Asymptomatic esophageal varices should be endoscopically treated

Endoscopic treatment has generally been accepted in the management of bleeding esophageal varices. Both the control of acute variceal bleeding and elective variceal eradication to prevent recurrent bleeding can be achieved via endoscopic methods. Endoscopic obliteration therapy using the tissue glue Histoacryl (B Braun, Melsungen, Germany) is the best method for controlling acute variceal bleeding and for eradicating large fundic varices. For the initial eradication of esophageal varices, endoscopic band ligation appears to be superior to conventional sclerotherapy because of its lower procedural morbidity. But varices tend to recur more frequently after band ligation than after sclerotherapy because of the insufficient fibrosing effect of band ligation. To achieve adequate long term variceal eradication, sclerotherapy is indispensable in most cases. In contrast to acute and elective treatment, the role of endoscopic therapy in asymptomatic patients who have never had variceal bleeding remains controversial because of the rather disappointing results obtained from prophylactic sclerotherapy over the past 15 years. Most published randomized controlled trials showed that prophylactic sclerotherapy had no effect on survival. In some studies, neither survival rate nor bleeding risk was improved. In most of these studies, sample sizes were relatively small and patients too heteroge-

Natural course studies of cirrhotic patients with esophageal varices have shown that the mean two-year incidence of variceal bleeding is 30% and that the mortality rate within six weeks of the first bleeding is approximately 50% (1,2).However, only 20% to 30% of deaths of cirrhotic patients are caused by variceal bleeding (3,4).The risk of variceal bleeding in cirrhotic patients is related to the severity of liver disease, size of varices and presence of red signs on varices.According to the study of the North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices (5), the predicted one-year bleeding risk for patients in Child's stage C, with large varices and cherry red spots, was 76%, and for patients in Child's stage A, with small varices but without red signs, was 6%.

COST OF VARICEAL BLEEDING PROPHYLAXIS
In a recent study by Teran et al (6), a Markov model was used to estimate the cost-effectiveness of variceal bleeding prophylaxis with propranolol, sclerotherapy and shunt surgery.The analysis was based on data from the literature published mostly until the early 1990s.The incidence of variceal bleeding, mortality due to variceal bleeding, total mortality, life expectancy and quality-adjusted life expectancy were analyzed, and the three predictive bleeding risk factors -Child's class, size of varices and presence of red signs -were considered.
The study found that propranolol is the only costeffective form of prophylactic therapy for preventing initial variceal bleeding in cirrhotic patients, except for patients with Child's class A with small varices.Prophylactic sclerotherapy caused significantly higher cumulative costs than propranolol due to the higher incidence of bleeding episodes (between US$6,000 and US$11,000 for five years, depending on the patient's bleeding risk).Based on data from three meta-analyses, the relative reduction of bleeding risk with sclerotherapy was 30% (7-9); it was 50% with propranolol (7,8,10).
Weighing the benefit of prophylactic sclerotherapy and the risk of bleeding if varices are untreated, one should also consider the risk of sclerotherapy.Sclerotherapy is a userdependant endoscopic method that may carry considerable procedural complications including bleeding (mainly from sclerotherapy induced ulcers), esophageal stricture, mediastinitis and perforation.According to a meta-analysis of seven studies of elective sclerotherapy, complications of sclerotherapy occurred in 35% of patients, with 12% of patients having esophageal strictures, 7% pneumonia, 13% bleeding and 4% perforation (11).The mean procedurerelated mortality reported in the literature is approximately 1%.
There is no doubt that physician expertise influences the results of sclerotherapy.In our previous controlled trial on prophylactic sclerotherapy comprising two periods (23), the complication rate in the first period, between 1980 and 1982, was 20% (bleeding from sclerotherapy-induced ulcers, stricture and perforation); in the second period, between 1985 and 1989, it was only 2.5% (one stricture in 40 patients).
Another possible influencing factor is the sclerosant used.In a meta-analysis study, Fardy and Laupacis (24) found heterogeneity when the trials were pooled in subgroups based on the sclerosant used.Trials using polidocanol (Kreussler; Wiesbaden, Germany) showed a highly significant benefit in prophylactic sclerotherapy in terms of overall mortality compared with results from trials using sodium tetradecyl.It remains unclear whether injection technique and treatment protocol may also influence the results of sclerotherapy.

ENDOSCOPIC VARICEAL LIGATION
Since the beginning of the 1990s, endoscopic therapy of esophageal varices has significantly evolved.Today, endoscopic variceal ligation (EVL) has replaced sclerotherapy as the initial method of choice in the elective treatment of esophageal varices.Sclerotherapy remains as an adjuvant measure for achieving long term eradication in cases where residual varices are refractory to EVL.Both the complication rate and the rebleeding rate have been reported to be lower for EVL than for sclerotherapy in several clinical trials.In those studies, the rebleeding rates for EVL ranged from 20% to 36%, whereas rates for sclerotherapy were between 27% and 53% (25).Owing to the lower rebleeding rate, prophylactic EVL may be beneficial in patients with asymptomatic esophageal varices.
In three recent controlled studies, the results favoured endoscopic prophylaxis using EVL.Sarin et al (26) studied the efficacy and safety of EVL for primary prophylaxis of variceal bleeding in patients with high risk varices.Esophageal varices were obliterated by EVL in 3.2±1.2sessions within 4.9±2.2weeks.Three patients in the EVL group (8.6%) and 13 in the control group (39.4%) bled during a mean followup of 14.1±5.0months (P<0.01).The cumulative probability of the patients remaining free from bleeding was higher in the EVL group than in controls (P<0.01).Bleeding-related mortality was significantly lower in the EVL group versus the control group (2.9% versus 15.2%).The overall mortality rate was 11.4% in the EVL group and 24.2% in the control group.The difference was, however, not statistically significant.
In another controlled study, Lo and Lai ( 27) reported similar results in terms of variceal bleeding (7.7% versus 22%) and mortality rate (11% versus 24%).No significant complications occurred in the EVL group except for two cases of EVL-induced ulcer bleeding.
Lay et al (28) randomly assigned 126 cirrhotic patients with high risk esophageal varices to either prophylactic ligation using the Stiegmann-Goff single ligator or no treat-ment.Their results also showed that prophylactic ligation significantly reduced the rates of variceal bleeding (19% versus 60%) and overall mortality (28% versus 58%).Ten per cent of patients in the EVL group died of hepatic failure; only 2% died of esophageal hemorrhage.In this study, Child's class A patients benefitted more than those with advanced disease from the prophylactic variceal banding.
Prophylactic EVL is justified because of its low complication rate and capability of reducing the risk of variceal bleeding.With the multiple band ligation devices, endoscopic variceal banding has become easier and safer.More studies are needed to reproduce the encouraging results of the three controlled trials.
The efficacy of prophylactic EVL needs to be weighed against the risk of first variceal bleeding.We now know more about the risk factors for bleeding.Patients in Child's stage C with large varices and red signs on the varices are more likely to bleed.Whether the incidence of first variceal bleeding can be significantly reduced by ligation therapy, particularly in patients at high bleeding risk, is unknown.However, patients with decompensated liver disease are more likely to die from the first bleeding episode.For this subgroup of patients, any prophylactic treatment modality must therefore provide safety in terms of treatment-induced bleeding.
There have been some reservations about the use of band ligation alone in Child's stage C patients with coagulopathy, due to relatively high risk of bleeding from ligation-induced mucosal necrosis (29).The combined use of obliteration therapy using tissue glue (30) and EVL in Child's stage C patients with large varices should therefore be discussed.
Comparative studies with beta-blockers are also warranted in order to demonstrate whether band ligation can achieve a reduction in bleeding episodes and mortality.

TABLE 1 Prospective controlled trials: Rates of variceal bleeding and bleeding-related mortality
348Can J Gastroenterol Vol 12 No 5 July/August 1998 Soehendra