Comparative efficacy of early TIPS, Non-early TIPS, and Standard treatment in patients with cirrhosis and acute variceal bleeding: a network meta-analysis

Background: Cirrhosis is a chronic disease characterized by chronic liver inflammation and diffuse fibrosis. A combination of vasoactive drugs, preventive antibiotics, and endoscopy is the recommended standard treatment for patients with acute variceal bleeding; however, this has been challenged. We compared the effects of early transjugular intrahepatic portosystemic shunt (TIPS), non-early TIPS, and standard treatment in patients with cirrhosis and acute variceal bleeding. Materials and Methods: The present network meta-analysis was conducted in accordance with the criteria outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses and Assessing the methodological quality of systematic reviews guidelines. The review has been registered with the International Prospective Register of Systematic Reviews. The PubMed, Embase, Cochrane Library, ClinicalTrials.gov, and World Health Organization-approved trial registry databases were searched for randomized controlled trials (RCTs) evaluating early TIPS, non-early TIPS, and standard treatment in patients with cirrhosis and acute variceal bleeding. Results: Twenty-four RCTs (1894 patients) were included in the review. Compared with standard treatment, early TIPS [odds ratio (OR), 0.53; 95% credible interval (Cr), 0.30–0.94; surface under the cumulative ranking curve (SUCRA), 98.3] had a lower risk of all-cause mortality (moderate-to-high-quality evidence), and early TIPS (OR, 0.19; 95% CrI, 0.11–0.28; SUCRA, 98.2) and non-early TIPS (OR, 0.30; 95% CrI, 0.23–0.42; SUCRA, 1.8) were associated with a lower risk of rebleeding (moderate-to-high-quality evidence). Early TIPS was not associated with a reduced risk of hepatic encephalopathy, and non-early TIPS (OR, 2.78; 95% CrI, 1.89–4.23, SUCRA, 0) was associated with an increased incidence of hepatic encephalopathy (moderate-to-high-quality evidence). There was no difference in the incidence of new or worsening ascites (moderate-to-high-quality evidence) among the three interventions. Conclusion: Based on the moderate-to-high quality evidence presented in this study, early TIPS placement was associated with reduced all-cause mortality [with a median follow-up of 1.9 years (25th–75th percentile range 1.9–2.3 years)] and rebleeding compared to standard treatment and non-early TIPS. Although early TIPS and standard treatment had a comparable incidence of hepatic encephalopathy, early TIPS showed superiority over non-early TIPS in this aspect. Recent studies have also shown promising results in controlling TIPS-related hepatic encephalopathy. However, it is important to consider individual patient characteristics and weigh the potential benefits against the risks associated with early TIPS. Therefore, we recommend that clinicians carefully evaluate the patient’s condition, considering factors such as severity of variceal bleeding, underlying liver disease, and overall clinical status, before making a treatment decision. Further well-designed RCTs comparing early TIPS with non-early TIPS are needed to validate these findings and provide more definitive guidance.


Introduction
Cirrhosis is a chronic disease characterized by chronic liver inflammation and diffuse fibrosis.Consequently, the normal liver structure is replaced by regenerative liver nodules, leading to portal hypertension [1] .Gastrointestinal bleeding is a serious complication of portal hypertension and is the most common cause of bleeding; it is also the main cause of death in patients with liver cirrhosis [2] .The prevalence of esophageal varices in patients with liver cirrhosis is 50-60%, increasing to as high as 85% in patients with decompensated cirrhosis, and the risk of death is 15-30% within 6 weeks, especially higher when infection is present [3,4] .Therefore, early control of variceal bleeding is essential to reduce mortality in patients with non-transplant cirrhosis [5] .At present, a combination of vasoactive drugs, preventive antibiotics, and endoscopic techniques is the recommended standard treatment for patients with acute variceal bleeding [6,7] .
However, the standard treatment has been challenged because transjugular intrahepatic portosystemic shunt (TIPS) has been demonstrated by several original experiments to be superior to the standard treatment in rebleeding prevention.Yet, early TIPS has gained wide application in recent years, as patients undergoing TIPS treatment have demonstrated clear benefits in reducing allcause mortality and incidence of hepatic encephalopathy [5][6][7] .This may be related to the timing of TIPS implementation or severity of underlying liver disease.Specifically, cases in which TIPS was ineffective or inadequate included those of patients who died before receiving TIPS, patients who died soon after rescue TIPS, patients deemed too ill for TIPS due to liver failure, and patients in whom TIPS did not sufficiently reduce portal pressures [5,7] .
There is still some controversy regarding the application of early TIPS in cirrhotic portal hypertension.Clinicians still have some confusion when choosing this treatment, and there is an urgent need for more reliable evidentiary support from evidence-based medicine research.As a systematic review and meta-analysis, this study aims to comprehensively summarize the high-quality published research to provide more clear evidence-based medical evidence for clinical decision-making.
Early TIPS is TIPS performed within 72 h after diagnostic endoscopy [5] .With the recent introduction and application of early TIPS, current studies have shown its advantages over the standard treatment of all-cause mortality [5,7,8] .However, three network meta-analyses (NMAs) [9][10][11] only explained the effects of endoscopic therapy, drug therapy, and surgery on mortality and adverse events and paid little attention to TIPS, either including only one article or not including any.Furthermore, there was no discussion on early TIPS.
This study aimed to compare the effects of early TIPS, nonearly TIPS, and standard treatment in patients with cirrhosis and acute variceal bleeding.Owing to the paucity of direct comparison studies for some interventions, we used NMA to enable indirect comparisons across all treatments, allowing comprehensive utilization of all study data to assess the relative efficacy of the interventions, and it allows for direct comparisons between early TIPS and non-early TIPS, as well as indirect comparisons with the standard treatment group.By including studies where the control group received standard treatment, the NMA approach enables a larger sample size than traditional meta-analyses, providing a more robust analysis.

Protocol
The present NMA was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [12] and Assessing the Methodological Quality of Systematic Reviews (AMSTAR) [13] Guidelines.The protocol of this review has been registered with the International Prospective Register of Systematic Reviews, but due to the double-blind peer review process, we are unable to provide the registration number.The NMA component of this study was designed and conducted by a statistical expert, an author with epidemiology and biostatistics degrees and extensive experience in NMA.

Literature search
Through computer and manual search of the PubMed, Embase, Cochrane Library, ClinicalTrials.gov,and World Health Organization-approved trial registry databases, the search time was the establishment of the database to 31 October 2022; the search terms were "TIPS" and "transjugular intrahepatic portosystemic shunt", among others.No language restrictions were applied.The detailed search strategy is shown in SDC, Supplementary Material.

Inclusion criteria
(1) Patients: Patients with cirrhosis and acute variceal bleeding.
(2) Intervention: Early TIPS, defined as early TIPS, which included all patients who underwent TIPS treatment within 72 h after the diagnostic endoscopy regardless of whether they received endoscopic sclerotherapy or any other form of hemostatic treatment during the initial endoscopic procedure.(3) Comparator: Non-early TIPS, refers to TIPS procedures performed beyond the 72-h timeframe following the diagnostic endoscopy.It includes patients who may have received endoscopic sclerotherapy or other hemostatic treatments during the initial endoscopic procedure but subsequently underwent TIPS treatment after the 72-h window.Standard treatment, defined as endoscopic techniques plus drug treatment or endoscopic techniques [7] .(4) Outcomes: The main outcomes were all-cause mortality and rebleeding.The secondary outcome was hepatic encephalopathy and new or worsening ascites.(5) Study designs: Randomized controlled trials (RCTs).

HIGHLIGHTS
Early transjugular intrahepatic portosystemic shunt (TIPS) has a better therapeutic effect in patients with cirrhosis and acute variceal bleeding.Early TIPS is associated with a lower risk of rebleeding and mortality than standard treatment or non-early TIPS.Early TIPS is associated with a lower risk of hepatic encephalopathy than non-early TIPS.

Exclusion criteria
(1) Non-RCTs, cohort studies, crossover trials, case-control studies, retrospective studies, protocols, reviews, case reports, and ongoing trials lacking results.(2) Studies with an unidentified or unclear design or duplicate publications.
(3) The time from bleeding to TIPS was not described.(4) Full text was unavailable.

Data extraction and risk of bias assessment
The data extraction process was conducted by two independent reviewers who adhered to the eligibility criteria and reached a consensus in cases of discrepancy.The data extracted from all eligible studies, including the first author, publication year, country, sample size, age of participants, intervention type, Child-Pugh class, bleeding to TIPS time, and follow-up time, were systematically documented.In case of incomplete information, the corresponding author was contacted through e-mail to obtain any missing data.Any disagreements in the extracted data were resolved through discussion or, if necessary, adjudicated by a third reviewer.
Two reviewers independently used the Cochrane risk-of-bias tool 2 (RoB-2) [14] to determine the individual bias for all included RCTs, which contained the following five items: bias related to the randomization process, deviations from intended interventions, missing outcome data, outcome measurement, and selection of the reported result.Each item was classified as 'low risk', 'high risk', or 'some concerns'.To ensure the robustness of the risk of bias assessment, discrepancies were resolved through rigorous discussion between the reviewers or, if deemed necessary, by adjudication from a third reviewer.

Certainty of evidence
We used the Confidence in Network Meta-Analysis (CINeMA) tool to assess the risk of bias within individual studies and evaluate the confidence in treatment comparisons in our NMA [15,16] .The CINeMA tool employs a framework that considers six key domains of bias or uncertainty: bias within studies, reporting bias, indirectness, imprecision, heterogeneity, and incoherence.We used the summary risk of bias rating to evaluate within-study bias.The confidence rating was classified as 'high' (low risk of bias), 'moderate' (moderate risk of bias), 'low' (high risk of bias), or 'very low' (very high risk of bias).

Outcome definitions
The primary outcomes included all-cause mortality and rebleeding.The secondary outcomes were hepatic encephalopathy and new or worsening ascites.
All outcomes were binary, and the estimates are presented as odds ratio (OR) with 95% credible interval (CrI).Unlike traditional confidence intervals, which are based on frequentist statistical theory, CrIs are derived from Bayesian analysis theory.They incorporate both the information from the sample data and prior distribution information to express the uncertainty range of parameters.In this study, we chose to use CrIs because they excel at integrating multiple sources of information, including external data, and incorporating the researcher's prior knowledge into parameter estimation, which traditional confidence intervals do not possess.We employed a non-informative uniform prior distribution in our CrI estimation, which means that no subjective prior knowledge was included, ensuring objectivity in our analysis [17] .Treatment success was defined as an OR and 95% CrI of <1.Following this, we calculated the relative ranking of therapies and ranked all therapies using surface under the cumulative ranking curve (SUCRA) value (range 0-100%), with higher values indicating a higher likelihood of an intervention ranking higher in efficacy.

Pairwise meta-analysis
The present study employed the ADDIS software to conduct random-effects model pairwise meta-analyses of all comparisons, encompassing standard treatment, for every outcome of interest.In addition, heterogeneity analysis was performed using the aforementioned software.Specifically, an I 2 value of <40% was considered indicative of the low heterogeneity, whereas a higher I 2 value was considered indicative of heterogeneity [14] .

NMA
Network plots were created using StataSE 15.0 (64-bit), which presents the connection between different interventions.
R software was used to perform heterogeneity analysis [14] .Heterogeneity was assessed based on I 2 ; an I 2 value of <40% was considered indicative of low heterogeneity, whereas a higher I 2 value was considered indicative of heterogeneity [14] .
NMA was performed using GeMTC software package, which is a Bayesian NMA package.The primary analysis was conducted using a random-effects model, which allows for heterogeneity across studies and provides estimates of treatment effects in a more conservative manner.Node-splitting tests were initially employed to assess the consistency and transitivity of effect estimates.In cases where closed loops could not be formed, rendering the consistency analysis infeasible, a visual examination of direct and indirect effect estimates was conducted to identify any notable discrepancies [18] .The analyses were performed with four Markov chains, and a potential scale reduction factor (PSRF) analysis was subsequently performed to generate results [19] .

Subgroup and sensitivity analyses
In the subgroup analysis, we evaluated the impact of study period on outcomes by dividing the included RCTs into two subgroups based on the year of study publication, using 2007 as the cutoff point.This cutoff was chosen as it was the median publication year for all included RCTs.The two subgroups were pre-2007 RCTs before 2007 and post-2007 RCTs after 2007.This subgroup analysis was performed to assess whether improvements in techniques and experience over time influenced outcomes based on the publication year of trials.Sensitivity analyses were conducted using a change analysis model (fixed and random models).The implementation of sensitivity analyses was premeditated or undertaken upon the request of the reviewer or adherence to established protocol.

Publication bias
StataSE 15.0 (64-bit) was used to analyze potential publication bias across the included studies through the evaluation of comparisonadjusted funnel plots.Symmetric distribution of data points (representing individual studies) on both sides of the null line in a funnel plot indicates the absence of publication bias and small study effects [14] .

Study selection
In the preliminary search phase, a comprehensive search of relevant databases yielded a total of 1842 citations.Subsequently, an automated duplicate removal process was implemented, which resulted in the exclusion of 627 duplicate records.Next, 1174 articles were discarded through a meticulous screening process based on their title and abstract.Following this, a thorough assessment of the full texts of the remaining articles was conducted, leading to the exclusion of 17 articles.Among these, two articles were deemed ineligible because of the inability to retrieve the original manuscript or establish communication with the corresponding author to procure the necessary data.Consequently, a total of 24 RCTs [5,7,8, fulfilled the prespecified inclusion criteria and were subsequently incorporated into the NMA. The seection process for these studies is depicted in the PRISMA process (Fig. 1), providing a comprehensive visualization of the methodology employed for study inclusion.

Study characteristics
Twenty-four articles with 1894 participants were included in the study.Nine articles compared early TIPS with standard treatment, whereas 15 compared non-early TIPS with standard treatment.The duration of follow-up in the included studies was 1-2.63 years.The basic characteristics of the included studies are listed in SDC Table 1 (Supplemental Digital Content 1, http:// links.lww.com/JS9/B279).

Quality assessment
The RoB-2 findings for 24 eligible RCTs are shown in SDC Figure 1 (Supplemental Digital Content 1, http://links.lww.com/JS9/B279).All included studies were deemed to have a 'low risk' of bias regarding the random sequence generation.This was attributed to the detailed account provided by each study regarding their methods of randomization, such as employing a random number table, balanced block randomization, or software-generated randomized number.Moreover, all studies were appraised as having a 'low risk' of bias concerning the 'randomization process' as they explicitly delineated their allocation approach, which involved using sealed envelopes.

Rebleeding
Twenty-four articles [5,7,8, reported rebleeding. The thee nodes represent early TIPS, non-early TIPS, or standard treatment.The three line-linked nodes represent direct comparison, which were 9 for the early TIPS to standard treatment and 15 for the non-early TIPS to standard treatment (Fig. 2B).Upon conducting 50 000 simulations, the PSRF was determined to be 1.00.4B) reduced the risk of rebleeding (moderate-to-high-quality evidence; SDC Fig. 3, Supplemental Digital Content 1, http://links.lww.com/JS9/B279).Low heterogeneity was observed (I 2 = 0%; Fig. 4B).Moreover, given the inability to form loop connections, the assessment of consistency could not be operationalized, but the results from the consistency model and the inconsistency model showed negligible differences in the obtained ORs and confidence intervals.This implies an arguably high level of consistency across studies (SDC Table 2, Supplemental Digital Content 1, http://links.lww.com/JS9/B279).We found consistent results between the pairwise and NMAs (SDC Table 3, Supplemental Digital Content 1, http://links.lww.com/JS9/B279).

Subgroup and sensitivity analyses
In the subgroup analysis, we divided the studies into two groups, pre-2007 and post-2007, using the median publication year of all RCTs as the cutoff point.Subsequently, we conducted subgroup analyses for the primary outcome measures.For all-cause mortality, our results indicated no statistically significant differences between the pre-2007 and post-2007 groups when compared after standard treatment.However, when the results were pooled, early TIPS was superior to standard treatment.In the case of hepatic encephalopathy, early TIPS showed an advantage over standard treatment in the pre-2007 subgroup analysis, but the pooled results did not show a statistically significant difference.Overall, we did not find a significant effect of the publication year of trials on efficacy.Sensitivity analysis results aligned with the main results (SDC Tables 4, 5, Supplemental Digital Content 1, http://links.lww.com/JS9/B279).

Discussion
On the basis of moderate-to-high-quality evidence, we concluded that early TIPS is associated with reduced all-cause mortality and rebleeding.Although early TIPS and standard treatment had a comparable incidence of hepatic encephalopathy, early TIPS was superior to non-early TIPS in this regard.Some recent studies have also reported that TIPS-related hepatic encephalopathy can be controlled [41,42] .In addition, one study found that early TIPS placement improved survival at admission in patients at high risk for hepatic encephalopathy, suggesting that hemorrhage may be a precipitating factor for hepatic encephalopathy and that early TIPS placement helps control bleeding and thus address a precipitating factor for hepatic encephalopathy rapidly [43] .To date, there have been no trials comparing early TIPS and nonearly TIPS.Hence, this study, to our best knowledge, is the first, and the results show that early TIPS is superior, not only to standard treatment, but also to non-early TIPS.
Considering technological advancements, we hypothesized that later trials might yield slightly better results compared with earlier trials.However, our subgroup analysis did not observe this trend, possibly due to the small sample size.Overall, our study did not find the publication year of the trials significantly affected efficacy.Nevertheless, this potential effect is worth further evaluation in future studies with larger sample sizes.
Acute variceal bleeding frequently complicates cirrhosis and carries a high mortality risk, urgently requiring prompt and effective control methods that can significantly reduce mortality and complications [44,45] .For acute variceal bleeding, the current standard treatment remains a combination of vasoactive agents, prophylactic antibiotics, endoscopic therapy, and TIPS, which is often used as a salvage treatment or second-line therapy, but several multicenter studies have shown that only 7-10% of patients had early TIPS placement [46,47] .
Recently, the application of early TIPS has become a frequently discussed topic, and multiple RCTs have indicated that early TIPS reduces the risk of rebleeding from acute variceal bleeding in patients with cirrhosis compared with standard treatment.However, other studies have reported conflicting results regarding the effect of early TIPS and standard treatment on mortality and hepatic encephalopathy [7,8,40] .Moreover, three studies showed that early TIPS significantly improved survival in patients with cirrhosis and acute variceal bleeding compared to standard treatment.Significant survival benefits are particularly evident in high-risk patients (Child-Pugh B and Child-Pugh C) with active bleeding during endoscopy [48][49][50] , which is consistent with the conclusion of this study.However, both studies included fewer than five RCTs and included a large number of observational studies.Our study further found that early TIPS did not increase the risk of hepatic encephalopathy compared with standard treatment.A study published in 2022 indicated that early TIPS may not reduce mortality at 6 months, which contradicts our findings.It is worth noting that the meta-analysis published in 2022 included a limited number of studies, with only four RCTs out of nine included.The small sample size might explain the discrepancy in the results [51] .Regarding NMA, a recent study published in Cochrane [11] showed that compared with endoscopic treatment, TIPS has advantages in improving rebleeding but may increase the risk of other complications, which contradicts the conclusion of our study.However, that study included very few RCTs in which the intervention was TIPS, and there were no RCTs on early TIPS.
However, there are many clinical factors that should be considered when selecting TIPS: non-malignant portal vein thrombosis, refractory ascites, recurrent symptomatic hepatic hydrothorax, failed treatment of gastroesophageal varices with nonselective beta-blockers (NSBBs) and variceal ligation, Child class C with score <14 or Child class B with active bleeding, bridge treatment in patients listed for liver transplantation, transfusion-dependent portal hypertensive gastropathy cases in which NSBBs fail or are not tolerated, treatment of gastroesophageal varices type 2 or isolated varices type 1 with TIPS placement with or without embolization, hepatorenal syndrome type 2 associated with refractory/recidivant ascites, failure of anticoagulation and mechanical revascularization in Budd-Chiari syndrome cases with persistent ascites or other related complications, and patients with porto-sinusoidal vascular disorder [52,53] .
Before performing a TIPS procedure, it is essential to consider whether the patient exhibits the following factors: very advanced liver disease (Child-Pugh score > 13 points), recurrent episodes of overt hepatic encephalopathy without an identifiable precipitating factor, heart failure, pulmonary hypertension, hepatic vein obstruction, or anatomical alterations due to prior surgical procedures [54] .
Nonetheless, our study has some limitations.First, this study employed NMA to compare treatment effects, but this approach comes with inherent limitations related to the combination of diverse comparisons, such as heterogeneity arising from different studies that may affect the results.Additionally, the inclusion of a limited number of studies also restricts the applicability of the method.Second, it is difficult to quantify specific factors, such as inconsistencies in the measures of standard treatment across different eras and studies, such as variations in endoscopic hemostasis methods, and treatment intervals; these factors may introduce potential heterogeneity and affect the results, and they lead to unknown bias.Third, some studies with small sample sizes resulted in low power.Finally, the included clinical trials in this study exhibited variability in follow-up times, preventing us from conducting stratified analyses for different standardized follow-up time points.This could potentially affect the assessment of long-term efficacy.Future research should routinely report clinical outcomes at standard follow-up times to achieve more reliable comparisons of long-term efficacy.

Conclusion
Based on the moderate-to-high quality evidence presented in this study, early TIPS placement was associated with reduced allcause mortality [with a median follow-up of 1.9 years (25th-75th percentile range 1.9-2.3years)] and rebleeding compared to standard treatment and non-early TIPS.Although early TIPS and standard treatment had a comparable incidence of hepatic encephalopathy, early TIPS showed superiority over non-early TIPS in this aspect.Recent studies have also shown promising results in controlling TIPS-related hepatic encephalopathy.However, it is important to consider individual patient characteristics and weigh the potential benefits against the risks associated with early TIPS.Therefore, we recommend that clinicians carefully evaluate the patient's condition, considering factors such as the severity of variceal bleeding, underlying liver disease, and overall clinical status, before making a treatment decision.Further, welldesigned RCTs comparing early TIPS with non-early TIPS are needed to validate these findings and provide more definitive guidance.

Figure 1 .
Figure 1.PRISMA process.PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Figure 2 .
Figure 2. Network plots correlating TIPS and standard treatment for patients with cirrhosis and acute variceal bleeding.Each circle represents an intervention, called node.Node size is related to the number of patients receiving an intervention.The line between nodes represents a direct comparison, and its thickness is proportional to the number of tests for each comparison.TIPS, transjugular intrahepatic portosystemic shunt.

Figure 3 .
Figure 3. Forest plot of each treatment to the comparator.The plot shows the OR value and 95% CrI between each treatment to comparator; CrI, credible interval; OR, odds ratio; TIPS, transjugular intrahepatic portosystemic shunt.

Figure 4 .
Figure 4. SUCRA (surface under the cumulative ranking curve) value graph.The curves represent interventions.The larger the area under the curve, the higher the likelihood that the intervention is the optimal choice.TIPS, transjugular intrahepatic portosystemic shunt.