Propranolol or endoscopic sclerotherapy in the prevention of recurrence of variceal bleeding

JournalofHepatology, 1988; 7:175-185 175 Elsevier

HEP 00440

Propranolol or endoscopic sclerotherapy in the prevention of recurrence of variceal bleeding

A prospective, randomized controlled trial*

Paula T. Alexandrino, M. Martins Alves and J. Pinto Correia Department of Medicine 2, Center of Gastroenterology (INIC), University Hospital of Santa Maria, Lisbon (Portugal)

'Received 30 December 1987) (Accepted 26 April 1988)

Summary

Endoscopic sc lerotherapy (ES) and continuous propranolol (P) t rea tment have both been proposed as useful

methods to prevent recurrent esophageal variceal bleeding. We report a prospective randomized trial in 65 pa-

tients with a previous history of endoscopical ly proven esophageal variceal bleeding_ Patients were randomized

by sealed envelopes stratified for Child 's A and B groups to receive either endoscopic sclerotherapy (n = 31) or

propranolo l (n = 34). The dose of oral propranolol was based on a reduction of the resting pulse rate by 25%. In-

t ravascular e thanolamine oleate was used for the endoscopic sclerotherapy in a 3-week schedule. The follow-up

per iod ranged from 17 to 57 months (median: ES = 31; P = 28 months). There was no difference in the cumula-

tive percentages of patients free of rebleeding from any source: esophageal and gastric varices, acute esophageal

and gastric ulcers or erosions (ES = 37%; P = 16%). Also, there was no difference in the cumulative survival (ES

= 69% ; P = 54%). However , patients in the propranolol group had significanily more variceal rebleeding from

the esophagus (n = 21) than did those in the sclerotherapy group (n = 9). The cumulative percentages of patients

free of esophageal variceal rebleeding after inclusion were 67% in the endoscopic sc lerotherapy group and 25%

in the propranolo l group ( log-rank test, P < 0.02). These differences indicated that sclerotherapy should be used

in 29% of the propranolol patients who rebled. Based on these results we recommend elective sc lerotherapy as

long-term therapy for preventing rebleeding of esophageal varices.

Introduction

Recurrent variceal bleeding is one of the main

causes of morbidi ty and poor quality of life in cirrhot-

ics; it is also responsible for their high mortal i ty [1].

Disappoint ing results were obta ined in trials compar-

* Part of the results were presented at the EASL and AASLD Meetings in 1986. Correspondence: Prof. J. Pinto Correia, Dept. Med. 2, Hospital Santa Maria, 1699 Lisbon Codex, Portugal.

0168-8278/88/$03.50(~) 1988 Elsevier Science Publishers B.V. (Biomedical Division)

176 P.T. ALEXANDRINO et al.

ing non-selective portal-systemic shunts with routine

medical therapy [2-4].

Lebrec et al. [5] demons t ra ted a significant reduc-

tion of recurrent variceal and non-variceal bleeding

and an increased survival in selected well-compen-

sated cirrhotic patients t reated with propranolol .

However , three subsequent studies in a less selected

populat ion [6-8] failed to confirm these results. On

the other hand, a few control led studies [9-11] re-

ported a beneficial effect of endoscopic sclerothera-

py in reducing the number of episodes of variceal re-

bleeding, and one of them claimed the addit ional

benefit of long-term survival [10].

Thus, in recent times, the most commonly used

non-surgical t rea tment al ternatives for prevention of

recurrent variceal bleeding are ei ther endoscopic

sclerotherapy or the long-term administrat ion of oral

propranolol . We designed a prospect ive randomized

trial to compare the short- and long-term results of

both therapies in patients with portal hypertension

who had bled from esophageal varices.

were a l ready in a sc lerotherapy group program, and

35 did not meet the above-ment ioned inclusion crite-

ria. Ano the r 27 were lost to immedia te follow-up_

The remaining 48 (17%) plus 17 referred from other

hospitals (meet ing the same selection criteria) in a total of 65 were randomly al located ei ther to sclero-

therapy (n = 31) or to propranolol (17 = 34).

The characteristics of the two groups randomized

are summarized in Table 1_ There were no significant

differences in age and sex distr ibution, cause of por-

tal hyper tension (confirmed histologically in 61 of 65

pat ients) , previous episodes of bleeding or severi ty of

liver damage graded on randomizat ion by Camp-

bell 's modificat ion of the Chi ld-Turcot te classifica-

tion [12].

Study design Patients with clinically significant bleeding (hema-

temesis and/or melena requiring blood transfusion or

with a fall in hemoglobin of 3 g/dl or greater) under-

went emergency endoscopy.

Materials and Methods

Patients The study populat ion comprised 65 patients with

portal hypertension with at least one episode of

esophageal variceal bleeding. Cri ter ia for selection

to this study were: (a) recent episode of bleeding

from esophageal varices confirmed by emergency en-

doscopy (patients with suspicion of hemorrhage from

gastric erosions or o ther gastric or duodenal patholo-

gy were not included); (b) Child 's A and B groups;

(c) no contraindicat ions to the use of fl-blocking

agents, such as asthma or airway obstruct ion, left

ventricular failure, d iabetes mellitus type I; (d) no

previous sclerotherapy; (e) acceptance by the pa-

tients and/or his doctor to be included in the trial.

During a 3.5-year period, 277 patients with rup-

tured esophageal varices were admit ted to the

Gastroenterological Intensive Care Unit of our Hos-

pital_ Eighty-six died during the hospital admission

and 116 were excluded from randomizat ion: 81 were

submit ted to urgent endoscopic sc lerotherapy or

TABLE 1

CHARACTERISTICS OF THE TWO GROUPS AT ENTRY

Sclerolherapy Propranolol (n = 31) ( . = 34)

Age (yr. mean + S.D.) 44.9 + 11.3 49.9 + 11).9 Sex (M/F) 24/7 28/9 Diagnosis

alcoholic cirrhosis 26 26 non-alcoholic cirrhosis 2 6 nodular rcgcnerative hyperplasia I I

idiopathic portal thrombosis 2 1 Child's classification

Group A 23 24 Group B 8 10

No. o f previous b]eedings mean + S.D. 2.6 + 1.5 2.9 + 1.8 No. of patients 22 28

Transfusions (units) during the index bleeding episode (mean + S.D.) 2.8 + 2.4 3.8 _+ 2.8

Cardiac index" 5.(1 + 1.4 5.0 _+ 1.4 Portal pressure

(WHVP-FHVP)" 17.9 + 6.5 18.2 + 5 1 (n = 19) (n = 22)

Only in a subset of patients with liver cirrhosis.

PREVENTION OF VARICEAL REBLEEDING 177

Esophageal varices were considered the source of

bleeding when one of the following criteria was ob-

served: active bleeding from varices; signs of recent bleeding (clot or fibrin plug); no signs of recent

bleeding but no other potential source of bleeding. The hemorrhage was controlled by intravenous infu-

sion of vasopressin and/or Sengstaken-Blakemore balloon tamponade. Child's classification was estab-

lished at least 15 days after the end of the index bleeding episode, when the patients were hemodyna-

mically stable_ At that time, patients were classified

according to a numerical score: Child's A, 5 - 8 points; Child's B, 9-11 points [12]. Randomization

was done by a list of random numbers allocated to sealed envelopes in two separate groups for Child's A

and B patients: even numbers, sclerotherapy (n =

31); odd numbers, propranolol (n = 34). The mean interval between the end of bleeding and randomiza-

tion was 17.9 + 9.7 (mean + S.D.) days in the sclerotherapy group and 19.7 + 8.8 days in the propranolol

group.

After randomization, cardiac output and hepatic and systemic hemodynamic parameters were mea-

sured by a transjugular approach before starting

therapy, in 19 of 28 (68%) patients with cirrhosis in the sclerotherapy group and in 22 of 32 (69%) pa-

tients with cirrhosis in the propranolol group (Table 1). These studies were repeated after treatment in

some patients in both groups, and the results have

been published in part [13]. Sclerotherapy was carried out with intravascular

injection of 5% ethanolamine oleate through a later-

al-view Olympus endoscope (GIF-K 2 or Kl0 ) and NM-IK needle injector (Olympus)_ The procedure was performed under diazepam sedation, with a 'free-hand' technique, every 3 weeks, until the

esophageal varices had been obliterated at the

gastro-esophageal junction or were too small to receive further injections. Endoscopic follow-up was performed every 3 months in the first year, and every

6 months after, and varices were reinjected if they re- curred, at similar intervals to the initial injections.

Propranolol was given orally, twice a day, in in- creasing doses, until the resting pulse rate was reduced by 25%. Patients were seen every week for 1

month and then at intervals of 3 months and the dos-

age of propranolol was adjusted, if necessary, to maintain a reduction in resting pulse of approximate-

ly 25%. This expected response to the drug was used

as a measure of compliance which was also assessed by the discrepancy between the number of tablets is-

sued to the patients and patients' requests for further

supplies. All patients had a clinical and laboratory follow-up

and were systematically asked about drinking habits.

All potential adverse side-effects were looked for

and registered. All clinical rebleeding episodes (hematemesis or

melena after a 24 h free interval) requiring blood transfusion or with a fall in hemoglobin of 3 g/dl or greater were treated in our GE Intensive Care Unit,

and endoscopically assessed to confirm the cause of

bleeding. In a case of no active bleeding from esoph-

ageal varices but where gastric varices or acute gastric or duodenal erosions were bleeding at the time of

endoscopy, the cause of rebleeding was ascribed to these associated lesions. Otherwise, the source of

bleeding was considered' unknown_ In the sclerotherapy group, esophageal variceal rebleeding was

treated with standard medical treatment and urgent

sclerotherapy, avoiding balloon tamponade when possible. In the propranolol group patients were treated with standard medical treatment and with ur-

gent sclerotherapy when they met the following criteria: either one single episode with shock or requiring

4 units of blood trarrsfusions or with a fall in hemoglo-

bin below 8 g/dl; or in any case after the second episode of rebleeding. Patients who required urgent

sclerotherapy underwent further elective injections. Two patients were submitted to surgery after failure

of both therapies.

Statistical analysis This was performed on the basis of an 'intention to

treat' principle, so including in each group all clinically evident rebleeding episodes and deaths after ran-

domization, irrespective of compliance and regular attendance. The data of 11 patients assigned to propranolol, 10 of whom required emergency sclerothe-

rapy and 1 of whom had surgery, were analysed with


the. data of the propranolol group. The data of one

patient assigned to the sclerotherapy group who was submitted to surgery was analysed with the data of

the sclerotherapy group. Two patients in the propra-

nolol group stopped taking the drug because of side-

effects and another two were lost to follow-up. They were censored at the time of stopping the drug or

when lost to follow-up. No patients in the sclerothe-

rapy group withdrew or were lost to follow-up. The percentages of patients free of rebleeding and

of patients surviving during the period after inclusion were calculated according to the Kaplan-Meier

method and were compared by means of a log-rank

test [14]. Results of continuous variables were ana-

lysed using Student's t test for paired or unpaired data, as indicated. The ~ test was used when analy-

sing discrete variables. Differences in P values less

than 0.05 were considered significant.

Results

The 65 patients were followed for between 17 and 57 months. Mean duration of follow-up (until pa-

tients were last seen, lost, withdrawn or dead) was

similar in both groups (sclerotherapy group 30.7 +

17.5; propranolol group 28.2 + 14.9 months). Of the 31 patients treated by sclerotherapy, vari-

ceal obliteration was achieved in 27 (87%) with a

mean of 4.5 + 1.5 courses in 3.7 + 2.8 months. The

remaining 4 patients died prior to this_ Of the 27 patients with obliterated varices, 3 died without further

recurrence of varices. In 8 of the 24 (33%) patients

alive no further recurrence of varices was observed

over a median follow-up period of 35.9 + 15.5 months (range, 12-55 months). In 16 (66%), new

varices were observed, in 9 after a short interval (3-6

months) and in 7 after a long interval (12-30

months). A median of 1.8 + 1.6 further sclerotherapy courses were required for reobliteration, although

in 5 of the 16 patients further varices developed again after a median interval of 11.4 months (ranged from 6

to 24 months). A 25% decrease from base-line pulse rate was ob-

tained in 32 out of 34 patients in the propranolol

group (after 1.3 + 0.2 months) and was maintained

throughout the trial period with an average dosage of 130 + 65 mg daily (range 40- 360 mg daily). Mean

systolic blood pressure fell significantly from 124.0 +

12.9 mmHg to 105_3 + 13.2 mmHg. Failure to abstain from alcohol was observed in 11

of the 26 (42%) patients in the sclerotherapy group and in 9 of 26 (35%) patients in the propranolol

group. Alcoholic consumption was not significantly

different in patients who rebled or died in both

groups.

Rebleeding Rebleeding from esophageal varices was observed

in 9 of 31 (29%) patients in the sclerotherapy group

and in 21 of 34 (62%) in the propranolol group (Table

2). The percentages of patients free of rebleeding

from esophageal varices at 1, 2, 3 and 4 years after in-

clusion were 77%, 73%, 67% and 67%, respectively, in the sclerotherapy group and 62%, 42%, 38% and

25%, respectively, in the propranolol group (Fig. 1). The overall differences between the two groups were

significant (~ = 5.746, P < 0_02), as well the differences at 2, 3 and 4 years.

In the sclerotherapy group 7 patients had rebleed-

ing from esophageal varices within the first 6 months,

all before complete variceal obliteration. The two other patients had variceal rebleeding after 10 and

30 months following initial obliteration. In one, new

varices were not detected because he failed to attend

TABLE 2

FREQUENCY OF THE FIRST REBLEEDING EPISODE, ACCORDING TO CAUSES (NO. OF PATIENTS)

Sclerotherapy Propranolol (n = 31) (n = 34)

Esophageal varices 9 21 a'b

Gastric varices 2 ~ 0 Gastric erosions 5 2 Esophageal ulcer 1 0 Unknown 0 2

Total 17 25

One patient in each group underwent surgery because of un-- controlled bleeding.

b Ten patients changed to sclerotherapy.


96

100 -

80.

6 0 -

40

20

L,

t I

N

k

i

I

4.

t

t - - - 3

h I

q . . . . - I~ L I - - - ~ I

1 Dead

* Lost to fo l low-up

"~ Withdrawn

• Surgery

' ' ' 1 +, ,,

. . . . 4 . 1 - - - - - 4 . . . . . . . - r l - . . . . . + l - . n

I i

I I I I

i

k . . . . . . . . i

ES J I I I

p<O,02

P P- . . . . " 4

years

P - 34 20 11 6 2

ES - 31 21 16 11 4

Fig. 1. Percentage of patients free of rebleeding from esophageal varices vs. time after randomization. ES = sclerotherapy (solid line); P = propranolol (hatched line). The short vertical bars represent trial times for patients free of rebleeding when last seen, lost to follow-up, withdrawn, submitted to surgery or dead. The numbers below the abscissa denote the number of patients at risk during the ob-

servation period in each group. The solid and hatched vertical bars represent the 95% confidence level (= 2 S_E.).

for follow-up endoscopies. In contrast, rebleeding in

the propranolol group was observed throughout the

42-month follow-up period. Two non-compliant pa-

tients were alcoholic and had early variceal rebleed-

ing complicated by death.

The 9 patients in the sclerotherapy group who had

variceal bleeding had a total of 15 episodes with a

mean of 0.06 rebleeding episodes per patient per

month of follow-up, and the 21 patients in the propra-

nolol group had a total of 39 episodes with a mean of

0.15 rebleeding episodes per patient per month of

follow-up. There was a significant difference be-

tween both the number of episodes of bleeding (P <

0.03) and the mean values per month (P < 0.05).

Mean blood transfusion requirements for each epi-

sode of rebleeding were similar for the two groups

(3.5 + 2.4 and 3.1 + 3.5 units, sclerotherapy and pro-

pranolol, respectively; Table 3).

TABLE 3

ANALYSIS OF THE REBLEEDING EPISODES


Mean follow-up (months) 30.7 + 17.5 28.2 __. 14.9

Rebleeding rate esophageal varices 33%* 75%* all sources 63% ~ 84% ~

No. of episodes of rebleeding esophageal varices 15"* 39** all sources 46 ~ 76 ~

Mean episodes/patient/month esophageal varices 0.06*** 0.15"** all sources 0.17 ~ 0.188

*P < 0.02.

**P < 0.03.

***P < 0.05

§ r l . s .


Because of severe or repeated bleeding, 10 pa-

tients in the propranolol group received urgent and

elective sclerotherapy. In 8 patients variceal obliteration was achieved: 2 died after 12 and 19 months due

to rebleeding from 'new' varices and hepatic failure; in the 6 patients surviving the mean follow-up was 27

months from the time of initial sclerotherapy to the fi-

nal analysis of the trial_ In 2 patients obliteration was

not achieved: one of these died from variceal bleeding after three sessions of sclerotherapy and one was

submitted to surgery because of uncontrolled bleed-

ing. Two patients in the sclerotherapy group rebled

from gastric varices but none in the propranolol

group. In one of them, a severe rebleeding occurred

after complete obliteration of esophageal varices and

the patient was submitted to surgery. The percent-

ages of patients free of both esophageal and gastric variceal rebleeding in the sclerotherapy group at 1, 2,

3 and 4 years after inclusion were 74%, 66%, 61%

and 61%, respectively (Fig. 2). The overall differences between the two groups were again significant

(X 2 = 3.863, P < 0.05). In the two figures, the curves of the sclerotherapy groups show a marked change at

6 months after entry, corresponding to the time when

complete esophageal variceal obliteration was achieved.

Gastric erosions were the origin of rebleeding in 5

patients (with seven episodes) with obliterated

esophageal varices in the sclerotherapy group and in

two fl-blockaded patients (with four episodes) in the propranolol group (Table 2). Half of the patients in

each group had heavy alcoholic consumption. In one

patient an esophageal ulcer causing three severe epi-

sodes of bleeding was observed 2 days after the initial sclerotherapy course and was complicated by an epi-

gastric peritonitis (see side-effects).

Analysing all sources of rebleeding (Table 2), 17 of

%

1 0 0 -

8 0 -

6 0 -

4 0

2 0 ¸

-f

i1 t l L - - - ~

L - - -

-f JJ JI

I .% L . . - 1

% H . . . . -I-i

L 4 - - - 1

1" Dead

Lost to f o l l ow-up

A W i t hd rawn

I I I ] t II

L _ ~

. . . . . ~x_ _ _ _ I - . . . . . . . . ! . . . . . . . I - I - - 1

i i ' ,, '

i .L

I I JI I I

P < 0 , 0 5 ',

i i i i , P

~. . . . . . . . ' t 4 . . . . . --t i I i I I

J .

ES l i

I I I I 1 2 3 4

y e a r s

P - 3 4 2 0 11 6 2

ES - 31 2 1 16 11 4

Fig. 2. Percentage of patients free of rebleeding from esophageal and gastric varices vs. time from randomization. ES = sclerotherapy (solid line); P = propranolol (hatched line). The short vertical bars represent trial times for patients free of rebleeding when last seen,

lost to follow-up, withdrawn or dead. The solid and hatched vertical bars represent the 95% confidence level (= 2 S.E.).


31 (54.8%) patients in the sclerotherapy group rebled, compared with 25 of 34 (73.6%) patients in the propranolol group. The percentages of patients free of rebleeding from any source at 1, 2, 3 and 4 years after inclusion were 60%, 43%, 37% and 37%, respectively, in the sclerotherapy group and 53%, 33%, 16% and 16%, respectively, in the propranolol group (Fig. 3). Although the curves show a distinct separation after 6 months, the difference between them was not significant ( ~ = 1.29, P < 0.21). In the sclerotherapy group a total of 46 episodes of rebleeding were observed in 17 patients (0.17 episodes per patient per month), whilst in the propranolol group there were 76 episodes in 25 patients (0.18 episodes per patient per month). This difference was not significant. The total and mean blood transfusion requirements for each episode of rebleeding were similar for the two groups (1284 and 1485 units; 2.9 + 2.5 and 3.0 +_ 3.0 units for sclerotherapy and proprano- Iol, respectively).

Survival

In the sclerotherapy group 9 patients died: 4 died of liver failure (3 without further rebleeding 4.0, 5.5

and 20.0 months after inclusion; and one who had bled from gastric erosions, 36 months after entry); 4 died of complications related to a rebleeding episode 2.0, 2.5, 3.0 and 7.5 months after inclusion; and one died from uncontrolled hemorrhage 10 months after inclusion.

In the propranolol group 11 patients died after inclusion: 2 of neoplasia (hepatocellular carcinoma and metastatic colon cancer) at 33 and 25 months; one of hepatic failure at 45.4 months; 3 from uncontrolled rebleeding at 6, 8 and 10 months; 5 from complications related to rebleeding at 3.6, 4, 7, 12 and 22.3 months (Table 4).

The percentages of patients surviving at 1, 2, 3 and 4 years after inclusion were 77%, 74%, 69% and 69%, respectively, in the sclerotherapy group and 85%, 78%, 65% and 54%, respectively, in the pro-

%

1 0 0 -

8 0

6 0

4 0

2 0 -

L 1

L 1

L._.< 1. !l

.... t I " JL

i L

i

-]- D e a d

"A- L o s t t o f o l l o w - u p

, , ']

P 34

ES 31

. . . . I- . . . . i- l i i . . . . . . . . .

I I

i i i L . . . . . . .

i i

ES I I I I

NS

P I - . . . . -'1

Fig. 3. Percentage of patients free of rebleeding from any source vs. time after randomization. ES = sclerotherapy (solid line); P = propranolol (hatched line). No significant difference was found.

I I 2 3 4

years

17 9 4 2

17 10 6 5

182 P.T. A L E X A N D R I N O et al.

TABLE 4

MORTALITY TABLE, WITH CAUSES OF D E A T H IN B O T H G R O U P S


Total deaths 9 11 Group A (23/24) 6 6 Group B (8/10) 3 5

Causes of death bleeding 1 3" complications related to

bleeding 4 5 ~ hepatic failure 4 1 ~ others 0 2

" In each group, 1 patient after changing to sclerotherapy.

pranolol group (Fig. 4). These differences were not significant at any time. When differences in survival

were analysed according to Child's classification,

there was no statistically significant difference.

Side-effects In the propranolol group 2 patients stopped taking

the drug, because of bronchospasm and hepatic en-

cephalopathy in one, and left ventricular failure in the other. There was no rebleeding upon cessation or

subsequently. Both died, one from hepatic failure and the other from bladder cancer.

Side-effects occurred early in the course of treat-

ment: 3 patients had transient hepatic encepha- lopathy, 2 developed ascites, and 10 had dizziness

and fatigue. Two male patients experienced reduced sexual activity. Transient bronchospasm was ob-

served in 7 patients. No notable deterioration was

observed in routine liver function tests and serum electrolytes. Urea and creatinine did not change sig-

nificantly from base-line values at 1 or 2 years after inclusion.

In the sclerotherapy group ulceration was the most

frequent complication, occurring in 8 patients (5

%

lOO - ~ - - , ~, - - ~ , ~ * t

. . . . . . . H I - q - - I" ~" I

8 0 - I III I

60 l

4 0 -

Lost to f o l l o w - u p 20-

z~ Wi thdrawn

t F o l l o w - u p a f te r su rge ry

L I - - - I ~ i

I - - . - , } - - , , ~ I I I I t l I I I I I I • " , • i

L. . . . . . F - -~ - . . . . H - f - - - h . . . . NS

:" p i L - - - J - q - I ' - . . . . I-"~

F o l l o w - u p a f te r s c l e r o t h e r a p y

I I I I

1 2 3 4 years

P - 3 4 2 8 2 0 12 4

ES 31 2 4 2 0 14 4

Fig. 4. Cumulative survival curves for patients randomly assigned to sclerotherapy (ES) or propranolol (P). The short vertical bars represent trial times for patients when last seen (after surgery or after sclerotherapy in propranolol patients), lost to follow-up, or with-

drawn. No significant difference was found.


were asymptomatic, in 2 ulceration was confirmed endoscopically to be the site of rebleeding, following rebleeding from esophageal varices, and in 1 the bleeding ulcer was complicated by localised peritoni-

tis; this resolved with conservative management). Transient fever (/>38 °C) developed in 7 patients; cultures of blood and urine were sterile and the chest X-ray was normal. Seven patients had transient chest pain; 5 had pleural effusions, which were transient and asymptomatic in 4 but required medical therapy in one. Three patients had transient dysphagia due to narrowing of the esophageal lumen, diagnosed by re- sistance to passage of the endoscope. In two patients a prolonged hemorrhage after needle injection ne- cessitated blood transfusion. One patient developed aspiration pneumonia after emergency sclerotherapy. There were no fatal side-effects in any of the

groups.

Discussion

The present trial shows that long-term oral propranolol is less effective in reducing the frequency of variceal rebleeding as compared to repeated sclerotherapy. However, the frequency of rebleeding, whatever the sourcc, is the same with either propranolol or sclerotherapy. There was no difference in

survival over a 4-year period. Similar results have been observed in a trial recent-

ly published by Fleig et al. [15]. In this trial the re-

bleeding rate in both groups was much lower, around 30%, but the follow-up time was approximately only

half as long as in our study. The majority of our propranolol-treated patients

(60%) had variceal rebleeding distributed throughout the period of follow-up. This is similar to the re- suits in propranolol-treated patients of three recent randomized controlled trials reported by Burroughs et al. [6] in London, Villeneuve [7] in Montreal or Queuniet [8] in Rouen. Our results differ from those reported by Lebrec et al. [5] in Paris, despite the patients' characteristics being very similar: 80% were alcoholic cirr~,otics in a good condition; 72% of patients were class A and 28% class B, using the Child-

Campbell score [12]; the interval between the initial episode of bleeding and randomization and the mean dose of propranolol used were similar. Our selection was due to the inclusion criteria, and in addition ex-

cluded almost 30% of the total population already submitted to other therapies. When the study was designed, propranolol was supposed to be used only after the bleeding episode was controlled and the patient was hemodynamically stable. This excluded patients who died in the first 2 weeks of the index episode, approximately 30%. In the remaining Child's C patients, the potential danger offl-receptor blockade producing hypotension [16] was another reason for exclusion. Patients who had known contraindications to fl-receptor blockade were similarly excluded

(13%). The discrepancy between the results of the three

studies comparing propranolol with placebo [6-8] and the study of Lebrec et al. [5] has been related to the difference in patient selection: Burrough's and Villeneuve's patients had more advanced cirrhosis and the majority had nonalcoholic cirrhosis [6,7]. This does not apply to our study. The major difference between our trial and the French trial [5] is the inclusion of patients who had bled from nonvariceal lesions. However, exclusion of this subgroup still shows effective prevention of variceal rebleeding in patients randomized by Lebrec et al. [5]. Their patients who bled but did not die were withdrawn from the study and this may explain the differences observed [17].

In both treatment groups we analysed rebleeding (both esophageal variceal rebleeding or all clinically important rebleeding) in terms of the first occurrence of rebleeding subsequent to randomization, irrespective of compliance, as noncompliance is an integral part of therapy [17]. Clearly, with a predominantly alcoholic population it is likely that strict compliance was not observed all of the time. Whatever the reason, all the 39 rebleeding episodes in the propranolol group occurred before changing to sclerotherapy.

Our results showed that sclerotherapy was su- perior to propranolol for the prevention of both esophageal and gastric variceal rebleeding. The number of patients who rebled and the number of ep-


isodes were significantly less. The benefit of sclerotherapy in reducing the incidence of variceal rebleeding has also been reported in a number of controlled

trials comparing long-term sclerotherapy with place-

bo and/or medical therapy [9-11]. However, we did

not observe any difference between sclerotherapy and propranolol when the total rebleeding episodes,

from varices and other causes, were analysed. Nei-

ther the number of patients with rebleeding nor the number of episodes was statistically different. A dis-

tinct separation of the curves of patients free of rebleeding is observed after the first year of follow-up.

The overall rebleeding rate during sclerotherapy is

63%, and 84% whilst taking propranolol_ The causes

of rebleeding were different: 84% of the propranolol

patients who rebled did so from varices and 50% of

the rebleeding sclerotherapy patients rebled from varices, but patients with obliterated varices rebleed-

ing from gastric erosions represent 29% of all re-

bleeders. This was unrelated to the continued consumption of alcohol.

It is a controversial issue as to whether bleeding

gastric erosions are related to portal hypertension.

The majority of controlled trials comparing sclerotherapy with other therapies did not consider it as a

• cause of rebleeding. Shunt surgery was proposed as a

therapy of bleeding gastritis, but on the other hand surgical trials report an incidence of rebleeding from

nonvariceal causes ranging from 9% [18], 13% [3],

15% [2] and 19% [19], despite patency of the shunts in the checked patients.

In a recent paper comparing sclerotherapy with distal splenorenal shunt [20], the authors reported an

overall rebleeding rate with sclerotherapy of 53% at

4 years of follow-up, with 26% of the total number of

patients rebleeding due to 'portal hypertension gas-

tritis'. Our patients who rebled from gastric erosions did well with medical treatment and all but one are

alive at the time of the trial analysis.

The cumulative survival of both groups of patients

was calculated according to the group into which they

were randomized, even if the patients who rebled had been offered other treatments. This was the case

for 10 patients in the propranolol group submitted to

chronic sclerotherapy after rebleeding and one pa-

tient in each group submitted to surgery. There is no difference in cumulative survival rate at any time or

between the Child's classes between the two thera-

pies.

The results are also no different if the two patients withdrawn from propranolol are included, or if the

analysis is performed according to treatment received rather than an intention to treat. It must be re-

membered, however, that this trial deals with a sub-

group of patiens who survived for 2 weeks after the

index bleeding episode and with a good general condition, which increases survival [1]. In general our

patients are not dissimilar to those reported in a num-

ber of controlled trials of other forms of elective ther-

apy of variceal bleeding. The characteristics of our

population explain the 2-year survival rate of over 70% observed in both the sclerotherapy and propra-

nolol groups. These figures are very similar to the 2-

year survival rate in sclerotherapy groups reported

by Westaby et al. [10] and the Copenhagen group,

among patients who survive the initial period [21], all of them reporting that sclerotherapy markedly im-

proved survival. Although propranolol patients rebled more from esophageal varices than sclerothe-

rapy patients, if they were then treated with an alter-

native therapy, they had the same survival. As there

was no standard medically treated (placebo) group, it

cannot be ascertained whether both therapies are similar with respect to survival, because both are su-

perior to standard medical treatment, or whether the

reduction in variceal rebleeding did not affect surviv-

al in this particular subgroup of patients_ The other

clinical trial comparing propranolol with paravari-

ceal sclerotherapy [15] did not show any difference in variceal rebleeding and survival during 2 years of follow-up.

A final word about the number of patients- It

would be very difficult to design a trial with a suffi-

cient number of patients to avoid both type I and type

II errors: to demonstrate a reduction of 20% in all

causes of rebleeding we would need 212 patients, in retrospective analysis we admit a type II error in rela-

tion to that difference, but there is power enough in

the significant difference between bleeding from esophageal and/or gastric varices.


Since 2 9 % of the p a t i e n t s u n d e r p r o p r a n o l o l re-

q u i r e d e m e r g e n c y s c l e r o t h e r a p y for acu t e r e b l e e d -

ing, we r e c o m m e n d the use of s c l e r o t h e r a p y as long-

t e r m e lec t ive t h e r a p y in p a t i e n t s w h o h a v e b led f r o m

e s o p h a g e a l var ices . Th i s s e e m s to be the bes t m e t h o d

of p r e v e n t i n g va r i cea l r e b l e e d i n g .

Acknowledgement

W e are g ra t e fu l to Dr . A . B u r r o u g h s for r e a d i n g

and c o r r e c t i n g the E n g l i s h of this p a p e r .

References

1 Graham DY, Smith JL. The course of patients after variceal hemorrhage. Gastroenterology 1981~ 80: 800-809.

2 Reynolds TB, Donovan A J, Mikkelsen WP, et al. Results of a 12 year randomized trial of portocaval shunl in patients with alcoholic cirrhosis and bleeding varices. Gastroenter- ology 1981; 80:1005-1011.

3 Resnick RH, Iber FL, Ishihara AM, et al. A controlled study of the therapeutic portocaval shunt. Gastroenter- ology 1974; 67: 843-857.

4 Rueff, B, Prandi D, Degos F, et al. A controlled study of therapeutic portocaval shunt in alcoholic cirrhosis. Lancet 1976; i: 655-659.

5 Lebrec D, Poynard .T, Bernuau J, et al. A randomized controlled study of propranolol for prevention of recurrent gastrointestinal bleeding in patients with cirrhosis: a final report. Hepatology 1984; 4: 355-358.

6 Burroughs AK, Jenkins W J, Sherlock S, et al. Controlled trial of propranolol for the prevention of recurrent variceal hemorrhage in patients with cirrhosis. N Engl J Med 1983: 309: 1539-1542.

7 Villeneuve JP, Layrargues GP, Rivard CI, et al. Proprano- 1ol for the prevention of recurrent variceal hemorrhage; a controlled trial. Hepatology 1986; 6: 1239-1243.

8 Queuniet AM, Czernichow P, Lereboughrs E, Ducrotte P, Tranvouez JL, Colin R. Etude controllde du propranolol dans la prdvention des r~cidives h6morragiques chez les patients cirrhotiques. Gastroentero[ Clin Biol 1987; 11: 41-47.

9 Terblanche J, Bornman PC, Kahn D, et al. Failure of repeated injection sclerotherapy to improve tong-term survival after oesophageal variceal bleeding: a five year prospective controlled trial. Lancet 1983; ii: 1328-1332.

10 Westaby D, MacDouga[l BRD, Williams R. Improved survival following injection sclerotherapy for oesophageal vat- ices: final analysis of a controlled trial. Hepato[ogy 1985; 5: 827-830.

11 Korula J, Balart LA, Radvan G, et al. A prospective randomized controlled trial of chronic oesophageal variceal sclerotherapy. Hepatology 1985; 5: 584-589.

12 Campbell DP, Parker DE, Anagustapoulos CE. Survival prediction in porto-caval shunts: a computorised statistical analysis. Am J Surg 1973; 126: 748-751.

13 Grima N, Alexandrino P, Diogo AN, Alves MM, Cotter J, Pinto Correia J. Hemodynamic effects of endoscopic sclerotherapy (ES) in patients with portal hypertension [Ab- stract]. Dig Dis Sci 1986; 31 (suppl): 196S.

14 Peto R, Pike MC, Armitage P, et al. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. II. Analysis and examples. Br J Cancer 1977; 35: 1-39.

15 Fleig WE, Stange EF, Hunecke R, et al. Prevention of recurrent bleeding in cirrhotics with recent variceal hemorrhage: prospective, randomized comparison of propranolol and sclerotherapy. Hepatology 1987; 7: 355-361.

16 Wilkinson SP. Propranolol and portal hypertension in cirrhosis [Letter]. Lancet 1980; ii: 429.

17 Corm H. Propranolol in portal hypertension: problems in paradise? Hepatology 1984; 4: 560-564.

18 Harley HA J, Morgan T, Redeker AG, et al. Results of a randomized trial of end-to-side portocaval shunt and distal splenorenal shunt in alcoholic liver disease and variceal bleeding. Gastroenterology 1986; 91: 802-809.

19 Conn HQ, Resnick RH, Grace ND, et al. Distal splenorenal shunts vs portal-systemic shunt: current status of a controlled trial. Hepatolo~v 1981: l- |'~1-160.

20 Warren WD, Henderson JM, Millikan W J, et al. Distal splenorenal shunt versus endoscopic sclerotherapy [or long-term management of variceal bleeding. Ann Surg 1986; 203: 454-462.

21 The Copenhagen Oesophageal Varices Sclerotherapy Pro- ject. Sclerotherapy after first variceal hemorrhage in cirrhosis. A randomized multicenter trial. N Engl J Meal 1984; 311: 1594-1600.

Documents

Propranolol or endoscopic sclerotherapy in the prevention of recurrence of variceal bleeding