During endovascular thrombectomy for acute ischemic stroke caused by large vessel occlusion, continuing the procedure for up to six clot retrieval attempts confers a statistically significant functional benefit compared with terminating the intervention after unsuccessful reperfusion. Analyzing multicenter registry data through the lens of an infinite-horizon optimal stopping problem, researchers established that the expected clinical utility of attempting an additional pass consistently outweighed the utility of accepting persistent arterial occlusion through six sequential attempts. At the seventh attempt, this net expected benefit ceased to be statistically significant, identifying an empirical threshold where further recanalization efforts may no longer provide superior expected functional independence compared to stopping.
Endovascular thrombectomy represents the established standard of care for selected patients presenting with acute ischemic stroke secondary to large vessel occlusion in the anterior circulation. Clinical investigations consistently demonstrate that achieving prompt and complete reperfusion, defined as a modified Treatment in Cerebral Infarction score of 3 on the initial attempt, produces the most favorable functional trajectories. Despite advances in endovascular devices and techniques, persistent arterial occlusions classified as modified Treatment in Cerebral Infarction grades 0 to 2a occur in up to 54 percent of patients after the first retrieval pass. Consequently, neurointerventionalists routinely encounter intraprocedural scenarios where the initial deployment fails to re-establish blood flow, necessitating sequential real-time choices regarding whether to perform another retrieval attempt or conclude the operation.
Deciding whether to pursue another retrieval attempt presents a critical clinical dilemma during emergency neurovascular procedures. While additional passes offer an incremental opportunity to achieve successful reperfusion and subsequent functional recovery, repeated mechanical manipulations inside fragile cerebral arteries introduce procedural hazards. Previous medical literature notes that an increasing total number of thrombectomy passes correlates with poorer functional recovery, heightened rates of vessel perforation or dissection, clot fragmentation, distal embolization, vasospasm, and symptomatic intracranial hemorrhage. Interventional operators have therefore lacked rigorous forward-looking quantitative guidance to determine whether the risk of complications from multiple passes outweighs the severe clinical disability associated with abandoning recanalization efforts.
To provide empirical intraprocedural decision support, investigators examined clinical and angiographic records from the German Stroke Registry Endovascular Treatment cohort enrolled between June 2015 and December 2023. The registry represents an ongoing multicenter observational study conducted across 25 medical centers in Germany. Eligible participants for the present analysis included adult patients aged 18 years or older who presented with acute ischemic stroke secondary to vessel occlusion in the anterior circulation, underwent endovascular thrombectomy involving between 1 and 15 retrieval passes, and possessed complete baseline, procedural, and follow-up outcome data. The primary clinical endpoint was good functional outcome at 90 days, defined as a score of 0 to 2 on the modified Rankin Scale.
Rather than evaluating patient outcomes retrospectively by grouping cases according to their final total pass count, the researchers formulated endovascular thrombectomy as a forward-looking infinite-horizon optimal stopping framework. At each procedural pass, designated as decision node k, the operator must determine whether to proceed or terminate based solely on the current angiographic state. The net expected benefit of continuation, denoted as delta k, was calculated as the difference between the expected utility of proceeding to an additional pass, incorporating all downstream outcomes and subsequent passes, and the utility of stopping the procedure with persistent occlusion at the current pass. A positive and statistically significant delta value indicates that continuing the thrombectomy attempt provides superior expected functional outcomes over stopping.
The study cohort comprised 9,495 patients who satisfied all inclusion parameters. Within this population, the median age was 77 years with an interquartile range of 66 to 83 years, and 54 percent were women. Documented medical comorbidities included hypertension in 7,238 individuals representing 76 percent, diabetes mellitus in 2,151 patients representing 23 percent, dyslipidemia in 3,989 patients representing 42 percent, and atrial fibrillation in 4,165 patients representing 44 percent. The median baseline National Institutes of Health Stroke Scale score was 14, and the median Alberta Stroke Program Early Computed Tomography Score was 9. Intravenous thrombolysis was administered prior to or during intervention in 4,477 patients, accounting for 47 percent of the cohort.
Across the entire registry cohort, successful reperfusion, defined as achieving a modified Treatment in Cerebral Infarction score of 2b to 3, was attained in 8,405 patients, representing an overall recanalization rate of 89 percent. Overall functional independence at 90 days, measured as a modified Rankin Scale score of 0 to 2, was achieved by 3,428 patients, or 36 percent of the population. The median number of thrombectomy passes across all procedures was 2. Following the first pass, successful reperfusion with immediate procedural termination occurred in 4,136 patients, yielding a 90-day good functional outcome rate of 44.4 percent. In contrast, 198 patients had their procedure terminated after a single unsuccessful pass, among whom only 22.2 percent achieved a good functional outcome.
In 5,161 patients, representing 54.4 percent of the cohort, endovascular intervention was continued beyond the first retrieval attempt. Among these continued cases, the expected probability of achieving a modified Rankin Scale score of 0 to 2 at 90 days was 30.0 percent, which was significantly higher than the 22.2 percent observed among patients whose procedures were terminated after pass one with persistent occlusion. The net expected benefit of continuation remained positive and statistically significant across passes one through six, with p-values below 0.05 at each consecutive stage. However, at pass seven, the expected benefit became negative and lost statistical significance, with an observed functional independence rate of 5.5 percent for termination versus an expected rate of 7.9 percent for further attempts, yielding a non-significant p-value of 0.53.
Analysis of angiographic performance demonstrated that the probability of achieving successful reperfusion declined with each consecutive pass, decreasing from 39.5 percent on the second retrieval attempt to 24.4 percent for patients requiring seven or more passes. Concurrently, the rate of procedure-related adverse events increased alongside pass numbers, recorded at 8.0 percent among patients undergoing a single attempt and rising to 24.6 percent among those receiving seven or more passes. Reported procedural complications across the entire cohort included vasospasm in 367 individuals representing 4 percent, clot migration or distal embolization in 382 patients representing 4 percent, vessel dissection or perforation in 260 patients representing 3 percent, and intracranial hemorrhage in 345 patients representing 4 percent.
The study authors emphasize several critical methodological and clinical limitations inherent to observational registry evaluations. Because the analysis was restricted to complete cases with all required data points, selection bias could arise from the exclusion of individuals with missing documentation or those lost to follow-up. The statistical calculations were unadjusted for potential baseline prognostic confounders, such as age, admission stroke severity, or collateral vessel status, which may have influenced operator decisions to abort procedures early. Furthermore, the registry did not document specific intraprocedural rationales for procedural termination, preventing distinction between technical futility and procedural complications. Finally, the subgroup requiring seven or more attempts remained relatively small at 258 patients, reducing the statistical power to detect smaller potential outcome differences at higher pass numbers.
In formalizing intraprocedural choices, the mathematical framework relied on an empirical realization of the Bellman continuation value, a recursive decision-theoretic concept that summarizes the expected clinical utility across all future procedural nodes following any given attempt. For each attempt k, the expected utility of continuation incorporated both the empirical likelihood of achieving successful recanalization on subsequent passes and the possibility of persistent occlusion, without requiring downstream procedural steps to be simulated in isolation. Using R version 4.1.3, investigators evaluated between-group differences in ninety-day functional independence between individuals proceeding to pass k plus one and those whose procedures were stopped at pass k with persistent occlusion via two-sample Welch t-tests.
The registry findings challenge earlier observational reports suggesting that endovascular attempts become futile after the third, fourth, or fifth pass. In this cohort, 8,599 patients underwent up to four passes, while substantial cohorts had interventions concluded after five passes (394 patients), six passes (244 patients), or seven or more passes (258 patients). Successful reperfusion rates remained between 24.4 percent for seven or more attempts and approximately 32 percent for passes five and six. Among patients achieving successful reperfusion at these advanced stages, rates of functional independence reached 28.1 percent at five passes, 22.8 percent at six passes, and 31.7 percent at seven or more passes, notably exceeding functional independence rates seen in earlier cohorts such as those documented by Uchida and colleagues.
The investigators also addressed a potential interpretative pitfall regarding procedural stopping thresholds. Although the expected rate of ninety-day functional independence when continuing beyond three passes was 20.0 percent—numerically lower than the 22.2 percent observed among patients whose procedures ended after one failed pass—the authors noted that comparing these figures is clinically invalid. Because thrombectomy decisions operate strictly in an anterograde sequence, patients reaching a third unsuccessful attempt cannot return to the baseline state of an earlier pass. Instead, intraprocedural guidance must rely entirely on the net expected benefit relative to the current angiographic state, which remained positive through six attempts. The researchers additionally observed that the registry did not record the specific modalities used across centers to measure ninety-day modified Rankin Scale scores.
Subject of Research: Sequential intraprocedural decision-making and optimal stopping thresholds in endovascular thrombectomy for acute ischemic stroke
Article Title: Sequential decision-making in endovascular stroke therapy: benefit and harm of the next retrieval attempt after failed reperfusion
Article References: Meucci, L., Meyer, L., Broocks, G., Bechstein, M., Thaler, C., Heitkamp, C., Winkelmeier, L., Geest, V., Heitkamp, A., Peter, G., Schlicht, F., Brekenfeld, C., Maros, M. E., Schell, M., Bester, M., Thomalla, G., Fiehler, J., Gellißen, S., Flottmann, F., … for the German Stroke Registry – Endovascular Treatment (GSR-ET) (2026). Sequential decision-making in endovascular stroke therapy: benefit and harm of the next retrieval attempt after failed reperfusion. Neurological Research and Practice, 8(1), Article 88. https://doi.org/10.1186/s42466-026-00543-y
Image Credits: AI Generated
DOI: 10.1186/s42466-026-00543-y
Keywords: endovascular thrombectomy, acute ischemic stroke, large vessel occlusion, optimal stopping problem, retrieval attempts, modified Rankin Scale, mTICI score, reperfusion success, German Stroke Registry, intraprocedural decision-making, procedural complications, functional outcome, interventional neuroradiology, Bellman continuation value, cerebral infarction, per-pass analysis, clinical utility
News Source: Cassandra Pierce. (November 17, 2026). Stroke Thrombectomy Yields Significant Net Benefit Through Six Retrieval Attempts. Scienmag.



