However, endoscopic sclerotherapy is associated with various local and systemic complications that may limit its effectiveness. Therefore, endoscopic ligation was developed in an attempt to provide a treatment at least as effective as sclerotherapy but with fewer adverse side effects [11-13]. Only one randomized, controlled trial evaluating ligation treatment is available, and this suggested that variceal ligation may not only decrease complications but also improve survival compared with endoscopic sclerotherapy [14]. We describe a second randomized, controlled trial comparing endoscopic ligation and sclerotherapy for the treatment of esophageal variceal bleeding.

Methods

Top Methods Results Discussion Conclusion Author & Article Info References

Patients

Patients with chronic liver disease who had not received sclerotherapy in the past 6 months were eligible if they had 1) witnessed hematemesis, bloody nasogastric aspirate, melena, or hematochezia; 2) systolic blood pressure 90 mm Hg, heart rate 110/min, or orthostatic change in blood pressure of 20 mm Hg or in heart rate of 20/min; transfusion of 2 units of blood; or a decrease in hematocrit level of 0.06 within 12 hours; and 3) endoscopy within 24 hours of admission showing active variceal bleeding or grade 2 to 4 esophageal varices (grading scale of Korula and colleagues [15]) without any other lesion in the upper gastrointestinal tract.

Patients were excluded if gastric varices or findings of severe portal hypertensive gastropathy were present, if they were unable or unwilling to sign an informed consent, if they had hepatocellular carcinoma or other malignancy, or if they were homeless. The study was done at Los Angeles County + University of Southern California Medical Center, and patients were enrolled between May 1990 and June 1992.

Treatment

Sclerotherapy was done using 3% sodium tetradecyl sulfate mixed in equal volumes with 50% dextrose in water. Intravariceal injections of 0.5 to 2.5 mL were given based on the size of the varix. Ligation was done using an endoscopic ligating device (Bard Interventional Products, Tewksbury, Massachusetts) [13, 14]. At each treatment session, either sclerotherapy or ligation was begun in the region of the gastroesophageal junction with subsequent injections or ligatures applied more proximally. A plastic endoscopic esophageal overtube (Bard Interventional Products), 25 cm in length, was used in all treatment sessions for both sclerotherapy and ligation.

Patients qualifying for the study were randomly assigned to receive sclerotherapy or ligation at the initial endoscopic examination according to a computer-generated randomization sequence. Treatment was repeated weekly until variceal obliteration was achieved. If extensive ulceration or stricture was found, the treatment was withheld, although endoscopy was continued on a weekly basis. All patients received oral sucralfate, 1 g four times per day, during treatment until eradication was achieved. After eradication, patients had endoscopic examinations every 3 months or for any episode of rebleeding. Recurrent varices were treated with the originally assigned form of endoscopic therapy. This protocol was approved by the Research Committee of the Los Angeles County + University of Southern California Medical Center. Participants signed a written consent before endoscopic examination and treatment.

End points of patient follow-up included death, patient lost to follow-up, treatment failure, and definitive surgical treatment of portal hypertension [for example, portocaval shunt, liver transplantation]. Treatment failure was defined as 1) death related to esophageal variceal bleeding or to a complication of bleeding or endoscopic treatment; 2) three rebleeding episodes due to esophageal varices or treatment-induced esophageal ulcers that required hospitalization and transfusion; and 3) further bleeding in a single hospitalization that failed to respond to endoscopic therapy and required transfusion of at least 6 units of blood after endoscopic therapy. Rebleeding was ascribed to a treatment-induced esophageal ulcer when active bleeding was seen or stigmata of recent hemorrhage were present in the ulcer base. An esophageal stricture was diagnosed if a patient complained of dysphagia and had endoscopic evidence of narrowing; dilatation was done for all esophageal strictures.

Statistical Analysis

Quantitative data (expressed as mean ± SE) were compared using Wilcoxon nonpaired rank-sum tests. Proportional data were compared with the Fisher exact test. Kaplan-Meier estimations of time to the first rebleeding episode and time to death were done, and comparisons between treatment groups were done using the Gehan Wilcoxon test for equality of the tables [16]. The Cox proportional-hazards model was used to examine the relation of treatment to rebleeding and death with adjustment for the independent variables, severity of liver disease (Child-Pugh score), and variceal size, which were not distributed evenly between the two treatment groups. A two-tailed P value less than 0.05 was considered significant.

Results

Top Methods Results Discussion Conclusion Author & Article Info References

Seventy-seven patients were enrolled in the study. Nineteen patients meeting the inclusion criteria were excluded for the following reasons: 5 refused to participate; 2 had cancer; 5 could not consent (3 had encephalopathy, 2 were intubated and were receiving ventilatory support); 5 were homeless; and, for 2 patients, the physicians doing the study were not called in at the time of initial endoscopic examination. In addition, 14 patients who otherwise met inclusion criteria were excluded because they also had gastric varices.

Characteristics of Two Groups

Selected characteristics of the two study groups are shown in Table 1. Patients assigned to ligation had more severe liver disease as quantified by the Child-Pugh score [17], and more patients in the ligation group were classified as Child-Pugh class C (13 of 38 [34%] compared with 5 of 39 [13%]; P = 0.03). The mean prothrombin time was closer to normal in the sclerotherapy group than in the ligation group (56% ± 2% compared with 46% ± 2%; P = 0.009); platelet counts were not statistically different.

More patients in the ligation group had grade 4 varices (20 of 38 [53%] compared with 12 of 39 [31%]; P = 0.07) in the sclerotherapy group. Red color signs (red wale markings, hematocystic spots, or cherry red spots seen on the varices at endoscopy) were present in 27 (69%) of the patients receiving sclerotherapy and in 20 (53%) of those treated with ligation (P = 0.16). The mean follow-up period was 307 days for patients treated with sclerotherapy and 295 days for those receiving ligation therapy (no statistical difference); the range was 1 to 815 days.

Actively bleeding varices were present in 23% of patients in the sclerotherapy group and in 24% of the ligation group. Both treatments stopped active variceal bleeding in all but one case (Table 2). During the initial hospitalization, the two treatment groups were comparable in the length of hospital stay and in the post-treatment transfusion requirements (Table 2).

Varices were eradicated in 69% of patients treated with sclerotherapy and 59% of those treated with ligation. However, 9 (75%) of the 12 patients without eradication in the sclerotherapy group and 13 (81%) of the 16 without eradication in the ligation group received 4 treatments. The patients whose varices were not eradicated consisted of 13 who were lost to follow-up, 11 in whom treatment failed (including 6 deaths), and 4 whose treatment was still ongoing (2 without eradication despite more than 6 sessions, 1 delayed by treatment complications, and 1 in whom hepatocellular carcinoma developed).

The mean number of treatment sessions required to achieve eradication was greater with sclerotherapy than with ligation (6.2 ± 0.4 compared with 4.1 ± 0.3; difference, 2.1; 95% [CI, 1.3 to 2.9]; P < 0.001). Further, because of an increased rate of local complications (discussed below) necessitating postponement of therapy at the time of endoscopy, the difference in the number of endoscopic visits between the two therapies was even greater (sclerotherapy, 7.2 ± 0.5; ligation, 4.3 ± 0.3; P < 0.001). Means of 9.3 mL of sclerosant and 5.6 injections were used at each sclerotherapy session, and an average of 3.9 bands were placed at each ligation session.

Rebleeding and Treatment Failures

The proportion of patients rebleeding from esophageal varices or treatment-induced esophageal ulcers after entering the trial was 26% in the ligation group and 44% in the sclerotherapy group (difference, 17% [CI, –4% to 38%]), and the relative risk was 0.6 (CI, 0.3 to 1.1; P = 0.15) (see Table 2). No statistical difference was seen between the two treatment groups in Kaplan-Meier analysis of time to rebleeding (Figure 1); nor were statistical differences between treatments noted on analysis of the subgroups of patients with Child-Pugh class A-B, Child-Pugh class C, variceal grade 2 to 3, or variceal grade 4. Cox proportional-hazards analysis of time to rebleeding, controlling for Child-Pugh score and variceal grade, showed a relative risk of 0.6 [CI, 0.2 to 1.3; P = 0.18] for ligation compared with sclerotherapy.

View larger version (15K): [in this window] [in a new window]   Figure 1. Kaplan-Meier estimation of time to first rebleeding episode due to esophageal varices or esophageal ulcers in the ligation and sclerotherapy groups. The number of patients still under observation (without rebleeding or reaching an end point of follow-up) at 200-day intervals is shown at the bottom of the figure. P > 0.2 for comparison of ligation and sclerotherapy groups.

Three additional patients rebled only from sources other than an esophageal varix or ulcer: in the sclerotherapy group, one patient had a Mallory-Weiss tear and one patient had no endoscopy done; one patient treated with ligation bled from a peptic ulcer. After the initial hospitalization, the number of hospital days and the units of blood transfused for either bleeding esophageal varices or complications of therapy were not statistically different between the two treatment groups. Furthermore, the total number of hospital days and the total units of blood transfused after entry into the trial did not differ statistically between the treatment groups (see Table 2).

Treatment failures occurred in eight (21%) of the patients receiving sclerotherapy and seven (18%) receiving ligation. In the sclerotherapy group, three patients died due to bleeding, four had three separate episodes of recurrent variceal bleeding, and one had persistent bleeding from a sclerotherapy-induced esophageal ulcer (this patient had one variceal rebleeding episode followed by two separate rebleeding episodes from the ulcer). In the ligation group, three patients died due to variceal bleeding, one died due to aspiration and adult respiratory distress syndrome during the index hospitalization, two had three episodes of recurrent variceal bleeding, and one had persistent bleeding from esophageal varices that did not respond to treatment. In addition, one patient in the ligation group, who had no recurrent bleeding, had liver transplantation for primary biliary cirrhosis, and another patient in this group had a portacaval shunt done for bleeding gastric varices (this patient had one early rebleeding episode from esophageal varices and later, after eradication of the esophageal varices, bleeding gastric varices developed).

Mortality Rate and Treatment-associated Complications

The mortality rate was 11% in the ligation group and 15% in the sclerotherapy group (difference, 5%; CI, –10% to 20%) and a relative risk of 0.7 (CI, 0.2 to 2.2) (see Table 2). Three patients [8%] in each group died due to bleeding. The other three deaths in the sclerotherapy group were due to hepatocellular carcinoma; head trauma; and a brain abscess (treated with surgical drainage) in combination with tuberculosis, postoperative pneumonia, and decompensated liver disease. The one other death in the ligation group was due to aspiration and subsequent adult respiratory distress syndrome during the index hospitalization. Kaplan-Meier estimation of survival curves showed no statistical difference between the two treatment groups (Figure 2); nor were there statistical differences between treatments noted on analysis of the subgroups of patients with Child-Pugh class A-B, Child-Pugh class C, variceal grade 2 to 3, or variceal grade 4. Cox proportional-hazards analysis of mortality controlling for Child-Pugh score and variceal grade showed a relative risk of 0.5 (CI, 0.1 to 2.3; P > 0.2) for ligation compared with sclerotherapy.

View larger version (14K): [in this window] [in a new window]   Figure 2. Kaplan-Meier estimation of survival curves in the ligation and sclerotherapy groups. The number of patients still under observation (without reaching an end point of follow-up) at 200-day intervals is shown at the bottom of the figure. P > 0.2 for comparison of ligation and sclerotherapy groups.

A summary of treatment-associated complications is shown in Table 3. Esophageal ulcers were seen at follow-up endoscopy at some time during the course of therapy in 82% of patients receiving sclerotherapy and 94% of those receiving ligation. However, bleeding from esophageal ulcers (requiring hospitalization and transfusion) occurred in five patients in the sclerotherapy group and in one in the ligation group. One additional patient in the sclerotherapy group required hospitalization for severe odynophagia and dysphagia due to sclerotherapy-induced esophageal ulceration. The single complicated ulcer in the ligation group was due to insertion of the overtube. A giant ulcerated area, comprising almost one half of the circumference of the esophagus, extended 5 to 6 cm distally from just below the upper esophageal sphincter. This ulcer was identified 2 days after the initial procedure and occurred in the area covered by the overtube, far above the region in which bands were placed. All other ulcers, noted with either sclerotherapy or ligation, were seen in the distal portion of the esophagus (well below the distal end of the overtube) in the area in which treatment was applied.

Symptomatic esophageal strictures were more common with sclerotherapy than with ligation (33% compared with 0%; difference, 33% [CI, 19% to 48%]; P < 0.001). No patient with a stricture required hospitalization specifically for dysphagia or inadequate oral intake. As mentioned above, the presence of strictures or ulcers or both led to postponement of sclerotherapy more frequently than ligation. No statistical differences were present between the two treatment groups for complications such as pneumonia or spontaneous bacterial peritonitis. Although no patients had gastric varices at entry, gastric varices were identified during follow-up in 10 patients in the sclerotherapy group and in 2 patients in the ligation group.

Discussion

Top Methods Results Discussion Conclusion Author & Article Info References

Endoscopic ligation of esophageal varices is a new technique developed by Stiegmann and coworkers [11-13] in an attempt to provide an endoscopic treatment for bleeding esophageal varices that is at least as effective as sclerotherapy but that has a lower incidence of complications. Ligation of varices is based on the widely used technique of band ligation of hemorrhoids. The mucosa and submucosa of the esophagus (containing the variceal channels) are ensnared, leading to strangulation, sloughing, and eventual fibrosis—ideally with obliteration of the varices [11-13].

Efficacy

The first randomized trial comparing sclerotherapy and ligation was recently published [14] and showed statistical differences in favor of ligation for two variables: complications and survival. The somewhat surprising survival advantage of ligation may have been attributable to the lower rate of complications combined with a trend toward less rebleeding in the ligation group. No statistical differences were noted for other variables, such as cessation of active bleeding, transfusion requirements, or length of hospitalization.

Our trial also was designed to compare the safety and efficacy of endoscopic ligation with sclerotherapy in the treatment of bleeding esophageal varices. All patients enrolled had clinical evidence of acute, severe upper gastrointestinal hemorrhage due to esophageal varices. Patients were randomly assigned to receive either ligation or sclerotherapy weekly, until eradication of varices was achieved, and were then to return every 3 months for endoscopic examination and, if necessary, treatment of recurrent varices. Comparison of Kaplan-Meier curves of time to rebleeding (see Figure 1 showed no difference between treatments, although rebleeding was more frequent with sclerotherapy than with ligation [difference, 17% (CI, –4% to 38%); P = 0.15]. This difference may suggest a benefit for ligation in terms of rebleeding—albeit not definitively (the 95% CI of the difference includes zero, and the P value for the difference between the groups is >0.05). Results were comparable in the ligation and sclerotherapy groups for blood transfusions, length of hospitalization, and survival, providing no suggestion of a benefit for ligation in these important outcome variables.

The patients assigned to ligation in our trial had more severe liver disease, as indicated by a statistically greater Child-Pugh score, and a higher proportion of grade 4 varices than did the patients who received sclerotherapy. This maldistribution might have favored the sclerotherapy group. However, comparisons of ligation and sclerotherapy, controlled for Child-Pugh score and variceal size, with a Cox proportional-hazards model yielded no meaningful changes in results for rebleeding or risk of death. Comparison of Kaplan-Meier curves of similar subgroups of patients in each treatment group (based on severity of liver disease or size of varices) also showed no statistical differences between treatments in terms of rebleeding or death.

Complications and Treatment Sessions

Statistical differences in favor of ligation therapy were noted in the incidence of local complications (esophageal strictures and complicated esophageal ulcers) and the number of treatment sessions required for variceal eradication. Furthermore, because of the greater incidence of local complications with sclerotherapy, the number of endoscopic visits at which no therapy could be given was higher in the patients receiving sclerotherapy, leading to an even greater difference between ligation and sclerotherapy in the number of endoscopic treatments required to achieve variceal eradication. Our 33% rate of esophageal stricture with sclerotherapy is somewhat higher than average (for example, Stiegmann and colleagues [14] reported an incidence of 12%). Several factors may provide an explanation for our higher rate of strictures: our sclerosant of 3% sodium tetradecyl sulfate mixed in a 1:1 ratio with 50% dextrose may cause relatively more tissue damage than sclerosants used in other trials (Stiegmann and colleagues [14] used 3% tetradecyl diluted to 1% with saline); we did sclerotherapy every week rather than at less frequent intervals; even if small ulcers were present, we injected at sites away from the ulcers; and other studies may have had more stringent criteria for the diagnosis of a stricture (most studies do not provide specific criteria).

Comparison with Previous Trial

The results of our trial and the single, previous randomized comparison of sclerotherapy and ligation [14] are not dissimilar. Both trials showed a trend to more rebleeding with sclerotherapy than with ligation (differences between treatment groups: present study = 17% [CI, –4% to 38%]; Stiegmann and colleagues = 12% [CI, –5% to 29%]) but comparable results in the other related variables such as cessation of bleeding, transfusions, and days of hospitalization. Stiegmann and colleagues [14] reported that the total number of patients with any complication was 20% greater with sclerotherapy than with ligation (CI of the difference, 10% to 30%), whereas we found a difference of 32% (CI of difference, 12% to 53%). The only complication that was more common with sclerotherapy in the study of Stiegmann and colleagues was esophageal stricture (difference, 12%; CI, 4% to 20%). We found that both symptomatic esophageal strictures and complicated esophageal ulcers were more frequent after treatment with sclerotherapy (stricture difference, 33% [CI, 19% to 48%]; complicated ulcer difference, 13% [CI, 0 to 25%]) but that no differences were present in the incidence of complications, such as pneumonia or bacterial peritonitis. The previous study [14] reported that ligation required one less treatment session than did sclerotherapy to achieve eradication (CI, 0 to 2 sessions; P = 0.056). We found that ligation required 2.1 fewer treatment sessions for variceal obliteration (CI, 1.3 to 2.9 sessions; P < 0.001).

At first glance, the two studies appear to be dissimilar in terms of mortality risks. Stiegmann and colleagues reported a statistical difference in favor of ligation after comparison of Kaplan-Meier estimates of survival curves; the difference in crude survival rates was 16% (CI, 0% to 33%). Mortality rates were low in both of our treatment groups, and the Kaplan-Meier estimates of survival curves were similar for the two treatment groups. The difference in survival rates between ligation and sclerotherapy in our study was 5% (CI, –10% to 20%).

How can the apparent difference in mortality between the two studies be reconciled? Because there is a 66% overlap of the confidence intervals for survival differences from the two studies, the variation in survival differences between the trials easily could have occurred by chance. Patients with cirrhosis (primarily alcoholic cirrhosis in these studies) have highly variable courses so that different trials using the same therapy in the same target population may have widely divergent outcomes. Survival in patients with alcoholic cirrhosis and bleeding varices may relate primarily to the severity of the underlying liver disease rather than to the presence of bleeding, leading to disparate results from study to study. In addition, our trial included only patients who could give informed consent, whereas the previous trial [14] allowed the next of kin to sign the consent, possibly leading to the inclusion of a subgroup of sicker patients in the trial of Stiegmann and colleagues. The fact that mortality rates in the previous trial were higher than in our trial, despite a similar duration of follow-up, supports the idea that the patients of Stiegmann and colleagues may have been more ill. Differences in mortality rates between the therapies might only be present in sicker patients, or a relatively large number of deaths may be required to show a difference.

Conclusion

Top Methods Results Discussion Conclusion Author & Article Info References

The results of our randomized comparison of sclerotherapy and ligation for bleeding esophageal varices showed that ligation induced statistically fewer local complications (esophageal strictures and complicated esophageal ulcers) than did sclerotherapy. Furthermore, ligation achieved variceal eradication statistically more rapidly than did sclerotherapy—approximately two fewer treatment sessions were required for variceal eradication. Our study indicated that the two therapies were comparable for outcome variables such as blood transfusions, hospitalization, treatment failure, and survival. The results of our study, coupled with that of Steigmann and colleagues [14], suggest that endoscopic ligation is a viable alternative to sclerotherapy and may have several advantages over sclerotherapy. Trials with a mean follow-up beyond 1 year will be necessary to evaluate further the long-term efficacy of ligation therapy, and studies assessing the use of combined sclerotherapy and ligation [18, 19] may establish further refinements in the endoscopic treatment of bleeding esophageal varices.