Does Elective Sclerotherapy Improve the Efficacy of Long-term Propranolol for Prevention of Recurrent Bleeding in Patients with Severe Cirrhosis? A Prospective Multicenter, Randomized Trial OLMER INK,’ THIERRYMARTIN,’ THIERRY POYNARD,3 h h R C REVILLE,4 hhRIE-LAURE ANCIAUX,5 CLAUDE LENOIR,~ JEAN-LUC MARILL,’ HELENELABADIE,8 CLAUDE MASLIAH,2 DANIELPERRIN,’ AND JEAN-CLAUDE CHAPUT,3 DENISVETTER,4 CLAUDE EUGENE,5 LOUISLEBODIC,7 HENRILICHT~ JEAN-PIERRE ETIENNE~ ‘Service des Maladies du Foie et de 1’Appareil Digestif, Hdpital Bidtre, 94270 Le Kremlin Bidtre; 2Service d’Ht!patogastroentt!rologie, Hdtel-Dieu, 44030 Nantes; 3Service d’Hkpatologie et de Gastroentkrologie, Hdpital Antoine Biclere, 92140 Clamart; 4Service d’Hipatogastroentt!rologie,Hdpital Central, 67090 Strasbourg; 5Service de Mkdecine Interne et de Gastroentirologie, Centre Hospitalier Intercommunal, 78300 Poissy; %Servicede Mkdecine C, Centre Hospitalier, 58000 Nevers; 7Clinique des Maladies de I’Appareil Digesti; Hdpital Guillaume et Rent! Laennec, 44030 Nantes; %‘ervice d’Ht!patogastroentt!rologie, Hdpital Delafontaine, 93200 Saint-Denis, France

We conducted a prospective, multicenter, randomized trial to compare the efficacy of sclerotherapy plus propranolol with that of propranolol alone in the prevention of recurrent gastroesophageal bleeding in severely cirrhotic patients. For 2 yr (1987 to 1988) 131 patients (96%of whom were alcoholic)with Child-Pugh class B or C cirrhosis (66% were class B and 44% were class C) were randomly assigned to one of our two treatment groups after cessation of variceal bleeding, without hemostatic sclerosis, and were observed for at least 2 yr. Treatment observance was good in 89% of cases; alcohol withdrawal was observed in 62%of cases. Sclerotherapy was performed weekly with 1% polidocanol, and variceal obliteration was obtained in 83% of cases, in a mean number of four sessions. The cumulative percentages (expressed as mean f S.D.) of recurrent bleeding at 2 yr were 42% f 6% for propranolol plus sclerotherapy and 69% f 6% for propranolol alone (a nonsignificant difference). Twentyeight patients from the propranolol group but only 12 patients from the propranolol-plus-sclerotherapy group had recurrent bleedingfrom esophageal variceal rupture (p c 0.01). The total number of blood units per patient with recurrent bleeding was slightly but not significantlymore important in the propranolol group (8 f 7) than in the propranolol-plus-sclerotherapy group (6 f 5; p = 0.09). There were no statistical differences in the cumulative survival rate at 2 yr (propranolol plus sclerotherapy,74% f 6%and propranolol alone, 64% f 6%)or in the number of patientswho died of repeat bleeding (propranolol plus sclerotherapy, 13% f 4% and propranolol alone, 17% f 5%). Among

Received December 9, 1991;accepted May 14,1992. Address reprint requesta to:0. Ink, M.D.,Centre Hospitalier de Soissons, 46 Avenue du Gn6ral de Gaulle, 02209 Soissons Cedex, France. 31/1/39465

the surviving patients, cirrhosis improved during the follow-up;Child-Pugh classification was the following at 1 mo: 35% for class A, 60% for class B and 15% for class C and the followingat 2 yr: 68%for classA, 38%for class B and 4% for class C. In conclusion, elective sclerotherapy does not significantly decrease the rate of recurrent bleeding or death in severely cirrhotic patients who are treated with propranolol and who mainly abstain from drinking alcohol. (HEPATOLOGY 1992;16912-919.)

In many studies on the secondary prevention of gastroesophageal bleeding in cirrhosis, P-blockade and sclerotherapy have been shown to have equivalent results on recurrent bleeding or overall mortality (1-8). Some recent trials have suggested that propranolol can reduce recurrent gastroesophageal bleeding in cirrhotic patients treated with elective sclerotherapy (9-12). On the contrary, the effectiveness of sclerotherapy among patients already treated with propranolol has rarely been shown (13) but deserves further attention, especially in severely cirrhotic patients. Decision making is rather difficult for these patients, because the risks of both recurrent bleeding and death are high (14-16), P-blockade is not very effective (17-19) and complications of sclerotherapy occur frequently and are sometimes life-threatening (20). Therefore the aim of the present study was to compare the effectiveness of P-blockade alone with that of (3-blockadeassociated with elective sclerotherapy in the prevention of recurrent gastroesophageal bleeding in severely cirrhotic Child-Pugh class B or C (21) patients. PATIENTS AND METHODS Patients. All adult cirrhotic patients seen in eight French centers, from both academic and general hospitals, were

912

Vol. 16,No. 4, 1992

DOES SCLEROTHERAPY IMPROVE THE EFFICACY OF PROPRANOLOL?

considered eligible for inclusion if they met the following criteria: (a) a recent episode of bleeding from esophageal varices that was confirmed by emergency endoscopy (endoscopic stigmata of recent variceal bleeding or esophageal varices with no other pathological condition present to explain major upper gastrointestinal trace bleeding); (b)bleeding that had stopped for at least 24 hr without any sclerotherapy session; (c) a Child-Pugh score greater than 6 (21);and (d) oral acceptance to participate from the patient or from the next of kin if the patient was too ill to consent. Noninclusion criteria included the following: (a) previous treatment with propranolol or sclerotherapy for portal hypertension; (b) contraindication to the use of propranolol because of asthma, cardiac insufficiencyor use of insulin or sulfamides indlcating diabetes mellitus, disturbance of atrioventricular conduction or Raynaud’s syndrome, (c) contraindication to sclerotherapy because of severe encephalopathy, previous esophageal surgery, hiatal hernia longer than 4 cm or esophagealstenosis; (d) existence of HCC or serious illness reducing life expectancy (for example, ongoing cancer, hepatic coma or prothrombin time less than 20%);or (e) unfeasibility of regular surveillance (for reasons of distance or apparent indiscipline). Informed consent was given by all patients and the study protocol was approved by the Ethics Committee of the HBpital de Bidtre. If systematic ultrasound screening or measurement of serum a-fetoprotein levels detected an HCC less than 1 mo after inclusion, the patient was excluded from the study. Study Protocol. After inclusion, the patients in each center were randomly assigned to their treatment groups by sealed opaque envelopes and assigned to receive either propranolol alone or propranolol plus sclerotherapy. Propranolol was given twice daily and the titration was adjusted in an attempt to reduce the resting pulse rate by 25%.Injection sclerotherapy was performed with a flexible endoscope, and the sclerosant, 1%polidocanol, was administered by intravariceal and extravariceal iqjections in the distal 10 cm of the esophagus (22). The quantity used varied between 4 and 6 ml per injection, according to the size of the esophageal varices, for a total volume not greater than 40 to 60 ml per session. The first two to three sessions were performed at 7-day intervals, and subsequent injections were performed on an outpatient basis at 1-mo intervals until the visible varices were eradicated. Ranitidine could be prescribed if an esophageal ulcer was observed during the sessions. After disappearance of the esophageal varices, the results were verified endoscopically every 3 mo for 1 yr and every 6 mo thereafter. In case of recurrence sclerotherapy was repeated according to the same treatment modality. Gastric varices were not injeded. At the beginning of the trial, a questionnaire was filled out containing clinical and laboratory data. The origin of cirrhosis and the present drinking habits of the patients were recorded. The severity of liver damage was graded by Pugh modification of the Child score and divided into two classes: class B was 7 to 9 points and class C was 10 to 15 points. Variceal initial bleeding was classified as certain (active bleeding from varices or signs of recent bleeding-clot or fibrin plug) or probable (varices with no sign of recent bleeding but no other potential source of bleeding). No endoscopic classification was used for variceal size because endoscopic examination was not repeated in the patients receiving propranolol. The delay between the dates of the index bleeding episode and the dates of randomization was recorded as a probable prognostic variable (23,241. Patients were seen 1 mo after inclusion in the trial and then a t 2- to 3-mo intervals. Clinical data, compliance and tolerance to the treatments were recorded at each examination. The modifications of drinking habits were noted; abstinence was

913

defined as the total arrest of alcohol consumption as stated by the patient when considered to be truthful by the practitioner. To complete clinical judgement about alcoholic abstinence, we regularly recorded biological data (mean corpuscular volume and y-glutamyltranspeptidase level). When necessary, we adjusted the dosage of propranolol to obtain the desired heart rate, but no dosage of propranolol or alcohol level was systematically prescribed. The laboratory data were reevaluated at 1 mo and at 2 yr, allowing new classifications of the severity of the liver disease. For the surviving patients a fiberoptic examination was performed at 2 yr; the variceal size and the presence of congestive gastric disease or gastric varices were recorded. The patients were also screened at 2 yr for HCC with ultrasonography and a-fetoprotein assay. When patients failed to attend the follow-up examinations, attempts were made to contact them by mail, by telephone or through their family doctor. If these methods failed, public records were consulted to determine whether the patients were still living. Treatment and follow-up were planned to last 2 yr. The total duration of hospital stays during the first 3 mo of follow-upwere recorded and compared between the two treatment groups. Patients with gastroesophageal bleeding underwent endoscopic examination as soon as possible and were given standard therapy, excluding emergency sclerosis. In the propranolol group the assigned treatment was interrupted only if the recurrent bleeding episode occurred in a compliant patient who had more than a 25% reduction in heart rate at the time of the repeat bleeding episode or who had severe side effects. In most cases chronic sclerotherapy was then performed. In the propranololplus-sclerotherapy group, the assigned treatment was interrupted only if recurrent bleeding occurred after obliteration of the varices, and surgery (usually portocaval shunt surgery) was considered. If the patient bled from causes other than variceal rupture, the assigned treatment was continued as soon as possible. The patients who were withdrawn from the assigned treatment were nonetheless followed until the end of the trial for survival analysis. Statistical Analysis. On the basis of an intention-to-treat principle, statistical analysis was performed from the date of randomization. Two major end points were considered recurrent gastroesophageal bleeding, ascertained by the occurrence of hematemesis or by the endoscopic examination of recurrently bleeding patients, and death. We assumed that 50%of patients in the propranolol group would be free from repeat bleeding after 1yr of follow-up and that sclerotherapy must increase this figure to 75% to be worthwhile in high-risk patients; therefore 62 patients per treatment group were required for a type I error of 5% and a power of 80%, if the comparison was made with the one-sided log-rank test (25). The percentages of patients who survived and were free from bleeding during the 2 yr of follow-up were estimated by means of the Kaplan-Meier method and were compared by means of a log-rank test. Severe repeat bleeding, that is, recurrent gastroesophageal bleeding that required the transfusion of more than 5 units of blood or that led to death, and death from recurrent bleeding were also compared between the two treatment groups. Biological and clinical characteristics of the patients at entry to the trial were compared between the two treatment groups by use of Student’s t test for quantitative variables and the x2 test for qualitative variables. Quantitative variables were expressed as mean t S.D. Differences in p values less than 0.05 were considered significant. The multivariate Cox model was used to identify independent factors predictive of gastroesophageal bleeding and survival and to compare the treatments after adjustment for possible prog-

914

INK ET AL.

HEPATOLOGY

TABLE1. Characteristics of the two groups at entry Characteristics

Propranolol

Propranolol plum sclerotherapy

No. of patients Sex (% male) Age b 9 " Alcoholic cirrhosis (%) Alcohol withdrawal at least 3 mo before index bleeding History of previous bleeding (%) &cites (%) Encephalopathy (%) Bilirubin (bmol/L)" Prothrombin time (%)" Albumin (gm/L)" Child-Pugh class (%) B C Variceal bleeding classified as certain (%) No. of units of blood transfused during index bleeding Use of gastroesophageal tamponade (%) Delay between index bleeding and randomization date (days)" Initial pulse rate (beats/min)"

66 74

65 80

"Data expressed as mean

52

2

10

95 9 14 58 38 75 54 30

2 -t

*

54

* 11

97 19 25 60 25 86 15 4

47 53 39 4 2 3 35 6 f 6 86 2 9

46 2 34 56 f 13 30 2 5 65 35 49 4 2 3 40 5-c4 87 8

*

p Value

0.43 0.27 0.66 0.12 0.11 0.78 0.11 0.01 0.33 0.62 0.04 0.26 0.56 0.54 0.21 0.49

* S.D.

nostic variables. Because of an asymmetrical distribution at randomization, bilirubin was always forced into the proportional hazards model. To symmetrize distributions, we used the logarithms of bilirubin, prothrombin and albumin. Statistical analyses were performed with Programme Informatique de Gestion et #Analyses Statistique (26) and Biomedical Data Package softwares (27) at the INSERM Unit6 292, Department de Statistiqueset d'InformatiqueMBdicales, HBpital de Bicgtre.

RESULTS Patients. Between December 1986 and January 1989, 543 consecutive cirrhotic patients who had bled from esophageal varices were considered for inclusion in this trial. During this 2-yr period, 412 patients were excluded for the following reasons: continual bleeding or death before randomization (18%);classification as Child-Pugh class A (15%); previous treatment with propranolol or sclerotherapy for portal hypertension (21%); contraindication to p-receptor blockade (12%); a history of HCC (8%);a history of other cancers (2%);and refusal to participate or inability to be followed up (10%). Two noninclusion criteria were observed in 10%of these patients. The remaining 131 patients were enrolled in the study (7to 31 patients per participating center) and were randomly assigned to receive either propranolol alone (n = 66) or propranolol plus sclerotherapy (n = 65). The characteristics of the two groups randomized are summarized in Table 1. There were no significant differences with regard to the clinical and biological recorded data between the two groups of patients, apart from bilirubin levels that were significantly higher in the propranolol group (75 f 86 FmoVL) than in the propranolol-plus-sclerotherapygroup (46 k 34 FmoVL; p = 0.01). In particular, bilirubin levels were higher

than 100 FmoVL in the blood of 15 patients allocated to propranolol but in only the blood of 2 patients allocated to propranolol plus sclerotherapy. Consequently, significantly more propranolol patients (53%) than propranolol-plus-sclerotherapypatients (35%;p 0.05) belonged to the Child-Pugh class C group. The main criteria (incidence of repeat bleeding and death) were therefore compared after adjustment for initial bilirubin levels. Follow-up. Propranolol was adjusted over a mean period of 5 k 5 days (mean & S.D.), up to a maximum initial dosage of 120 ? 80 mg/day, and p-blockade was considered obtained in 82% of patients at the end of the hospital stay. No significant difference was observed between the two treatment groups with regard to propranolol treatment. Sclerotherapy was begun 2 ? 2 days after the time of randomization. Variceal obliteration was achieved in 83% of patients, with a mean of 4 2 2 courses in 3 2 1 mo. The total duration of hospital stay during the 3 first mo of follow-up was 20 k 24 days in the propranolol group and 17 2 22 days in the propranolol-plus-sclerotherapygroup, which was considered not significant. Twenty patients (15%) were lost to follow-up after a period of 15 8 mo (12in the propranolol group and 8 in the propranolol-plus-sclerotherapygroup). With the exception of these patients, patients were monitored for at least 2 yr or until death. In the propranolol group severe side effects occurred in 10 patients (15%). Three patients had dyspnea, two had cardiac insufficiency, three had orthostatic hypotension and one had hypoglycemia. Treatment was discontinued in seven of these patients. Nine patients (14%) did not regularly take propranolol. In the propranolol-plus-sclerotherapy group, severe side effects were observed in 21 patients (33%). These side effects included dyspnea in three

-=

*

Vol. 16, No. 4, 1992

DOES S C L E R O T H E W Y IMPROVE THE EFFICACY OF PROPRANOLOL?

915

OF PATIENTS FREE OF REBLEEDING (96)

PROPRANOUX

T + SCLEROTHERAW

60-

i P=0,20

5 0

-h

PROPRANOLOL *.-0

-* 6

12

months

14

FIG.1. Cumulative percentages of patients free from repeat bleeding in the propranolol (P)and propranolol-plus-sclerotherapy(P + S) groups, from the randomization date (Kaplan-Meier estimates & 1 S.D.). Symbols at bottom represent the following: 0 = patients lost to follow-up before repeat bleeding; A = patients withdrawn from the study before repeat bleeding because of severe, propranolol-related side effects; and = patients dead before repeat bleeding.

*

patients, cardiac insufficiency in one, hepatic coma in one, orthostatic hypotension in three, hypoglycemia in one, pleuropulmonary infection in one, severe esophageal ulcer in seven and dysphagia with esophageal stenosis in four, but discontinuation of the treatment was required in only five patients. Five patients (8%) neither took propranolol regularly nor followed the sclerotherapy program. Alcohol withdrawal was observed during follow-up in 62% of alcoholic patients, whatever the treatment allocated. Resultu at 1 Mo. During the first month after randomization, 16 patients had recurrent bleeding episodes (8 in the propranolol group and 8 in the propranolol-plus-sclerotherapygroup). Seven patients died (3in the propranolol group and 4 in the propranololplus-sclerotherapy group), 6 of whom died of recurrent bleeding (2 in the propranolol group and 4 in the propranolol-plus-sclerotherapygroup). New clinical and biological data were available at 1 mo for calculation of Child-Pugh score in 115 patients (57 in the propranolol group and 58 in the propranolol-plus-sclerotherapy group). The Pugh score had improved at least 2 points in 56% of the propranolol patients and in 50% of the propranolol-plus-sclerotherapypatients, which was not significant. At 1 mo, both treatment groups were comparable for any of the Pugh variables, and Child-Pugh classification was the following: 35% were in class A, 50% were in class B and 15%were in class C. Recurrent Bleeding. Sixty patients had repeated gastroesophageal bleeding during the study period; 35 were in the propranolol group, and 25 were in the

propranolol-plus-sclerotherapy group, which was not significant. The cumulative percentages of patients free from additional bleeding episodes at 1 and 2 yr after inclusion were 60% t 6% and 41% -C 6% respectively, in the propranolol group and 72% t 6%and 58% ? 6% respectively, in the propranolol-plus-sclerotherapy group (Fig. 1).The overall differences between the two groups were not significant before (p = 0.07) and after (p = 0.20) adjustment for the degree of bilirubinemia. Taking into account these results, we found that the p-risk error (type I1 error) was 30%. The cumulative percentages of patients who were free from repeat bleeding 3 yr after inclusion were 37% k 7% in the propranolol group and 55% ? 7% in the propranololplus-sclerotherapy group, which was not significant. Among the 60 patients who had at least one recurrent bleeding episode, the numbers and the types of bleeding episodes during the study period were not significantly different between the two treatment groups (Table 2). However, 28 patients from the propranolol group and only 12 patients from the propranolol-plussclerotherapy group had additional bleeding episodes caused by esophageal variceal rupture (p < 0.01).In the propranolol group 19 of 35 patients with repeat bleeding subsequently underwent a course of chronic sclerotherapy. In the propranolol-plus-sclerotherapygroup, 14 of 25 patients bled again before variceal obliteration. The cumulative mean number of blood units per patient with recurrent bleeding was 8 t 7 in the propranolol group and 5 ? 5 in the propranolol-plus-sclerotherapy group (p = 0.09). Twenty-four patients had severe

INK ET AL.

916

HEPATOLOGY

TABLE 2. No. of gastroesophagealbleeding episodes according to treatment and causes Parameters

Bleeding episodes Esophageal varices Erosive gastritis Gastric varices Esophageal ulcer Other

Unknown

Propranolol (n = 36)"

Propranolol plus sclerotherapy (n = 26)"

p Value

1.9 ? 0.5 1.2 t 1.1 0.2 t 0.4 0.1 ? 0.7 0.1 t 0.2 0.1 % 0.3 0.2 % 0.6

1.6 % 0.3 0.7 ? 1.1 0.2 ? 0.4 0.1 ? 0.3 0.2 % 0.4 0.04 t 0.2 0.4 t 0.8

0.27 0.07 0.91 0.86 0.12 0.16 0.56

"Data expressed as mean t S.D.

episodes of recurrent bleeding, 14 from the propranolol group and 10 from the propranolol-plus-sclerotherapy group, which was not significant. The multivariate Cox model indicated that the level of bilirubin in the blood and the delay between the index bleeding episode and the randomization date were independent prognostic factors for the onset of recurrent gastroesophageal bleeding (Table 3). The statistical significance of treatment for prediction of repeat gastroesophagealbleeding was 0.12 after adjustment for these two factors. Survival. The cumulative percentages of patients surviving 1 and 2 yr after inclusion were, respectively, 7670 & 5% and 64% k 6% in the propranolol group and 82% 5% and 72% k 6% in the propranolol-plussclerotherapy group (Fig. 2). The overall differences between the two groups were not significant before (p = 0.26) and after (p = 0.77) adjustment for the degree of bilirubinemia. The cumulative 3-yr survival rates were 55% f 7% in the propranolol group and 64% k 7% in the propranolol-plus-sclerotherapygroup, which was not significant. The occurrence of repeat bleeding was a major prognostic factor for prediction of survival during the study period; 50% of the patients with recurrent bleeding episodes (30 of 60) but only 14% of the patients without recurrent bleeding episodes died (10 of 71) (p c 0.0001). The multivariate Cox model indicated that the occurrence of repeat bleeding, the degree of bilirubinemia and the level of albumin were independent prognostic factors for survival (Table 3). The statistical significanceof treatment for prediction of survival was 0.89 after adjustment for these three factors. In the propranolol group 23 patients died (10 of recurrent bleeding, 9 of hepatic insufficiency, 3 of other causes and 1 of unknown causes). In the propranololplus-sclerotherapy group, 17 patients died (8 of recurrent bleeding, 5 of hepatic insufficiency and 4 of other causes). Five patients from each treatment group died during their first recurrent bleeding episode. The cumulative percentages of patients who died of recurrent bleeding were, at 2 yr, 17% 3% in the propranolol group and 13% 4% in the propranololplus-sclerotherapy group (Fig. 3). No significant difference was observed between the two treatment groups before (p = 0.59) and after (p = 0.90) adjustment for the level of bilirubin in the blood.

*

*

End Point Analysis. Seventy-one patients (31 in the propranolol group and 40 in the propranolol-plussclerotherapy group) were evaluated 2 yr after inclusion (delay was 24 & 7 mo). Sixty-four patients still received propranolol, 44 patients still followed a sclerotherapy program (6 in the propranolol group and 38 in the propranolol-plus-sclerotherapy group) and 5 patients had been operated on (4 underwent portocaval anastomosis and 1 underwent liver transplantation). Forty patients were abstinent (62% of the propranolol group and 56% of the propranolol-plus-sclerotherapygroup). According to the Child-Pugh classification, 58% of patients belonged to group A, 38%to group B and 4% to group C. A gastroesophageal endoscopy was performed in 62 patients and showed that esophageal varices were present in 56% of the propranolol patients and 5% of the propranolol-plus-sclerotherapy patients (p < 0.01). Gastric varices were observed in 37%of the propranolol patients and 11%of the propranolol-plus-sclerotherapy patients (p c 0.05). Congestive gastric disease was observed in 56% of both groups. During the study period HCC was diagnosed in eight cases and was associated with recurrent bleeding or death in five cases. DISCUSSION

This trial shows that elective sclerotherapy does not improve the efficacy of propranolol in the prevention of recurrent gastroesophagealbleeding in Child-Pugh class B or C cirrhotic patients. There was no difference in survival over a 2- to 3-year period. Both treatments, propranolol alone or propranolol plus sclerotherapy, were equally effective during the first month after the index bleeding episode, when the risks of additional episodes of bleeding and death are very high (14,161. In our patients who had already been treated with propranolol, sclerotherapy did not significantly reduce the incidence of severe recurrent bleeding or death from recurrent bleeding, the number of repeat bleeding episodes or the number of blood units transfused during the study period. Our unique positive result was a significantly lower frequency of recurrent esophageal variceal bleeding in the propranolol-plus-sclerotherapy patients, when compared with that of the propranolol patients, and a nonsignificant reduction of the number of blood units transfused after randomization. From many studies on the secondary prevention of gastroesophageal bleeding in cirrhosis, some major

Vol. 16, No. 4, 1992

DOES SCLEROTHERAPY IMPROVE THE EFFICACY OF PROPRANOLOL?

CUMULATIVE

917

PERCENTAGES

SURVIVING PATIENTS

(96

T

90

80;

70

-

NS

60!

rnonlhs

v 6

11

14

FIG.2. Cumulative percentages of surviving patients treated with propranolol or propranolol plus sclerotherapy from the randomization date (Kaplan-Meier estimates f 1 S.D.). NS = not significant.

TABLE 3. Prognostic factors for recurrent gastroesophagealbleeding and death identified by the Cox model p Value

Model" ~~~~

~

Recurrent gastroesophageal bleeding model Bilirubinemia Delay between index bleeding and randomization (days, Death model Recurrent bleeding Bilirubinemia Albumin

~~

Relative risk

~

0.002 0.03

More than 100 FrnolL vs. less = 3.2 Less than 5 days vs. more = 1.5

0.0001 0.0001 0.004

Yes vs. no = 4.6 More than 100 FmoVL vs. less = 5.2 Less than 28 grwl vs. more = 1.5

"Variables are shown in the order of entry to the model.

conclusions have already been established. First, because of the constantly high risk of repeat bleeding and because there have been a large number of conclusive trials, especially concerning P-blockers and sclerotherapy (28-30),no further trials should be performed in which some patients receive no specific treatment (30). Second, predictive data for the highest risk of recurrent bleeding, such as severity of liver disease (15,19,31,32), endoscopic criteria (33), time delay from the index bleeding episode or admission to the start of analysis (23, 24) and compliance and alcohol withdrawal (34, 35), must be considered as confounding variables in the analysis of future trials, or they could be used for the selection of patients for specific treatments. Third, p-blockade significantly reduces the occurrence of recurrent gastroesophageal bleeding, death from recurrent bleeding and overall mortality (28) but is less effective in Child class C patients (17, 19); however, propranolol has the advantage of being safe and costeffective. Fourth, sclerotherapy significantly reduces the risk of recurrent variceal bleeding and of mortality, and most comparative trials have shown both treatments to

have nonsignificantly hfferent results (1-8, 36). However, one study suggested that apart from emergency treatment for bleeding, elective sclerotherapy had no effect on the prevention of recurrent variceal bleeding (37).During sclerotherapy, variceal eradication does not depend on the functional hepatic reserve (38). Fifth, recent studies have suggested that propranolol can reduce recurrent gastroesophageal bleeding in cirrhotic patients treated with elective sclerotherapy (9-121, but only one prospective trial evaluated the efficacy of combined sclerotherapy and propranolol treatment compared with propranolol alone (13). Our sclerotherapy program was among the most commonly recommended (391, with no significant discrepancy between our eight different centers. Our results concerning the variceal eradication and its delay from the start of the sclerotherapy sessions are very similar to previous data, as published in recent controlled studies (1-3,6, 7). The frequencies of esophageal and gastric, mostly junctional (401, varices were significantly decreased at 2 yr in our propranolol-plussclerotherapy group. Because no technique of chronic

918

INK ET AL.

HEPATOLOGY

-

lPROPRANOLOL

-

+ SCLEROTHERAPV

L

I months

sclerotherapy has been proved to be more effective than another for the prevention of recurrent gastroesophaged bleeding, our negative results cannot be assigned to a specific lack of efficiency of our sclerotherapy program. Our observed rates of compliance with treatment and alcohol withdrawal were comparable with previously published rates, which were, respectively, 80%to 85%(3, 7) and 40% to 70% (1,3, 7). In our multivariate analysis, alcohol abstinence was correlated neither with recurrent bleeding nor with death, and it was a less comprehensive result. Alcohol withdrawal was probably an important contributing factor in the compliance rate and in the time-related decrease of the severity of liver disease observed in our patients. However, because abstinence is difficult to prove conclusively in an individual patient, these conclusions should be drawn only after a prospective and specific study. Our results confirm the close relationship between the risk of repeat bleeding and the severity of liver disease in cirrhotic patients. Our multivariate analysis showed that bilirubinemia, a factor reflecting hepatic reserve, was highly and independently predictive of recurrent bleeding and that recurrent bleeding was highly and independently predictive of mortality. Previous studies have shown that severe cirrhotic liver deficiency was correlated with recurrent bleeding in historical series (15) in propranolol- (19) or sclerotherapy-treated patients (15, 31, 32, 41) and in comparative trials (7). In other studies the survival of cirrhotic patients with esophageal varices was significantly influenced by the recurrence of bleeding, especially in the 2-week (42) to 6-month period (14, 16). Elective sclerotherapy is probably not very effective in the prevention of early recurrent variceal bleeding, as previously suggested in controlled trials (43). In the only published trial that compared randomized propranolol-treated patients with propranolol-plus-sclerotherapy-treated patients (13), the initial sclerotherapy session was performed early (within 48 hr of bleeding) for most patients. Signifi-

cantly more patients receiving propranolol had severe repeat bleeding episodes in the fist month after randomization than did patients receiving sclerotherapy plus propranolol. This difference in severe recurrent bleeding was sustained until 15 mo after randomization but not thereafter. The ability of early sclerotherapy to stop active bleeding and to prevent early recurrent variceal bleeding is consistent with our negative results, because we performed only elective sclerotherapy programs without hemostatic sessions. From our results, we believe that in severely cirrhotic patients, propranolol alone is the most rational first-line treatment for the prevention of recurrent bleeding from variceal rupture and portal hypertensive gastric disease (44). Early sclerosiscould benefit patients in the prevention of recurrent short- and middle-term variceal bleeding. Elective sclerotherapy should be initiated in patients who have contraindication to p-blockade or when propranolol has already failed. The most promising approach in cirrhotic patients could be the prevention of the first gastrointestinal bleeding episode with p-blockade (45). Acknowledgment: We thank Mrs. Catherine Corbet for her secretarial assistance. REFERENCES 1. Alexandrino PT, Alves MM, Pinto Correia J. Propranolol or endoscopic sclerotherapy in the prevention of recurrence of variceal bleeding: a prospective, randomized controlled trial. J Hepatol 1988;7:175-185. 2. Dollet JM,Champigneulle B, Patris A, Bigard MA, Gaucher P. Endoscopic sclerotherapy versus oral propranolol after variceal hemorrhage in cirrhosis: results of a 4-year prospective randomized trial. Gastroenterol Clin Biol 1988;12:234-239. 3. Fleig WE,Stange EF, Hunecke R, Schonborn W, Hurler U, Rainer K, Gaus W, et al. Prevention of recurrent bleeding in cirrhotics with recent variceal hemorrhage: prospective, randomized com1987;7: parison of propranolol and sclerotherapy. HEPATOLOGY 355-361. 4. Ink 0,Buffet C, Fritsch J, Pelletier G, Honein K, Attali P, Etienne JP.Prevention of recurrent bleeding in cirrhotic patients: is sclerotherapy better than propranolol? Presse Med 1988;17:615-619.

Vol. 16,No. 4,1992

DOES SCLEROTHERAPY IMPROVE THE EFFICACY OF PROPRANOLOL?

5. Martin T, Taupignon A, Lavignolle A, Perrin D, Lebo&c L.

Prevention of recurrent bleeding in cirrhotics: result of a controlled trial of propranolol versus endoscopic sclerotherapy. Gastroenterol Clin Biol 1991;15:833-837. 6. Qureshi H, Zuberi SJ, Alam E. Efficacy of oral propranolol and injection sclerotherapy in the long-term management of variceal bleeding. Digestion 1990;46:193-198. 7. Rossi V, Cales P, Burtin P. Charneau J, Person B, Pujol P, Valentin S, et al. Prevention of recurrent variceal bleeding in alcoholic cirrhotic patients: prospective controlled trial of propranolol and sclerotherapy. J Hepatol 1991;12:283-289. 8. Westaby D, Polson RJ, Gimson AES, Hayes PC, Hayllar K, Williams R. A controlled trial of oral propranolol compared with injection sclerotherapy for the long-term management of variceal 1990;11:353-359. bleeding. HEPATOLOGY 9.Avgerinos A, Rekoumis G, Klomis C, Papadimitriou N, Gouma P, Poumaras S, Raptis S. Endoscopic sclerotherapy (ES) versus the combination of ES and propranolol for the prevention of recurrent gastrointestinal bleeding in cirrhotics: a prospective study [Abstract]. Gastroenterology 1991;100:A717. 10. Jensen LS,K r m p N. Propranolol in the prevention of rebleeding from esophageal varices during the course of endoscopic sclerotherapy. Scand J Gastroenterol 1989;24:339-345. 11. Lundell L, Leth R, Lind T, Lonroth H, Sjovall M, Olbe L. Evaluation of propranolol for prevention of recurrent bleeding from esophageal varices between sclerotherapy sessions. Acta Chir Scand 1990;156:711-715. 12. Vine1 JP,Lamouliatte H, Cales P, Combis JM, Desmorat H, Roux D, Quinton A, et al. Propranolol reduces the rebleeding rate during injection sclerotherapy: final results of a randomized study [Abstract]. Gastroenterology 1990;98:A644. 13. O’Connor KW, Lehman G, Yune H, Brunelle R, Christiansen P, Host J, Compton M, et al. Comparison of three non surgical treatments for bleeding esophageal varices. Gastroenterology 1989;96:899-906. 14. Cortez-Pinto H, Abrantes A, Esteves A, Almeida H, Pinto Correia J. Long-term prognosis of patients with cirrhosis of the liver and upper gastrointestinal bleeding. Am J Gastroenterol 1989;84: 1239-1242. 15. Dimagno EP, Zinsmeister AR, Larson DE, Viggiano TR, Clain JE, Laughlin BL, Hughes RW. Influence of hepatic reserve and cause of esophageal varices on survival and rebleeding before and after the introduction of sclerotherapy: a retrospective analysis. Mayo Clin Proc 1985;60:149-157. 16. Graham DY,Lacey Smith J . The course of patients after variceal hemorrhage. Gastroenterology 1981;80:800-809. 17. Braillon A, Cales P, Valla D, Gaudy D, Geoffroy P, Lebrec D. Influence of the degree of liver failure on systemic and splanchnic haemodynamics and on response to propranolol in patients with cirrhosis. Gut 1986;27:1204-1209. 18. Garden OJ, Mills PR, Birnic GG, Murray GD, Carter DC. Propranolol in the prevention of recurrent variceal hemorrhage in cirrhotic patients: a controlled trial. Gastroenterology 1990;98: 185-190. 19. Ink 0, Servent L, Attali P, Pelletier G, Buffet C, Etienne JP. Propranolol in the prevention of variceal rebleeding: the deleterious influence of ascites and jaundice. Gastroenterol Clin Biol 1985;9:819-823. 20. &human BM, Beckman JW,Tedesco FJ, Griffin JW,Assad RT. Complications of endoscopic injection sclerotherapy: a review. Am J Gastmnterol 1987;82:823-830. 21. Pugh RNH, Murray-Lyon IM, Dawson JL,Pietroni MC, Williams R. Transsedion of the oesophagus for bleeding oesophageal varices. Br J Surg 1973;60:646-649. 22. Soehendra N,De Heer K, Kempeneers I, Frommelt L. Morphological alterations of oesophagus after endoscopic sclerotherapy of varices. Endoscopy 1983;15:291-296. 23. Bullimore D. Variceal bleeding: does variable delay in trial entry invalidate comparison between trials? Gut 1987;28:653-656. 24. Burroughs AK, Mezzanotte G, Phillips A, McCormick PA, McIntyre N. Cirrhotics with variceal hemorrhage: the importance of the time interval between admission and the start of analysis for survival and rebleeding rates. HEPATOLOGY 1989;9:801-807. 25. Casagrande JT,Pike MC, Smith PG. An improved approximate

919

formula for calculating sample sizes for comparing two binomial distributions. Biometrics 1978;34:483-486. 26. Wartelle M, Kramara A, Jan P, Kruger D. PIGAS. An interactive statistical data base management system. In: Proceedings of the second international workshop on statistical data base management. Los Altos, CA:1983. 27. Dixon WJ. BMDP statistical software. Los Angeles: University of California Press, 1981. 28. Hayes PC, Davis JM, Lewis JA, Bouchier IAD.Meta-analysis of value of propranolol in prevention of variceal haemorrhage. Lancet 1990;336:153-156. 29. Infante-Rivard C, Esnaola S, Villeneuve JP. Role of endoscopic variceal sclerotherapy in the long-term management of variceal bleeding: a meta-analysis. Gastroenterology 1989;96:10871092. 30. Pagliaro L, Burroughs AK, Sorensen TL4, Lebrec D, Morabito A, D’Amico G, Tine F. Therapeutic controversies and randomized controlled trials: prevention of bleeding and rebleeding in cirrhosis. Gastroenterol Int 1989;2:71-84. 31. Sauerbruch T, Weinzierl M, Ansari H, Paumgartner G. Injection sclerotherapy of oesophageal variceal haemorrhage: a prospective long-term follow-up study. Endoscopy 1987;19:181-184. 32. Garret KO,Reilly JJ, Schade RR, Van Thiel DH. Sclerotherapy of esophageal varices: long-term results and determinants of survival. Surgery 1988;104:813-818. 33. Sarin SK, Sundaram KR, Ahuja RK. Predictors of variceal bleeding: an analysis of clinical, endoscopic, and haemodynamics variables, with special reference to intravariceal pressure. Gut 1989;30:1757-1764. 34. Poynard T, Lebrec D, Hillon P, Sayegh R, Bernuau J , Naveau S, Chaput JC, et al. Propranolol for prevention of recurrent gastrointestinal bleeding in patients with cirrhosis: a prospective 1987;7: study of factors associated with rebleeding. HEPATOLOGY 447-451. 35. Colombo M, De Franchis R, Tommasini M, Sangiovanni A, Dioguardi N. Beta-blockade prevents gastrointestinal bleeding in well-compensated patients with alcoholic cirrhosis: a multicenter randomized controlled trial. HEPATOLOGY 1989;9:433-438. 36. Fleig WE,Stange EF, Schonborn W, Wordehoff D, Preclick G, Nubon R, Rainer K, et al. Final analysis of a randomized trial of propranolol vs. sclerotherapy for the prevention of recurrent variceal hemorrhage in cirrhosis [Abstract]. HEPATOLOGY 1988;8: 1220. 37. Burroughs AK, McCormick PA, Siringo S, Phillips A, Sprengers D, McIntyre N. Prospective randomized trial of long-term sclerotherapy for variceal rebleeding using the same protocol to treat rebleeding in all patients: final report [Abstract]. HEPATOLOGY 1989;10:579. 38. Chin K, Korula J , KOY, Yamada S. Factors influencing the efficacy of chronic endoscopic sclerotherapy in variceal hemorrhage [Abstract]. Gastroenterology 1988;94:A68. 39. Terblanche J , Burroughs AK, Hobbs KEF. Controversies in the management of bleeding esophageal varices. N Engl J Med 1989;320:1469-1475. 40. Hosking SW,Johnson AG. Gastric varices: a proposed classification leading to management. Br J Surg 1988;75:195-196. 41. Van Hootegem P, Van Besien K, Broeckaert L, Rutgeerts P, Fevery J. Endoscopic sclerotherapy of esophageal varices: long-term follow-up, recurrence and survival. J Clin Gastroenterol 1988;10:368-372. 42. Smith JL,Graham DY. Variceal hemorrhage: a critical evaluation of survival analysis. Gastroenterology 1982;82:968-973. 43. The Copenhagen Esophageal Varices Sclerotherapy Project. Sclerotherapy after first variceal hemorrhage in cirrhosis: a randomized multicenter trial. N Engl J Med 1984;311:1594-1600. 44. Perez-Ayuso RM,Pique JM, Bosch J , Panes J , Gonzales A, Perez R, Rigau J, et al.Propranolol in prevention of recurrent bleeding from severe portal hypertensive gastropathy in cirrhosis. Lancet 1991;337:1431-1434. 45. h d r e m i T,Poupon RE, B a h u BJ, Trinchet JC, Grange JD, Peigney N, Beaugrand M, et al. Preventive therapy of first gastrointestinal bleeding in patients with cirrhosis: results of a controlled trial comparing propranolol, endoscopic sclerotherapy 1990;12:1413-1419. and placebo. HEPATOLOGY