GASTROENTEROLOGY
1991;101:1087-1093
Prophylactic Sclerotherapy in High-Risk Cirrhotics Selected by Endoscopic Criteria A Multicenter ROBERTO PAOLO
Randomized
DE FRANCHIS,
G. ARCIDIACONO,
PASQUALE
VITAGLIANO,
RAFFAELE
ARCIDIACONO,
ALESSANDRO GIANCARLO GIORGIO
ZAMBELLI, CALETTI,
BATTAGLIA,
The North Italian Endoscopic
MASSIMO PAOLO MARIA FELICE
SERGIO
PRIMIGNANI,
C. VAZZOLER, ROSSI,
COSENTINO,
BRUNATI,
and GIORGIO
GERUNDA
Club for the Study and Treatment
lthough the first variceal bleeding in patients with liver cirrhosis has an extremely high mortality rate (l--4), prophylaxis of variceal bleeding is still a
Trial
M. RIZZI,
ALFRED0
Controlled trials of sclerotherapy for the prevention of the first variceal hemorrhage in cirrhotics have given conflicting results. In the present study, 106 cirrhotics were randomized to sclerotherapy (55 patients) or control group (51 patients). Admission criteria were no history of previous variceal hleeding and the presence of high-risk varices, i.e., a variceal score 10 according to Beppu et al. Sclerotherapy sessions were performed at time zero, 7 days, 30 days, and then monthly until eradication. Follow-up endoscopies were performed at 6-month intervals thereafter. Control patients underwent repeat endoscopy at 6-month intervals. Bleeding episodes were treated by sclerotherapy in both groups, whenever possible. Mean follow-up was 24 months. Analysis of the results was performed by the intention-to-treat method. Variceal bleeding occurred in 19 sclerotherapy patients (34.5%) and in 17 controls (35.4%, P = NS). Overall mortality was 34.5% in sclerotherapy patients and 50% in controls (P = NS). Seven of the 19 sclerotherapy patients (36.8%) and 11 of the 17 controls (64.7%) who bled died of hemorrhage (P < 0.05, log-linear model). It is concluded that prophylactic sclerotherapy does not reduce the incidence of first variceal bleeding in cirrhotics. However, there seems to be a trend toward a lower bleeding-related mortality in sclerotherapy patients than in controls.
A
Controlled
of Esophageal
Varices
matter of controversy. Prophylactic shunt operations are no longer performed because of the negative results of four randomized controlled trials performed in the 1960s (24). Three controlled trials of propranolo1 (5-7) and two of nadolol(8,9) have shown that p blockers decrease the incidence of first variceal hemorrhage but do not seem to improve survival. Prophylactic endoscopic sclerotherapy (EVS) has been investigated in several randomized controlled trials and has been shown to be extremely effective in some studies (l&14), of no benefit in others (15--19), and even harmful in others (20,21). These conflicting results might be related to different selection of patients in the various studies or to different treatment of variceal bleeding in patients assigned to EVS as compared with controls. It has been pointed out that the higher the bleeding rate in the control groups, the more effective is sclerotherapy in preventing first bleeding (22). With the above considerations in mind, we designed the present randomized controlled trial to test the efficacy of EVS in preventing the first variceal bleeding and in reducing mortality in cirrhotics with high-risk varices identified by endoscopic criteria. Because emergency EVS has been shown to be lifesaving in variceal bleeding (23,24), we decided that EVS should be used whenever possible to treat acute
Abbreviations used in this study: EVS, endoscopic variceal sclerotherapy; NIEC, North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices. o 1991 by the American Gastroenterological Association 0016~5065/91/$3.00
1088
GASTROENTEROLOGY
DE FRANCHIS ET AL.
bleeding both in patients assigned lactic EVS and in controls.
to receive
prophy-
Patients and Methods Criteria for Admission From April 1985 to February 1987, all patients seen at the participating institutions with a diagnosis of cirrhosis of the liver based on history, physical examination, liver chemistry, and/or liver histology were eligible for the study if they met the following criteria. (a) They had no prior history of upper gastrointestinal bleeding, including bleeding from nonvariceal sources; (b) they had high-risk varices by endoscopic criteria (see below); (c) they had no other disease (e.g., cancer) reducing life expectancy to < 1 year; (d) they had no gastrointestinal ulcers at the time of randomization; (e) they had received no treatment for esophageal varices; and (f) they gave informed consent to participate in the study.
Definition of High-Risk Varices Varices were classified according to the suggestions of the Japanese Research Society for Portal Hypertension (25,261. They include recording of several features of the varices observed at endoscopy, such as color (blue or white); size (less than one third, up to two thirds, and more than two thirds of the esophageal radius); the longitudinal extension of the varices along the esophagus; and the presence and grade of several red signs on the variceal wall (red wale markings, cherry-red spots, hematocystic spots). Interobserver variation between the endoscopists participating in this study was decreased by intensive training, which has been described elsewhere (27). The risk of bleeding was estimated according to the discriminant equation of Beppu et al. (26), which was developed by using the above mentioned classification system. A score of < 0 according to Beppu’s calculations was chosen as the cutpoint to define high-risk varices. This corresponds to the presence of blue varices with red wale markings, and/or cherry-red spots, and/or hematocystic spots. According to Beppu’s retrospective calculations, this corresponds to a bleeding risk of 264%. However, prospective validation of Beppu’s score performed by our group has shown that the actual 2-year risk of bleeding of such patients is about 40% (27).
Randomization At the beginning of the study, each center was given a computer-generated randomization list, which was kept by physicians not directly involved in the study. Randomization took place on the day of diagnostic endoscopy. Baseline Clinical Assessment At entry, a full medical history, physical examination, hepatitis B virus markers, routine liver chemistries, and prothrombin activity were obtained for each patient. The degree of hepatic decompensation of the patients was
assessed by the Child’s classification Pugh’s criteria (28). Methods of Sclerotherapy Schedule
Vol. 101, No. 4
modified according
to
and Follow-up
Treatment was performed with Olympus GIF-D3 or GIF-Q scopes (Olympus Optical Co., Tokyo, Japan) under light sedation (diazepam, 5 mg IV). Polidocanol 1% (Atossisclerol, Kreussler Pharma, Wiesbaden, Germany) was used as the sclerosant in most instances. Etanolamine oleate 5% was used in four patients. Injections were intravariceal and paravariceal and varied in volume between 1 and 5 mL. The mean volume of sclerosant injected for each sclerotherapy session was 21.5 mL for first sessions (range, 15-60 mL) and decreased progressively for subsequent sessions in individual patients, because less sclerosant was injected as the varices became smaller as a consequence of sclerotherapy. With minor variations between centers, the treatment schedule was time zero, 7 days, 30 days, and then monthly until eradication, and then the patients underwent endoscopy every 6 months to detect recurrence of varices. Eradication was defined as disappearance of all varices from the distal 3 cm of the esophagus or reduction of varices to small white columns with no red signs. In addition to their scheduled visits to the liver clinic, control patients had follow-up phone interviews every 3 months and endoscopic control examinations every 6 months. At the beginning of the study all alcoholic patients were advised to stop drinking. At each visit, an attempt to assess compliance was made by questioning the patients and/or their relatives. Follow-up was closed on March 31,1989. Bleeding Whenever, during follow-up, a patient was admitted to one of the participating institutions for upper gastrointestinal hemorrhage, emergency endoscopy was performed to establish the source of bleeding. If a diagnosis of variceal bleeding was made, emergency sclerotherapy was performed as specified above, regardless of the treatment group to which the patient belonged. In any other instance, the source of bleeding was defined as “unknown,” and the patients were treated conservatively, unless an actively bleeding lesion other than a varix was observed. If a patient died before reaching the hospital or was admitted to a hospital not participating in the study, every effort was made to obtain information about the cause of death and the details of the bleeding episode. Sample Size Calculation and Statistical Analysis According to our previous observation, the expected 2-year incidence of first episodes of variceal hemorrhage was about 40% (27). Assuming a reduction of this risk to 10% by sclerotherapy, approximately 50 patients for each group would be required for a type I error of 5% and a type II error of 20%.
October
1991
The entry point of the study was the date of randomization, always performed on the day of variceal assessment. The variables of interest were compared by either x2 test or Student’s t test for independent samples. End points of the study were the first episode of bleeding, death of all causes, and bleeding-related death. Patients who survived a first episode of variceal hemorrhage, however, underwent further regular follow-up. The bleeding rates and the death rates were estimated by life-table analysis (Kaplan-Meier method) and compared by the log rank test (29). Bleeding-related death rates in relation to treatment group and to the modality of emergency treatment of bleeding were evaluated by a log-linear model (30). Adjustment for imbalance in prognostic variables was performed by Cox’s regression analysis in which the prognostic variables were included together with the treatment variable (31). The analysis was performed according to an intention-totreat strategy. Results When recruitment was completed in February had been considered for the study 1987, 126 patients by the eight participating hospitals. Five centers contributed equally to the treatment groups, 3-26 patients to each group. In the remaining three centers, slightly more patients were. assigned to EVS than to the control group (1 excess EVS patient in two centers, and 2 excess EVS patients in the third center). At the end of recruitment, 65 patients had been assigned to EVS, 61 to the control group. We subsequently found out that 20 patients (15.9%) did not fulfill the criteria for admission and, thus, excluded them from the study. The reasons for exclusion were Beppu’s variceal score > 0 (7 patients, 1 EVS, and 6 controls), upper gastrointestinal bleeding before randomization (3 patients, 2 EVS, and 1 control), prior treatment of portal hypertension (p blockers, 3 patients and all controls), and lack of consent to EVS after randomization (7 patients). Thus, a total of 106 patients remained in the study; 55 of them belonged to the EVS group, 51 to the control group. Table 1 shows the main demographic and clinical features of the patients. The groups were relatively well balanced in age, sex distribution, and Child-Pugh classification. However, there was a significant excess of alcoholic patients in the control group (P = 0.035). Conversely, Beppu’s variceal score tended to be worse in EVS patients (P = 0.034). At the beginning of the study, EVS was performed on hospitalized patients; the patients were admitted the day of the EVS session and discharged the day after. As a result, in 15 patients EVS could not be started within 2 days from randomization, as for the others, because in some instances there was a waiting list for admissions for elective procedures. In these 15 patients, the delay between the date of randomization
PROPHYLACTIC
SCLEROTHERAPY
IN CIRRHOTICS
1089
Table I. Demographic, Clinical, and Endoscopic Features of Patients at Entrv, According to Study Group Study
group
Sclerotherapy
Patients Age, mean (yr) Sex (M:F) Alcoholics (%) Child’s class
P
Controls
55
51
56.2
55.1
NS
31:24
38:13
NS
15 (27.3)
25 (49.0)
0.035
A (%I
14 (25.5)
18 (35.3)
NS
B (%)
31 (56.3)
22 (43.1)
NS
c (%I
10 (18.2)
11 (21.6)
NS
19 (34.5)
29 (56.8)
0.034
22 (40.0)
16 (31.4)
NS
14 (25.5)
6 (12.8)
NS
Beppu’s score 0 to -0.38 (%) -0.38
to -1.14
< -1.14
(%)
(%)
and that of the first EVS session lasted between 14 and 90 days (mean, 26.5 days). However, according to our intention-to-treat strategy, the time between randomization and start of EVS was included in the analysis. Unfortunately, during this time, 3 patients assigned to EVS died before EVS could be started. One of them died of variceal hemorrhage, and the remaining 2 died of liver failure. Of the 52 patients who actually received sclerotherapy, 37 (71%) achieved eradication of their varices after a mean of four sclerotherapy sessions, corresponding to about 2 months. Three controls were lost to follow-up and were thus excluded from analysis. Bleeding
The mean duration of follow-up was 24 months (range, l-48 months). During this time, 19 controls and 20 EVS patients bled (Table 2). Two controls and 1 EVS patient bled from gastric erosions. Thus, a total of 17 controls bled from varices. This figure corresponds to a 2-year rate of bleeding of 39% (KaplanMeier estimate) and is identical to that expected on the basis of our prospective validation of Beppu’s
Table 2. Bleeding and Mortality Sclerotherapy n = 55 Sources of bleeding Varices (%) Esophageal ulcers (%) Esophageal hematoma (%) Gastric erosions (%) Total (%) Causes of death Hemorrhage (%) Liver failure (%] Hepatocellular carcinoma (%) Extrahepatic causes (%) Total I%) Note. Differences
between
groups
14 (25.4)
Controls n = 48
17 (35.4)
4 (7.3) l(1.8) l(l.8)
-2 (4.2)
20 (36.3)
19 (39.6)
7 (12.7)
11 (22.9)
5 (9.1)
8 (16.7)
2 (3.6)
l(2.1)
5 (9.1)
4 (8.3)
19 (34.5)
24 (50.0)
are not statistically
significant
1090
GASTROENTEROLOGY
DE FRANCHIS ET AL.
variceal score. In the sclerotherapy group, 19 patients bled either from varices (14 patients) or because of complications of sclerotherapy (5 patients, Table 2). Figure 1 shows the cumulative proportion of patients with no bleeding in the two groups (Kaplan-Meier plot). During the early phase of the study, there was a slightly higher incidence of bleeding in the EVS group than in the control group. In this period, 4 EVS patients bled from esophageal ulcers on varices, and 1 bled from a sclerotherapy-induced esophageal hematoma. The difference between the two curves began to decrease after the sixth month and had disappeared completely by 18 months. (Odds ratio for bleeding at 2 years, 0.96; 95% confidence intervals, 0.43-2.17; P = NS.)
01
0 patients
Overall Mortality Twenty-four controls died compared with 19 EVS patients (Table 2). Deaths in the EVS group included also the 3 patients who died before EVS was begun (1 died of variceal hemorrhage and 2 of liver failure). The difference in the cumulative proportion of patients surviving in the two groups (Kaplan-Meier plot: Figure 2) was not statistically significant. (Odds ratio for intention to treat, 0.53; 95% confidence intervals, 0.24-1.16; P = NS. Analysis according to the “treatment received” strategy showed similar results.) Adjustment for Imbalance Adjustment for imbalance in prognostic variables was performed by Cox’s regression analysis in which the prognostic variables were included together with the treatment variable. In addition to
months
Figure 1. Cumulative proportion of patients with no bleeding (Kaplan-Meier plot). a-0, Sclerotherapy; O-O, controls: [O], number of patients at risk control group; (O), number of patients at risk sclerotherapy group.
at risk
I
1
I
I
IO
20
30
40
30
24
15
6
27
20
17
6
Vol. 101, No. 4
I
50 months
sclerotherapy controls
Figure 2. Cumulative proportion of patients surviving (KaplanMeier plot; O-O, controls: O-O, sclerotherapy) and adjusted estimated survival curves (adjustment for imbalance, Cox model) for EVS (A-A) and control [A-A) patients based on the final Cox’s model for alcoholic etiology, Child’s class B, and a Beppu’s score ranging between -0.38 and -1.14.
treatment, the final Cox’s regression model included the etiology of cirrhosis, Child’s class, and Beppu’s variceal score. In this analysis, survival of EVS patients was better than that of controls, but the difference failed to reach statistical significance (P = 0.0758, Figure 2). Treatment of Acu te Bleeding and Bleeding-Related Deaths Our protocol called for emergency EVS for all patients acutely bleeding from varices. However, only 6 of 19 EVS patients and 9 of 17 control patients actually received emergency EVS. All these patients were treated with intravariceal and paravariceal injections of Polidocanol 1%. The volumes of sclerosant used per single injection varied between 1 and 5 mL; the total volumes injected per session ranged between 15 and 40 mL (median, 23 mL). In the control group, the explanation of the low proportion of patients receiving EVS is that eight patients were admitted to hospitals not participating in the study, where no emergency EVS was available at the time. In the EVS group, eight patients were admitted to hospitals where no emergency EVS was available. In all these patients, conservative treatment [Sengstaken tube with or without vasoactive drugs, i.e., Glypressin (Ferring AB, Malmo, Sweden) or Somatostatin (Serono, Rome, Italy), was used]. The remaining five patients did not receive emergency EVS, because they were bleeding from esophageal ulcers on varices (four patients) and from an esoph-
October
PROPHYLACTIC
199 1
ageal hematoma (one patient). In these cases, conservative treatment (Glypressin) was deemed more appropriate than emergency EVS. Nearly 65% of the patients in the control group who bled from varices died of hemorrhage compared with 36.8% of patients in the EVS group. This difference is significant when analyzed by log-linear model (P < 0.05, Table 3). Regardless of the modality of treatment of bleeding episodes, the mortality rates were lower in EVS patients than in control patients (Table 3). This suggests that bleeding is less severe in EVS patients, although the transfusion requirements were not significantly different in the two groups (EVS: mean, 3.75 U per patient; range, O-10 U; controls: mean, 3.3 U per patient; range, O-8 U). Discussion Although prophylactic sclerotherapy has been studied in many randomized controlled trials (lo21), the role of this technique in preventing the first variceal hemorrhage in patients with cirrhosis of the liver has not been established so far. The main reason for this is the extreme variability in the results of the above trials. For example, the bleeding rate in the control groups of such trials ranged between < 16% (20) and > 70% (13). This clearly indicates that different criteria of patients selection were applied in the various studies. It has been pointed out that the higher the bleeding rate in the control groups, the more effective is sclerotherapy in preventing first bleeding (22). When our study was begun in 1985, we decided to make an attempt to select patients with a high risk of bleeding. At that time, we had only preliminary data from our study on the prediction of the first variceal hemorrhage in cirrhotics (27). Those data showed that the endoscopic score proposed by Beppu et al. (26) was a valuable tool in identifying high-risk patients and that patients with a score of 10 had a a-year chance of bleeding of about 40%. We speculated that this level of risk was sufficiently high to allow the beneficial effect of sclerotherapy in the prevention of first bleeding to emerge, if such an effect existed. Indeed, the control patients in our study had a 2-year rate of bleeding of 39% by Kaplan-Meier estimate, which
Table 3. Bleeding-Related Deaths According to the Modality of Treatment of Acute Variceal Bleeding Sclerotherapy Deaths/total (%)
Treatment Emergency
EVS
Controls Deaths/total
(%)
4/9 (44.4)
O/6 (0.0)
No EVS
7/13 (53.8)
718 (87.5)
Total
7/19 (36.8)”
11/17 (64.7)
“P < 0.05 vs. controls,
log-linear
model
(30).
SCLEROTHERAPY
IN CIRRHOTICS
lo91
closely corresponded to the expected 40%. However, during the early phase of the study, we observed that the rate of first bleeding in EVS patients was higher than that in controls. In fact, during the first 6 months, some patients in the EVS group bled because of complications of the procedure (esophageal ulcers and esophageal hematoma). After the sixth month of the study, when eradication of the varices was achieved in patients in the EVS group, the bleeding rate in this group decreased to almost zero, whereas patients in the control group continued to bleed. As a result, the final proportions of patients bleeding in the two groups were very similar. Therefore, in our hands, prophylactic sclerotherapy did not change the incidence of first variceal bleeding in cirrhotics. Our findings are in contrast with the results of some other published studies that show a beneficial (10-14) or a harmful (20-21) effect of EVS on bleeding; however, they confirm the results of other controlled trials, in which only minor changes in the incidence of first episodes of variceal bleeding were observed in patients undergoing prophylactic EVS as compared with controls (15-19). It is difficult to explain such discrepancies, possibly related to differences in the selection of patients or to differences in EVS technique, that led to less effectiveness or to more side effects in some studies compared with others. Overall mortality was higher in control patients than in EVS patients (50% vs. 35%), but the difference was not statistically significant. Because the two groups differed in the distribution of some variables of potential prognostic significance, we performed a Cox’s regression analysis to adjust for imbalance in such variables. By this analysis, the difference in survival between EVS patients and controls increased markedly compared with the raw data (Figure 2) but barely missed statistical significance. Thus, prophylactic EVS does not seem to improve overall survival. The difference between the two groups may have been reduced by the life-saving effect of emergency EVS in bleeding control patients: however, we thought it was unethical to withold such an effective emergency measure in our controls. About two thirds of our control patients who bled from varices died because of the hemorrhage. This agrees with previous data from other trials (ll12,15,18) and confirms that the first variceal hemorrhage in cirrhotics has high mortality. In the present study, the intention was to treat all variceal hemorrhages in both treatment groups by emergency sclerotherapy; in practice, emergency EVS could be performed in only 31.5% of patients in the EVS group and in 52.9% of patients in the control group. Nevertheless, bleeding-related mortality rates were significantly lower in the EVS group. This suggests that,
1092
DE FRANCHIS
GASTROENTEROLOGY
ET AL.
although prophylactic EVS does not reduce the incidence of first variceal hemorrhage, it does reduce the severity of bleeding and, hence, the bleeding-related mortality. However. in our studv, this did not result in a significant improvement of overall survival. The reason for this was that only 23% of our controls died of a first episode of variceal hemorrhage; these were the only deaths that could conceivably be prevented by prophylactic sclerotherapy. Had EVS reduced bleeding-related mortality to near zero, our study would have shown a significant difference. However, because EVS reduced such mortality to 13%, 231 patients per study group (type I error, 5%; type II error, 20%) would have been required to show the significance of this difference. Because prophylactic EVS does not reduce the incidence of first variceal hemorrhage but seems to reduce bleeding-related mortality, only patients with an extremely high risk of bleeding (and, hence, an extremely high risk of dying of hemorrhage) should be treated by this technique. The problem thus arises of where to set the level of risk at which patients qualify for EVS. In the present study, by using Beppu’s variceal score (26), we selected a population of patients with a 2year risk of bleeding of about 40%. Although the observed incidence of bleeding matched closely the expected values, the influence of prophylactic EVS on overall survival approached, but did not reach, statistical significance. On the other hand, by using the same selection method but a higher cutoff for the expected risk of bleeding, Piai et al. (12) showed a beneficial effect of EVS on both bleeding and overall survival. Thus, it appears that patients with a a-year risk of bleeding > 50% might be suitable candidates for prophylactic EVS. However, the recruitment of patients for studies aimed at assessing this point may prove very difficult. In fact, we recently published a prognostic index [the North Italian Endoscopic Club (NIEC) index] (27)based on clinical and endoscopic parameters, which provides a simple tool to obtain a reliable estimate of the risk of bleeding of individual patients. Use of the NIEC index has shown that the patients with a 2-year risk of bleeding in excess of 50% represent < 20% of all cirrhotics with varices. In conclusion, with present-day injection techniques, EVS is not a suitable method to prevent the first variceal hemorrhage in cirrhotics with varices. Because it seems to reduce bleeding-related mortality, prophylactic EVS could be used to improve survival in patients with an extremely high risk of bleeding: however, further studies are needed before such a policy may be recommended for general clinical application.
Vol. 101, No. 4
References 1. Graham DY, Smith JL. The course of patients after variceal hemorrhage.
Gastroenterology
1981;80:800-809.
2. Jackson FC, Perrin EB, Smith AG, Dagradi AE, Nadal HM. A clinical investigation of the portacaval shunt. Survival analysis of the prophylactic operation. Am J Surg 1968;11:22-42. 3. Resnick RI-I, Chalmers TC, Ishihara AM, et al. A controlled study of the prophylactic portacaval shunt. A final report. Ann Intern Med 1969;70:675-688. 4. Conn HO, Lindenmuth WW, May CJ, Ramsby GR. Prophylactic portacaval anastomosis: a tale of two studies. Medicine 1972;51: 27-40. 5. Pascal JP, Cales P, Multicenter Study Group. Propranolol in the prevention of first gastrointestinal hemorrhage in patients with cirrhosis. N Engl J Med 1988;319:8-14. 6. The Italian Multicentric
Project for Propranolol
in the preven-
tion of Bleeding. Propranolol for prevention of bleeding in cirrhotic patients with large varices: a prophylactic multicentric randomized
clinical trial. Hepatology 1988;8:1-5.
7. Bosch J, Groszman RJ, Grace N, Mastai R, Conn HO, Rodes J. Propranolol in the prevention of the first hemorrhage from esophageal cooperative
varices: results of a randomized double clinical trial. J Hepatol 1988;7(Suppl):19.
blind
8. Ideo G, Bellati G, Fesce E, Grimoldi D. Nadolol can prevent the first gastrointestinal bleeding in cirrhotics: a prospective randomized study. Hepatology 1988;8:6-9. 9. Lebrec D, Poynard T, Capron JP, Hillon P, Geoffroy P, Roulot D, Chaput JL, Rueff B, Benhamou JP. Nadolol for prophylaxis of gastrointestinal bleeding in patients with cirrhosis. J Hepatol 1988;7:118-125. 10. Paquet KJ. Prophylactic endoscopic sclerosing treatment of the esophageal wall in varices. A prospective controlled randomized trial. Endoscopy 1982;14:4-5. 11. Witzel L, Wolbergs E, Merki H. Prophylactic endoscopic sclerotherapy of oesophageal varices: a prospective controlled trial. Lancet 1985;1:773-775. 12. Piai G, Cipolletta L, Claar M, Marone G, Bianco MA, Forte G, Iodice G, Mattera D, Minieri M, Rocco P, Santoro LM, Mazzacca G. Prophylactic sclerotherapy of high-risk esophageal varices: results of a multicentric prospective controlled trial. Hepatology 1988;8:1495-1500. 13. Saggioro A, Pallini P, Chiozzini G, Gilio AD. Prophylactic sclerotherapy for esophageal varices: preliminary results of a controlled prospective trial (abstr). Base1 Liver Week, Munich. Lancaster, England: MTP, 1986:218. 14. Wordehoff D, Spech HJ. Prophylaktische Osophagusvarizensklerosierung. Dtsch Med Wochenschr 1987;112:947-951. 15. Koch H, Henning H, Grimm H, Soehendra N. Prophylactic sclerosing of esophageal varices: results of a prospective controlled
study. Endoscopy
1986;18:40-43.
16. Triger DR, Johnson AG. Prophylactic sclerotherapy of oesophageal varices: a preliminary report (abstr). Gut 1985;3A:lOZ. 17. Kobe VE, Schentke KU. Endoskopische Sklerotherapie (EST) von osophagusvarizon Studien zur Teknik und zum Krankheitsverlauf. Dtsch Z Verdau Stoffwechsel Krankh 1987;47:151157. 18. Sauerbruch T, Wotzka R, Kopke W, Harlin M, Heldwein W, Bayerdorffer E, Sander R, Ansari H, Starz I, Paumgartner G. Prophylactic sclerotherapy before the first episode of variceal hemorrhage in patients with cirrhosis. N Engl J Med 1988;319: 8-14. 19. Fleig WE, Stange EF, Wordehoff D, Preclik G, Nuber R, Rainer K, Ditschuneit H. A randomized trial comparing prophylactic (PS) and therapeutic sclerotherapy in cirrhotic patients with
October
PROPHYLACTIC
1991
large oesophageal varices and no previous hemorrhage (abstr). J Hepatol 1988;7(Suppl):S128. 20. Gregory P, Hartigan P, Amodeo D, Baum R, Camara D, Colcher H, Fye C, Gebhard R, Goff J, Kruss D, McPhee M, Meier P, Rankin R, Reichelderfer J, Sanowski R, Shields D, Silvis S, Weesner R, Winship D, Young H. Prophylactic sclerotherapy for variceal bleeding: preliminary results of a controlled clinical trial (abstr). Gastroenterology 1987;92:1414. 21. Santangelo WC, Dueno MI, Estes BL, Krejs GJ. Prophylactic sclerotherapy of large esophageal varices. N Engl J Med 1988; 318:814-818. 22. Pagliaro L, Burroughs AK, Soerensen TIA, Lebrec D, Morabito A, D’Amico G, Tine’ F. Therapeutic controversies and randomised controlled trials (RCTs): prevention of bleeding and rebleeding in cirrhosis. Gastroenterol Int 1989;2:71-84. 23. Larson AW, Cohen H, Zweiban B, Chapman D, Gourdji M, Korula J, Weiner J. Acute esophageal variceal sclerotherapy. Results of a prospective randomized controlled trial. JAMA 1986;255:497-500. 24. Paquet KJ, Feussner H. Endoscopic sclerosis and esophageal balloon tamponade in acute hemorrhage from esophagogastric varices: a prospective controlled randomized trial. Hepatology 1985;5:580-583. The general 25. Japanese Research Society for Portal Hypertension. rules for recording endoscopic findings of esophageal varices. Jpn J Surg 1980;10:84-87. K, Koyanagi N, Nakayama S, Sakata H, 26. Beppu K, Inokuchi Kitano S, Kobayashi M. Prediction of variceal hemorrhage by esophageal endoscopy. Gastrointest Endosc 1981;27:213-218. Club for the Study and Treat27. The North Italian Endoscopic ment of Esophageal Varices. Prediction of the first variceal hemorrhage in patients with cirrhosis of the liver and esophageal varices. A prospective multicenter study. N Engl J Med 1988;389:983-989. IM, Dawson JL, Pietroni MC, Williams 28. Pugh RNH, Murray-Lyon R. Transection of the oesophagus for bleeding oesophageal varices.BrJ Surg 1973;60:646-649. 29. Peto R, Pike MC, Armitage P, Breslow NE, Cox DR, Howard SV,
SCLEROTHERAPY
IN CIRRHOTICS
1093
Mantel N, McPherson K, Peto T, Smith PG. Design and analysis of randomized clinical trials requiring prolonged observation. II. Analysis and examples. Br J Cancer 1977;35:1-39. 30. Jenrich R, Sampson P. In: Dixon WJ, ed. BMDP statistical software. Berkeley, CA: University of California, 1983:143206. E. Multivariate survival analysis using Cox’s regres31. Christensen sion model. Hepatology 1987;7:1346-1358. Received May 4,199O. Accepted February 18, 1991. Address requests for reprints to: Roberto de Franchis, M.D., Istituto di Medicina Interna, University of Milan, Via Pace 9. 20122 Milan, Italy. The following physicians also participated in the study: Fabrizio Antoniozzi, Ermanno Ancona, Federica Bedendo, Marta Bini, Gianfranc0 Brambilla, Emilio Brocchi, Marco Broggi, Renzo Curioni, Antonio Cusumano, Ersilio Del Ninno, Claudio Grosso, Laura La Mantia, Guido Lupinacci, Elisabetta Morandi, Daniele Neri, Angelika Nolte, Lorenzo Norberto, Graziano Pisano, Gianattilio Puerari, Giovanni Rubis Passoni, Domenico Scibetta. and Fabio Zangrandi. The following institutions are members of the NIEC: Servizio di Endoscopia Digestiva, Clinica Medica I, University of Bologna; Servizio di Endoscopia e Fisiopatologia Digestiva, Ospedale Maggiore, Crema; Servizio di Gastroenterologia ed Endoscopia Digestiva, Ospedale G. Fornaroli, Magenta; Istituto di Medicina Interna, University of Milan; Servizio di Endoscopia Digestiva, Ospedale Maggiore, Niguarda, Milan; Servizio di Endoscopia Digestiva, Ospedale S. Paolo, Milan: Servizio di Endoscopia Digestiva, Clinica Chirurgica I, University of Padova; and Patologia Chirurgica II. University of Padova, Italy. The authors are indebted to Dr. Benito Chinea for invaluable assistance in the statistical analysis. Preliminary reports of the present study were presented at the annual meeting of the American Gastroenterological Association, New Orleans, Louisiana, 1988, and in the annual meeting of the American Association for the Study of Liver Disease, Chicago, Illinois, 1989.
1991;101:1087-1093
Prophylactic Sclerotherapy in High-Risk Cirrhotics Selected by Endoscopic Criteria A Multicenter ROBERTO PAOLO
Randomized
DE FRANCHIS,
G. ARCIDIACONO,
PASQUALE
VITAGLIANO,
RAFFAELE
ARCIDIACONO,
ALESSANDRO GIANCARLO GIORGIO
ZAMBELLI, CALETTI,
BATTAGLIA,
The North Italian Endoscopic
MASSIMO PAOLO MARIA FELICE
SERGIO
PRIMIGNANI,
C. VAZZOLER, ROSSI,
COSENTINO,
BRUNATI,
and GIORGIO
GERUNDA
Club for the Study and Treatment
lthough the first variceal bleeding in patients with liver cirrhosis has an extremely high mortality rate (l--4), prophylaxis of variceal bleeding is still a
Trial
M. RIZZI,
ALFRED0
Controlled trials of sclerotherapy for the prevention of the first variceal hemorrhage in cirrhotics have given conflicting results. In the present study, 106 cirrhotics were randomized to sclerotherapy (55 patients) or control group (51 patients). Admission criteria were no history of previous variceal hleeding and the presence of high-risk varices, i.e., a variceal score 10 according to Beppu et al. Sclerotherapy sessions were performed at time zero, 7 days, 30 days, and then monthly until eradication. Follow-up endoscopies were performed at 6-month intervals thereafter. Control patients underwent repeat endoscopy at 6-month intervals. Bleeding episodes were treated by sclerotherapy in both groups, whenever possible. Mean follow-up was 24 months. Analysis of the results was performed by the intention-to-treat method. Variceal bleeding occurred in 19 sclerotherapy patients (34.5%) and in 17 controls (35.4%, P = NS). Overall mortality was 34.5% in sclerotherapy patients and 50% in controls (P = NS). Seven of the 19 sclerotherapy patients (36.8%) and 11 of the 17 controls (64.7%) who bled died of hemorrhage (P < 0.05, log-linear model). It is concluded that prophylactic sclerotherapy does not reduce the incidence of first variceal bleeding in cirrhotics. However, there seems to be a trend toward a lower bleeding-related mortality in sclerotherapy patients than in controls.
A
Controlled
of Esophageal
Varices
matter of controversy. Prophylactic shunt operations are no longer performed because of the negative results of four randomized controlled trials performed in the 1960s (24). Three controlled trials of propranolo1 (5-7) and two of nadolol(8,9) have shown that p blockers decrease the incidence of first variceal hemorrhage but do not seem to improve survival. Prophylactic endoscopic sclerotherapy (EVS) has been investigated in several randomized controlled trials and has been shown to be extremely effective in some studies (l&14), of no benefit in others (15--19), and even harmful in others (20,21). These conflicting results might be related to different selection of patients in the various studies or to different treatment of variceal bleeding in patients assigned to EVS as compared with controls. It has been pointed out that the higher the bleeding rate in the control groups, the more effective is sclerotherapy in preventing first bleeding (22). With the above considerations in mind, we designed the present randomized controlled trial to test the efficacy of EVS in preventing the first variceal bleeding and in reducing mortality in cirrhotics with high-risk varices identified by endoscopic criteria. Because emergency EVS has been shown to be lifesaving in variceal bleeding (23,24), we decided that EVS should be used whenever possible to treat acute
Abbreviations used in this study: EVS, endoscopic variceal sclerotherapy; NIEC, North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices. o 1991 by the American Gastroenterological Association 0016~5065/91/$3.00
1088
GASTROENTEROLOGY
DE FRANCHIS ET AL.
bleeding both in patients assigned lactic EVS and in controls.
to receive
prophy-
Patients and Methods Criteria for Admission From April 1985 to February 1987, all patients seen at the participating institutions with a diagnosis of cirrhosis of the liver based on history, physical examination, liver chemistry, and/or liver histology were eligible for the study if they met the following criteria. (a) They had no prior history of upper gastrointestinal bleeding, including bleeding from nonvariceal sources; (b) they had high-risk varices by endoscopic criteria (see below); (c) they had no other disease (e.g., cancer) reducing life expectancy to < 1 year; (d) they had no gastrointestinal ulcers at the time of randomization; (e) they had received no treatment for esophageal varices; and (f) they gave informed consent to participate in the study.
Definition of High-Risk Varices Varices were classified according to the suggestions of the Japanese Research Society for Portal Hypertension (25,261. They include recording of several features of the varices observed at endoscopy, such as color (blue or white); size (less than one third, up to two thirds, and more than two thirds of the esophageal radius); the longitudinal extension of the varices along the esophagus; and the presence and grade of several red signs on the variceal wall (red wale markings, cherry-red spots, hematocystic spots). Interobserver variation between the endoscopists participating in this study was decreased by intensive training, which has been described elsewhere (27). The risk of bleeding was estimated according to the discriminant equation of Beppu et al. (26), which was developed by using the above mentioned classification system. A score of < 0 according to Beppu’s calculations was chosen as the cutpoint to define high-risk varices. This corresponds to the presence of blue varices with red wale markings, and/or cherry-red spots, and/or hematocystic spots. According to Beppu’s retrospective calculations, this corresponds to a bleeding risk of 264%. However, prospective validation of Beppu’s score performed by our group has shown that the actual 2-year risk of bleeding of such patients is about 40% (27).
Randomization At the beginning of the study, each center was given a computer-generated randomization list, which was kept by physicians not directly involved in the study. Randomization took place on the day of diagnostic endoscopy. Baseline Clinical Assessment At entry, a full medical history, physical examination, hepatitis B virus markers, routine liver chemistries, and prothrombin activity were obtained for each patient. The degree of hepatic decompensation of the patients was
assessed by the Child’s classification Pugh’s criteria (28). Methods of Sclerotherapy Schedule
Vol. 101, No. 4
modified according
to
and Follow-up
Treatment was performed with Olympus GIF-D3 or GIF-Q scopes (Olympus Optical Co., Tokyo, Japan) under light sedation (diazepam, 5 mg IV). Polidocanol 1% (Atossisclerol, Kreussler Pharma, Wiesbaden, Germany) was used as the sclerosant in most instances. Etanolamine oleate 5% was used in four patients. Injections were intravariceal and paravariceal and varied in volume between 1 and 5 mL. The mean volume of sclerosant injected for each sclerotherapy session was 21.5 mL for first sessions (range, 15-60 mL) and decreased progressively for subsequent sessions in individual patients, because less sclerosant was injected as the varices became smaller as a consequence of sclerotherapy. With minor variations between centers, the treatment schedule was time zero, 7 days, 30 days, and then monthly until eradication, and then the patients underwent endoscopy every 6 months to detect recurrence of varices. Eradication was defined as disappearance of all varices from the distal 3 cm of the esophagus or reduction of varices to small white columns with no red signs. In addition to their scheduled visits to the liver clinic, control patients had follow-up phone interviews every 3 months and endoscopic control examinations every 6 months. At the beginning of the study all alcoholic patients were advised to stop drinking. At each visit, an attempt to assess compliance was made by questioning the patients and/or their relatives. Follow-up was closed on March 31,1989. Bleeding Whenever, during follow-up, a patient was admitted to one of the participating institutions for upper gastrointestinal hemorrhage, emergency endoscopy was performed to establish the source of bleeding. If a diagnosis of variceal bleeding was made, emergency sclerotherapy was performed as specified above, regardless of the treatment group to which the patient belonged. In any other instance, the source of bleeding was defined as “unknown,” and the patients were treated conservatively, unless an actively bleeding lesion other than a varix was observed. If a patient died before reaching the hospital or was admitted to a hospital not participating in the study, every effort was made to obtain information about the cause of death and the details of the bleeding episode. Sample Size Calculation and Statistical Analysis According to our previous observation, the expected 2-year incidence of first episodes of variceal hemorrhage was about 40% (27). Assuming a reduction of this risk to 10% by sclerotherapy, approximately 50 patients for each group would be required for a type I error of 5% and a type II error of 20%.
October
1991
The entry point of the study was the date of randomization, always performed on the day of variceal assessment. The variables of interest were compared by either x2 test or Student’s t test for independent samples. End points of the study were the first episode of bleeding, death of all causes, and bleeding-related death. Patients who survived a first episode of variceal hemorrhage, however, underwent further regular follow-up. The bleeding rates and the death rates were estimated by life-table analysis (Kaplan-Meier method) and compared by the log rank test (29). Bleeding-related death rates in relation to treatment group and to the modality of emergency treatment of bleeding were evaluated by a log-linear model (30). Adjustment for imbalance in prognostic variables was performed by Cox’s regression analysis in which the prognostic variables were included together with the treatment variable (31). The analysis was performed according to an intention-totreat strategy. Results When recruitment was completed in February had been considered for the study 1987, 126 patients by the eight participating hospitals. Five centers contributed equally to the treatment groups, 3-26 patients to each group. In the remaining three centers, slightly more patients were. assigned to EVS than to the control group (1 excess EVS patient in two centers, and 2 excess EVS patients in the third center). At the end of recruitment, 65 patients had been assigned to EVS, 61 to the control group. We subsequently found out that 20 patients (15.9%) did not fulfill the criteria for admission and, thus, excluded them from the study. The reasons for exclusion were Beppu’s variceal score > 0 (7 patients, 1 EVS, and 6 controls), upper gastrointestinal bleeding before randomization (3 patients, 2 EVS, and 1 control), prior treatment of portal hypertension (p blockers, 3 patients and all controls), and lack of consent to EVS after randomization (7 patients). Thus, a total of 106 patients remained in the study; 55 of them belonged to the EVS group, 51 to the control group. Table 1 shows the main demographic and clinical features of the patients. The groups were relatively well balanced in age, sex distribution, and Child-Pugh classification. However, there was a significant excess of alcoholic patients in the control group (P = 0.035). Conversely, Beppu’s variceal score tended to be worse in EVS patients (P = 0.034). At the beginning of the study, EVS was performed on hospitalized patients; the patients were admitted the day of the EVS session and discharged the day after. As a result, in 15 patients EVS could not be started within 2 days from randomization, as for the others, because in some instances there was a waiting list for admissions for elective procedures. In these 15 patients, the delay between the date of randomization
PROPHYLACTIC
SCLEROTHERAPY
IN CIRRHOTICS
1089
Table I. Demographic, Clinical, and Endoscopic Features of Patients at Entrv, According to Study Group Study
group
Sclerotherapy
Patients Age, mean (yr) Sex (M:F) Alcoholics (%) Child’s class
P
Controls
55
51
56.2
55.1
NS
31:24
38:13
NS
15 (27.3)
25 (49.0)
0.035
A (%I
14 (25.5)
18 (35.3)
NS
B (%)
31 (56.3)
22 (43.1)
NS
c (%I
10 (18.2)
11 (21.6)
NS
19 (34.5)
29 (56.8)
0.034
22 (40.0)
16 (31.4)
NS
14 (25.5)
6 (12.8)
NS
Beppu’s score 0 to -0.38 (%) -0.38
to -1.14
< -1.14
(%)
(%)
and that of the first EVS session lasted between 14 and 90 days (mean, 26.5 days). However, according to our intention-to-treat strategy, the time between randomization and start of EVS was included in the analysis. Unfortunately, during this time, 3 patients assigned to EVS died before EVS could be started. One of them died of variceal hemorrhage, and the remaining 2 died of liver failure. Of the 52 patients who actually received sclerotherapy, 37 (71%) achieved eradication of their varices after a mean of four sclerotherapy sessions, corresponding to about 2 months. Three controls were lost to follow-up and were thus excluded from analysis. Bleeding
The mean duration of follow-up was 24 months (range, l-48 months). During this time, 19 controls and 20 EVS patients bled (Table 2). Two controls and 1 EVS patient bled from gastric erosions. Thus, a total of 17 controls bled from varices. This figure corresponds to a 2-year rate of bleeding of 39% (KaplanMeier estimate) and is identical to that expected on the basis of our prospective validation of Beppu’s
Table 2. Bleeding and Mortality Sclerotherapy n = 55 Sources of bleeding Varices (%) Esophageal ulcers (%) Esophageal hematoma (%) Gastric erosions (%) Total (%) Causes of death Hemorrhage (%) Liver failure (%] Hepatocellular carcinoma (%) Extrahepatic causes (%) Total I%) Note. Differences
between
groups
14 (25.4)
Controls n = 48
17 (35.4)
4 (7.3) l(1.8) l(l.8)
-2 (4.2)
20 (36.3)
19 (39.6)
7 (12.7)
11 (22.9)
5 (9.1)
8 (16.7)
2 (3.6)
l(2.1)
5 (9.1)
4 (8.3)
19 (34.5)
24 (50.0)
are not statistically
significant
1090
GASTROENTEROLOGY
DE FRANCHIS ET AL.
variceal score. In the sclerotherapy group, 19 patients bled either from varices (14 patients) or because of complications of sclerotherapy (5 patients, Table 2). Figure 1 shows the cumulative proportion of patients with no bleeding in the two groups (Kaplan-Meier plot). During the early phase of the study, there was a slightly higher incidence of bleeding in the EVS group than in the control group. In this period, 4 EVS patients bled from esophageal ulcers on varices, and 1 bled from a sclerotherapy-induced esophageal hematoma. The difference between the two curves began to decrease after the sixth month and had disappeared completely by 18 months. (Odds ratio for bleeding at 2 years, 0.96; 95% confidence intervals, 0.43-2.17; P = NS.)
01
0 patients
Overall Mortality Twenty-four controls died compared with 19 EVS patients (Table 2). Deaths in the EVS group included also the 3 patients who died before EVS was begun (1 died of variceal hemorrhage and 2 of liver failure). The difference in the cumulative proportion of patients surviving in the two groups (Kaplan-Meier plot: Figure 2) was not statistically significant. (Odds ratio for intention to treat, 0.53; 95% confidence intervals, 0.24-1.16; P = NS. Analysis according to the “treatment received” strategy showed similar results.) Adjustment for Imbalance Adjustment for imbalance in prognostic variables was performed by Cox’s regression analysis in which the prognostic variables were included together with the treatment variable. In addition to
months
Figure 1. Cumulative proportion of patients with no bleeding (Kaplan-Meier plot). a-0, Sclerotherapy; O-O, controls: [O], number of patients at risk control group; (O), number of patients at risk sclerotherapy group.
at risk
I
1
I
I
IO
20
30
40
30
24
15
6
27
20
17
6
Vol. 101, No. 4
I
50 months
sclerotherapy controls
Figure 2. Cumulative proportion of patients surviving (KaplanMeier plot; O-O, controls: O-O, sclerotherapy) and adjusted estimated survival curves (adjustment for imbalance, Cox model) for EVS (A-A) and control [A-A) patients based on the final Cox’s model for alcoholic etiology, Child’s class B, and a Beppu’s score ranging between -0.38 and -1.14.
treatment, the final Cox’s regression model included the etiology of cirrhosis, Child’s class, and Beppu’s variceal score. In this analysis, survival of EVS patients was better than that of controls, but the difference failed to reach statistical significance (P = 0.0758, Figure 2). Treatment of Acu te Bleeding and Bleeding-Related Deaths Our protocol called for emergency EVS for all patients acutely bleeding from varices. However, only 6 of 19 EVS patients and 9 of 17 control patients actually received emergency EVS. All these patients were treated with intravariceal and paravariceal injections of Polidocanol 1%. The volumes of sclerosant used per single injection varied between 1 and 5 mL; the total volumes injected per session ranged between 15 and 40 mL (median, 23 mL). In the control group, the explanation of the low proportion of patients receiving EVS is that eight patients were admitted to hospitals not participating in the study, where no emergency EVS was available at the time. In the EVS group, eight patients were admitted to hospitals where no emergency EVS was available. In all these patients, conservative treatment [Sengstaken tube with or without vasoactive drugs, i.e., Glypressin (Ferring AB, Malmo, Sweden) or Somatostatin (Serono, Rome, Italy), was used]. The remaining five patients did not receive emergency EVS, because they were bleeding from esophageal ulcers on varices (four patients) and from an esoph-
October
PROPHYLACTIC
199 1
ageal hematoma (one patient). In these cases, conservative treatment (Glypressin) was deemed more appropriate than emergency EVS. Nearly 65% of the patients in the control group who bled from varices died of hemorrhage compared with 36.8% of patients in the EVS group. This difference is significant when analyzed by log-linear model (P < 0.05, Table 3). Regardless of the modality of treatment of bleeding episodes, the mortality rates were lower in EVS patients than in control patients (Table 3). This suggests that bleeding is less severe in EVS patients, although the transfusion requirements were not significantly different in the two groups (EVS: mean, 3.75 U per patient; range, O-10 U; controls: mean, 3.3 U per patient; range, O-8 U). Discussion Although prophylactic sclerotherapy has been studied in many randomized controlled trials (lo21), the role of this technique in preventing the first variceal hemorrhage in patients with cirrhosis of the liver has not been established so far. The main reason for this is the extreme variability in the results of the above trials. For example, the bleeding rate in the control groups of such trials ranged between < 16% (20) and > 70% (13). This clearly indicates that different criteria of patients selection were applied in the various studies. It has been pointed out that the higher the bleeding rate in the control groups, the more effective is sclerotherapy in preventing first bleeding (22). When our study was begun in 1985, we decided to make an attempt to select patients with a high risk of bleeding. At that time, we had only preliminary data from our study on the prediction of the first variceal hemorrhage in cirrhotics (27). Those data showed that the endoscopic score proposed by Beppu et al. (26) was a valuable tool in identifying high-risk patients and that patients with a score of 10 had a a-year chance of bleeding of about 40%. We speculated that this level of risk was sufficiently high to allow the beneficial effect of sclerotherapy in the prevention of first bleeding to emerge, if such an effect existed. Indeed, the control patients in our study had a 2-year rate of bleeding of 39% by Kaplan-Meier estimate, which
Table 3. Bleeding-Related Deaths According to the Modality of Treatment of Acute Variceal Bleeding Sclerotherapy Deaths/total (%)
Treatment Emergency
EVS
Controls Deaths/total
(%)
4/9 (44.4)
O/6 (0.0)
No EVS
7/13 (53.8)
718 (87.5)
Total
7/19 (36.8)”
11/17 (64.7)
“P < 0.05 vs. controls,
log-linear
model
(30).
SCLEROTHERAPY
IN CIRRHOTICS
lo91
closely corresponded to the expected 40%. However, during the early phase of the study, we observed that the rate of first bleeding in EVS patients was higher than that in controls. In fact, during the first 6 months, some patients in the EVS group bled because of complications of the procedure (esophageal ulcers and esophageal hematoma). After the sixth month of the study, when eradication of the varices was achieved in patients in the EVS group, the bleeding rate in this group decreased to almost zero, whereas patients in the control group continued to bleed. As a result, the final proportions of patients bleeding in the two groups were very similar. Therefore, in our hands, prophylactic sclerotherapy did not change the incidence of first variceal bleeding in cirrhotics. Our findings are in contrast with the results of some other published studies that show a beneficial (10-14) or a harmful (20-21) effect of EVS on bleeding; however, they confirm the results of other controlled trials, in which only minor changes in the incidence of first episodes of variceal bleeding were observed in patients undergoing prophylactic EVS as compared with controls (15-19). It is difficult to explain such discrepancies, possibly related to differences in the selection of patients or to differences in EVS technique, that led to less effectiveness or to more side effects in some studies compared with others. Overall mortality was higher in control patients than in EVS patients (50% vs. 35%), but the difference was not statistically significant. Because the two groups differed in the distribution of some variables of potential prognostic significance, we performed a Cox’s regression analysis to adjust for imbalance in such variables. By this analysis, the difference in survival between EVS patients and controls increased markedly compared with the raw data (Figure 2) but barely missed statistical significance. Thus, prophylactic EVS does not seem to improve overall survival. The difference between the two groups may have been reduced by the life-saving effect of emergency EVS in bleeding control patients: however, we thought it was unethical to withold such an effective emergency measure in our controls. About two thirds of our control patients who bled from varices died because of the hemorrhage. This agrees with previous data from other trials (ll12,15,18) and confirms that the first variceal hemorrhage in cirrhotics has high mortality. In the present study, the intention was to treat all variceal hemorrhages in both treatment groups by emergency sclerotherapy; in practice, emergency EVS could be performed in only 31.5% of patients in the EVS group and in 52.9% of patients in the control group. Nevertheless, bleeding-related mortality rates were significantly lower in the EVS group. This suggests that,
1092
DE FRANCHIS
GASTROENTEROLOGY
ET AL.
although prophylactic EVS does not reduce the incidence of first variceal hemorrhage, it does reduce the severity of bleeding and, hence, the bleeding-related mortality. However. in our studv, this did not result in a significant improvement of overall survival. The reason for this was that only 23% of our controls died of a first episode of variceal hemorrhage; these were the only deaths that could conceivably be prevented by prophylactic sclerotherapy. Had EVS reduced bleeding-related mortality to near zero, our study would have shown a significant difference. However, because EVS reduced such mortality to 13%, 231 patients per study group (type I error, 5%; type II error, 20%) would have been required to show the significance of this difference. Because prophylactic EVS does not reduce the incidence of first variceal hemorrhage but seems to reduce bleeding-related mortality, only patients with an extremely high risk of bleeding (and, hence, an extremely high risk of dying of hemorrhage) should be treated by this technique. The problem thus arises of where to set the level of risk at which patients qualify for EVS. In the present study, by using Beppu’s variceal score (26), we selected a population of patients with a 2year risk of bleeding of about 40%. Although the observed incidence of bleeding matched closely the expected values, the influence of prophylactic EVS on overall survival approached, but did not reach, statistical significance. On the other hand, by using the same selection method but a higher cutoff for the expected risk of bleeding, Piai et al. (12) showed a beneficial effect of EVS on both bleeding and overall survival. Thus, it appears that patients with a a-year risk of bleeding > 50% might be suitable candidates for prophylactic EVS. However, the recruitment of patients for studies aimed at assessing this point may prove very difficult. In fact, we recently published a prognostic index [the North Italian Endoscopic Club (NIEC) index] (27)based on clinical and endoscopic parameters, which provides a simple tool to obtain a reliable estimate of the risk of bleeding of individual patients. Use of the NIEC index has shown that the patients with a 2-year risk of bleeding in excess of 50% represent < 20% of all cirrhotics with varices. In conclusion, with present-day injection techniques, EVS is not a suitable method to prevent the first variceal hemorrhage in cirrhotics with varices. Because it seems to reduce bleeding-related mortality, prophylactic EVS could be used to improve survival in patients with an extremely high risk of bleeding: however, further studies are needed before such a policy may be recommended for general clinical application.
Vol. 101, No. 4
References 1. Graham DY, Smith JL. The course of patients after variceal hemorrhage.
Gastroenterology
1981;80:800-809.
2. Jackson FC, Perrin EB, Smith AG, Dagradi AE, Nadal HM. A clinical investigation of the portacaval shunt. Survival analysis of the prophylactic operation. Am J Surg 1968;11:22-42. 3. Resnick RI-I, Chalmers TC, Ishihara AM, et al. A controlled study of the prophylactic portacaval shunt. A final report. Ann Intern Med 1969;70:675-688. 4. Conn HO, Lindenmuth WW, May CJ, Ramsby GR. Prophylactic portacaval anastomosis: a tale of two studies. Medicine 1972;51: 27-40. 5. Pascal JP, Cales P, Multicenter Study Group. Propranolol in the prevention of first gastrointestinal hemorrhage in patients with cirrhosis. N Engl J Med 1988;319:8-14. 6. The Italian Multicentric
Project for Propranolol
in the preven-
tion of Bleeding. Propranolol for prevention of bleeding in cirrhotic patients with large varices: a prophylactic multicentric randomized
clinical trial. Hepatology 1988;8:1-5.
7. Bosch J, Groszman RJ, Grace N, Mastai R, Conn HO, Rodes J. Propranolol in the prevention of the first hemorrhage from esophageal cooperative
varices: results of a randomized double clinical trial. J Hepatol 1988;7(Suppl):19.
blind
8. Ideo G, Bellati G, Fesce E, Grimoldi D. Nadolol can prevent the first gastrointestinal bleeding in cirrhotics: a prospective randomized study. Hepatology 1988;8:6-9. 9. Lebrec D, Poynard T, Capron JP, Hillon P, Geoffroy P, Roulot D, Chaput JL, Rueff B, Benhamou JP. Nadolol for prophylaxis of gastrointestinal bleeding in patients with cirrhosis. J Hepatol 1988;7:118-125. 10. Paquet KJ. Prophylactic endoscopic sclerosing treatment of the esophageal wall in varices. A prospective controlled randomized trial. Endoscopy 1982;14:4-5. 11. Witzel L, Wolbergs E, Merki H. Prophylactic endoscopic sclerotherapy of oesophageal varices: a prospective controlled trial. Lancet 1985;1:773-775. 12. Piai G, Cipolletta L, Claar M, Marone G, Bianco MA, Forte G, Iodice G, Mattera D, Minieri M, Rocco P, Santoro LM, Mazzacca G. Prophylactic sclerotherapy of high-risk esophageal varices: results of a multicentric prospective controlled trial. Hepatology 1988;8:1495-1500. 13. Saggioro A, Pallini P, Chiozzini G, Gilio AD. Prophylactic sclerotherapy for esophageal varices: preliminary results of a controlled prospective trial (abstr). Base1 Liver Week, Munich. Lancaster, England: MTP, 1986:218. 14. Wordehoff D, Spech HJ. Prophylaktische Osophagusvarizensklerosierung. Dtsch Med Wochenschr 1987;112:947-951. 15. Koch H, Henning H, Grimm H, Soehendra N. Prophylactic sclerosing of esophageal varices: results of a prospective controlled
study. Endoscopy
1986;18:40-43.
16. Triger DR, Johnson AG. Prophylactic sclerotherapy of oesophageal varices: a preliminary report (abstr). Gut 1985;3A:lOZ. 17. Kobe VE, Schentke KU. Endoskopische Sklerotherapie (EST) von osophagusvarizon Studien zur Teknik und zum Krankheitsverlauf. Dtsch Z Verdau Stoffwechsel Krankh 1987;47:151157. 18. Sauerbruch T, Wotzka R, Kopke W, Harlin M, Heldwein W, Bayerdorffer E, Sander R, Ansari H, Starz I, Paumgartner G. Prophylactic sclerotherapy before the first episode of variceal hemorrhage in patients with cirrhosis. N Engl J Med 1988;319: 8-14. 19. Fleig WE, Stange EF, Wordehoff D, Preclik G, Nuber R, Rainer K, Ditschuneit H. A randomized trial comparing prophylactic (PS) and therapeutic sclerotherapy in cirrhotic patients with
October
PROPHYLACTIC
1991
large oesophageal varices and no previous hemorrhage (abstr). J Hepatol 1988;7(Suppl):S128. 20. Gregory P, Hartigan P, Amodeo D, Baum R, Camara D, Colcher H, Fye C, Gebhard R, Goff J, Kruss D, McPhee M, Meier P, Rankin R, Reichelderfer J, Sanowski R, Shields D, Silvis S, Weesner R, Winship D, Young H. Prophylactic sclerotherapy for variceal bleeding: preliminary results of a controlled clinical trial (abstr). Gastroenterology 1987;92:1414. 21. Santangelo WC, Dueno MI, Estes BL, Krejs GJ. Prophylactic sclerotherapy of large esophageal varices. N Engl J Med 1988; 318:814-818. 22. Pagliaro L, Burroughs AK, Soerensen TIA, Lebrec D, Morabito A, D’Amico G, Tine’ F. Therapeutic controversies and randomised controlled trials (RCTs): prevention of bleeding and rebleeding in cirrhosis. Gastroenterol Int 1989;2:71-84. 23. Larson AW, Cohen H, Zweiban B, Chapman D, Gourdji M, Korula J, Weiner J. Acute esophageal variceal sclerotherapy. Results of a prospective randomized controlled trial. JAMA 1986;255:497-500. 24. Paquet KJ, Feussner H. Endoscopic sclerosis and esophageal balloon tamponade in acute hemorrhage from esophagogastric varices: a prospective controlled randomized trial. Hepatology 1985;5:580-583. The general 25. Japanese Research Society for Portal Hypertension. rules for recording endoscopic findings of esophageal varices. Jpn J Surg 1980;10:84-87. K, Koyanagi N, Nakayama S, Sakata H, 26. Beppu K, Inokuchi Kitano S, Kobayashi M. Prediction of variceal hemorrhage by esophageal endoscopy. Gastrointest Endosc 1981;27:213-218. Club for the Study and Treat27. The North Italian Endoscopic ment of Esophageal Varices. Prediction of the first variceal hemorrhage in patients with cirrhosis of the liver and esophageal varices. A prospective multicenter study. N Engl J Med 1988;389:983-989. IM, Dawson JL, Pietroni MC, Williams 28. Pugh RNH, Murray-Lyon R. Transection of the oesophagus for bleeding oesophageal varices.BrJ Surg 1973;60:646-649. 29. Peto R, Pike MC, Armitage P, Breslow NE, Cox DR, Howard SV,
SCLEROTHERAPY
IN CIRRHOTICS
1093
Mantel N, McPherson K, Peto T, Smith PG. Design and analysis of randomized clinical trials requiring prolonged observation. II. Analysis and examples. Br J Cancer 1977;35:1-39. 30. Jenrich R, Sampson P. In: Dixon WJ, ed. BMDP statistical software. Berkeley, CA: University of California, 1983:143206. E. Multivariate survival analysis using Cox’s regres31. Christensen sion model. Hepatology 1987;7:1346-1358. Received May 4,199O. Accepted February 18, 1991. Address requests for reprints to: Roberto de Franchis, M.D., Istituto di Medicina Interna, University of Milan, Via Pace 9. 20122 Milan, Italy. The following physicians also participated in the study: Fabrizio Antoniozzi, Ermanno Ancona, Federica Bedendo, Marta Bini, Gianfranc0 Brambilla, Emilio Brocchi, Marco Broggi, Renzo Curioni, Antonio Cusumano, Ersilio Del Ninno, Claudio Grosso, Laura La Mantia, Guido Lupinacci, Elisabetta Morandi, Daniele Neri, Angelika Nolte, Lorenzo Norberto, Graziano Pisano, Gianattilio Puerari, Giovanni Rubis Passoni, Domenico Scibetta. and Fabio Zangrandi. The following institutions are members of the NIEC: Servizio di Endoscopia Digestiva, Clinica Medica I, University of Bologna; Servizio di Endoscopia e Fisiopatologia Digestiva, Ospedale Maggiore, Crema; Servizio di Gastroenterologia ed Endoscopia Digestiva, Ospedale G. Fornaroli, Magenta; Istituto di Medicina Interna, University of Milan; Servizio di Endoscopia Digestiva, Ospedale Maggiore, Niguarda, Milan; Servizio di Endoscopia Digestiva, Ospedale S. Paolo, Milan: Servizio di Endoscopia Digestiva, Clinica Chirurgica I, University of Padova; and Patologia Chirurgica II. University of Padova, Italy. The authors are indebted to Dr. Benito Chinea for invaluable assistance in the statistical analysis. Preliminary reports of the present study were presented at the annual meeting of the American Gastroenterological Association, New Orleans, Louisiana, 1988, and in the annual meeting of the American Association for the Study of Liver Disease, Chicago, Illinois, 1989.
