refusal by the patient to continue treatment, and a decision by the physician responsible to use an alternative treatment.

End Points

The primary end point of the study was the incidence of complications other than bleeding that were associated with endoscopic treatment. Complications were defined as any untoward events (e.g., pneumonia, pleural effusion, bacterial peritonitis, and esophageal stricture or perforation) that required active treatment or prolonged hospitalization. Transient dysphagia, retrosternal pain, and other symptoms that did not merit investigation or treatment were not considered complications. The incidence of complications other than bleeding was chosen as the primary end point, because when the trial was designed we did not anticipate the occurrence of substantial differences between the two treatments with regard to more traditional end points, such as survival and recurrent bleeding.

Secondary end points included the control of active variceal bleeding, the incidence of recurrent bleeding, the efficacy of therapy for eradicating varices, the incidence of treatment failure, and mortality.

Study Termination and Statistical Analysis

The size of the study sample was based on a rate of complications that was anticipated to be 25 percent in the sclerotherapy group and 6 percent in the ligation group. Most complications were expected to occur during the patients' first 4 months of treatment, when endoscopic treatments were performed at intervals of 5 to 21 days. For a two-sided test to be used, 170 patients were required.¹⁶ Patient accrual and major end points were reviewed at 12-month intervals.¹⁷ Significant differences in the incidence of complications and a significant increase in mortality in one cohort were found at the 24-month review, after the enrollment of 130 patients. Enrollment in the study was therefore terminated.

Student's t-test was used to compare the means of continuous variables, the Wilcoxon rank-sum test was used for skewed or ordinal data, and the chi-square test was used for discrete variables. Kaplan–Meier analysis was used to examine the time to death and to a first recurrence of bleeding. The log-rank test was used to determine whether times to outcomes differed between groups. A proportional-hazards model was used to test for differences in complications, survival, and rates of recurrent bleeding between the two cohorts after we controlled for the severity of liver disease, as measured by the Child—Pugh classification (in which A denotes good hepatic function, B intermediate function, and C poor function).¹⁸ The Breslow-Day test (for the homogeneity of odds ratios) was used to test for differences in complications, survival, and rates of recurrent bleeding with control for the study site. All the analyses were based on the intention-to-treat principle, and all P values were two-sided. The level of significance was P<0.05. Calculations were performed with the SAS statistical program, except for the analysis of survival and rebleeding, which was performed with True Epistat.

Of approximately 230 patients screened over a 27-month period, 130 patients entered the study. In virtually all the excluded patients, the reason was previous endoscopic treatment. Sixty-six patients were randomly assigned to treatment with sclerotherapy, and 64 to treatment with ligation. One patient in the sclerotherapy group was inappropriately randomized and was not included in the analysis. At the four study sites, 11, 30, 33, and 56 patients were enlisted in the study.

Table 1. Table 1. Base-Line Characteristics of Patients with Cirrhosis and Bleeding Esophageal Varices Treated by Endoscopic Sclerotherapy or Endoscopic Ligation.*

The base-line characteristics of the two groups are shown in Table 1. Most patients were men with alcohol-induced cirrhosis. The severity of liver disease (i.e., the modified Child-Pugh score¹⁸) was similar in both groups, as were other characteristics. Sixty-four percent of the patients entered the study at the time of their first episode of bleeding. For the 39 patients in the sclerotherapy group and the 49 patients in the ligation group who remained in the study more than 30 days, the mean (±SD) duration of follow-up was 325±214 and 317±208 days, respectively.

Control of Active and Recurrent Bleeding and Variceal Eradication

Table 2. Table 2. Control of Bleeding and Recurrent Bleeding and Eradication of Esophageal Varices in Patients Treated by Endoscopic Sclerotherapy or Endoscopic Ligation.*

Figure 1. Figure 1. Absence of Recurrent Bleeding after Initial Treatment of Variceal Bleeding by Endoscopic Sclerotherapy or Endoscopic Ligation in Patients with Cirrhosis.

The difference between groups was not statistically significant (P = 0.072).

One fifth of the patients in each group were bleeding actively at the time of the index treatment. Both treatments were equally effective for the control of active hemorrhage (Table 2). The likelihood of recurrent bleeding was smaller in the patients treated with endoscopic ligation, but the difference was not statistically significant (P = 0.072) (Fig. 1). The site of recurrent bleeding was most often an esophageal varix (Table 2). Treatment-induced recurrences of bleeding as a result of ulceration or mucosal sloughing occurred in four patients in the sclerotherapy group and five patients in the ligation group. Patients in whom the site of a recurrence was undetermined either had a number of potential bleeding sites at the time of endoscopy (e.g., treatment-induced ulceration, gastritis, and varices) or did not have endoscopy. There was no correlation between the incidence of recurrences of bleeding and either the severity of liver disease or the individual study site. There were also no differences between the groups in the number of endoscopic examinations or treatments, the number of transfusions, or the length of hospitalization.

The extent of eradication of esophageal varices was assessed in patients who remained in the study more than 30 days (Table 2). Both techniques were found to be effective for this purpose, but eradication with ligation required, on average, one less treatment per patient than sclerotherapy (P = 0.056). Esophageal varices recurred more often after their initial eradication in the patients in the sclerotherapy group, but the difference was not statistically significant.

Survival

Figure 2. Figure 2. Survival in Patients with Cirrhosis and Bleeding Esophageal Varices Treated with Endoscopic Sclerotherapy or Endoscopic Ligation.

The survival rate in patients treated with endoscopic ligation was significantly higher (P = 0.041) than that in patients treated with endoscopic sclerotherapy.

Figure 3. Figure 3. Survival in Patients with Cirrhosis and Bleeding Esophageal Varices Treated with Endoscopic Sclerotherapy or Endoscopic Ligation, According to Severity of Liver Disease (Child-Pugh Class).

Among the patients in classes A and B (solid lines) who were treated with endoscopic ligation, survival was significantly better than that in the patients treated with sclerotherapy (P = 0.046). The survival rate in patients in class C (broken lines) did not differ significantly between groups. See the Methods section for a description of the Child—Pugh classification system.

More than half the deaths in each group occurred within 30 days of a patient's entry into the study. The overall rate of survival was higher in the patients treated by ligation than in those treated by sclerotherapy (P = 0.041) (Fig. 2). The patients in Child-Pugh classes A and B benefited the most from ligation treatment (Fig. 3). Among the patients in Child-Pugh class C, there was no difference in survival between study groups, nor were there differences in survival between study sites. Liver failure was the most common cause of death, followed by exsanguination. Other causes included pneumonia, spontaneous bacterial peritonitis, multiple organ failure, pancreatitis, myocardial infarction, and cerebrovascular accident.

Complications

Table 3. Table 3. Complications Associated with the Use of Endoscopic Sclerotherapy and Endoscopic Ligation to Treat Patients with Cirrhosis and Bleeding Esophageal Varices.

The complication rates were 22 percent in the sclerotherapy group and 2 percent in the ligation group (P<0.001) (Table 3). The most common complication after sclerotherapy was esophageal stricture, which was treated in all patients with a mean of two sessions of bougienage (range, one to four). Pneumonia or bacterial peritonitis occurred in 11 percent of the patients in the sclerotherapy group and 2 percent of those in the ligation group. Four patients in the sclerotherapy group who had pneumonia or peritonitis and one such patient in the ligation group died from these complications. There was no correlation between the severity of liver disease (Child-Pugh classification) and the incidence of complications, nor was there a difference between study sites in the incidence of complications.

Treatment Failure

Table 4. Table 4. Causes of Treatment Failure and Death in Patients with Cirrhosis and Bleeding Esophageal Varices Treated by Endoscopic Sclerotherapy and Endoscopic Ligation.*

Eight patients in the sclerotherapy group and nine patients in the ligation group had an alternative treatment during the study (Table 4). Of these patients, five in the sclerotherapy group and eight in the ligation group received such treatment on an emergency basis for uncontrolled bleeding. Four patients had an elective alternative treatment (liver transplantation or shunt insertion). Among the patients who received emergency alternative therapy, four in the sclerotherapy group (80 percent) and six in the ligation group (75 percent) survived for 30 days. Five and seven patients in the respective groups were lost to follow-up or declined to participate further in the study.

Endoscopic sclerotherapy and ligation were equally effective for the control of active variceal bleeding. Recurrent bleeding, however, was less common in the patients treated by endoscopic ligation than in those treated by sclerotherapy. In the latter group, the incidence of recurrent bleeding in this study was similar to that in other reports.^{3 4 5} This benefit of ligation is not explained by differences in recurrent bleeding from sites of endoscopic treatment, since there were few instances of such recurrence in either group. Similar trends of less recurrent bleeding in patients treated by endoscopic ligation than in those treated by sclerotherapy have been reported in two other ongoing trials^{19 , 20}; in these studies, when ligation rather than sclerotherapy was used, significantly fewer treatments were needed to eradicate varices. In our study, one less treatment with ligation was needed than the number of treatments required with sclerotherapy for the varices to be eradicated. In this and other trials, recurrent bleeding has commonly resulted from varices not yet eradicated. The more efficient eradication with ligation may therefore contribute to the lower incidence of recurrent bleeding.

The benefit for survival conferred by endoscopic ligation predominantly affected patients in Child-Pugh classes A and B. Death from exsanguination was nearly three times more frequent among the patients in the sclerotherapy group, and mortality was disproportionately higher than the overall increase in the incidence of recurrent bleeding in this group. Liver failure, the most common cause of death in both groups, was often precipitated by a recurrence of bleeding. Infection-related complications associated with treatment resulted in additional deaths in the sclerotherapy group. Fewer recurrences of bleeding and fewer infection-related complications seem the likely explanation for the higher rate of survival among the patients in the ligation group.

The incidence of treatment-related complications associated with endoscopic ligation was significantly lower than that associated with sclerotherapy. The incidence and type of complications in the current group of patients undergoing sclerotherapy were similar to those described in previous reports.^{3 4 5 6 7 8} Half the complications were esophageal strictures without apparent long-term sequelae. Pneumonia and bacterial peritonitis, on the other hand, caused four deaths in this group. The patients treated with endoscopic ligation had few pulmonary complications, possibly because the endoscopic overtube prevented tracheal aspiration. Alternatively, sclerotherapy is known to result in transient bacteremia, a finding dismissed as inconsequential by most investigators and not yet studied in patients treated by ligation.^{21 22 23} Such bacteremia may trigger infectious complications in small numbers of patients treated by sclerotherapy.

A recent report compared prophylactic sclerotherapy and sham sclerotherapy for the prevention of a first variceal hemorrhage,²⁴ but that trial was terminated early because of a doubling in mortality in the sclerotherapy-treated patients. This increase occurred despite a reduction in the risk of bleeding in the patients treated with sclerotherapy, and among these patients many deaths resulted from infection. We hypothesize that an injection of sclerosant may cause subtle adverse effects on defense mechanisms in the host or may directly foster infection-related sequelae. Such effects could counterbalance the benefit provided by sclerotherapy for the control of bleeding, and they may in part explain the survival advantage in the patients treated by endoscopic ligation in our study.

We conclude that with respect to a lower incidence of complications other than bleeding, endoscopic ligation is superior to sclerotherapy in the treatment of bleeding esophageal varices, and the benefit of endoscopic ligation for survival in this trial resulted in part from fewer recurrences of bleeding and fewer complications.