Factors influencing functional recovery in posterior circulation stroke following mechanical thrombectomy: a systematic review and meta-analysis Authors: Elvan Wiyarta, Rakhmad Hidayat, Mohammad Kurniawan, Iskandar Purba Geraldi, Gideon Hot Partogi Sinaga, Gamaliel Wibowo Soetanto, Gilang Nispu Saputra, Nyoman Angga Krishna Pramana Published: 2026 ABSTRACT AbstractIntroductionPosterior circulation stroke (PCS) in the vertebrobasilar system carries high morbidity and mortality. While mechanical thrombectomy (MT) is effective in selected patients, functional recovery after MT for PCS remains variable and may be influenced by clinical, imaging and procedural factors.ObjectivesTo summarise clinical, imaging and procedural factors associated with functional recovery, defined as modified Rankin Scale 0–2 at 90 days, after MT in PCS.MethodsThis systematic review and meta-analysis searched EMBASE, MEDLINE, PubMed, EBSCO, Scopus, CNKI, SSRN, bioRxiv and medRxiv from inception until 31 December 2025. Inclusion criteria were adult patients with PCS who underwent MT in clinical trials or observational studies. Risk Of Bias In Non-randomised Studies-of-Exposure was assessed. Meta-analysis used a random-effects model to pool ORs with 95% CIs, assessing heterogeneity via I2 and Cochran’s Q test (PROSPERO: CRD42024626263).ResultsWe analysed 12 094 patients from 75 studies. Predictors of favourable outcomes included higher posterior circulation-Alberta Stroke Programme Early CT Score (OR=1.80, 95% CI 1.59 to 2.03, p<0.001, I2=0.0%), intravenous thrombolysis (OR=1.66, 95% CI 1.31 to 2.1, p<0.001, I2=21.7%), first-pass effect (OR=2.36, 95% CI 1.84 to 3.03, p<0.001, I2=9.28%) and modified thrombolysis in cerebral infarction 2b/3 (OR=3.58, 95% CI 2.31 to 5.55, p<0.001, I2=58.43%). Poor outcomes were associated with higher baseline National Institutes of Health Stroke Scale (OR=0.92, 95% CI 0.89 to 0.95, p<0.001, I2=85.09%), diabetes mellitus (OR=0.56, 95% CI 0.43 to 0.72, p<0.001, I2=24.63%), hypertension (OR=0.73, 95% CI 0.59 to 0.91, p=0.010, I2=0.0%) and general anaesthesia use (OR=0.56, 95% CI 0.36 to 0.89, p=0.010, I2=64.53%).ConclusionsStroke severity, imaging scores and procedural success are factors consistently reported to be associated with functional outcomes in PCS after MT. Although the independence of these factors cannot be determined from this analysis, these findings may help summarise current evidence on reported prognostic factors and support future prospective validation using standardised protocols. WHAT IS ALREADY KNOWN ON THIS TOPIC Mechanical thrombectomy has become an important treatment for posterior circulation large vessel occlusion, and recent randomised trials, trial-level meta-analyses and individual patient data meta-analysis have supported its benefit in selected patients. However, despite successful reperfusion in many cases, functional recovery remains variable, leaving uncertainty about which reported clinical, imaging and procedural factors are most consistently associated with outcome. Mechanical thrombectomy has become an important treatment for posterior circulation large vessel occlusion, and recent randomised trials, trial-level meta-analyses and individual patient data meta-analysis have supported its benefit in selected patients. However, despite successful reperfusion in many cases, functional recovery remains variable, leaving uncertainty about which reported clinical, imaging and procedural factors are most consistently associated with outcome. WHAT THIS STUDY ADDS This systematic review and meta-analysis of 75 studies involving 12 094 patients showed that successful reperfusion, first pass effect, higher posterior circulation-Alberta Stroke Programme Early CT Score, intravenous thrombolysis, higher Glasgow Coma Scale and distal basilar occlusion were associated with better functional recovery after thrombectomy for posterior circulation stroke. In contrast, higher baseline National Institutes of Health Stroke Scale, higher pre stroke disability, diabetes mellitus, hypertension, general anaesthesia, large artery atherosclerosis and first line stent retriever use were the most consistently reported factors associated with worse functional outcome across included studies, though their independence from one another cannot be established from this analysis. This systematic review and meta-analysis of 75 studies involving 12 094 patients showed that successful reperfusion, first pass effect, higher posterior circulation-Alberta Stroke Programme Early CT Score, intravenous thrombolysis, higher Glasgow Coma Scale and distal basilar occlusion were associated with better functional recovery after thrombectomy for posterior circulation stroke. In contrast, higher baseline National Institutes of Health Stroke Scale, higher pre stroke disability, diabetes mellitus, hypertension, general anaesthesia, large artery atherosclerosis and first line stent retriever use were the most consistently reported factors associated with worse functional outcome across included studies, though their independence from one another cannot be established from this analysis. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY These findings may support risk stratification and procedural decision-making in patients with posterior circulation stroke undergoing mechanical thrombectomy, particularly by emphasising the importance of imaging selection and procedural efficiency. They also highlight the need for prospective studies using standardised definitions and protocols to validate prognostic factors and optimise workflow in posterior circulation thrombectomy. These findings may support risk stratification and procedural decision-making in patients with posterior circulation stroke undergoing mechanical thrombectomy, particularly by emphasising the importance of imaging selection and procedural efficiency. They also highlight the need for prospective studies using standardised definitions and protocols to validate prognostic factors and optimise workflow in posterior circulation thrombectomy. Introduction Posterior circulation (PC) stroke accounts for approximately one-fifth of all ischaemic strokes.1 In a hospital-based ischaemic stroke cohort, large-vessel occlusion (LVO) was identified in 11.3% of ischaemic stroke events, of which 24% involved PC vessels.2 PC LVO, though less common than anterior circulation LVO, remains clinically important given its diagnostic challenges and high morbidity and mortality.3 Clinical presentation in PC stroke can be subtle and variable, often delaying diagnosis and treatment, contributing to poorer outcomes compared with anterior circulation strokes, even when reperfusion is achieved.4 The success of mechanical thrombectomy (MT) in anterior LVOs has transformed stroke care and led to its increasing use in PC occlusions.5–7 Large prospective studies and meta-analyses have shown that successful reperfusion in the PC improves functional recovery.8 9 However, randomised controlled trials have produced mixed evidence. While early trials such as BEST7 and BASICS10 failed to show a clear benefit due to recruitment difficulties and crossover, more recent studies, including BAOCHE11 and ATTENTION,12 have demonstrated the superiority of endovascular therapy over medical management.7 10–12 Still, even with intervention, pooled meta-analysis from BEST, BASICS, BAOCHE and ATTENTION found that more than half of patients remain severely disabled or deceased at 90 days.13 The VERITAS individual patient data meta-analysis provided patient-level evidence supporting endovascular therapy for acute vertebrobasilar occlusion and showed heterogeneity of treatment effect according to baseline stroke severity and occlusion site. Therefore, the remaining uncertainty addressed by the present review does not concern whether endovascular therapy is beneficial in appropriately selected patients, but rather which clinical, imaging and procedural factors have been most consistently reported to be associated with functional recovery after thrombectomy in the broader PC literature.14 Currently, prognostic factors associated with recovery after MT in PC-LVOs have begun to be widely researched in several studies.15 Gao et al,15 found that factors such as a 1-point increase in the 24-hour National Institutes of Health Stroke Scale (NIHSS) score and a 1-point decrease in the posterior circulation-Alberta Stroke Programme Early CT Score (pc-ASPECT) score increased the likelihood of poor outcomes.15 Another study also found that factors such as the use of aspiration techniques and achieving a thrombolysis in cerebral infarction (TICI) of 2b/3 were associated with good outcomes.16 Meta-analyses have collectively shown that MT is technically effective for PC stroke.14 MT technical success should be associated with relatively better outcomes; however, there are other factors that may influence the patient’s clinical outcomes.17 To address these remaining uncertainties, we conducted a systematic review and meta-analysis of patients with PC-LVO who underwent MT, aiming to summarise clinical, imaging and procedural factors reported to be associated with functional independence. Our goal was to synthesise current evidence from observational studies and clinical trials to estimate the strength of association between various clinical, radiological and procedural factors and the likelihood of achieving functional independence, which is defined as a modified Rankin Scale (mRS) score of 0–2 at 90 days. Methods Protocol and reporting standards The systematic review and meta-analysis were conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement.18 The review protocol was registered in PROSPERO (CRD42024626263). Information sources and search strategy Studies reporting predictors of good functional outcomes for PC stroke following MT were included. A systematic search was conducted across electronic databases (EMBASE, MEDLINE, PubMed, EBSCO, Scopus, CNKI, SSRN) and grey literature (BioRxiv and MedRxiv) from database inception until 31 December 2025. No language restrictions were applied. To minimise omissions, we also conducted backward citation tracking of all relevant articles. Full search strategy could be seen in online supplemental file 1. Eligibility criteria We included studies that met all of the following criteria: Population: Adults (≥18 years) with acute PC-LVO (basilar, vertebral or posterior cerebral artery) treated with MT. Population: Adults (≥18 years) with acute PC-LVO (basilar, vertebral or posterior cerebral artery) treated with MT. Study designs: Randomised controlled trials or observational (cohort or case–control) studies reporting predictors of functional outcomes. Study designs: Randomised controlled trials or observational (cohort or case–control) studies reporting predictors of functional outcomes. Outcomes: Good functional recovery, defined as an mRS score of 0–2 at 90 days, with effect estimates (eg, ORs and 95% CIs) for at least one predictor. Outcomes: Good functional recovery, defined as an mRS score of 0–2 at 90 days, with effect estimates (eg, ORs and 95% CIs) for at least one predictor. Timing and setting: No restrictions on publication year or healthcare setting. Timing and setting: No restrictions on publication year or healthcare setting. We included basilar, vertebral and posterior cerebral artery occlusions because they all fall within the PC vascular territory and may coexist or extend across contiguous vertebrobasilar segments in clinical practice. We excluded case reports, small case series (<10 patients), conference abstracts without full data, studies limited to anterior circulation strokes, paediatric populations and non-human research. Study selection Two reviewers independently screened titles and abstracts in Rayyan. Records deemed potentially eligible advanced to full-text review against the inclusion criteria. Discrepancies were resolved by discussion or, when necessary, by a third reviewer. The study selection process was documented in a PRISMA 2020 flow diagram. Data extraction We piloted a standardised form using spreadsheet programmes (Microsoft Excel, USA) to capture study details (author, year, country, design, sample size), patient demographics (age, sex), baseline stroke severity (NIHSS, pc-ASPECTS), comorbidities and procedural details (time to groin puncture, device type, TICI score, intravenous tissue plasminogen activator use). The primary objective was good functional outcomes defined as mRS score 0–2 at 90 days. Additional outcomes included mortality, recanalisation rates and symptomatic intracranial haemorrhage. Predictor variables of interest were classified into five domains, including patient‐related, vascular risk, baseline severity, imaging‐derived and treatment/workflow variables. Two reviewers extracted data independently, with any discrepancies reconciled by consensus. 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 (V.2.0) tool for randomised controlled trial,19 encompassing five bias domains and yielding low/some-concerns/high-risk judgements. At the same time, the Risk Of Bias In Non-randomised Studies-of-Exposure (ROBINS-E) tool was assessed for observational studies, which evaluates seven bias domains via structured signalling questions. Risk of bias was evaluated by two reviewers, with any discrepancies reconciled by consensus. Effect measures and data synthesis The primary effect measure was the OR with 95% CIs for each predictor of good functional outcome. Before pooling, all extracted ORs were checked and harmonised so that OR >1 consistently indicated higher odds of 90-day functional independence, defined as mRS 0–2, whereas OR <1 indicated lower odds of 90-day functional independence. For binary predictors, the comparison direction was checked against the source article and harmonised to the comparison label used in the forest plot. Where a source article reported the reciprocal comparison, the estimate was inverted before pooling. For continuous predictors, estimates were pooled on the reported per-unit scale whenever available. Each predictor was synthesised separately using aggregate-level pairwise random-effects pooling. We did not perform multivariable meta-analysis, network meta-analysis or individual patient data meta-analysis; therefore, this approach could not model intervariable relationships, collinearity, mediation or the independence of one predictor from another. When three or more studies reported on the same predictor, we pooled adjusted ORs using a random-effects model (DerSimonian-Laird). As a sensitivity analysis, we repeated the eligible random-effects meta-analyses using restricted maximum likelihood estimation of between-study variance with Hartung-Knapp-Sidik-Jonkman CIs. This analysis was performed to assess whether the direction and statistical significance of the pooled estimates were robust to a more conservative random-effects approach, particularly for predictors informed by a small number of studies. Because this review primarily pooled adjusted ORs from multivariable models, corresponding raw 2×2 data were not consistently available across studies and could not be uniformly presented in the forest plots. Heterogeneity was quantified with I² and Cochran’s Q. Formal assessment of publication bias was not performed because many pooled predictors were informed by a limited number of studies, making funnel plots and Egger’s test unreliable. Subgroup analyses according to risk of bias category were considered but were not feasible for most predictors because too few studies contributed to individual pooled estimates to support meaningful stratification. All analyses were performed in R using the ‘meta’ and ‘metafor’ packages. Results Study selection and characteristics We retrieved 452 articles from databases and registries, along with 23 additional articles identified through hand-searching. After screening, retrieval and eligibility assessment, 75 studies were included,3 20–93 with a total of 12 094 participants. The PRISMA flow diagram detailing the literature search and selection process is shown in figure 1. The existing studies were obtained from various centres worldwide. Most of the studies found from the literature search were retrospective (online supplemental file 1) provides the detailed patient characteristics in the included studies. The included studies were conducted across multiple countries and consisted predominantly of retrospective observational cohorts, with smaller contributions from prospective studies and randomised trials (online supplemental file 1) summarises the key study level characteristics, including design, setting, sample size and major baseline clinical variables. Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 flow diagram of study identification, screening, eligibility and inclusion. Risk of bias We evaluated all 75 included cohort and registry studies using the ROBINS-E tool, covering seven domains. More than one-quarter of included studies were judged to have high or very high overall risk of bias, mainly because of residual confounding, incomplete outcome data and lack of blinded or independent outcome assessment. Most studies (80%) appropriately adjusted for key confounders, and nearly all clearly defined the thrombectomy technique used. However, several concerns emerged—particularly regarding deviations from intended interventions, such as crossovers or rescue procedures, often influenced by operator preference. In addition, over 5% of the cohorts had missing 90-day mRS data, outcome assessment was frequently unblinded or lacked independent or core laboratory evaluation, and many studies did not report a prespecified analysis protocol. A detailed summary of the ROBINS-E assessment is shown in online supplemental file 1. Results of syntheses We evaluated 25 candidate predictors of 90-day functional independence after posterior-circulation thrombectomy, of which 16 showed statistically significant associations (table 1; figures 2–6). We found neither advancing age (per 1-year increment; figure 2A) nor female sex (figure 2B) carried a statistically significant effect on outcome (both p>0.050), suggesting that demographic characteristics alone did not predict recovery. Table 1Pooled effect estimates for predictors of 90-day functional independence after posterior circulation thrombectomyNoVariableComparisonOR (95% CI)I² (%)P value1Diabetes mellitusvs no diabetes0.56 (0.43 to 0.72)24.63<0.0012Hypertensionvs no hypertension0.73 (0.59 to 0.91)0.000.013Current/former smokingvs never smoked1.48 (1.00 to 2.18)35.700.054Alcohol consumptionvs no alcohol1.15 (0.37 to 3.58)59.730.815Large artery atherosclerosisvs no atherosclerosis0.76 (0.59 to 0.97)0.000.036Pre-stroke mRS scoreper 1 score increment0.66 (0.51 to 0.84)0.00<0.0017GCS at admissionper 1 score increment1.13 (1.01 to 1.26)33.480.048Baseline NIHSS scoreper 1 score increment0.92 (0.89 to 0.95)85.09<0.0019pc-ASPECTSper 1 score increment1.80 (1.59 to 2.03)0.00<0.00110PCCS >5vs ≤51.78 (1.19 to 2.67)0.000.0111Distal basilar occlusionvs non-distal1.63 (1.03 to 2.57)59.360.0412IV-tPA administrationvs no IV-tPA1.66 (1.31 to 2.10)21.67<0.00113General anaesthesiavs conscious sedation0.56 (0.36 to 0.89)64.530.0114Stent retrievervs contact aspiration0.52 (0.36 to 0.76)1.95<0.00115First pass effectvs no FPE2.36 (1.84 to 3.03)9.28<0.00116Final mTICI 2b/3vs mTICI 0–2a3.58 (2.31 to 5.55)58.43<0.001FPE, first pass effect; GCS, Glasgow Coma Scale; IV-tPA, intravenous tissue plasminogen activator; mRS, modified Rankin Scale; mTICI, modified thrombolysis in cerebral infarction; NIHSS, National Institutes of Health Stroke Scale; pc-ASPECTS, posterior circulation Alberta Stroke programme Early CT Score; PCCS, posterior-circulation collateral score. Pooled effect estimates for predictors of 90-day functional independence after posterior circulation thrombectomy FPE, first pass effect; GCS, Glasgow Coma Scale; IV-tPA, intravenous tissue plasminogen activator; mRS, modified Rankin Scale; mTICI, modified thrombolysis in cerebral infarction; NIHSS, National Institutes of Health Stroke Scale; pc-ASPECTS, posterior circulation Alberta Stroke programme Early CT Score; PCCS, posterior-circulation collateral score. Demographic factors associated with functional independence after posterior-circulation mechanical thrombectomy: (A) advancing age (per 1-year increase); (B) female sex. ES, effect size; REML, restricted maximum likelihood. Vascular comorbidities factors associated with functional independence after posterior-circulation mechanical thrombectomy: (A) diabetes mellitus; (B) hypertension; (C) atrial fibrillation; (D) hyperlipidaemia; (E) chronic kidney disease; (F) prior stroke or transient ischaemic attack; (G) current or former smoking; (H) alcohol consumption. ES, effect size; REML, restricted maximum likelihood. Stroke-severity measures associated with functional independence after posterior-circulation mechanical thrombectomy: (A) pre-stroke modified Rankin Scale per 1-point increase; (B) baseline National Institutes of Health Stroke Scale per 1-point increase; (C) Glasgow Coma Scale at admission per 1-point increase. ES, effect size; REML, restricted maximum likelihood. Imaging-derived factors associated with functional independence after posterior-circulation mechanical thrombectomy: (A) posterior-circulation Alberta Stroke Programme Early CT Score per 1-point increase; (B) posterior-circulation collateral score >5; (C) presence of distal basilar artery occlusion; (D) presence of large-artery atherosclerosis. ES, effect size; REML, restricted maximum likelihood. Procedural, treatment and workflow factors associated with functional independence after posterior-circulation mechanical thrombectomy: (A) final modified thrombolysis in cerebral infarction 2b/3 reperfusion; (B) first-pass reperfusion; (C) intravenous thrombolysis; (D) first-line stent retriever alone versus contact aspiration; (E) combined stent retriever and aspiration vs contact aspiration alone; (F) general anaesthesia versus conscious sedation; (G) door-to-groin puncture time (per 1-minute increase); (H) puncture-to-recanalisation time (per 1-minute increase). ES, effect size; REML, restricted maximum likelihood. For vascular comorbidity, diabetes mellitus (figure 3A) and hypertension (figure 3B) each showed significantly lower odds of good outcome (OR=0.56, 95% CI 0.43 to 0.72, I²=24.6%, p<0.001; and OR=0.73, 95% CI 0.59 to 0.91, I²=0%, p=0.010, respectively), whereas atrial fibrillation (figure 3C), hyperlipidaemia (figure 3D), chronic kidney disease (figure 3E) and prior stroke or TIA (figure 3F) did not reach significance. Among lifestyle-related factors, neither current or former smoking (figure 3G) nor alcohol consumption (figure 3H) was significantly associated with functional independence (OR=1.48, 95% CI 1.00 to 2.18, I²=35.7%, p=0.050; and OR=1.15, 95% CI 0.37 to 3.58, I²=59.7%, p=0.810, respectively). Within stroke-severity measures, each 1-point increase in pre-stroke mRS was associated with a 34% reduction in odds of favourable outcome (OR=0.66, 95% CI 0.51 to 0.84, I²=0%, p<0.001; figure 4A), and every additional point on the baseline NIHSS scores was related to poor recovery (OR=0.92 per point, 95% CI 0.89 to 0.95, I²=85.1%, p<0.001; figure 4B), although with substantial heterogeneity. In conjunction, each 1-point increase in admission Glasgow Coma Scale (GCS) modestly improved odds of functional independence after posterior-circulation thrombectomy (OR=1.13, 95% CI 1.01 to 1.26, I²=33.5%, p=0.040; figure 4C). For imaging-derived factors, each point increase in pc-ASPECTS nearly doubled the odds of functional independence (OR=1.80, 95% CI 1.59 to 2.03, I²=0%, p<0.001; figure 5A), and a posterior-circulation collateral score above five similarly favoured good outcome (OR=1.78, 95% CI 1.19 to 2.67, I²=0%, p=0.010; figure 5B). Patients with distal basilar artery occlusion also had higher odds of functional independence compared with other locations (OR=1.63, 95% CI 1.03 to 2.57, I²=59.4%, p=0.040; figure 5C), although with substantial heterogeneity. In conjunction, patient with large-artery atherosclerosis also predicted worse recovery (OR=0.76, 95% CI 0.59 to 0.97, I²=0%, p=0.030; figure 5D). In the treatment and workflow domain, final modified thrombolysis in cerebral infarction (mTICI) 2b/3 reperfusion was the strongest procedural predictor (OR=3.58, 95% CI 2.31 to 5.55, I²=58.4%, p<0.001; figure 6A), with achieving first-pass reperfusion doubling the likelihood of independence (OR=2.36, 95% CI 1.84 to 3.03, I²=9.3%, p<0.001; figure 6B). Administration of intravenous thrombolysis (IVT) also showed higher odds of functional independence (OR=1.66, 95% CI 1.31 to 2.10, I²=21.7%, p<0.001; figure 6C). Meanwhile, first-line use of stent retriever alone was significantly less effective compared with contact aspiration (OR=0.52, 95% CI 0.36 to 0.76, I²=1.9%, p<0.001; figure 6D). However, their combination did not show a statistically significant difference compared with contact aspiration alone (OR=0.52, 95% CI 0.23 to 1.19, I²=12.42%, p=0.120; figure 6E). General anaesthesia was also associated with lower odds of good outcome than conscious sedation (OR=0.56, 95% CI 0.36 to 0.89, I²=64.5%, p=0.010; figure 6F). Neither door-to-groin puncture (figure 6G) nor puncture-to-recanalisation (figure 6H) per 1-minute increment reached statistical significance. Sensitivity analysis using restricted maximum likelihood estimation with Hartung-Knapp-Sidik-Jonkman CIs was broadly consistent with the primary DerSimonian-Laird analysis for several key findings. Baseline NIHSS, pc-ASPECTS, IVT, first-pass effect, final mTICI 2b/3 reperfusion, diabetes mellitus, hypertension, general anaesthesia and first-line stent retriever versus contact aspiration remained statistically significant in the same direction. In contrast, admission GCS, posterior-circulation collateral score (PCCS) >5, distal basilar occlusion and large-artery atherosclerosis were attenuated and were no longer statistically significant under the Restricted Maximum Likelihood-Hartung–Knapp–Sidik–Jonkman (REML-HKSJ) approach. The full sensitivity analysis is presented in online supplemental table S1. Discussion In this aggregate-level systematic review and meta-analysis of 75 studies, the factors most consistently reported with favourable 90-day functional independence after MT for PC stroke were successful reperfusion, first-pass reperfusion, higher pc-ASPECTS, PCCS >5, higher admission GCS and distal basilar occlusion. In contrast, higher baseline NIHSS, higher pre-stroke mRS, diabetes mellitus, hypertension, large artery atherosclerosis, general anaesthesia and first-line stent retriever use were more often reported with lower odds of functional independence. These findings should be interpreted as a study-level summary of reported associations rather than evidence that these variables are independent prognostic factors, because the included studies used different adjustment models and our meta-analysis pooled each predictor separately. Several estimates showed substantial heterogeneity and should therefore be interpreted as hypothesis-generating. The REML-HKSJ sensitivity analysis supported the direction of several primary findings but also showed that some borderline associations were not robust to a more conservative random-effects approach. In particular, admission GCS, PCCS >5, distal basilar occlusion and large-artery atherosclerosis were no longer statistically significant in the sensitivity analysis. This reinforces the need to interpret these variables cautiously, particularly when estimates are based on few studies, heterogeneous populations or non-uniform adjustment models. The most consistent findings across the primary and sensitivity analyses were related to baseline NIHSS, pc-ASPECTS, IVT, first-pass effect, final mTICI 2b/3 reperfusion, diabetes mellitus, hypertension, general anaesthesia and first-line stent retriever versus contact aspiration. However, because this was an aggregate-level synthesis of variably adjusted estimates, these findings should not be interpreted as independent prognostic effects. Regarding demographic factors, age and gender in this study analysis were not found to be significant for patient functional outcomes. In general, Rhiner et al found that increasing age is associated with a higher prevalence of age-related comorbidities, a decline in functional independence outcomes and an increased risk of mortality in MT procedures.94 95 Gender in our analysis also did not significantly influence patient functional outcomes after MT. So further research is needed to determine the influence of gender on patient functional outcomes after MT.96 Although current or former smoking and alcohol consumption showed a trend toward higher odds of functional independence, these counterintuitive findings were not statistically significant and should not be interpreted as protective effects. The included studies were not designed to evaluate lifestyle exposures, and these variables were likely affected by residual confounding, survivor bias, incomplete capture of comorbidities, differences in baseline severity and non-uniform adjustment models across predominantly retrospective studies. Therefore, these findings should be interpreted with caution and are not considered clinically actionable prognostic factors. Regarding imaging factors, pc-ASPECTS score and PCCS score above 5 are associated with a significantly increased odds of post-MT functional outcomes. Other studies have also found similar findings, indicating that pc-ASPECTS is a significant independent predictor of functional outcomes.32 In the BAOCHE11 and ATTENTION97 trials, the selection of research subjects used the pc-ASPECTS score ≥6, and in the BAOCHE trial, the population included patients with 6–24 hours after the onset of stroke symptoms.96 Good PC collateral flow is known to prolong the time of decline in brainstem tissue function and prevent necrosis.98 99 In the BAOCHE11 trial, MT was found to still show good outcomes in patients with basilar artery occlusion even though it was up to 24 hours after onset.11 This may indicate that the potential of the pc-ASPECTS score and collateral evaluation can be used in determining patients who may still benefit from MT action in conditions that are late in onset.11 67 100 Distal basilar artery occlusion is associated with good functional outcomes. Distal occlusion is more likely to have better collateral status, experience fewer neurological symptoms, lower clot burden and a higher probability of successful recanalisation.22 Distal basilar occlusion is also said to be associated with the most common aetiology of cardioembolism, which is more easily recanalised.77 However, substantial heterogeneity suggests the need for more comprehensive research regarding clot location and functional outcome after MT. Regarding procedural factors, successful reperfusion (achieving mTICI 2b/3) and first-pass reperfusion are also associated with good functional outcomes. This is similar to the study by Zaidat et al, which found that patients who achieved a TICI 2b/3 on the first pass had better clinical outcomes.101 Studies have shown better neurological clinical improvement at discharge, better 90-day clinical outcomes and a reduced risk of distal embolisation in patients who achieve first-pass efficacy.41 101 102 Regarding first-line thrombectomy technique, first-line stent retriever use was associated with lower odds of functional independence compared with contact aspiration in the pooled analysis. This finding should be interpreted cautiously because device generation, operator preference, rescue therapy, clot aetiology, vascular anatomy and centre-level workflow may have influenced technique selection. Previous studies have reported mixed findings when comparing aspiration and stent retriever thrombectomy in PC stroke, and some differences may reflect heterogeneity in devices, operator experience and rescue strategies rather than a true superiority of one technique. Therefore, the present result should not be taken as definitive evidence that one first-line technique is superior to another. In a meta-analysis, Sheng et al showed no significant difference in clinical outcomes between the two techniques.103 In theory, the aspiration technique may be beneficial due to shorter procedure times, damage to the vessel wall is less frequent and the risk of perforator artery occlusion is lower.28 41 104 But, stent retrieval is also less frequently performed in the posterior arteries due to the smaller vertebral arteries and the presence of collateral flow.96 In our analysis, IVT was associated with higher odds of functional independence. This finding is consistent with some previous studies reporting favourable outcomes with bridging therapy in PC stroke. However, IVT use may be influenced by onset-to-treatment time, transfer status, eligibility criteria, contraindications and local workflow. Therefore, this result should be interpreted as an aggregate association rather than definitive evidence that bridging therapy is superior to direct MT. Weber et al showed that IVT as bridging therapy was associated with a lower risk of death in patients with posterior stroke undergoing MT.21 The study found that MT alone was as effective as the combination of IVT+MT regarding functional outcomes at 90 days.105 Prior treatment with IVT can lead to delayed MT, which is associated with poorer clinical outcomes in the setting of a drip-and-ship protocol.105 General anaesthesia has been found to result in lower odds of a good outcome compared with conscious sedation, with quite a heterogeneity. This may be related to the selection of patients for general anaesthesia, which is typically associated with poorer functional status, higher stroke rates or a greater burden of comorbidities. Furthermore, physician and patient preferences may also play a role.106 Complications from heavy manipulation are a concern in patients with conscious sedation. Patient movement while decreased consciousness or due to pain from the stent retriever can lead to manipulation errors and can lead to vessel perforation or perforator artery rupture due to sudden displacement.107 This systematic review included a large sample and diverse study settings, and adjusted estimates were used to reduce bias. However, several limitations remain. Most included studies were retrospective observational studies, making the findings vulnerable to residual confounding because adjustment methods varied across studies. This is supported by the risk-of-bias assessment, which identified more than one-quarter of studies as having high or very high risk of bias. Several pooled estimates also showed substantial heterogeneity (I²=>50%), including baseline NIHSS, distal basilar occlusion, final mTICI 2b/3 and anaesthesia strategy, while some estimates were based on only a few studies, limiting precision and the ability to detect true heterogeneity. Conversely, for predictors synthesised from only a small number of studies, an I² value of 0% should not be interpreted as evidence of true homogeneity, because the power to detect between-study heterogeneity is limited when few studies are available. In addition, formal sensitivity analyses stratified by study design were not feasible for most predictors because randomised trials contributed sparse and inconsistently reported adjusted estimates. The retrospective design may also have restricted the assessment of important confounders, such as comorbidities and treatment selection. Additionally, lack of blinding in outcome assessment, variation in thrombectomy techniques and definitions and inconsistent reporting of time metrics may have influenced the results. Although no language restrictions were applied, relevant studies may have been missed because of database and indexing limitations. Publication bias cannot be excluded and could not be formally assessed for most predictors due to the small number of available studies. Finally, the use of aggregate-level rather than individual patient data may have limited the evaluation of patient-specific factors affecting outcomes. The use of MT in posterior stroke trials still resulted in a significant number of relatively poor functional outcomes.11 97 Therefore, the results of this meta-analysis can be a reference for clinicians in determining the prognosis of patients with posterior stroke admitted. Future prospective studies or further large-scale randomised clinical trials are needed, assessing functional outcomes based on uniform and clear protocols. Future studies should further evaluate workflow-related factors, including anaesthesia strategy and approaches to achieving first-pass reperfusion, using standardised definitions and prospective data collection. Conclusion In this aggregate-level systematic review and meta-analysis, successful reperfusion, first-pass effect, higher pc-ASPECTS, IVT, lower baseline NIHSS, lower pre-stroke mRS and selected procedural factors were the most consistently reported factors associated with 90-day functional independence after mechanical thrombectomy for PC stroke. These findings should be interpreted as a study-level summary of reported associations rather than evidence that these variables are independent prognostic factors, because the included studies used different adjustment models and our meta-analysis pooled each predictor separately. Several estimates showed substantial heterogeneity or were attenuated in the REML-HKSJ sensitivity analysis and should therefore be interpreted as hypothesis-generating. Future prospective studies using standardised definitions, harmonised adjustment sets and individual patient-level data are needed to validate these prognostic factors. Supplementary material