Endovascular thrombectomy (EVT) has become the standard of care for patients with acute ischemic stroke caused by large vessel occlusion. Current guidelines advocate for a combined approach of EVT with best medical management, such as intravenous thrombolysis (IVT), specifically for large vessel occlusions in the anterior circulation (1–3).

The location of arterial occlusion, influenced by the thrombus origin and individual vascular anatomy, plays a critical role in the decision to perform EVT and affects clinical outcomes. Thrombus migration (TM), defined as the distal movement of thrombus, is occasionally observed in patients with large vessel occlusion and has been increasingly recognized with the widespread use of EVT. However, its clinical relevance and predictive factors remain unclear. TM may result from spontaneous displacement or partial thrombus resolution, especially after IVT. Considering this mechanism, factors known to predict recanalization after IVT in large vessel occlusion, such as thrombus length, composition, and time to treatment (4–6), may influence the likelihood of TM. Previous studies have suggested that IVT (3, 7) and thrombus burden (8) may be associated with the occurrence of TM. Additionally, TM may be influenced by various hemodynamic factors, collateral status, or premorbid antithrombotic treatment.

Despite these insights, few studies have comprehensively evaluated the combined effect of these variables on TM. We hypothesized that thrombus characteristics and time parameters may be the key factors for TM. Therefore, we aimed to investigate the natural course and independent predictors of TM and their impact on outcomes in patients undergoing EVT through detailed clinical and radiological evaluation.

In this study, 22.1% of patients undergoing EVT developed TM. TM was significantly associated with HAS, diastolic blood pressure, and shorter onset-to-arrival time, but not with IVT. Patients with TM showed greater improvement in NIHSS scores and higher rates of ENI after EVT, though this did not affect 3-month functional outcomes. While TM was linked to slightly higher rates of FPE, successful reperfusion rates remained similar between groups. However, patients with TM had significantly higher rates of parenchymal hemorrhage.

Our data suggest that thrombus composition, as indicated by HAS, may be a strong predictor of TM. HAS on non-contrast CT represents erythrocyte-rich clots (20, 21) and correlates with better recanalization response to reperfusion therapy (22–24). Importantly, erythrocyte-rich clots exhibit relatively less friction against vessel walls compared with fibrin-rich clots (25). This reduced friction may enable thrombi with HAS to migrate more readily to distal arteries while also demonstrating better response to both physiologic and pharmacologic thrombolysis. This mechanism is supported by previous studies showing that TM occurs more frequently with erythrocyte-rich thrombi, as confirmed by histological examination in patients with acute ischemic stroke (26, 27). These findings align with our observations. As HAS is an effective imaging biomarker of thrombus composition that is relatively easy to assess, it could be a valuable clinical tool for predicting TM and informing procedural planning strategies. Although histological validation was not available in our cohort, HAS remains a robust surrogate marker, as it reflects native clot characteristics unaffected by periprocedural interventions.

Beyond composition, we examined temporal factors influencing TM. Few studies have explored the relationship between time parameters and TM. One previous study found that longer intervals between initial CT angiography and DSA were associated with TM, possibly due to the time needed for thrombolysis after intravenous therapy or delayed detection of migration itself (28). Conversely, our study found that the TM group had shorter onset-to-arrival times (median 75 vs. 135 min), with shorter times independently associated with increased TM risk. This finding aligns with previous reports of shorter onset-to-imaging or puncture times in TM-positive cases (3, 7, 28).

This temporal relationship can be explained by thrombus maturation dynamics. Fresh thrombi have a core-shell architecture with a fibrin/platelet-rich outer shell surrounding an erythrocyte-rich core (29, 30). Early after occlusion, the erythrocyte-rich core maintains greater deformability, making the thrombus more susceptible to migration. Over time, thrombus composition shifts toward increased fibrin content throughout, particularly reinforcing the outer shell and reducing erythrocyte proportion in the core. This structural evolution increases stability and rigidity, thereby decreasing migration tendency. Additionally, shorter onset-to-arrival times correlate with an increased use of IVT, which may promote TM through partial thrombolysis of the vulnerable erythrocyte-rich core.

Our results revealed that lower diastolic blood pressure was independently associated with an increased risk of TM. Lower diastolic blood pressure has been related to various cardiac pathologies, including impaired coronary perfusion and myocardial injury, leading to coronary artery disease and heart failure (31–33). These cardiac conditions may create a prothrombotic environment that facilitates intracardiac thrombus formation, as well as predisposition to atrial fibrillation. Consistent with these findings, previous studies have reported that lower diastolic blood pressure was associated with cardioembolic stroke, in which thrombi are typically less adherent to vessel walls compared to the in situ thrombi from atherosclerotic disease (34, 35). Moreover, lower diastolic blood pressure may reflect reduced peripheral vascular resistance or diminished distal perfusion pressure, potentially altering hemodynamic forces at the thrombus-vessel interface and promoting thrombus dislodgement (36). This novel association suggests the need for further research into the hemodynamic mechanisms underlying TM and may have important implications for blood pressure management strategies during the hyperacute phase of ischemic stroke before reperfusion therapy.

Several factors previously reported to be associated with TM, including IVT, collateral circulation, and CBS, showed no significant association with TM in our study. Previous studies have shown conflicting results regarding the association between IVT and TM, with some reporting a positive association (3, 7, 28, 37), while others found no relationship (38). Contrary to expectations, our analysis did not demonstrate an independent association between IVT and TM adjustment for confounders. This discrepancy from previous studies may be attributable to two temporal factors. First, our cohort had markedly shorter image-to-puncture times compared to prior studies, which closely approximated the time from IVT administration to puncture. Consequently, TM assessment on DSA may have occurred before IVT could exert its full therapeutic effect, potentially underestimating its influence on TM. Second, few studies investigating the relationship between IVT and TM have reported onset-to-arrival times in the IVT versus no-IVT groups (3, 28). In our cohort, patients who received IVT arrived much earlier than those who did not [median 66 vs. 224 min]. As thrombus characteristics change over time, this difference suggests that intrinsic clot characteristics, rather than IVT itself, may influence TM occurrence.

The absence of associations with collateral circulation status and CBS provides additional insights into TM mechanisms. These findings suggest that hemodynamic factors and clot size may be less important than intrinsic thrombus composition in determining migration likelihood.

In our study, patients with TM showed greater NIHSS improvement at 24 h and trends toward higher rates of ENI and first-pass effect. However, these early advantages did not translate into significant differences in the 3-month functional outcomes between groups. The presence of TM may indicate a more fragile or mobile thrombus composition that facilitates clot retrieval, potentially explaining the improved early neurological recovery. By definition, TM leads to more distal occlusion sites, which can alter EVT strategy, such as the use of smaller devices or even the decision to attempt EVT in very distal lesions. Nevertheless, successful reperfusion (mTICI 2b–3) was comparable between groups, indicating that EVT effectiveness was not diminished by distal migration. Previous research on TM’s clinical impact has yielded contradictory findings. Some studies report better functional outcomes with TM despite lower complete reperfusion rates (3, 7), while others found reduced successful reperfusion and consequently diminished neurological improvement in patients with TM (38). This paradox may reflect TM’s dual nature. On the one hand, TM can hinder recanalization when thrombi migrate to distal, less accessible vessels, potentially worsening outcomes. Conversely, distal migration may relieve proximal obstruction and enhance collateral perfusion, contributing to early symptom improvement. In our cohort, TM did not compromise reperfusion success and appeared to positively influence early neurological outcomes, possibly through improved distal perfusion.

Notably, TM was significantly associated with higher rates of parenchymal hemorrhage (13.0% vs. 5.2%). The underlying mechanism remains unclear. TM may not directly cause hemorrhage but rather indicate thrombus characteristics that predispose to bleeding complications. Erythrocyte-rich thrombi, which are more migration-prone as shown by our HAS findings, may respond more aggressively to thrombolytic therapy, increasing hemorrhage risk. Alternatively, distal migration may contribute to hemorrhagic complications when thrombi migrate to smaller vessels with compromised blood–brain barrier integrity. Subsequent reperfusion of these vulnerable territories could precipitate hemorrhagic transformation. Additionally, patients with TM had shorter onset-to-arrival times and may have received higher IVT during the early period when hemorrhage risk was elevated. Given the increased hemorrhage risk associated with TM, enhanced monitoring protocols may be warranted for patients presenting with predictive factors such as HAS and early onset.

Despite the higher frequency of PH in the TM group, the incidence of symptomatic ICH was similar between groups. In our study, symptomatic ICH was defined as any ICH, including SAH, that was accompanied by clinical worsening, whereas PH represents only a subtype of intracerebral hemorrhage. Accordingly, PH and symptomatic ICH do not necessarily coincide, and many PH cases may not result in significant neurological deterioration. Furthermore, distal migration in the TM group may have led to smaller or non-eloquent hemorrhages, which could also explain the absence of a higher symptomatic ICH rate despite the increased occurrence of PH.

In this study, 22.1% of patients who underwent EVT developed TM, a rate consistent with previous reports ranging from 15 to 25% (3, 8, 38). TM was most commonly observed in patients with initial ICA and MCA M1 segment occlusions, with the most frequent migration patterns being M1 to M2 (19 patients) and ICA to M1 (13 patients). The absence of migration from M2 segments to more distal branches in our EVT-treated cohort likely reflects our inclusion criteria, as patients with very distal migrations may not have met indications for mechanical thrombectomy. This selection bias inherent to EVT studies may underestimate the true incidence of distal TM in the broader stroke population. Additionally, our findings may not be generalizable to patients managed with medical therapy alone, where different migration patterns might occur.

There are several limitations in this study. First, we cannot exclude the possibility of hidden confounders because of the retrospective study design. However, we attempted to include all possible variables to find factors that were independently associated with TM. Second, because our study only included patients who underwent EVT, we could not evaluate patients who did not undergo EVT in case of TM to the far distal artery. However, it is inconclusive whether EVT is advisable for medium or distal arteries because the smaller the occluded vessel, the smaller the infarct size, the smaller the sequelae, and the higher the risk of periprocedural complications. Our study showed that TM did not adversely affect the outcome of patients undergoing EVT. Third, we evaluated TM or HAS by visual assessment without measuring migration distances or Hounsfield units. These aspects may be subject to some error. However, our study is significant in that we have assessed TM in a practical way and identified radiologic predictors for TM that can be applied in practice. Fourth, because we defined TM as any distal movement, our classification may have included both clinically meaningful migrations and subtle positional changes. This challenge is especially relevant for intrasegmental migration, where interpretation is more difficult. Developing a standardized grading system for TM severity could improve comparability across future studies. Fifth, our study did not collect data on patient transfer status, which could potentially influence TM patterns and time-related variables. Finally, we did not assess thrombus perviousness on CT angiography or perform histological analysis of retrieved thrombi, both of which could have provided additional insights into thrombus composition and its relationship to migration patterns.

In this study, 22.1% of patients with intracranial large vessel occlusion developed TM during EVT. HAS, lower diastolic blood pressure, and shorter onset-to-arrival time were independently associated with TM occurrence, suggesting that thrombus composition and maturation play a key role in migration susceptibility. While TM was associated with slightly better neurological improvement and higher rates of parenchymal hemorrhage, overall procedural success and 3-month functional outcomes were comparable between the TM and non-TM groups. These findings indicate that TM has minimal impact on EVT effectiveness and should not alter treatment decisions for patients with large vessel occlusion.