Evaluating the effectiveness and safety of initial thrombectomy approaches for acute anterior large vessel occlusions: a comparative network analysis of techniques Authors: Elvan Wiyarta, Mohammad Kurniawan, Rakhmad Hidayat, Affan Priyambodo Permana, Gideon Hot Partogi Sinaga, Refael Alfa Budiman, Gamaliel Wibowo Soetanto Published: 2025 ABSTRACT AbstractBackgroundAcute anterior circulation large vessel occlusion (ACLVO) strokes represent a critical challenge in stroke management. Advancements in thrombectomy strategies, including contact aspiration (CA), stent retrievers (SR), and their combination (SRA), particularly with the addition of balloon guide catheters (BGC), aim to enhance outcomes. This systematic review and Bayesian network meta-analysis evaluates the efficacy and safety of these first-line thrombectomy strategies.MethodsA systematic search was performed across nine databases up to March 15, 2025, yielding 44 studies with 19 268 participants. The risk of bias was assessed using RoB 2.0 for randomized controlled trials and ROBINS-I for observational studies. Bayesian random-effects network meta-analysis was conducted, presenting odds ratios (OR) with 95% CIs and ranking interventions via surface under the cumulative ranking curve (SUCRA). Outcomes included recanalization success (modified Thrombolysis in Cerebral Infarction (mTICI) 2b–3), first-pass effect (FPE), functional independence (modified Rankin Scale ≤2 at 90 days), symptomatic intracranial hemorrhage (sICH), and 90-day mortality.ResultsSRA+BGC was most effective for achieving mTICI 2b–3 (OR 2.87, 95% CI 1.64 to 5.20, SUCRA 90.1) and FPE (OR 7.33, 95% CI 1.71 to 39.71, SUCRA 91.0). CA+BGC yielded the highest odds of functional independence at 90 days (OR 1.96, 95% CI 1.16 to 3.41, SUCRA 88.1) and the lowest 90-day mortality (OR 0.42, 95% CI 0.20 to 0.95). Safety outcomes favored SR+BGC and SRA+BGC for reducing sICH (OR 0.59, 95% CI 0.37 to 0.91, SUCRA 85.3). CA alone was associated with significant improvements over SR in functional outcomes (OR 1.41, 95% CI 1.09 to 1.83, SUCRA 78.5), but SRA alone was linked to higher mortality (OR 1.74, 95% CI 1.06 to 2.66).ConclusionThrombectomy strategies incorporating BGC, particularly CA+BGC and SRA+BGC, achieve superior efficacy and safety outcomes in patients with acute ACLVO stroke.Keywords: Stroke, Thrombectomy, Vessel Wall, Catheter WHAT IS ALREADY KNOWN ON THIS TOPIC Mechanical thrombectomy is the established standard of care for acute anterior circulation large vessel occlusion (ACLVO) strokes. Techniques such as stent retrievers (SR), contact aspiration (CA), and their combination (SRA) have individually demonstrated efficacy in randomized trials. The use of balloon guide catheters (BGC) has been associated with reduced embolization and improved reperfusion, yet comparative data evaluating all first-line strategies with and without BGC remain limited. Mechanical thrombectomy is the established standard of care for acute anterior circulation large vessel occlusion (ACLVO) strokes. Techniques such as stent retrievers (SR), contact aspiration (CA), and their combination (SRA) have individually demonstrated efficacy in randomized trials. The use of balloon guide catheters (BGC) has been associated with reduced embolization and improved reperfusion, yet comparative data evaluating all first-line strategies with and without BGC remain limited. WHAT THIS STUDY ADDS This comprehensive Bayesian network meta-analysis synthesizes data from 44 studies involving over 19 000 participants and shows that thrombectomy strategies incorporating BGC, especially SRA+BGC and CA+BGC, yield superior outcomes across multiple domains. These include higher recanalization rates (modified Thrombolysis in Cerebral Infarction 2b–3), improved first-pass effect, greater functional independence at 90 days, and lower rates of symptomatic intracranial hemorrhage and 90-day mortality. The study also shows that SRA without BGC may be associated with increased mortality, highlighting the critical role of BGC in optimizing safety. This comprehensive Bayesian network meta-analysis synthesizes data from 44 studies involving over 19 000 participants and shows that thrombectomy strategies incorporating BGC, especially SRA+BGC and CA+BGC, yield superior outcomes across multiple domains. These include higher recanalization rates (modified Thrombolysis in Cerebral Infarction 2b–3), improved first-pass effect, greater functional independence at 90 days, and lower rates of symptomatic intracranial hemorrhage and 90-day mortality. The study also shows that SRA without BGC may be associated with increased mortality, highlighting the critical role of BGC in optimizing safety. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE, OR POLICY These findings support prioritizing BGC-enhanced thrombectomy approaches in clinical protocols for acute ACLVO stroke, particularly CA+BGC and SRA+BGC. Future guidelines may consider recommending BGC as a standard adjunct in first-line mechanical thrombectomy. Additionally, this analysis identifies a need for standardized reporting and further high-quality randomized trials to refine the selection of thrombectomy techniques based on individual patient and occlusion characteristics. These findings support prioritizing BGC-enhanced thrombectomy approaches in clinical protocols for acute ACLVO stroke, particularly CA+BGC and SRA+BGC. Future guidelines may consider recommending BGC as a standard adjunct in first-line mechanical thrombectomy. Additionally, this analysis identifies a need for standardized reporting and further high-quality randomized trials to refine the selection of thrombectomy techniques based on individual patient and occlusion characteristics. Introduction Acute ischemic stroke (AIS) remains a leading cause of morbidity and mortality worldwide, with large vessel occlusions (LVOs) accounting for a substantial proportion of stroke-related disability and death. LVOs in the anterior circulation, particularly involving the middle cerebral artery (MCA) and internal carotid artery (ICA), demand prompt and effective revascularization to mitigate neurological damage and improve clinical outcomes. Over the past decade, mechanical thrombectomy has emerged as the gold standard for treating AIS secondary to LVO, replacing earlier reliance on IV thrombolysis, particularly for cases refractory to pharmacological treatment.1 2 Epidemiological data underscore the pressing need for optimizing thrombectomy techniques. Globally, stroke affects approximately 12 million individuals annually, with ischemic strokes constituting nearly 87% of cases.3 In anterior circulation LVO (ACLVO) strokes, the use of mechanical thrombectomy has shown remarkable improvements in modified Rankin Scale (mRS) outcomes at 90 days, ranging from slight to no disability (mRS ≤2) to complete recovery (mRS ≤1).4 However, variations in patient response to different thrombectomy modalities necessitate comparative analyses to identify optimal strategies. The advent of stent retrievers (SR) and contact aspiration (CA) revolutionized mechanical thrombectomy, demonstrating superior efficacy compared with traditional methods.4 5 Randomized controlled trials such as ASTER and COMPASS established that both SR and CA achieved high rates of recanalization, functional independence, and reduced mortality.6,8 Recent strategies incorporating a combination of stent retriever and aspiration techniques (SRA), often augmented with balloon guide catheters (BGC), have further advanced the efficacy of thrombectomy. These innovations are supported by evidence suggesting improved reperfusion success, reduced embolization in new territories (ENT), and lower rates of symptomatic intracranial hemorrhage (sICH).9 To date, no comprehensive study has systematically compared the clinical and reperfusion outcomes of first-line SR, CA, and their combination approaches across multiple studies in acute ACLVO strokes. This gap in the literature presents a critical challenge, as clinicians lack consolidated evidence to guide optimal modality selection tailored to patient and procedural factors. In this study we aim to elucidate the relative efficacy and safety of these approaches, contributing to informed clinical decision-making in acute ACLVO stroke management to improve thrombectomy protocols and enhance patient outcomes in diverse healthcare settings. Methods Study design This systematic review and Bayesian network meta-analysis followed the PRISMA extension statement for network meta-analysis.10 The research protocol has been registered in PROSPERO (CRD42024523068). This study was designed to answer the following clinical question using the Population, Intervention, Comparison, and Outcome (PICO) framework11: in adult patients with acute ACLVO stroke (Population), how do different first-line thrombectomy strategies—contact aspiration (CA), stent retriever (SR), or their combination (SRA), with or without balloon guide catheter (BGC) (Intervention)—compare with each other (Comparison) in terms of efficacy and safety outcomes (Outcome)? The efficacy outcomes include successful reperfusion (mTICI 2b–3 and mTICI 3), first-pass effect (FPE), and functional independence at 90 days (mRS score ≤2). The safety outcomes include sICH, ENT, and 90-day mortality. Search strategy Studies reporting reperfusion and clinical outcomes of different first-line strategies for acute ACLVO stroke were included. We systematically searched nine databases from inception to March 15, 2025: PubMed, Cochrane CENTRAL, EBSCO CINAHL, Scopus, ProQuest, MedRxiv, BioRxiv, SSRN, and ClinicalTrials.gov. Search terms included keywords related to acute ACLVO stroke and mechanical thrombectomy strategies such as CA, SR, and SRA. For brevity, we refer to this combined approach as ‘SRA’ throughout the manuscript. The details of each database, along with their specific search strategy, are shown in online supplementay file 1. To ensure comprehensive coverage and avoid omissions, we also performed a ‘snowball’ search by reviewing the references of relevant articles to identify additional studies. Selection criteria Researches that met the following inclusion criteria were included: (a) study type: randomized controlled trial (RCT), cohort, or case–control study; (b) participants: adult patients with acute ACLVO stroke; (c) intervention: comparisons between mechanical thrombectomy strategies (SR, CA, or SRA) as first-line treatments; (d) outcomes: any efficacy outcome assessing successful revascularization or degree of independence, and any safety outcome evaluating death, secondary intracranial hemorrhage, or adverse events. Given that standard medical treatment alone is typically less effective than mechanical thrombectomy for acute ACLVO stroke, we excluded studies comparing mechanical thrombectomy solely with standard medical treatment to focus on direct comparisons between thrombectomy strategies. We also excluded studies meeting the following criteria: (a) single-arm studies and case reports; (b) thrombectomy performed outside the current therapeutic window; (c) use of experimental devices or surgical interventions. Study selection and data extraction Two independent authors separately searched the electronic databases and the references of relevant articles based on the inclusion and exclusion criteria outlined above. Duplicates were excluded by the automation tool EndNote 20. The full texts of these potentially relevant papers were collected and independently evaluated by the same two reviewers for adherence to the stipulated inclusion and exclusion criteria. After careful selection and assessment, the following data were extracted using a standardized extraction form: research features (author name, year of publication, country, and research design); characteristics of participants: age, sex ratio, National Institutes of Health Stroke Scale (NIHSS) score at admission, proportion of acute ACLVO stroke location, time from stroke onset to groin puncture; details on intervention and comparator procedures and outcomes. Any disagreements that arose in the process of study selection and data extraction were discussed with a third author to achieve consensus. To determine eligibility for inclusion in each synthesis we first categorized all included studies based on the thrombectomy strategies they evaluated (ie, SR, CA, SRA, and their combinations with BGC). Intervention characteristics were tabulated in a standardized data extraction form, allowing clear identification of first-line thrombectomy modalities applied in each study. These characteristics were then systematically compared with the pre-specified intervention groups defined in our PICO framework. Only studies that directly compared at least two of the defined strategies and reported relevant outcomes (eg, mTICI, FPE, mRS, sICH, mortality) were included in the synthesis for each outcome. Prior to synthesis, we harmonized outcome definitions and standardized all extracted data. Dichotomous outcomes were entered as event counts and total sample sizes; if only percentages were reported, absolute counts were back-calculated. For continuous outcomes like NIHSS scores, data were recorded as means and SD; when SDs were not reported, they were imputed using interquartile ranges or standard errors based on Cochrane Handbook guidance. No assumptions were made for completely missing outcome data, and such studies were excluded from the relevant synthesis. All conversions and derived estimates were double-checked by the data extraction team to ensure accuracy and transparency. Outcome measures We focused on two sets of outcomes: efficacy outcomes related to recanalization and safety outcomes related to clinical measures. Efficacy outcomes included final successful reperfusion, defined as mTICI grade 2b–3 and final complete reperfusion (mTICI grade 3). We also evaluated the FPE, defined as complete reperfusion achieved with a single pass; and the modified FPE (mFPE), defined as successful reperfusion with a single pass. Additional recanalization outcomes included functional independence, defined as mRS scores of ≤1 and ≤2 at 90 days. Safety outcomes included the NIHSS scores at 24 hours and discharge, mortality at 90 days, ENT, and sICH. Risk of bias assessment To evaluate the risk of bias we assessed the methodological quality of studies using the Cochrane Collaboration’s Risk of Bias (ROB 2.0) tool for RCT12 and the Risk of Bias in Non-Randomized Studies of Interventions (ROBINS-I) tool for observational studies.13 The ROB 2.0 tool evaluated five domains of bias including randomization, deviations from intended interventions, missing outcome data, measurement of outcomes, and selection of reported results. Each RCT was classified as having a ‘low’ risk of bias, ‘some concerns’, or a ‘high’ risk of bias based on the responses to the signaling questions. For RCTs, factors associated with a low risk of bias included well-defined selection criteria, comparable baseline stroke severity, and independent assessment of recanalization and clinical outcomes. For cohort studies, the ROBINS-I tool assessed seven domains, with specific criteria tailored for this review. A low risk of bias required addressing confounding through adjustment or matching, prospective participant selection, clear intervention protocols, minimal operator variability, transparent handling of missing data, independent outcome measurement, and pre-specified outcomes. Studies were rated as having a ‘low’, ‘moderate’, ‘serious’, or ‘critical’ risk of bias. Two authors independently evaluated all studies, with disagreements resolved by a third author. Statistical analysis We conducted a random-effects network meta-analysis using Markov chain Monte Carlo (MCMC) simulations within a Bayesian framework. Statistical analyses were performed in R (version 4.0.3) with the ‘gemtc’ package.14 Network plots showed direct comparisons among mechanical thrombectomy strategies, with line width indicating the number of studies. Results were presented as odds ratios (OR) with 95% CI for dichotomous data and mean difference with 95% CI for continuous data. To rank strategies, we calculated the surface under the cumulative ranking curve (SUCRA), visualizing ranks with Litmus Rank-O-Gram and Radial SUCRA,15 where higher SUCRA values indicated a higher likelihood of top ranking. Heterogeneity across studies was evaluated using the I² statistic, with I² >50% indicating substantial heterogeneity. However, due to the complexity of the network meta-analysis and the limited number of studies per specific comparison node, formal subgroup analyses or network meta-regression were not performed. Instead, heterogeneity was explored narratively in the discussion, contextualizing differences in outcomes based on study setting, population, and procedural variations. Consistency analyses between direct and indirect estimates were performed, with statistical inconsistency defined as P<0.05. To assess the robustness of the findings we explored consistency across study designs and risk of bias levels during interpretation. Outcome rankings and SUCRA values were examined for stability across multiple outcomes, supporting confidence in the synthesized results. Given the complexity of the network meta-analysis and the inclusion of both randomized and non-randomized studies, we assessed the risk of bias from missing results primarily through qualitative approaches. We evaluated whether eligible studies reported all pre-specified outcomes relevant to our synthesis by comparing the outcomes listed in the Methods sections with those reported in the Results. We also assessed whether outcome reporting appeared selective, particularly in studies that referenced protocols or were registered in clinical trial databases. Inconsistencies between pre-specified and reported outcomes were flagged as potential sources of reporting bias. Furthermore, we examined whether certain comparisons or outcomes were systematically under-reported across studies. Certainty of evidence We applied the GRADE (Grading of Recommendations Assessment, Development, and Evaluation) framework to evaluate the certainty of evidence for each comparison.16 Evidence was categorized into four levels: high, moderate, low, or very low certainty. For direct estimates we downgraded the certainty of evidence based on limitations such as risk of bias, inconsistency (heterogeneity), indirectness, imprecision, and publication bias. If no significant limitations were identified, the evidence was rated as high certainty. For indirect comparisons the certainty was assessed using the dominant first-order loop, with the rating determined by the lowest certainty among the direct comparisons contributing to that loop. If a first-order loop was unavailable, higher-order loops were used, following the same principle of using the lowest-rated direct comparisons. Additional downgrading was considered for intransitivity if the distribution of effect modifiers varied across the contributing direct comparisons. Finally, network meta-analysis results were graded based on the certainty of direct and indirect comparisons, along with assessments of consistency and precision. When direct and indirect estimates were available and consistent, the final network estimate was assigned the highest certainty rating. Results Study characteristics We retrieved 2393 articles from databases and registries and 28 additional articles identified through hand-searching. After screening, retrieval, and eligibility assessment, four RCTs13,16 and 40 cohort studies48 17,55 were included, with a total of 19 268 participants. The PRISMA flow diagram detailing the literature search and selection process is shown in figure 1. PRISMA flowchart. *Consider, if feasible to do so, reporting the number of records identified from each database or register searched (rather than the total number across all databases/registers). **If automation tools were used, indicate how many records were excluded by a human and how many were excluded by automation tools. The age of the participants ranged from 60.3 to 79 years, with the percentage of females ranging from 15.4% to 71.4%. Onset to groin puncture times varied from 139 min to 560.73 min, and NIHSS scores ranged from 10.5 to 21. Across the studies there were a total of 19 213 occlusions, with MCA-M1 (60.3%) and ICA (25.7%) being the most common occlusion locations. Only two studies exclusively focused on M2 occlusions.8 39 online supplementary file 2 shows the detailed patient characteristics in the included studies. Risk of bias We evaluated the risk of bias for each included RCT (figure 2). The study by Pop et al did not report blinding of outcome assessors, leading to an overall judgment of ‘some concern’.56 The remaining three RCTs were rated as low risk of bias. Risk of bias of randomized controlled trials. For cohort studies, the study by Zaidat et al was considered to have a low risk of bias and was therefore considered comparable to a well-conducted RCT.52 Given that only four of the 44 included studies were RCTs, the majority of evidence is derived from observational data, which are inherently susceptible to confounding and selection bias. In particular, operator experience, procedural preferences, and device availability may influence treatment allocation and outcomes. This underscores the importance of interpreting our results in the context of these limitations. In contrast, the majority of cohort studies were judged to have a moderate to serious risk of bias (figure 3). Sources of bias were primarily due to non-comparable baseline characteristics (confounding factors), influenced by availability or operator preferences without a standardized protocol (deviations from intended interventions), unreported handling of missing data, and lack of blinding in outcome measurement or independent core laboratory adjudication. Risk of bias of cohort studies. Bayesian network meta-analysis The Bayesian network meta-analysis explored the efficacy and safety of different mechanical thrombectomy approaches, including SR, CA, and SRA, with each intervention also combined with BGC to assess improvements in outcomes for acute anterior circulation strokes. Figure 4 and online supplementary files 3 and 4 present the network plots and league tables for efficacy and safety outcomes, respectively. Comparisons between treatments should be read from top to bottom. Bold type and underlining indicate significant differences between the two treatment strategies. The color of each cell indicates the certainty of evidence according to the GRADE methodology. Network plots of available direct comparisons: (A) modified Rankin Scale (mRS) score 0–2; (B) mRS 0–1; (C) Thrombolysis in Cerebral Infarction (mTICI 2b–3); (D) mTICI 3; (E) first pass effect (FPE); (F) modified FPE (mFPE); (G) National Institutes of Health Stroke Scale (NIHSS) scores at 24 hours; (H) NIHSS scores at discharge; (I) symptomatic intracranial hemorrhage; (J) all-cause mortality; (K) embolization in new territory. Efficacy In terms of reperfusion outcomes, the addition of a BGC to SRA substantially improved rates of successful recanalization. SRA combined with BGC significantly increased the odds of achieving mTICI 2b–3 compared with SR alone, with an OR of 2.46 (95% CI 1.10 to 5.49, high certainty) (see online supplementary file 3). For final complete reperfusion, SRA combined with BGC also outperformed both SR alone (OR 2.87, 95% CI 1.64 to 5.2, high certainty) and SR with BGC (OR 1.86, 95% CI 1.09 to 3.15, high certainty) (see online supplementary file 3), showing an advantage across both measures. Other thrombectomy methods such as CA and SRA were also notably more effective than SR alone in mTICI 3. However, no significant differences were observed among these strategies for both mTICI 2b–3 and mTICI 3 reperfusion rates beyond the comparisons mentioned. In first-pass recanalization outcomes (online supplementary file 3), SRA with and without BGC was superior in achieving mFPE compared with SR alone, with ORs of 2.56 (95% CI 1.40 to 4.66, high certainty) and 3.05 (95% CI 1.43 to 6.48, low certainty), respectively. Both these techniques also showed significantly higher odds of achieving mFPE compared with CA alone, with SRA alone achieving an OR of 2.48 (95% CI 1.29 to 4.85, high certainty) and SRA with BGC achieving an OR of 2.08 (95% CI 1.22 to 3.56, high certainty), respectively. Additionally, pairing SR with BGC improved mFPE compared with SR alone (OR 1.83, 95% CI 1.04 to 3.21, high certainty). In terms of FPE, only SRA with BGC was notably effective, achieving FPE at a significantly higher rate than SR alone (OR 7.33, 95% CI 1.71 to 39.71, moderate certainty). Beyond these findings, no statistically significant differences were found between other pairs of strategies for mFPE and FPE. Regarding functional outcomes at 90 days, the probability of slight or no disability (90-day mRS ≤2) was significantly higher with the addition of BGC: CA with BGC (OR 1.96, 95% CI 1.16 to 3.41, low certainty), SR with BGC (OR 1.79, 95% CI 1.34 to 2.45, high certainty), and SRA with BGC (OR 1.47, 95% CI 1.03 to 2.15, high certainty) compared with SR alone. Without BGC, only CA alone (OR 1.41, 95% CI 1.09 to 1.83, high certainty) showed a significant improvement over SR alone. For no significant disability (90-day mRS ≤1), all thrombectomy methods were significantly more effective than SR alone. Safety In terms of safety, NIHSS scores at 24 hours were significantly lower with all strategies compared with SRA alone (see online supplementary file 4). At discharge there were no statistically different scores between any two pairs of mechanical thrombectomy strategies. In terms of sICH, SRA with BGC showed a protective effect compared with SRA without BGC (OR 0.59, 95% CI 0.37 to 0.91, high certainty), indicating the role of BGC in minimizing hemorrhagic complications. All-cause mortality in 90-day outcomes further supported the benefits of BGC-enhanced strategies, with all BGC approaches (CA+BGC, SR+BGC, and SRA+BGC) associated with lower mortality than SRA without BGC. In addition, CA+BGC showed a significant reduction in mortality compared with all other thrombectomy strategies. Nevertheless, without BGC, CA alone (OR 0.54, 95% CI 0.36 to 0.84, moderate certainty) and SR alone (OR 0.61, 95% CI 0.4 to 0.96, moderate certainty) were associated with lower mortality than its combination. ENT was also notably lower with the addition of BGC. CA with BGC was a significant protective factor compared with all thrombectomy methods without BGC, including CA alone (OR 0.27, 95% CI 0.09 to 0.71, high certainty), SR alone (OR 0.19, 95% CI 0.06 to 0.55, low certainty), and SRA (OR 0.31, 95% CI 0.1 to 0.86, low certainty). In addition, SR with BGC was also a significant protective factor compared with SR alone (OR 0.5, 95% CI 0.27 to 0.88, high certainty). However, there was no significant difference between CA with BGC and SR with BGC (OR 0.39, 95% CI 0.13 to 1.05, high certainty). The SUCRA values (see online supplementary file 5) ranked the likelihood of each strategy performing favorably across these efficacy and safety outcomes. SRA with BGC held the highest probability for mTICI 2b–3 (SUCRA 0.68), mTICI 3 (SUCRA 0.75), and FPE (SUCRA 0.91), confirming its effectiveness in achieving optimal reperfusion. SRA with BGC also held the highest probability for the sICH protective factor (SUCRA 0.58) and lower NIHSS score at the final follow-up (SUCRA 0.39). For clinical outcomes at 90 days, CA with BGC showed the highest probability of achieving functional independence (mRS ≤2) (SUCRA 0.63), reducing mortality (SUCRA 0.96), and minimizing ENT (SUCRA 0.97). SR with BGC led to achieving 90-day mRS ≤1 (SUCRA 0.37). Meanwhile, without BGC, SRA alone ranked highest in mFPE (SUCRA 0.65) and NIHSS improvement at 24 hours (SUCRA 0.97). Discussion This network meta-analysis highlights critical advancements in the management of acute ACLVO stroke and provides a robust comparison of mechanical thrombectomy strategies. From a synthesis of 44 studies we observed that SRA, particularly with BGC, outperforms standalone SR and CA techniques in both efficacy and safety outcomes. These results, while aligned with prior studies, also highlight nuanced differences and areas for improvement. The results show that BGC-enhanced SRA strategies consistently achieve higher rates of successful recanalization (mTICI 2b–3) and functional independence (mRS ≤2) at 90 days compared with SR or CA alone. These findings align with prior randomized trials such as ASTER and COMPASS, which reported improved outcomes with combination techniques, especially in high clot burden scenarios.6,8 The addition of BGC was particularly beneficial, significantly increasing the FPE and mFPE. Our findings further expand on these results by quantifying the added benefit of BGC in enhancing first-pass success and reducing ENT, as previously noted in studies such as STRATIS and MR CLEAN.29 52 This advantage can be attributed to the reduction in ENT and the enhanced retrieval of thrombus fragments during the procedure. Additionally, the clinical outcomes at 90 days (mRS ≤2 and ≤1) reflect trends reported by Bernsen et al, who emphasized the critical role of patient-specific factors such as clot composition and occlusion site in determining the optimal thrombectomy approach.18 While our meta-analysis shows a clear preference for SRA with BGC across multiple efficacy metrics, it also aligns with observations from Hesse et al, which indicate comparable performance of SR and CA in specific occlusion contexts such as M2 branches, suggesting the need for more targeted evaluations in such subgroups.32 Safety metrics, including NIHSS improvement and reduced sICH, highlight the protective role of BGC-enhanced strategies. Importantly, the analysis showed a consistent protective effect of BGC use against ENT, with CA+BGC and SR+BGC significantly outperforming their non-BGC counterparts. This aligns with prior registry data suggesting that proximal flow arrest minimizes distal clot migration. Additionally, 24-hour NIHSS scores were significantly more favorable in BGC-enhanced approaches, supporting their role in improving immediate post-procedural outcomes. Although discharge NIHSS scores did not differ significantly, the early benefit suggests BGC use may have acute neuroprotective effects. Additional studies corroborate the safety advantages of BGC-enhanced strategies. For instance, the STRATIS registry showed that the use of BGC significantly reduced ENT and improved functional outcomes compared with non-BGC techniques.52 Similar findings from the MR CLEAN registry highlighted the protective effect of BGC in minimizing sICH and improving rates of functional independence at 90 days.29 The findings show lower all-cause mortality and fewer adverse events with these techniques, reinforcing their role as a preferred approach for acute ACLVO stroke. Our finding that CA alone demonstrated superior odds of functional independence compared with SR alone contrasts with the COMPASS trial,6 which found aspiration to be non-inferior to stent retriever use. This discrepancy may arise from the predominance of observational studies in our dataset where unadjusted confounding—such as selection bias or operator technique preference—may have influenced outcomes. Notably, SRA without BGC was associated with increased 90-day mortality compared with both its BGC-enhanced counterpart and other strategies. These findings suggest that, without BGC, the procedural complexity and embolic risk of SRA may outweigh its theoretical advantages. Several plausible mechanisms may underlie this finding. First, the absence of BGC eliminates the proximal flow arrest effect, which can increase the likelihood of distal embolization during clot retrieval. Second, the SRA technique itself may involve more device exchanges and repositioning, especially in anatomically tortuous or resistant clots, increasing procedural complexity and duration. These factors may result in delayed recanalization and greater ischemic burden. Finally, operator experience and center-level variability in applying SRA without BGC may introduce unmeasured procedural inefficiencies. Furthermore, while our findings reinforce the clinical utility of BGC in improving recanalization efficacy and safety, this analysis did not assess the cost-effectiveness of BGC-enhanced strategies. Given the potentially marginal differences in clinical benefit between combinations such as CA+BGC and SRA+BGC, economic considerations become increasingly relevant. Study limitations There are some limitations within this meta-analysis. The included studies vary widely in patient demographics, stroke severity, and occlusion characteristics. Specifically, baseline NIHSS scores ranged from 10.5 to 21, onset-to-groin puncture times varied between 139 min and 560 min, and the proportion of ICA versus MCA occlusions differed substantially across cohorts. Furthermore, only a small subset of included studies reported outcomes stratified by M2 occlusions, limiting the applicability of our findings to distal vessel thrombectomy. As the risk–benefit profile and technical considerations differ significantly in more distal anatomy, caution should be exercised in generalizing these results beyond large vessel occlusions (eg, ICA, proximal M1). These factors can introduce substantial clinical heterogeneity that may confound direct treatment comparisons and impact the external validity of the findings. While formal subgroup or meta-regression analyses (eg, stratification by occlusion site, NIHSS, or time metrics) would offer additional insight, these were not feasible due to insufficient subgroup data and network sparsity. Procedural success in mechanical thrombectomy is heavily operator-dependent. Differences in training, experience, and adherence to standardized protocols could introduce bias. Studies like Kang et al have highlighted that procedural nuances, including the use of proximal flow arrest with BGC, can significantly impact outcomes.54 Variability in the therapeutic windows across studies complicates comparisons. For instance, early presenters may benefit disproportionately from thrombectomy, as highlighted by Nguyen et al, while delayed presentations pose unique challenges not fully addressed in the existing literature.45 Additionally, the majority of included studies were conducted in high-volume stroke centers located in Europe and North America. These institutions often benefit from experienced neurointerventionists, streamlined workflows, and ready access to advanced thrombectomy devices, all of which may enhance outcomes. As such, our findings may not be fully generalizable to lower-volume centers or regions with limited neurointerventional resources. Common issues included confounding due to non-comparable baseline characteristics, inconsistent handling of missing data, and lack of blinding in outcome measurements. The predominance of observational studies also poses challenges regarding internal validity. Operator expertise, center-specific protocols, and non-randomized device selection may confound outcomes in ways not fully accounted for. While we considered performing sensitivity analyses restricted to the four available RCTs or low risk of bias studies, the limited number of such trials and heterogeneity in outcome reporting would have resulted in a fragmented and underpowered network. Consequently, our findings—though robust in scope—should be interpreted with caution given the methodological variability, and future high-quality multicenter RCTs remain essential to confirm these results. These biases may have affected the reliability of individual studies and consequently the network meta-analysis. The predominance of positive findings in the thrombectomy literature raises concerns about publication bias. The under-representation of studies with neutral or negative results may have skewed the overall conclusions of the meta-analysis, overestimating the efficacy of certain strategies. Variability in effect modifiers such as clot composition or occlusion site introduces intransitivity in the network. Another important source of heterogeneity not captured in this analysis is clot composition. Fibrin-rich clots tend to be more resistant to aspiration and stent retrieval, while erythrocyte-rich clots are generally softer and more easily removed.57 Since most included studies did not report thrombus histology, we could not stratify outcomes by clot type. Another important limitation involves the temporal heterogeneity in device technology. Many included studies span a decade or more, during which time significant improvements have been made to CA catheters, BGCs, and SRs. For instance, newer aspiration catheters offer larger bore sizes, improved navigation, and greater suction force, all of which may enhance procedural efficacy and safety. As such, outcomes reported in older studies may underestimate the performance of modern devices, particularly in CA-based strategies. While consistency checks were performed, discrepancies between direct and indirect evidence highlight potential issues in harmonizing data across studies with differing designs and populations. SUCRA rankings provide a valuable tool for comparing interventions across multiple outcomes. However, they should be interpreted with caution. These rankings are influenced by effect size and also by the distribution of certainty across contributing evidence. For example, some comparisons rely more heavily on indirect evidence or observational data, limiting their generalizability. Moreover, SUCRA does not account for patient-specific factors, operator expertise, or local resource availability—all of which play pivotal roles in treatment selection. Therefore, SUCRA rankings should be regarded as exploratory and hypothesis-generating rather than prescriptive. The findings of this network meta-analysis provide critical insights into the comparative effectiveness and safety of thrombectomy strategies for acute ACLVO stroke. However, the inherent biases and limitations outlined underscore the need for cautious interpretation. Future studies should prioritize standardizing protocols, reducing operator dependency, and addressing gaps in the evidence base to strengthen the validity and applicability of network meta-analysis findings in thrombectomy research. Conclusion This network meta-analysis underscores the superiority of SRA techniques augmented with BGC in achieving both efficacy and safety outcomes in the treatment of acute ACLVO stroke. These findings highlight the role of BGC in enhancing first-pass success, reducing ENT, and minimizing adverse events such as sICH. Despite these promising results, variability in study populations, procedural characteristics, and operator expertise underscores the need for cautious interpretation and application of these findings in clinical practice. Supplementary material Footnotes Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial, or not-for-profit sectors. Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: Not applicable. Patient and public involvement statement: Patients and/or the public were not involved in the design, conduct, reporting, or dissemination plans of this systematic review. Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information. References Associated Data Supplementary Materials Data Availability Statement All data relevant to the study are included in the article or uploaded as supplementary information.