Acute ischemic stroke (AIS) caused by large vessel occlusion (LVO) remains a leading cause of death and disability worldwide, with mechanical thrombectomy (MT) having emerged as the cornerstone of reperfusion therapy for eligible patients (1). In clinical practice, many patients who qualify for MT also meet criteria for intravenous thrombolysis (IVT), leading to the use of “bridging therapy” (BT)—IVT administration followed by MT—as a standard strategy recommended by international guidelines (2). The rationale behind BT lies in its potential to promote earlier recanalization, alter clot properties to facilitate mechanical retrieval, and address distal emboli that may be inaccessible to thrombectomy (3).

However, the clinical utility of BT compared with direct MT (dMT) has been the subject of considerable debate. While several observational studies have historically suggested that BT may be associated with better functional outcomes (4–8), randomized controlled trials (RCTs) designed to answer this question have yielded mixed results. To date, six RCTs have compared BT and dMT head-to-head, with most showing comparable efficacy and safety profiles between the two strategies (9–14). These trials have been synthesized in multiple recent meta-analyses (15, 16), which have increasingly guided clinical practice and policy recommendations.

Yet, RCTs, by design, enroll carefully selected patient populations under controlled conditions and may not fully reflect the complexity, comorbidities, and variations in care delivery present in real-world settings. Moreover, the existing RCTs differ considerably in design, population demographics, IVT protocols (e.g., alteplase vs. tenecteplase), and timing metrics, and some were underpowered or terminated early, further limiting their external validity. Real-world data—derived from observational studies, registries, and routine clinical practice—can provide valuable complementary insights, particularly concerning safety outcomes and generalizability.

A recent review by Qin et al. (17) synthesized real-world evidence from 12 registry-based studies, supporting the benefit of BT over dMT in terms of functional outcomes and mortality without a significant increase in hemorrhagic risk. However, the field has rapidly evolved, with new large-scale observational studies published in the interim. Furthermore, prior reviews often relied on unadjusted data or pooled estimates without fully addressing potential confounding through statistical matching or regression adjustment.

In this context, we conducted an updated meta-analysis focusing exclusively on non-randomized real-world studies comparing BT versus dMT in patients with AIS due to LVO. Our primary objective was to re-evaluate the safety and efficacy of BT using the most recent and methodologically robust observational evidence. Importantly, by limiting our analysis to non-RCT data, we aimed to capture a more accurate depiction of treatment performance in routine clinical practice and later compare these findings with the aggregated evidence from the six existing RCTs in the discussion.

This updated meta-analysis of over 93,000 real-world patients offers compelling evidence favoring the use of BT over dMT in AIS. Compared with dMT, BT was associated with significantly greater odds of achieving excellent and favorable functional outcomes at 90 days, higher rates of successful reperfusion, and lower mortality—without a corresponding increase in the risk of hemorrhagic complications. These findings are consistent with, and extend upon, prior real-world data (17) and provide critical complementary insights to those generated by RCTs (15).

Over the past few years, at least six major RCTs—DIRECT-MT (9), DEVT (10), SKIP (11), MR CLEAN-NO IV (12), SWIFT-DIRECT (13), and DIRECT-SAFE (14)—have attempted to resolve the clinical equipoise surrounding BT versus dMT in IVT-eligible patients. Pooled results from these RCTs generally support the non-inferiority of dMT for favorable functional outcomes (15, 16, 48), but their conclusions are tempered by regional heterogeneity, differences in imaging and selection protocols, and strict eligibility criteria. For example, Liu et al. (15) emphasized that while dMT met non-inferiority thresholds in some trials, others failed to demonstrate equivalence, underscoring unresolved uncertainty in specific subgroups.

In contrast, observational data—including the current study—reflect broader clinical settings, encompassing variations in hospital infrastructure, workflow logistics, and patient-level risk factors that are often underrepresented in RCTs. In this context, the term ‘real-world’ refers to several aspects of routine clinical practice that differ from the tightly controlled conditions of RCTs. These include broader and more heterogeneous inclusion criteria, encompassing patients with wider ranges of stroke severity, comorbidities, imaging profiles (including lower ASPECTS), and treatment windows. Real-world cohorts also reflect substantial variability in workflow metrics—such as onset-to-imaging, door-to-needle, and door-to-groin times—as well as differences in hospital infrastructure, transfer patterns, operator expertise, and local protocols for IVT and thrombectomy. Additionally, physician decision-making, system-level delays, and regional differences in prehospital logistics influence treatment delivery in ways not captured in trial settings. Together, these elements provide a more comprehensive and externally valid representation of how BT and dMT perform in everyday practice. Our findings align closely with those of Katsanos et al. (49), who similarly reported superior functional outcomes and reduced mortality with BT across 11,798 patients, even after adjustment for confounders. Likewise, a recent umbrella review by Campbell et al. (50) reiterated the modest yet consistent benefits of BT for functional recovery and reperfusion success, particularly in systems with efficient in-hospital workflows and minimal IVT-to-groin delay.

Importantly, this study contributes an up-to-date synthesis of real-world data at a larger scale than any previously published review (17). Compared to Waller et al. (51), who summarized older data from 2015 to 2020 and noted favorable BT outcomes in patients with anterior circulation LVOs, our analysis captures evolving practices, wider geographical representation, and improved adjustment for confounding via propensity-score matching or regression methods in 77% of included studies.

Concerns surrounding the potential risks of BT—particularly increased rates of thrombus fragmentation, distal embolization, or hemorrhagic transformation—remain important. However, in our meta-analysis, BT was not associated with increased rates of either symptomatic or asymptomatic intracerebral hemorrhage. However, this conclusion requires caution, particularly for aICH. The substantial heterogeneity among studies reporting aICH and the sensitivity finding—where exclusion of a single study shifted the effect toward significance—indicate that uncertainty remains. Differences in imaging criteria, reporting standards, and patient selection may contribute to this variability, making the true effect of BT on aICH less certain than the pooled summary suggests. This echoes the conclusions of Tsivgoulis et al. (52), who reported no excess bleeding risk with BT in their large observational cohort. Furthermore, the absence of heterogeneity for sICH across our included studies supports the robustness of this finding.

From a pathophysiological perspective, IVT may facilitate microvascular reperfusion, soften thrombus structure, and promote early partial recanalization before thrombectomy—mechanisms hypothesized in several translational studies and supported by improved TICI 2b/3 rates in both this and earlier meta-analyses (49, 53). Moreover, BT may act as a safety net in cases of failed or delayed thrombectomy, a scenario not uncommon in real-world settings but under-represented in RCTs. Additionally, intravenous thrombolysis may preferentially target the peripheral, branching projections of the clot—a phenomenon often described as the ‘finger-like thrombus’ effect—thereby reducing clot anchoring strength and improving device engagement during thrombectomy (54). IVT has also been shown to enhance microcirculatory reperfusion by dissolving distal microemboli and restoring capillary-level flow, effects that are not directly achieved by mechanical thrombectomy alone (55). Together, these mechanisms provide complementary biological pathways through which BT may improve successful reperfusion rates and functional outcomes, particularly in real-world settings where procedural delays or anatomical challenges may limit the effectiveness of dMT.

To further explore sources of heterogeneity and strengthen the interpretability of our findings, we performed meta-regression analyses across all primary outcomes using study-level covariates. Several demographics, clinical, imaging, and workflow-related characteristics were identified as significant modifiers of treatment effect (Table 3). Higher baseline ASPECTS and NIHSS scores were consistently associated with better odds of functional recovery, underscoring the importance of initial infarct size and neurological severity. Time intervals—particularly onset-to-imaging, imaging-to-groin, and door-to-groin metrics—also influenced effect estimates, highlighting the well-established impact of workflow efficiency on reperfusion success. In addition, risk-factor profiles such as hypertension, prior stroke, and smoking contributed to variations in mortality, sICH, or reperfusion outcomes. Although these associations should be interpreted cautiously given the ecological nature of study-level meta-regression, they provide important insights into factors that may shape real-world treatment performance and help explain some of the heterogeneity observed across included studies.

Still, our study is not without limitations. The non-randomized design of included studies introduces the possibility of residual confounding, although most employed statistical adjustments. The definition and adjudication of outcomes (e.g., functional independence, hemorrhage classification) varied across studies. Also, the predominance of data from high-income settings may limit generalizability to low-resource environments. Additionally, although most included studies performed statistical adjustment, the extent to which they accounted for thrombus characteristics or collateral circulation varied substantially. Several studies included imaging-based covariates such as collateral grade, mCTA collateral score, pial arterial collateral status, ASITN/SIR collateral grade, ASPECTS, and occlusion site, which partially capture thrombus burden and baseline tissue viability (see Supplementary Table S3). However, detailed thrombus morphology (e.g., clot length, clot perviousness) and standardized collateral scoring were not uniformly reported across studies. This heterogeneity may introduce residual confounding in reperfusion and functional outcomes. As such, while the majority of analyses adjusted for key clinical and imaging variables, incomplete adjustment for thrombus- and collateral-related factors remains an inherent limitation of real-world observational evidence.

Nevertheless, our findings offer a clinically meaningful perspective: in routine clinical practice, BT appears to confer modest yet consistent advantages in functional recovery and survival without added hemorrhagic risk. These real-world benefits contrast with the marginal differences reported in RCTs, raising important questions about the external validity of trial-based treatment algorithms. A structured comparison of RCT evidence versus real-world data is presented in Table 4 to illustrate these differences. Future research should prioritize individual patient data meta-analyses and explore tailored approaches to stroke triage, perhaps incorporating clot burden, collateral status, and time metrics into treatment selection.

This updated meta-analysis of real-world observational data reinforces the clinical value of bridging therapy in patients with acute ischemic stroke due to large vessel occlusion. Compared to direct mechanical thrombectomy, bridging therapy was associated with significantly improved functional outcomes, higher rates of successful reperfusion, and lower mortality—without an increased risk of hemorrhagic complications. These findings highlight the continued relevance of intravenous thrombolysis as a component of acute stroke care and suggest that, in real-world settings, the benefits of bridging therapy may be more pronounced than those observed in randomized controlled trials. As treatment paradigms evolve, individualized decision-making that incorporates both trial-based evidence and real-world data will be essential to optimizing outcomes for stroke patients.