Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Luca, A.
Right arrow Articles by Pagliaro, L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Luca, A.
Right arrow Articles by Pagliaro, L.
(Radiology. 1999;212:411-421.)
© RSNA, 1999

Special Report

TIPS for Prevention of Recurrent Bleeding in Patients with Cirrhosis: Meta-analysis of Randomized Clinical Trials1

Angelo Luca, MD, Gennaro D'Amico, MD, Roberto La Galla, MD, Massimo Midiri, MD, Alberto Morabito, MS and Luigi Pagliaro, MD

1 From the Departments of Radiology (A.L., R.L.G., M.M.) and Internal Medicine (G.D.) and the Institute of Internal Medicine and Pneumology (A.L., L.P.), University of Palermo, Ospedale V Cervello, Via Trabucco 180, 90146 Palermo, Italy, and the Institute of Medical Statistics and Biometry, University of Milan, Italy (A.M.). Received May 1, 1998; revision requested July 6; revision received August 26; accepted January 25, 1999. Address reprint requests to A.L.


   Abstract
TOP
 Abstract
Introduction
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 Appendix 1
 References
 
PURPOSE: To compare the effects of transjugular intrahepatic portosystemic shunt (TIPS) creation with those of endoscopic treatment with or without propranolol administration (ie, conventional treatment) on recurrent bleeding, encephalopathy, and mortality by using meta-analysis of 11 published randomized clinical trials.

MATERIALS AND METHODS: Data from 11 relevant studies were retrieved by means of computerized and manual search. The combinability of the studies was assessed in terms of clinical and statistical criteria. Data were extracted on the basis of the intention-to-treat principle, and treatment effects were measured as risk differences between TIPS creation and conventional treatment. Pooled estimates were computed according to a random-effects model.

RESULTS: A total of 750 patients were included in 11 trials. No significant heterogeneity was found for any of the outcomes. Pooled risk differences were recurrent bleeding, -31% (95% CI, -39%, -23%); encephalopathy, +16% (95% CI, +10%, +22%); death due to all causes, +2% (95% CI, -4%, +9%); and death due to bleeding, -5% (95% CI, -11%, +6%). Clinically important complications occurred in 22% of patients and were associated with both treatments. TIPS dysfunction occurred in 55% of patients.

CONCLUSION: TIPS creation markedly reduces risk of rebleeding but increases risk of encephalopathy without affecting survival. Therefore, TIPS creation may not be the best first-choice therapy for prevention of recurrent variceal bleeding. Criteria for selection of candidates for TIPS creation should be assessed in future prospective studies.

Index terms: Liver, cirrhosis, 761.794 • Liver, hemorrhage, 761.458. 761.75 • Liver, interventional procedures, 761.1299 • Radiology and radiologists, outcomes studies • Shunts, portosystemic, 95.1268 • Varices, 71.75


   Introduction
TOP
 Abstract
 Introduction
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 Appendix 1
 References
 
In 1990, Richter et al (1) reported preliminary clinical results of percutaneous transjugular intrahepatic portosystemic shunt (TIPS) creation with use of a balloon-expandable metallic stent. TIPS creation is an interventional radiologic procedure that results in decompression of the splanchnic venous system in patients with portal hypertension by creating a low-resistance channel between an intrahepatic branch of the portal vein and a main hepatic vein. This technique has gained wide popularity for management of portal hypertension because it is relatively noninvasive, has a low frequency of procedure-related complications, and does not preclude future liver transplantation. However, the initial enthusiasm for this procedure has been dampened because of results from several uncontrolled studies in which a remarkable frequency of stent dysfunction (2–5) and increased risk of encephalopathy (5–8) after TIPS creation were reported.

In an early randomized study (9), published in abstract form in 1994, TIPS creation was compared with endoscopic sclerotherapy for the prevention of recurrent variceal bleeding. The results of this study showed an increased mortality rate in patients treated with TIPS. In 1994, a consensus conference sponsored by the National Digestive Advisory Board (10) recommended guidelines for TIPS creation. It was recommended that until the role of TIPS has been identified in randomized clinical trials, the procedure should be clinically used only in patients with acute variceal bleeding that cannot be controlled with endoscopic and pharmacologic therapy and in patients with recurrent bleeding despite adequate treatment with ß-adrenergic blocking agents (hereafter, ß-blockers) and endoscopic therapy. Thus far, however, randomized clinical trials usually have included small sample sizes and have shown conflicting results, which hinder the drawing of conclusions for clinical practice.

The aim of this meta-analysis was to assess the effectiveness of TIPS as compared with sclerotherapy or banding ligation of esophageal varices, with or without administration of propranolol hydrochloride, for the prevention of recurrent bleeding. Because these endoscopic treatments and the use of propranolol are well established and widely used for the prevention of recurrent bleeding in patients with cirrhosis (11), they are referred to as conventional treatment throughout this article.


   MATERIALS AND METHODS
TOP
 Abstract
 Introduction
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 Appendix 1
 References
 
This meta-analysis was performed according to a protocol determined before the study, and we followed widely accepted methodological recommendations (12–15). Measurement of treatment effectiveness was determined on the basis of four clinically relevant outcomes: recurrence of gastrointestinal bleeding due to any cause, portosystemic encephalopathy, mortality due to any cause, and mortality due to recurrent bleeding. Although recurrent bleeding from esophageal varices may be a more clinically important measure of treatment effectiveness, we used recurrent bleeding from any source because, in some of the available studies, the information on the source of recurrent bleeding is incomplete and because TIPS is expected to reduce recurrent bleeding due to any portal hypertensive cause.

Selection of the Trials
Studies that fulfilled the following criteria were included in the present meta-analysis: (a) TIPS was compared with endoscopic treatment (sclerotherapy or banding ligation) alone or with propranolol; (b) patients were randomly assigned to the treatment regimen, and prospective follow-up was included; (c) patients with cirrhosis and previous variceal bleeding were included; (d) two or more of the following outcomes were assessed: (i) recurrent bleeding, (ii) encephalopathy, and (iii) death; and (e) results were published in English as abstracts or full reports.

Study Identification
Pertinent studies were retrieved from the MEDLINE database (January 1991–February 1998) by using the search terms "TIPS," "sclerotherapy," "variceal band ligation," and "beta-blockers" and by limiting the search to reports of clinical trials and studies with human patients. In addition, a manual search was performed by checking the reference lists from articles or reviews to identify studies not yet included in the MEDLINE database (16). Finally, the abstracts from international congresses on liver diseases underwent peer review to identify trials still in progress. When the results of a single study were reported in more than one publication, only the most recent and complete data were included in the meta-analysis. Eleven randomized clinical trials that fulfilled the criteria were identified, seven published in full form (17–23) and four published in abstract form (24–27).

Data Extraction
Data from each randomized clinical trial were extracted by two independent reviewers (A.L., G.D.). For each study and each type of treatment, the following data were extracted: number of patients, mean age, sex, prevalence of alcoholic cirrhosis, Child-Pugh score, presence of ascites, time between index bleeding and random assignment to treatment, type and modality of control treatment, rate of technical success of the procedures, portacaval pressure gradient before and after TIPS creation, mean length of follow-up, incidence of TIPS dysfunction, and definition and number of each outcome (Tables 1–4). Numeric discrepancies between the two independent data extractions were resolved after discussion.


View this table:
[in this window]
[in a new window]
  		TABLE 1. Patient Characteristics and Control Treatments in Randomized Clinical Trials in the Meta-analysis
 

View this table:
[in this window]
[in a new window]
  		TABLE 2. Technical Results of Treatment in Full Reports of Randomized Clinical Trials
 

View this table:
[in this window]
[in a new window]
  		TABLE 3. Outcome Rates in Randomized Clinical Trials and Pooled Risk Differences
 

View this table:
[in this window]
[in a new window]
  		TABLE 4. Methodologic Characteristics of Randomized Clinical Trials Published as Full Reports
 
Qualitative Assessment
The quality of randomized clinical trials published in full reports was assessed by two independent observers (A.L., G.D.) who used a previously reported (28) quality-assessment system that was partially modified according to the treatments under study (see Appendix). The system was based on the following principal criteria: clearly defined inclusion and exclusion criteria, blinded treatment assignment, baseline comparability of treatment groups, clearly defined outcome variables, treatment for acute bleeding equal in the two groups, and assessment of treatment side effects and complications. To assess baseline comparability of patient groups in each trial, we made an arbitrary and a priori decision to consider the trial groups to be similar at baseline if none of the baseline variables showed either a statistically significant difference or a difference of 40% or greater (28).

Any discrepancy in the qualitative assessment between the two observers was discussed, and a consensus was achieved. The total score of each trial also was expressed as a percentage of the maximum achievable score.

Statistical Methods
We assumed that variceal endoscopic sclerotherapy and banding ligation with or without the use of ß-blockers were equivalent control interventions. Although the results of previous meta-analyses (29,30) do not support this assumption, there were no significant differences between the control groups of the trials included in the present analysis in terms of the incidence of outcomes of interest during the study period (Table 3). Thus, in an attempt to improve the power of the meta-analysis, we considered the control treatments to be equivalent.

All comparisons were performed according to the randomly assigned treatment (intended-treatment analysis). Because of differing clinical characteristics among study groups, varying sample sizes, and different control treatments, we assumed that heterogeneity was present even when not statistically significant, and we decided to combine data by using a random-effects model (31) whenever the {tau}2 value (a measure of the between-study variability of the treatment effect) was greater than 0 (32,33), to achieve more conservative pooled estimates.

For all outcomes, the pooled risk difference was computed with the method of DerSimonian and Laird (31). Summary point estimates and 95% CIs, {chi}2 statistics for heterogeneity, and {tau}2 values are reported. For statistically significant risk differences (that is, if the 95% CI of the summary risk difference did not include 0), the number needed to treat was calculated (34). The number needed to treat is the number of patients who must be treated to avoid one unfavorable event; it is calculated by using the reciprocal of the weighted difference in rates between treatment and control groups. The 95% CIs for number needed to treat also are reported.

To determine the consistency of results, two sensitivity analyses were performed according to type of publication (abstract or full article) and control treatment. The data were processed by using the META-ANALYST program (version 0.989; Lau J, Boston, Mass).


   RESULTS
TOP
 Abstract
 Introduction
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 Appendix 1
 References
 
Reports of 11 pertinent randomized clinical trials were retrieved; these trials included a total of 750 patients: 372 treated with TIPS and 378 treated conventionally. Clinical and methodological characteristics of the 11 trials are summarized in Tables 1 and 4. In one trial (24), only patients assigned to Child-Pugh class C were enrolled. The time from cessation of bleeding to random assignment to treatment was less than 4 days in six trials (17,19,21,22,24,27), less than 3 weeks in three trials (18,20,25), and 1 day to 6 months in one trial (23) (Table 1).

TIPS creation was compared with sclerotherapy in five trials (17,20,21,23,24), with sclerotherapy and propranolol administration in three trials (18,19,24), and with band ligation in three trials (22,26,27). In one trial (18), eventual embolization of the varices was performed in half the patients in the TIPS group. The mean follow-up was 15.9 months in the TIPS groups and 16.3 months in the conventional treatment groups.

Treatment for episodes of recurrent bleeding was reported to be the same in the two treatment groups in one trial (19). In two trials (20,22), emergency sclerotherapy was performed in both groups, and balloon dilatation and/or further stent placement were also performed in the TIPS group. In the other trials, no information about treatment for recurrent bleeding was reported. The definition of death due to recurrent bleeding was reported in only two trials (19,23), although the number of patients who died of recurrent bleeding was reported in six of the seven studies published as full articles. The definitions of the other three outcomes in this meta-analysis were fairly comparable across the trials described in full reports (Table 4).

The quality score for the seven trials published as full reports and assessed for methodological quality ranged from 51% to 88%, which was indicative of fair to good quality. Details on the single components of the quality score for each trial are reported in the Appendix.

No significant heterogeneity of treatment effect was found for any of the assessed outcomes (even when the analyses were repeated by using odds ratios [data not shown]), and the {tau}2 value was greater than 0 in all analyses.

Technical Results and Complications of Treatment
TIPS creation was successful in 90%–100% (median, 98.5%) of patients. The reduction of the portosystemic pressure gradient ranged from 10.3 to 16.2 mm Hg (median, 13.0 mm Hg). At stent surveillance (Table 4), narrowing or occlusion of the stent lumen, defined as stent dysfunction (Table 4) and necessitating shunt revision, was found in 18%–76% (median, 55%) of patients (Table 2).

Endoscopic sclerotherapy or banding ligation of varices resulted in variceal obliteration in 51%–80% (median, 66%) of patients. Recurrence of varices in endoscopically treated patients was reported in three studies and was 28% (20), 32% (19), and 55% (17).

Complications of treatment published in the seven full reports are summarized in Table 5. After TIPS creation, 60 of 268 (22%) patients experienced complications, the most serious of which were hemorrhage in nine patients (intraperitoneal in three, intrahepatic in two, and hemobilia in four) and infections in 23 patients. Complications of endoscopic therapy occurred in 61 of 272 (22%) patients, including bleeding in 13 patients (due to esophageal ulcer in eight and immediately after a sclerotherapy session in five); infection in 20; and dysphagia in 14, nine of whom had esophageal stenosis necessitating dilation.


View this table:
[in this window]
[in a new window]
  		TABLE 5. Treatment Complications in Seven Randomized Clinical Trials Published as Full Reports
 
Recurrent Bleeding
Data on recurrent bleeding were available for all 11 trials (17–27) (Table 3). In five (17–19,21,22) of the full reports, the rate of recurrent bleeding from esophageal varices was reported; in one of these (18), the total number of patients who experienced recurrent bleeding from any source also was reported. In the remaining two studies (20,23), the rate of recurrent bleeding from any source was reported: In one of these (23), the mean number of episodes of recurrent bleeding from esophageal varices (but not the number of patients with recurrent bleeding from esophageal varices) also was reported. No detailed information on the source of recurrent bleeding was reported in the studies published in abstract form.

TIPS creation resulted in a significant reduction in the rate of recurrent bleeding in nine trials (17–19,21–23,25–27) and a nearly significant reduction in another (24) (Fig 1), whereas there was no effect in one trial (20). The overall weighted recurrent bleeding rate was 21% for TIPS creation and 52% for endoscopic treatment. The pooled risk difference was -31% (95% CI, -39%, -23%) (Fig 1). The {chi}2 value for heterogeneity was 11.79 (df = 10, P > .25). The "number needed to treat" (ie, number of patients who need to be treated with TIPS to prevent one episode of recurrent bleeding) was 3.3 (95% CI, 2.6, 4.4), which indicates that, compared to conventional therapy, two to five patients needed to be treated with TIPS to prevent one recurrent bleeding episode.



View larger version (20K):
[in this window]
[in a new window]
  		Figure 1. Point estimates (•) and 95% CIs (horizontal bars) of the risk difference between TIPS creation and conventional treatment show that TIPS creation significantly reduces the risk of recurrent bleeding. Risk differences represented to the left of the 0-difference line (the equivalence line) denote an advantage for TIPS creation; those to the right denote an advantage for conventional therapy (CT). CI bars that do not cross the 0-difference line denote a statistically significant advantage. "Overall" denotes the pooled risk difference. Trials reported in full articles are listed before those reported in abstracts, and all are listed in ascending order of year of publication. Each trial is identified by the name of the first author. z = z value, 2P = two-sided P value, #Pts = number of patients included in each trial.

 
Encephalopathy
The encephalopathy rate was reported in 10 studies (17–23,25–27) involving 685 patients, 340 treated with TIPS creation and 345 treated with conventional therapy (Table 3). TIPS creation increased the frequency of encephalopathy in all but one trial (25); we found a statistically significant increase in three trials (18,19,23) (Fig 2) and a nearly significant increase in two (17,20). The overall weighted encephalopathy rate was 35% after TIPS creation and 19% after conventional treatment. The pooled risk difference was 16% (95% CI, 10%, 22%). The {chi}2 value for heterogeneity was 6.87 (df = 9, P > .5). The number needed to treat was 6.0 (95% CI, 4.5, 10.0), which indicates that for every four to 10 patients treated with TIPS, one more episode of encephalopathy occurs more than if they had been treated conventionally.



View larger version (19K):
[in this window]
[in a new window]
  		Figure 2. Point estimates (•) and 95% CIs (horizontal bars) of the risk difference between TIPS creation and conventional treatment show that TIPS creation significantly increases the risk of portosystemic encephalopathy. Risk differences represented to the left of the 0-difference line (the equivalence line) denote an advantage for TIPS creation; those to the right denote an advantage for conventional therapy (CT). CI bars that do not cross the 0-difference line denote a statistically significant advantage. "Overall" denotes the pooled risk difference. Trials reported in full articles are listed before those reported in abstracts, and all are listed in ascending order of year of publication. Each trial is identified by the name of the first author. z = z value, 2P = two-sided P value, #Pts = number of patients included in each trial.

 
Mortality
Mortality data were reported for all 11 trials (17–27) (Table 3). TIPS creation resulted in a significant reduction in the mortality rate in one study (25) and had no significant effect in the others (Fig 3). The overall weighted mortality rate was 28% for TIPS creation and 26% for endoscopic treatment. The pooled risk difference was +2% (95% CI, -4%, +9%) (Figure 3). The {chi}2 value for heterogeneity was 9.53 (df = 10, P > .25). Causes of death are reported in Table 6.



View larger version (20K):
[in this window]
[in a new window]
  		Figure 3. Point estimates (•) and 95% CIs (horizontal bars) of the risk difference between TIPS creation and conventional treatment show that TIPS creation does not affect the risk of death due to all causes. Risk differences represented to the left of the 0-difference line (the equivalence line) denote an advantage for TIPS creation; those to the right denote an advantage for conventional therapy (CT). CI bars that do not cross the 0-difference line denote a statistically significant advantage. "Overall" denotes the pooled risk difference. Trials reported in full articles are listed before those reported in abstracts, and all are listed in ascending order of year of publication. Each trial is identified by the name of the first author. z = z value, 2P = two-sided P value, #Pts = number of patients included in each trial.

 

View this table:
[in this window]
[in a new window]
  		TABLE 6. Cause of Death in Full Reports of Randomized Clinical Trials
 
Death due to Recurrent Bleeding
Data from eight trials (17–20,22–25) involving 594 patients, 294 treated with TIPS creation and 299 treated conventionally, were available for this end point (Table 3). TIPS creation resulted in a significant reduction in the mortality rate due to recurrent bleeding in two studies (19,25), a nearly significant reduction in one (17), and no effect in the other five (18,20,22–24) (Fig 4). The overall weighted recurrent bleeding–related mortality rate was 5% for TIPS creation and 10% for endoscopic treatment. The pooled risk difference showed a trend (-5% [95% CI, -11%, +6%]) toward a reduction in mortality rate due to bleeding after TIPS creation, but this result was not statistically significant (Fig 4). The {chi}2 value for heterogeneity was 13.74 (df = 7, P > .05).



View larger version (17K):
[in this window]
[in a new window]
  		Figure 4. Point estimates (•) and 95% CIs (horizontal bars) of the risk difference between TIPS creation and conventional treatment show that TIPS creation does not significantly reduce the risk of death due to recurrent bleeding. Risk differences represented to the left of the 0-difference line (the equivalence line) denote an advantage for TIPS creation; those to the right denote an advantage for conventional therapy (CT). CI bars that do not cross the 0-difference line denote a statistically significant advantage. "Overall" denotes the pooled risk difference. Trials reported in full articles are listed before those reported in abstracts, and all are listed in ascending order of year of publication. Each trial is identified by the name of the first author. z = z value, 2P = two-sided P value, #Pts = number of patients included in each trial.

 
Treatment Crossover
In nine trials (17–23,26,27) involving 648 patients, the number of patients who underwent conventional treatment after TIPS creation or who underwent TIPS creation because of failure of conventional treatment (ie, treatment crossover) was reported. Of 321 patients randomly assigned to undergo TIPS creation, eight (2%) also underwent conventional treatment mainly because of failure to place the stent. Of 327 patients randomly assigned to undergo conventional treatment, 58 (18%) also underwent TIPS creation mainly because of uncontrolled or recurrent bleeding. Indications for crossover treatment in the trials published as full reports are summarized in Table 4.

Sensitivity Analyses
Sensitivity analyses showed consistency of results for type of publication (full report or abstract) and type of conventional treatment (sclerotherapy, sclerotherapy with propranolol administration, or endoscopic band ligation). However, the increased risk of encephalopathy after TIPS creation was not statistically significant in abstract reports and in trials in which banding ligation was the conventional therapy. Results of the sensitivity analyses are shown in Table 7.


View this table:
[in this window]
[in a new window]
  		TABLE 7. Results of Sensitivity Analyses
 

   DISCUSSION
TOP
 Abstract
 Introduction
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
Appendix 1
 References
 
In the present meta-analysis, we reviewed 11 randomized clinical trials in which TIPS creation was compared with endoscopic therapy (sclerotherapy in five trials, sclerotherapy plus administration of ß-blockers in three, and banding ligation in three) for the prevention of recurrent variceal bleeding in patients with cirrhosis. Overall results showed that TIPS creation significantly reduced the risk of recurrent bleeding, with no effect on survival. TIPS creation significantly increased the risk of encephalopathy, however. Sensitivity analysis results suggested that this interpretation holds true even if we considered only those trials published as full reports and when TIPS creation was compared with sclerotherapy, with sclerotherapy plus ß-blockers, or with banding ligation of esophageal varices. In fact, the lack of a statistically significant increase in the risk of encephalopathy after TIPS compared with that after banding ligation probably reflects a type II error due to insufficient sample size.

Although no significant heterogeneity was found for any of the assessed outcomes, differences in the ways in which the trials were conducted and in the methods used suggest that some caution is warranted as regards interpretation of the results of the pooled estimates. Although inclusion of data from trials published in abstract form (24–27) did not significantly affect the overall estimates of the effect of TIPS creation, the combinability of the included trials should be reassessed after full reports of all have been published.

The beneficial effect of TIPS creation on recurrent bleeding was observed in all but one of the trials. The average rate of recurrent bleeding was 20% in patients treated with TIPS creation, which suggests that this interventional radiologic procedure is highly effective for the prevention of recurrent bleeding. A comparably low frequency of recurrent bleeding was reported (35) only in trials where portosystemic shunts were surgically created. However, the comparability of the effect on recurrent bleeding rates associated with TIPS creation and that associated with surgical creation of other types of portosystemic shunt should be reassessed after longer follow-up in patients treated with TIPS creation.

Despite the low frequency of recurrent bleeding in patients treated with TIPS creation, the median rate of stent dysfunction reported in the seven full articles (17–23) was 55%. This finding suggests that other factors may contribute to the precipitation of recurrent bleeding. Of interest in this regard, Sanyal et al (4) recently reported that active alcoholism and high systemic venous pressure together with stent dysfunction are independent risk factors for recurrent bleeding in patients treated with TIPS creation. The wide variability in dysfunction rates across the studies is accounted for by the different definitions and surveillance modalities used, with higher rates found when dysfunction is defined on the basis of invasive hemodynamic measurements. A 2-year incidence of TIPS dysfunction of up to 90% has been reported (4) when invasive hemodynamic measurement was used. The best approach to monitoring of TIPS function in clinical practice is still unsettled, however, and further studies specifically designed for this aim are needed.

The most important side effect of TIPS creation is encephalopathy. Experience with the surgical creation of portosystemic shunts was predictive of this outcome (35). In fact, in hemodynamic terms, a TIPS behaves like a side-to-side portosystemic shunt: The greater the diversion of blood into the systemic circulation, the greater the reduction in portal venous pressure and, hence, the greater the protection against bleeding and the greater the risk of encephalopathy. For this reason, portosystemic shunt surgery, either total or selective, was almost completely abandoned in the past 2 decades. Other than ease of performance, TIPS creation has the advantage over shunt surgery of the ability to calibrate the shunt, which can be tailored in each patient to achieve the minimum portosystemic blood diversion with the optimal portal venous pressure reduction.

The 11 randomized clinical trials included in the present meta-analysis showed a significant increase in the frequency of encephalopathy in patients treated with TIPS creation. However, a conclusion on this crucial point is not possible for two important reasons. First, the optimal reduction in portal venous pressure still is undefined: A reduction below the threshold value of 12 mm Hg should represent maximal protection against the risk of bleeding (36). It has also been suggested (37) and prospectively validated (38) that a 20% reduction from baseline may be enough of a decrease; however, it is not known whether reductions of either more or less than 20% will provide better protection. In none of the 11 trials included in this meta-analysis was there an attempt to determine the optimal reduction in portal venous pressure, and the available data do not allow such an analysis.

The second reason we cannot draw conclusions about TIPS creation and encephalopathy is that encephalopathy was not blindly assessed in any of the studies included in this meta-analysis. Diagnostic bias, therefore, cannot be ruled out.

Thus, the frequency of encephalopathy after TIPS creation should probably be reassessed in further randomized clinical trials that have been more specifically designed to tailor portal venous pressure reduction to achieve the best decrease in the risk of bleeding with the lowest frequency of encephalopathy.

Until such studies are performed, it may be prudent to not perform TIPS creation or to use smaller-diameter stents in patients with known risk factors for encephalopathy, such as age older than 60 years, female sex, previous episodes of encephalopathy, nonalcoholic cause of cirrhosis, low albumin levels, and Child-Pugh class (6–8,17–19). Moreover, a portacaval pressure gradient of less than 10 mm Hg after TIPS creation is associated with a substantially higher risk of encephalopathy (8) and should be avoided. Finally, it is important to remark that in patients with severe and progressive encephalopathy, it is possible to reduce flow through the prosthesis by placing reducing stents or detachable balloons (39).

The overall incidence of complications that occurred during the reviewed trials was high and similar for the two types of treatment. Major complications were bleeding and sepsis for both, although bleeding was more frequent after endoscopic treatment, and sepsis was more frequent after TIPS creation. The available information does not allow assessment of whether the excess rate of encephalopathy after TIPS creation occurred in patients free of other complications. Thus, it is not possible to compare the numbers of patients without complications for each type of treatment.

The results of this meta-analysis showed that TIPS creation did not significantly affect survival, although the significant reduction in the rate of recurrent bleeding after TIPS creation was expected to result in a parallel reduction in deaths due to bleeding. In one study (25), a significant reduction in the risk of death was observed in patients treated with TIPS creation, but no data on cause of death are available because the study was reported in abstract form. In another study (20), a significant increase in the risk of death was reported; in this study, cumulative death proportions were compared by using the log rank test; however, the mortality rate related to recurrent bleeding after TIPS creation was unusually high (42% of all deaths). Overall, in patients who underwent TIPS creation, there was no significant trend toward a reduction in deaths due to bleeding, and there was an opposite trend toward an increase in deaths due to other causes. This combination of results seems to have balanced the overall mortality rate between the two treatment groups. The lack of a significant reduction in deaths caused by bleeding probably is due to type II error; also, the large number of patients who did not respond to conventional treatment and therefore crossed over to TIPS creation could have contributed to lessen this difference. However, whether TIPS creation has a substantial effect on cause of death should be reassessed when all the studies included in this meta-analysis have been published as full reports.

In conclusion, the results of this meta-analysis show that in patients with cirrhosis and previous variceal bleeding, TIPS creation is more effective than endoscopic therapy with or without propranolol administration for the prevention of recurrent bleeding but significantly increases the risk of encephalopathy and does not affect survival.

Because of the significant excess risk of encephalopathy, TIPS creation is far from the ideal therapy for prevention of recurrent bleeding and, in our opinion, cannot be considered a first-choice treatment. However, several considerations suggest that the efficiency of TIPS creation can probably be much improved. First, the optimal reduction in portal venous pressure to achieve the highest protection against bleeding and the lowest risk of encephalopathy still is unsettled. Second, there may be improvements in the selection criteria for patients at lower risk for encephalopathy who may benefit more from TIPS creation. Third, technical advances are expected to reduce the frequency of TIPS dysfunction. In some respects, further meta-analysis of the available trials, in which individual patient data are used, will be most helpful, and the authors who conduct such trials should be encouraged to combine their data.


   Appendix 1
TOP
 Abstract
 Introduction
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 Appendix 1
References
 
The following is the quality scoring system used in this meta-analysis. The number in parentheses for each item is the score assigned for that item. The score for each trial was reported as a percentage of the maximum possible score of 72 (Table A1).

  1. Randomization
    1. (6). By telephone contact with a central or independent source
    2. (4). By sealed opaque envelopes
    3. (0). Other or not reported
  2. Efficacy of randomization
    1. (6). Treated and control groups comparable for the following prognostic variables: age, sex, Child-Pugh class (or score), etiology of cirrhosis
    2. (3). Relative difference for Child-Pugh class (or score) >0.40, but not statistically significant
    3. (0). Child-Pugh class (or score) statistically different between TIPS or conventional treatment, or information not reported
  3. Time between randomization and treatment
    1. (6). Same in the two groups
    2. (0). Different interval of time in the two groups or no information reported
  4. Withdrawals
    1. (6). Intention-to-treat analysis; number of patients withdrawn from treatment and reasons of withdrawal reported; less than 15% of randomized patients lost to follow-up
    2. (3). One of the above conditions not fulfilled or information not reported
    3. (0). Two or all of the above conditions not fulfilled or information not reported
  5. Sample size
    1. (3). A priori estimate of sample size reported
    2. (0). A priori assessment of sample size not reported or not done
  6. Definition of treatment failure (as indication to other therapy)
    1. (6). Clear and reproducible
    2. (3). Incomplete information
    3. (0). No information reported
  7. Definition of hepatic encephalopathy
    1. (6). Clear and reproducible
    2. (3). Incomplete information
    3. (0). No information reported
  8. Definition of the necessity to modify the treatment
    1. (6). Clear and reproducible
    2. (3). Incomplete information
    3. (0). No information reported
  9. Treatment of the rebleeding
    1. (6). Same in the two groups
    2. (3). Information lacking for less than 20% of bleeding episodes
    3. (0). Information lacking for 20% or more than 20% of bleeding episodes. Different treatment in the two groups or no information reported
  10. Side effects and complications
    1. (6). Number, type, and outcome of side effects
    2. (3). Incomplete information
    3. (0). No information reported
  11. Prophylactic treatment of hepatic encephalopathy
    1. (6). Reported in the two groups
    2. (0). Reported only in one group or no information reported
  12. External validity
    1. (9). Inclusion and exclusion criteria clearly stated; adequate description of patient characteristics; source of population reported
    2. (6). Only two of the above conditions fulfilled
    3. (3). Only one of the above conditions fulfilled
    4. (0). No information reported


View this table:
[in this window]
[in a new window]
  		TABLE A1. Quality Scoring Components for Fully Reported Randomized Clinical Trials
 


   Acknowledgments
 
The present meta-analysis was carried out by using the META-ANALYST program (version 0.989, July 2, 1996) written by Joseph Lau, MD (Division of Clinical Care Research, New England Medical Center Hospitals, Boston, Mass). The authors thank Dr Lau for having made his program available.


   Footnotes
 
Abbreviation: TIPS = transjugular intrahepatic portosystemic shunt

Author contributions: Guarantors of integrity of entire study, A.L., G.D.; study concepts, A.L., G.D., L.P.; study design, G.D.; definition of intellectual content, A.L., G.D., L.P.; literature research, R.L.G., M.M.; data acquisition, R.L.G., M.M.; data analysis, A.L., G.D.; statistical analysis, A.M., G.D.; manuscript preparation and editing, G.D., A.L.; manuscript review, G.D.


   References
TOP
 Abstract
 Introduction
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 Appendix 1
 References
 

  1. Richter GM, Noeldge G, Palmaz JC, et al. Transjugular intrahepatic portocaval stent shunt: preliminary clinical results. Radiology 1990; 174:1027-1030.[Abstract]
  2. Lind CD, Malisch TW, Chong WK, et al. Incidence of shunt occlusion or stenosis following transjugular intrahepatic portosystemic shunt placement. Gastroenterology 1994; 106:1277-1283.[Medline]
  3. LaBerge JM, Somberg KA, Lake JR, et al. Two-year outcome following transjugular intrahepatic portosystemic shunt for variceal bleeding: results in 90 patients. Gastroenterology 1995; 108:1143-1151.[Medline]
  4. Sanyal AJ, Freedman AM, Luketic VA, et al. The natural history of portal hypertension after transjugular intrahepatic portosystemic shunts. Gastroenterology 1997; 112:889-898.[Medline]
  5. Barton RE, Rösch J, Saxon RR, Lakin PC, Petersen BD, Keller FS. TIPS: short and long-term results—a survey of 1750 patients. Semin Intervent Radiol 1995; 12:364-367.
  6. Somberg KA, Riegler JL, Doherty M, et al. Hepatic encephalopathy after transjugular intrahepatic portosystemic shunts: incidence and risk factors. Am J Gastroenterol 1995; 4:549-555.
  7. Sanyal AJ, Freedman AM, Shiffman ML, Purdum PP, III, Luketic VA, Cheatham AK. Portosystemic encephalopathy after transjugular intrahepatic portosystemic shunt: results of a prospective controlled study. Hepatology 1994; 20:46-55.[Medline]
  8. Riggio O, Merli M, Pedretti G, et al. Hepatic encephalopathy after transjugular intrahepatic portosystemic shunt: incidence and risk factors. Dig Dis Sci 1996; 41:578-584.[Medline]
  9. Sanyal AJ, Freedman AM, Purdum PP, et al. Transjugular intrahepatic portosystemic shunt (TIPS) vs sclerotherapy for prevention of recurrent variceal hemorrhage: a prospective trial (abstr). Gastroenterology 1994; 106:A975.
  10. Shiffman ML, Jeffers L, Hoofnagle JM, Tralka TS. The role of transjugular intrahepatic portosystemic shunt (TIPS) for treatment of portal hypertension and its complications: a conference sponsored by the National Digestive Advisory Board. Hepatology 1995; 22:1591-1597.[Medline]
  11. Pagliaro L, D'Amico G, Pasta L, et al. Efficacy and efficiency of treatments in portal hypertension. In: De Franchis R, eds. Portal hypertension II: proceedings of the Second Baveno International Consensus Workshop on Definitions, Methodology and Therapeutic Strategies. Oxford, England: Blackwell Science, 1996; 159-179.
  12. Sacks HS, Berrier J, Reitman D, Ancona-Berk VA, Chalmers TC. Meta-analyses of randomized controlled trials. N Engl J Med 1987; 316:450-455.[Abstract]
  13. Egger M, Smith GD, Phillips AN. Meta-analysis: principles and procedures. BMJ 1997; 315:1533-1537.[Free Full Text]
  14. Egger M, Smith GD. Meta-analysis: beyond the grand mean. BMJ 1997; 315:1610-1614.[Free Full Text]
  15. Pogue J, Yusuf S. Meta-analysis duet: overcoming the limitations of current meta-analysis of randomised controlled trials. Lancet 1998; 351:47-52.[Medline]
  16. Poynard T, Conn HO. The retrieval of randomized clinical trials in liver disease from the medical literature: a comparison of Medlars and manual methods. Control Clin Trials 1985; 6:271-279.[Medline]
  17. Cabrera J, Maynar M, Granados R, et al. Transjugular intrahepatic portosystemic shunt versus sclerotherapy in the elective treatment of variceal hemorrhage. Gastroenterology 1996; 110:832-839.[Medline]
  18. Rössle M, Deibert P, Haag K, et al. Randomised trial of transjugular-intrahepatic-portosystemic shunt versus endoscopy plus propranolol for prevention of variceal rebleeding. Lancet 1997; 249:1043-1049.
  19. Sauer P, Theilmann L, Stremmel W, Benz C, Richter GM, Stiehl A. Transjugular intrahepatic portosystemic stent shunt versus sclerotherapy plus propranolol for rebleeding. Gastroenterology 1997; 113:1623-1631.[Medline]
  20. Sanyal AJ, Freedman AM, Velimir A, et al. Transjugular intrahepatic portosystemic shunt compared with endoscopic sclerotherapy for the prevention of recurrent variceal hemorrhage: a randomized, controlled trial. Ann Int Med 1997; 126:849-857.[Abstract/Free Full Text]
  21. Cello JP, Ring EJ, Olcott EW, et al. Endoscopic sclerotherapy compared with percutaneous transjugular intrahepatic portosystemic shunt after initial sclerotherapy in patients with acute variceal hemorrhage: a randomized, controlled trial. Ann Int Med 1997; 126:858-865.[Abstract/Free Full Text]
  22. Jalan R, Forrest EH, Stanley AJ, et al. A randomized trial comparing transjugular intrahepatic portosystemic stent-shunt with variceal band ligation in the prevention of rebleeding from esophageal varices. Hepatology 1997; 26:1115-1122.[Medline]
  23. Merli M, Salerno F, Riggio O, et al. Transjugular intrahepatic portosystemic shunt versus endoscopic sclerotherapy for the prevention of variceal bleeding in cirrhosis: a randomized multicenter trial. Hepatology 1998; 27:40-45.
  24. Groupe d'Etude des Anastomoses Intra-Hépatic. TIPS vs sclerotherapy + propranolol in the prevention of variceal rebleeding: preliminary results of a multicenter randomized trial (abstr). Hepatology 1995; 22:A297.
  25. García-Villareal L, Martínez-Lagares F, Sierra A, et al. TIPS vs sclerotherapy for the prevention of variceal rebleeding: preliminary results of a randomized study (abstr). Hepatology 1996; 24:208A.
  26. Sauer P, Theilmann L, Benz C, Richter G, Stremmel W, Stiehl A. Transjugular intrahepatic portosystemic stent shunt versus endoscopic banding in the prevention of variceal rebleeding: interim analysis of a randomized study (abstr). Gastroenterology 1997; 111:A1374.
  27. Pomier-Layrargues G, Dufresne MP, Bui B, et al. TIPS vs endoscopic variceal ligation in the prevention of variceal rebleeding in cirrhotic patients: a comparative randomized clinical trial (interim analysis) (abstr). Hepatology 1997; 26:A137.
  28. Pagliaro L, D'Amico G, Sörensen TIA, et al. Prevention of first bleeding in cirrhosis: a meta-analysis of randomized trials of nonsurgical treatment. Ann Int Med 1992; 117:59-70.
  29. Laine L, Cook D. Endoscopic ligation compared with sclerotherapy for treatment of esophageal variceal bleeding: a meta-analysis. Ann Intern Med 1995; 123:280-287.[Abstract/Free Full Text]
  30. Bernard B, Lebrec D, Mathurin P, Opolon P, Poynard T. Propranolol and sclerotherapy in the prevention of gastrointestinal rebleeding in patients with cirrhosis: a meta-analysis. J Hepatol 1997; 26:312-324.[Medline]
  31. DerSimonian R, Laird N. Meta-analysis of randomized clinical trials. Control Clin Trial 1986; 7:177-188.[Medline]
  32. Whithead A, Whithead J. A general pragmatic approach to the meta-analysis of randomized clinical trials. Stat Med 1991; 10:1665-1677.[Medline]
  33. Lau J, Ioannidis JPA, Schmid HC. Summing up evidence: one answer is not always enough. Lancet 1998; 351:123-127.[Medline]
  34. Laupacis A, Sackett DL, Roberts AS. An assessment of clinically useful measures of the consequences of treatment. N Engl J Med 1988; 318:1728-1733.[Medline]
  35. D'Amico G, Pagliaro L, Bosch J. The treatment of portal hypertension: a meta-analytic review. Hepatology 1995; 22:332-354.[Medline]
  36. Groszmann RJ, Bosch J, Grace ND, et al. Hemodynamic events in a prospective randomized trial of propranolol versus placebo in the prevention of a first variceal hemorrhage. Gastroenterology 1990; 99:1401-1407.[Medline]
  37. Feu F, García-Pagán JC, Bosch J, et al. Relation between portal pressure response to pharmacotherapy and risk of recurrent variceal haemorrhage. Lancet 1995; 346:1056-1059.[Medline]
  38. Villanueva C, Balanzó J, Novella MT, et al. Nadolol plus isosorbide mononitrate compared with sclerotherapy for the prevention of variceal rebleeding. N Engl J Med 1996; 334:1624-1629.[Abstract/Free Full Text]
  39. Hauensten KH, Haag K, Ochs A, Lange M, Rössle M. The reducing stent: treatment for transjugular intrahepatic portosystemic shunt–induced refractory hepatic encephalopathy and liver failure. Radiology 1995; 194:175-179.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 QJMHome page
S. Masson, H.A. Mardini, J.D. Rose, and C.O. Record
Hepatic encephalopathy after transjugular intrahepatic portosystemic shunt insertion: a decade of experience
QJM, June 1, 2008; 101(6): 493 - 501.
[Abstract] [Full Text] [PDF]

Home page
 CMAJHome page
N. Dib, F. Oberti, and P. Cales
Current management of the complications of portal hypertension: variceal bleeding and ascites.
Can. Med. Assoc. J., May 9, 2006; 174(10): 1433 - 1443.
[Abstract] [Full Text] [PDF]

Home page
 GutHome page
D Tripathi, H F Lui, A Helmy, K Dabos, E Forrest, A J Stanley, R Jalan, D N Redhead, and P C Hayes
Randomised controlled trial of long term portographic follow up versus variceal band ligation following transjugular intrahepatic portosystemic stent shunt for preventing oesophageal variceal rebleeding
Gut, March 1, 2004; 53(3): 431 - 437.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
R. J. T. Owen, E. A. Ryan, K. O'Kelly, J. R. T. Lakey, M. C. McCarthy, B. W. Paty, D. L. Bigam, N. M. Kneteman, G. S. Korbutt, R. V. Rajotte, et al.
Percutaneous Transhepatic Pancreatic Islet Cell Transplantation in Type 1 Diabetes Mellitus: Radiologic Aspects
Radiology, October 1, 2003; 229(1): 165 - 170.
[Abstract] [Full Text] [PDF]

Home page
 GutHome page
L Bellis, E Moitinho, J G Abraldes, M Graupera, J C Garcia-Pagan, J Rodes, and J Bosch
Acute propranolol administration effectively decreases portal pressure in patients with TIPS dysfunction
Gut, January 1, 2003; 52(1): 130 - 133.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Roentgenol.Home page
P. Otal, T. Smayra, C. Bureau, J. M. Peron, V. Chabbert, P. Chemla, F. Joffre, J. P. Vinel, and H. Rousseau
Preliminary Results of a New Expanded-Polytetrafluoroethylene--Covered Stent-Graft for Transjugular Intrahepatic Portosystemic Shunt Procedures
Am. J. Roentgenol., January 1, 2002; 178(1): 141 - 147.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
G. Brancatelli, M. P. Federle, K. Pealer, and D. A. Geller
Portal Venous Thrombosis or Sclerosis in Liver Transplantation Candidates: Preoperative CT Findings and Correlation with Surgical Procedure
Radiology, August 1, 2001; 220(2): 321 - 328.
[Abstract] [Full Text] [PDF]

Home page
 GutHome page
A K BURROUGHS, D W PATCH, R JALAN, and P C HAYES
Management of variceal haemorrhage in cirrhotic patients Reply
Gut, May 1, 2001; 48(5): 738a - 740.
[Full Text] [PDF]

Home page
 Arch SurgHome page
W. S. Helton, R. Maves, K. Wicks, and K. Johansen
Transjugular Intrahepatic Portasystemic Shunt vs Surgical Shunt in Good-Risk Cirrhotic Patients: A Case-Control Comparison
Arch Surg, January 1, 2001; 136(1): 17 - 20.
[Abstract] [Full Text] [PDF]

Home page
 SEMIN CARDIOTHORAC VASC ANESTHHome page
S. M. Daves
The Liver and Vascular Surgery
Seminars in Cardiothoracic and Vascular Anesthesia, November 1, 2000; 4(4): 275 - 288.
[Abstract] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions