Efficacy and safety of 6-mm vs 8-mm covered stents for tips creation: A meta-analysis

Abstract

BACKGROUND

Hepatic encephalopathy (HE) remains a major complication of transjugular intrahepatic portosystemic shunt (TIPS) creation, occurring in approximately 30% of patients despite the widespread adoption of polytetrafluoroethylene (PTFE)-covered stents. Stent diameter influences the volume of portal blood diverted to the systemic circulation and may thereby modulate the risk of post-procedural HE. Whether 6-mm covered stents can reduce encephalopathy without compromising hemodynamic efficacy or survival relative to the standard 8-mm configuration is uncertain.

AIM

To compare the efficacy and safety of 6-mm vs 8-mm covered stents in adults undergoing TIPS for cirrhotic portal hypertension.

METHODS

A systematic review and meta-analysis was conducted in accordance with PRISMA 2020 guidance. MEDLINE, EMBASE, and the Cochrane Central Register of Controlled Trials were searched through the end of 2025. Studies comparing 6-mm and 8-mm PTFE-covered stents for TIPS creation in adults with cirrhotic portal hypertension were eligible. Risk of bias was assessed with RoB 2 for randomised trials and ROBINS-I for non-randomised studies. Pooled risk ratios (RR) and mean differences were estimated under a random-effects model. Certainty of evidence was rated with the Grading of Recommendations, Assessment, Development and Evaluations framework.

RESULTS

Five studies were included: One randomised controlled trial (Zhang et al) and four non-randomised cohort studies (Liu et al, Yan et al, Xia et al, Yao et al), enrolling a combined 2060 patients. The 6-mm stent was associated with a significant reduction in overall HE (RR = 0.56; 95%CI: 0.41-0.75; P = 0.0001; I² = 0%) and in overt HE (RR = 0.62; 95%CI: 0.41-0.93; P = 0.02; I² = 54%). Achievement of the target portal pressure gradient below 12 mmHg did not differ significantly between stent diameters (RR = 0.93; 95%CI: 0.83-1.03; P = 0.16; I² = 0%). Overall survival at 2 years (RR = 1.04; 95%CI: 0.96-1.13; P = 0.32), shunt dysfunction (RR = 0.97; 95%CI: 0.71-1.32; P = 0.83), and variceal rebleeding (RR = 1.28; 95%CI: 0.96-1.69; P = 0.09) were not significantly different.

CONCLUSION

A 6-mm effective shunt diameter was associated with a lower risk of HE after TIPS than an 8-mm diameter, without a detectable difference in survival, shunt patency, or portal decompression, and with a possible increase in variceal rebleeding. Because the certainty of evidence was low to very low and most data derived from non-randomised studies, these findings are hypothesis-generating and are not sufficient to recommend a preferred stent configuration; individualised consideration is required, particularly for patients at elevated baseline rebleeding risk.

Key Words: Transjugular intrahepatic portosystemic shunt; Stent diameter; Hepatic encephalopathy; Portal hypertension; Variceal rebleeding; Covered stent; Meta-analysis

Core Tip: This systematic review and meta-analysis of one randomised trial and four non-randomised studies compared 6-mm and 8-mm covered stents for transjugular intrahepatic portosystemic shunt. A 6-mm effective diameter was associated with a lower risk of hepatic encephalopathy, with no detectable difference in portal-pressure-gradient target achievement, survival or shunt dysfunction, and a possible increase in variceal rebleeding. Because the certainty of evidence was low to very low, these findings are hypothesis-generating and should not yet guide the choice of stent diameter in clinical practice.

INTRODUCTION

Transjugular intrahepatic portosystemic shunt (TIPS) is an established interventional procedure for managing complications of portal hypertension, most notably refractory variceal haemorrhage and refractory ascites[1,2]. By creating a low-resistance channel between the portal and hepatic venous systems, TIPS decompresses the portal circulation and prevents the sequelae of sustained portal hypertension. Since its introduction in 1989[3,4], the procedure has undergone substantial technical refinement, most consequentially the transition from bare-metal to polytetrafluoroethylene (PTFE)-covered stents, which dramatically improved long-term shunt patency and reduced the incidence of shunt dysfunction requiring reintervention[5,6]. Stent diameter, however, is not merely a technical detail: It directly determines the volume of portal blood diverted into the systemic circulation and therefore fundamentally modulates the risk of post-procedural hepatic encephalopathy (HE).

Despite these advances, HE remains the most clinically significant complication of TIPS, affecting approximately 30% of patients even in the modern covered-stent era[7,8]. Post-TIPS HE is driven by two interrelated mechanisms: The diversion of portal blood, carrying ammonia and other gut-derived neurotoxins, away from hepatic metabolism, and a reduction in residual hepatic parenchymal perfusion consequent to the shunt[9,10]. Because both mechanisms are directly proportional to shunt flow, the effective diameter of the implanted stent is a critical determinant of HE risk and has accordingly become a major focus of investigation.

The transition from 10-mm to 8-mm covered stents represented the first step in this diameter-reduction strategy. Randomised controlled trials (RCT) confirmed that 8-mm stents reduced post-TIPS HE compared with 10-mm stents without increasing variceal rebleeding or impairing survival[11,12], and this diameter shift accordingly became standard practice. Nevertheless, 8-mm covered stents continue to carry a meaningful residual HE burden[7,8], particularly in patients with advanced hepatic reserve impairment, high Child-Pugh or model of end stage liver disease (MELD) scores, or small atrophic livers, as are frequently encountered in hepatitis B-endemic regions[13,14]. These observations have renewed interest in further diameter reduction, either through deployment of true 6-mm covered stents or through the underdilation strategy, in which an 8-mm covered stent is intentionally expanded with a 6-mm balloon catheter to restrict effective shunt diameter.

Several recent studies have explored this question. Liu et al[15] first demonstrated that underdilation of 8-mm PTFE stents reduced HE without compromising portal pressure gradient (PPG) reduction or increasing rebleeding. Yan et al[14] evaluated a true 6-mm shunt strategy in patients with severe hepatic atrophy, reporting significantly lower overt HE incidence and preserved liver function. Yao et al[16] extended the underdilation finding to patients with prior splenectomy. Most recently, Zhang et al[17] conducted the first RCT directly comparing 6-mm and 8-mm covered TIPS stents, confirming a significantly lower incidence of overt HE with the 6-mm stent, albeit with a non-significant trend toward higher rebleeding.

Despite this emerging evidence base, the existing literature on 6-mm vs 8-mm TIPS stents remains scattered across heterogeneous study designs, patient populations, and follow-up durations, precluding straightforward cross-study comparison. Crucially, prior systematic reviews and meta-analyses in this area have focused almost exclusively on the 8-mm vs 10-mm comparison[11,12], establishing the central trade-off between HE reduction and haemodynamic adequacy at that diameter boundary, but the question of whether a further reduction from 8 mm to 6 mm can deliver additional encephalopathy benefit without an unacceptable loss of haemodynamic efficacy, increased rebleeding, higher shunt dysfunction rates, or worse survival has not been formally synthesised. Current guidelines-including the Baveno VII consensus, do not provide specific recommendations regarding optimal stent diameter for TIPS creation[1], leaving clinicians without evidence-based guidance for this technically consequential decision.

The present systematic review and meta-analysis was therefore conducted to fill this gap by quantifying, for the first time, the comparative effect of 6-mm vs 8-mm PTFE-covered stents, including both true 6-mm stents and 8-mm stents intentionally underdilated to an effective 6-mm diameter, on HE, portal pressure decompression, variceal rebleeding, shunt dysfunction, and overall survival in patients with cirrhotic portal hypertension undergoing TIPS creation[18].

MATERIALS AND METHODS

Protocol and registration

This systematic review and meta-analysis was designed and reported in accordance with the PRISMA 2020 statement[19]. The PRISMA flow diagram is provided as Supplementary Figure 1. PROSPERO registration code is CRD420261372870.

Eligibility criteria: (1) The study population comprised adult patients aged 18 years or older with cirrhotic portal hypertension undergoing elective or semi-elective TIPS creation; (2) The study compared a 6-mm PTFE-covered stent, or an 8-mm PTFE-covered stent intentionally underdilated to 6 mm using a 6-mm balloon catheter, against a standard 8-mm PTFE-covered stent expanded to its nominal diameter; (3) The study reported at least one pre-specified outcome of interest, including HE in any grade, shunt dysfunction, variceal rebleeding, PPG, or overall survival; and (4) The minimum follow-up duration was six months from the index TIPS procedure.

Studies were excluded if they were case reports, editorials, letters, or conference abstracts without full-text accompanying data. Studies evaluating uncovered or bare-metal stents, stents with a nominal diameter exceeding 8 mm in the comparator arm, or those that did not clearly distinguish stent diameter groups in their reported outcomes were also excluded. Studies evaluating paediatric populations or non-cirrhotic causes of portal hypertension were similarly excluded.

Search strategy

A systematic search was conducted in MEDLINE (via PubMed), EMBASE, and CENTRAL through March 2026 without language or date restrictions. MeSH terms and free-text keywords related to TIPS, portosystemic shunting, covered/PTFE stents, stent diameter (6 mm vs 8 mm), portal hypertension, and HE were combined using Boolean operators. Reference lists of included studies and relevant reviews were screened, and forward citation tracking was performed via Google Scholar. Authors were not contacted.

Study selection and data extraction

Records were deduplicated and screened independently by two reviewers at title/abstract and full-text levels using predefined criteria, with disagreements resolved by consensus or a third reviewer. Data were extracted using a standardized Excel form, including study characteristics, patient demographics, liver disease details, TIPS indication, baseline severity (Child-Pugh, MELD), PPG, stent details, follow-up, and outcomes. For underdilation studies, effective post-dilation diameter was recorded. Extraction was performed by one reviewer and verified by another.

Risk of bias assessment

The included RCT (Zhang et al[17]) was assessed using the Cochrane RoB 2 tool[20]. Cohort studies were evaluated using ROBINS-I[21] across seven bias domains, with overall ratings assigned. Additional quality assessment was performed using the Newcastle-Ottawa Scale (NOS)[22]. All assessments were conducted independently by two reviewers with consensus resolution.

Statistical analysis

All meta-analyses were conducted under a DerSimonian-Laird random-effects model[23] to account for anticipated between-study clinical and statistical heterogeneity. For dichotomous outcomes, the effect measure was the risk ratio (RR) with 95%CIs. For continuous outcomes, the effect measure was the mean difference (MD) with 95%CIs. Outcomes for which data were available from only one study were described narratively and not pooled. Statistical heterogeneity was quantified using the I² statistic, with values of 25%, 50%, and 75% representing low, moderate, and substantial heterogeneity, respectively[24]. For outcomes exhibiting moderate or substantial heterogeneity (I² ≥ 50%), pre-specified sensitivity analyses were conducted by sequentially excluding each contributing study and re-estimating the pooled effect to identify influential studies and assess robustness of the pooled estimate.

A pre-specified subgroup analysis for the primary outcome of HE was conducted by West Haven grade, stratifying events into Grade I-II (mild-to-moderate) and Grade III-IV (severe) categories. The test for subgroup differences was performed using a χ² test, with P < 0.10 taken as evidence of subgroup effect modification. All meta-analyses were performed in Review Manager (RevMan) version 5.4 (Cochrane Collaboration)[25] and visualised as forest plots. Statistical significance was defined as a two-tailed P value below 0.05 for all pooled analyses.

Certainty of evidence

The certainty of evidence for each pre-specified outcome was graded using the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) framework[26]. Evidence originating from RCT was initially rated as high certainty and could be downgraded by one or two levels for the following reasons: Serious or very serious risk of bias; inconsistency (unexplained heterogeneity, I² ≥ 50%); indirectness (population, intervention, or outcome differing importantly from the review question); imprecision (wide confidence intervals crossing the null or not excluding clinically meaningful effects); and publication bias (assessed informally by visual inspection of funnel plots where five or more studies contributed to a pooled analysis). Evidence from non-randomised studies was initially rated as low certainty and could be further downgraded or upgraded according to GRADE criteria. All GRADE judgements were made by consensus between reviewers.

RESULTS

Study characteristics

A systematic search identified 140 records, of which five met eligibility criteria (Liu et al[15], Yan et al[14], Xia et al[13], Yao et al[16], Zhang et al[17]; PRISMA flow in Supplementary Figure 1). The evidence base comprised one randomized controlled trial (Zhang et al[17]) and four non randomized cohort studies comparing TIPS creation with a 6 mm vs an 8 mm covered stent, with sample sizes ranging from 33 patients per arm (Yao et al[16]) to 1505 patients in the 8 mm arm of Xia et al[13] and follow up of at least 2 years across all studies. Baseline populations were broadly comparable, with cirrhosis of mixed etiology (predominantly viral hepatitis and alcohol related), Child Pugh class A and B distributions, and TIPS indications of refractory variceal bleeding or refractory ascites; the 6 mm and 8 mm groups within each study were comparable on reported parameters, with full baseline data summarized in Supplementary Table 1. Variability in case mix across studies, particularly the balance of Child Pugh class B vs C disease and of refractory ascites vs variceal hemorrhage, may contribute to between study clinical heterogeneity.

Quality assessment

The risk of bias assessment for the single randomized controlled trial (Zhang et al[17]) was performed using the Cochrane RoB 2 tool and is summarized in Supplementary Figure 2 (traffic light plot) and Supplementary Figure 3 (weighted bar chart). Zhang et al[17] was rated low risk in four of the five RoB 2 domains, including bias arising from the randomization process (D1), bias due to missing outcome data (D3), bias in measurement of the outcome (D4), and bias in selection of the reported result (D5). Some concerns were identified in the domain of bias due to deviations from intended interventions (D2), reflecting the unavoidable lack of blinding inherent to interventional radiology trials in which the operator and care team cannot be masked to stent diameter at the time of deployment. The overall risk of bias was therefore judged as some concerns.

The risk of bias assessment for the four non randomized cohort studies (Liu et al[15], Yan et al[14], Xia et al[13], and Yao et al[16]) was performed using the ROBINS-I tool and is summarized in Supplementary Figure 4 (traffic light plot) and Supplementary Figure 5 (weighted bar chart). The overall risk of bias was rated moderate in one study (Yao et al[16]) and serious in three studies (Liu et al[15], Xia et al[13], and Yan et al[14]). The principal sources of concern were bias due to confounding (D1) and bias in the selection of participants (D2), both reflecting inherent limitations of observational designs in which treatment allocation was not randomized and prognostic factors such as Child Pugh class, MELD score, and indication for TIPS may have differed systematically between groups. Xia et al[13] was additionally rated serious for bias due to missing data (D5), consistent with the registry style design and the imbalanced sample sizes between the two stent diameter groups. The remaining domains, including classification of interventions (D3), deviations from intended interventions (D4), measurement of outcomes (D6), and selection of the reported result (D7), were rated low risk across all four studies.

The four cohort studies were additionally evaluated using the NOS, with full marks awarded across the selection (4 of 4), comparability (2 of 2), and outcome (3 of 3) domains, yielding a total score of 9 out of 9 for each study and a classification of good quality (Supplementary Table 2). The discordance between the NOS classification and the ROBINS-I judgement reflects the differing sensitivities of the two tools, and the ROBINS-I judgements were therefore used to inform the GRADE certainty of evidence ratings, with NOS reported alongside for transparency.

Primary outcomes

Overall HE: Three of the five included studies reported this outcome. The pooled analysis for overall HE across all severity grades (Figure 1A) demonstrated a statistically significant reduction in risk with the 6 mm stent compared to the 8 mm stent (RR = 0.56; 95%CI: 0.41-0.75; P = 0.0001). The synthesized data exhibited complete statistical homogeneity (I² = 0%), with consistent directional effects across all three contributing trials. Zhang et al[17] contributed the largest weight (44.4%), followed by Yan et al[14] (33.7%) and Liu et al[15] (22.0%), consistent with an association between smaller stent diameter and a lower overall incidence of post-TIPS encephalopathy.

Figure 1 Forest plot. A: Risk ratio for overall hepatic encephalopathy, 6 mm vs 8 mm; B: Risk ratio for overt hepatic encephalopathy; C: Risk ratio for overall survival at 2 years; D: Risk ratio for achievement of target portal pressure gradient below 12 mmHg; E: Mean difference in post procedural portal pressure gradient (mmHg); F: Risk ratio for shunt dysfunction; G: Risk ratio for variceal rebleeding at 2 years.

HE by severity grade (subgroup analysis): The subgroup analysis stratified by West Haven grade (Supplementary Figure 6) demonstrated that the protective effect of the 6 mm stent was consistent across severity strata. In the Grade I-II subgroup, the pooled RR was 0.60 (95%CI: 0.42-0.87; P = 0.007; I² = 0%), and in the Grade III-IV subgroup the pooled RR was 0.53 (95%CI: 0.29-0.98; P = 0.04; I² = 0%). The test for subgroup differences was not significant (P = 0.72; I² = 0%), indicating no evidence of effect modification by encephalopathy severity. The benefit of smaller stent diameter therefore extended to both mild and severe presentations of post TIPS encephalopathy.

Overt HE: All five studies reported this outcome. The pooled analysis for overt HE (Figure 1B) demonstrated a statistically significant reduction in risk with the 6 mm stent (RR = 0.62; 95%CI: 0.41-0.93; P = 0.02). Moderate statistical heterogeneity was observed (I² = 54%), driven primarily by Yao et al[16], which was the only contributing trial directionally favoring the 8 mm stent (RR = 1.43). Xia et al[13] contributed the largest weight (30.7%), followed by Zhang et al[17] (22.8%), Liu et al[15] (16.5%), Yan et al[14] (15.3%), and Yao et al[16] (14.7%), with the remaining trial level estimates directionally favoring the 6 mm condition.

Overall survival at 2 years: All of the five included studies reported this outcome. The pooled analysis for overall survival at 2 years (Figure 1C) demonstrated no statistically significant difference between the two stent diameters (RR = 1.04; 95%CI: 0.96-1.13; P = 0.32). Moderate statistical heterogeneity was observed (I² = 47%), driven primarily by Xia et al[13], which contributed an unusually large 8 mm group (1505 patients) and reported a marginally significant directional effect favoring the 8 mm condition. Xia et al[13] contributed the largest weight (27.7%), followed by Liu et al[15] (21.1%), Yan et al[14] (20.2%), Zhang et al[17] (19.2%), and Yao et al[16] (11.8%). The pooled estimate confirmed survival equivalence between the two stent diameters at 2 years.

Secondary outcomes

Achievement of target PPG (PPG below 12 mmHg): Three of the five studies reported this outcome. The pooled analysis for achievement of target PPG (Figure 1D) demonstrated no statistically significant difference between the two stent diameters (RR = 0.93; 95%CI: 0.83-1.03; P = 0.16; I² = 0%). A total of 145 of 203 patients (71.4%) in the 6 mm group and 151 of 191 patients (79.1%) in the 8 mm group reached the hemodynamic target. Yan et al[14] contributed the largest weight (50.1%), followed by Liu et al[15] (26.4%) and Zhang et al[17] (23.5%), with all three trial level estimates near unity, supporting hemodynamic comparability between the two stent sizes.

Post procedural PPG (continuous outcome): Four of five studies reported this outcome. The pooled estimate for continuous post procedural PPG (Figure 1E) demonstrated no statistically significant difference between cohorts (MD = 0.82 mmHg; 95%CI: -1.18 to 2.81; P = 0.42). Due to substantial heterogeneity (I² = 85%), this pooled estimate should be interpreted with caution and is provided for exploratory purposes only. Substantial statistical heterogeneity was observed (I² = 85%), driven primarily by Zhang et al[17], which reported a notably larger mean difference of +3.50 mmHg in favor of the 8 mm group. Liu et al[15] (27.9%), Zhang et al[17] (27.0%), Yao et al[16] (24.4%), and Yan et al[14] (20.8%) contributed comparable weights, with directionally inconsistent trial level estimates that warranted formal sensitivity testing.

Shunt dysfunction

All five studies included this outcome. The pooled analysis for shunt dysfunction (Figure 1F) demonstrated no statistically significant difference between cohorts (RR = 0.97; 95%CI: 0.71-1.32; P = 0.83; I² = 0%). The synthesized data exhibited complete statistical homogeneity, with all five contributing trials directionally consistent with the null. Xia et al[13] contributed the largest weight (43.5%), followed by Zhang et al[17] (24.2%), Yan et al[14] (12.9%), Liu et al[15] (11.7%), and Yao et al[16] (7.7%), supporting comparable shunt patency between the 6 mm and 8 mm stent diameters in the modern PTFE covered stent era.

Variceal rebleeding at 2 years

All five studies reported this outcome. The pooled analysis for variceal rebleeding at 2 years (Figure 1F and G) demonstrated no statistically significant difference between the two stent diameters (RR = 1.28; 95%CI: 0.96-1.69; P = 0.09; I² = 0%), with a non significant directional trend toward higher rebleeding risk in the 6 mm group. The synthesized data exhibited complete statistical homogeneity. Xia et al[13] contributed the largest weight (61.0%), followed by Zhang et al[17] (17.0%), Yan et al[14] (9.9%), Liu et al[15] (7.4%), and Yao et al[16] (4.7%). Four of the five contributing trials directionally favored the 8 mm condition, consistent with the biologically expected reduction in portal decompression associated with smaller shunt diameter, although the magnitude did not reach statistical significance in the current evidence base.

Subgroup analysis by intervention type (true 6-mm vs underdilated 8-mm stents)

To address the mechanistic distinction between a true 6-mm PTFE-covered stent and an 8-mm stent underdilated with a 6-mm balloon, studies were stratified by intervention type (true 6-mm: Yan et al[14], Zhang et al[17], Xia et al[13]; underdilated 8-mm: Liu et al[15], Yao et al[16]), and between-subgroup differences were assessed using the Cochran Q interaction test (Supplementary Figures 7-9, Supplementary Table 3). The direction of effect was consistent across both strategies for every outcome, and the test for subgroup differences was non-significant throughout (all P ≥ 0.17). For overt HE, the RR was 0.61 (95%CI: 0.41-0.90) with true 6-mm stents and 0.72 (95%CI: 0.19-2.69) with underdilated stents (P = 0.81 for the subgroup difference). Because the underdilated subgroup comprised few studies and was imprecise, these analyses are exploratory, and the absence of a significant subgroup difference should not be interpreted as evidence of equivalence between the two strategies.

Grade assessment results

Certainty of evidence was graded using the GRADE framework, with ratings ranging from very low to high across the evaluated outcomes. Risk of bias was the most common reason for downgrading, reflecting the predominance of non randomized cohort studies in the evidence base, three of which were rated serious by ROBINS-I. Inconsistency was the second most common reason for downgrading, particularly for the continuous PPG outcome (I² = 85%, downgraded two levels) and for outcomes where moderate heterogeneity was driven by identifiable single study influences (overt encephalopathy and 2 years survival). Indirectness was not a concern for any outcome, as the population, intervention, comparator, and outcome definitions were directly applicable to the review question. The full GRADE judgement matrix is presented in Supplementary Table 4.

Sensitivity analysis

Pre specified sensitivity analyses were performed for the three outcomes demonstrating moderate to substantial statistical heterogeneity: Overt HE, overall survival at 2 years, and continuous post procedural PPG. For overt HE, exclusion of Yao et al[16] (the directional outlier) yielded a strengthened pooled effect (RR = 0.55; 95%CI: 0.38-0.81; P = 0.002), with heterogeneity reduced from I² = 54% to I² = 42% (Supplementary Figure 10). For overall survival at 2 years, exclusion of Xia et al[13] (the registry sized cohort) confirmed survival equivalence between the two stent diameters (RR = 1.00; 95%CI: 0.93-1.08; P = 0.94), with heterogeneity reduced from I² = 47% to I² = 6% (Supplementary Figure 11). For continuous PPG, exclusion of Zhang et al[17] (the visible outlier) confirmed hemodynamic equivalence between the two stent diameters (MD = 0.13 mmHg; 95%CI: -0.75 to 1.01; P = 0.77), with heterogeneity resolved completely (I² = 0%) (Supplementary Figure 12). The conclusions of the primary analyses were therefore robust to exclusion of the influential studies in each case, and in the case of overt HE were strengthened rather than weakened under sensitivity testing.

DISCUSSION

This systematic review and meta-analysis, synthesising data from 2060 patients across five comparative studies, shows that a 6-mm effective shunt diameter was associated with a lower incidence of both overall and overt HE than the standard 8-mm configuration, without a detectable difference in portal pressure decompression, shunt patency, variceal rebleeding, or overall survival. The magnitude of the HE reduction-a 44% decrease in overall risk (RR = 0.56) and a 38% decrease in overt risk (RR = 0.62) is clinically meaningful and consistent across severity subgroups. These findings extend the trajectory established by the 10-mm to 8-mm diameter-reduction step, in which randomised trials by Sauerbruch et al[11] and Riggio et al[12] confirmed that smaller stent diameters reduce post-TIPS HE without a significant survival penalty, and situate 6-mm stents as the logical next step in that continuum.

Shunt flow through a cylindrical channel scale with the fourth power of radius (Poiseuille’s law), such that a reduction in effective diameter from 8 mm to 6 mm reduces shunt conductance by approximately 80% at equivalent pressure gradients. This substantially curtails the volume of portal blood bypassing hepatic metabolism, thereby reducing the substrate for post-TIPS encephalopathy[9,10]. The preserved haemodynamic efficacy observed in the present analysis—with no significant difference in PPG target attainment, suggests that a 6-mm stent provides sufficient portal decompression in a majority of patients, at least over medium-term follow-up, consistent with the individual findings of Liu et al[15] and Yan et al[14].

The observed non-significant trend toward higher variceal rebleeding with 6-mm stents (RR = 1.28; 95%CI: 0.96-1.69) warrants careful consideration, as the confidence interval does not exclude a clinically important excess rebleeding risk. This finding is consistent with the mechanistic expectation that a smaller stent provides proportionally less portal decompression and may be insufficient to prevent variceal haemorrhage in patients with high-risk bleeding stigmata or particularly elevated portal pressures[2]. Importantly, this rebleeding signal mirrors a comparable pattern seen at the 10-mm to 8-mm boundary: Sauerbruch et al[11] observed that 8-mm stents achieved equivalent rebleeding prevention to haemodynamically guided medical therapy, but the transition to smaller diameters consistently generates uncertainty around bleeding endpoints. The results of Zhang et al[17], the only RCT in this analysis, similarly reported a non-significant trend toward higher rebleeding in the 6-mm arm, reinforcing the need for careful patient selection when further reducing stent diameter beyond 8 mm. Yao et al[16], whose outlying result drove the heterogeneity in the overt HE analysis, uniquely included post-splenectomy patients–a population with potentially different portal haemodynamics–highlighting that the rebleeding signal may be context-dependent.

The present findings have important implications for clinical practice. Current Baveno VII guidance does not differentiate recommendations by stent diameter, leaving clinicians to extrapolate from indirect evidence[1,18]. Because the certainty of evidence was low to very low, these data are not sufficient to recommend a preferred diameter; they may, however, support consideration of a 6-mm effective diameter as part of individualised decision-making in patients at high baseline HE risk (high Child-Pugh or MELD scores, prior HE episodes, sarcopenia) or with atrophic liver parenchyma. Conversely, in patients presenting with acute variceal haemorrhage, refractory high-risk bleeding, or very high baseline PPG, the relative.

The present results should be contextualised against the broader literature on TIPS stent diameter optimisation. The Bureau et al[5] landmark trial established PTFE-covered stents as the new standard by demonstrating superior patency over bare-metal devices, and the subsequent transition to 8-mm diameters built on RCT-level evidence that smaller shunts reduce encephalopathy without substantially increasing rebleeding risk[11,12]. Xia et al[13], whose large multicentre registry contributed the greatest patient volume in our analysis, provided real-world confirmation that diameter effects on HE are reproducible at scale across 6-, 8-, and 10-mm stent configurations. Liu et al[15] were among the first to operationalise this principle at the 6-mm threshold through underdilation, reporting reduced overt HE and preserved PPG reduction in their cohort-findings corroborated in a distinct clinical context (post-splenectomy patients) by Yao et al[16] and in severe hepatic atrophy by Yan et al[14]. Taken together, these studies trace a coherent dose–response relationship between effective shunt diameter and encephalopathy risk that our meta-analysis now synthesises at the 6-mm vs 8-mm boundary with formal statistical pooling. The only prior HE-focused TIPS diameter meta-analyses of which we are aware examined the 10-mm vs 8-mm comparison[11,12]; no systematic review had previously pooled 6-mm vs 8-mm data, confirming that the present analysis addresses a genuine and previously uncharted evidence gap.

Several limitations must be acknowledged. First, four of five included studies were non-randomised, and despite the use of propensity-matched cohorts or covariate adjustment in several, the risk of residual confounding is substantial, as confirmed by the ROBINS-I assessments. The single available RCT (Zhang et al[17]) is therefore the most important contributor to causal inference, though it alone is insufficiently powered for definitive conclusions on secondary endpoints. Second, meaningful clinical heterogeneity exists across studies in terms of TIPS indication, underlying liver disease aetiology, Child-Pugh distribution, and follow-up duration, which may influence the generalisability of pooled estimates. Third, the preponderance of Asian patient populations across included studies–reflecting the high prevalence of hepatitis B-related cirrhosis and liver atrophy in the relevant centres–may limit direct extrapolation to Western populations with predominantly alcohol-related or non-alcoholic fatty liver disease-related cirrhosis and different liver morphology. Fourth, formal assessment of publication bias was precluded for most outcomes by the small number of contributing studies.

Several further limitations warrant emphasis. First, two mechanistically distinct strategies were combined to define the 6-mm arm-true 6-mm PTFE-covered stents and 8-mm stents underdilated with a 6-mm balloon; because underdilated stents expand passively toward their nominal diameter within one to two weeks, the effective lumen may not remain 6 mm, and although the subgroup analysis showed no significant difference between strategies, the underdilated subgroup was small and the interaction tests were underpowered. Second, because four of five studies were non-randomised and only aggregate data were available, confounding by indication-whereby smaller shunts are preferentially chosen for patients at higher baseline HE risk-cannot be excluded, and the effect estimates should be read as associations rather than causal effects. Third, achievement of the target PPG may be influenced by intraoperative rescue dilation, which was not separately reported and could bias the 6-mm group toward apparent haemodynamic success. Fourth, functional shunt insufficiency-a gradual rise in PPG without overt stenosis prompting re-intervention-was not captured by the structural shunt-dysfunction endpoint, so any HE benefit must be weighed against re-intervention and cost that current evidence cannot quantify. Fifth, aggregate data preclude adjustment for the multiple established determinants of post-TIPS HE, including age, prior HE, sarcopenia, hyponatraemia, renal function, gut microbiota and medication adherence. Finally, shunt location is a potentially important unmeasured confounder-left portal-vein shunts bypass more hepatic parenchyma and carry higher HE rates than right-branch shunts (EASL Clinical Practice Guidelines on TIPS, 2025)-yet none of the included studies reported branch laterality, so this anatomical confounder cannot be excluded.

Future research should prioritise adequately powered RCT in diverse geographic and aetiological populations, with pre-specified stratification by TIPS indication, baseline HE risk score, and liver volume. Patient-reported outcomes, quality of life, and health economic analyses should be included as co-primary endpoints given the substantial morbidity and care burden associated with post-TIPS encephalopathy. The potential role of intermediate stent diameters and adjustable or constrained stent designs also merits investigation.

CONCLUSION

In patients with cirrhotic portal hypertension undergoing TIPS creation, a 6-mm effective shunt diameter was associated with a lower risk of HE than the 8-mm standard, with no detectable difference in overall survival, shunt patency, or portal pressure decompression, and a possible increase in variceal rebleeding. Because the certainty of evidence was low to very low and most data derived from non-randomised studies subject to confounding by indication, these findings are hypothesis-generating and are not sufficient to recommend a preferred stent configuration. Adequately powered randomised trials that pre-specify the stent strategy and report shunt location and re-intervention are needed to define optimal patient selection.