Mechanical thrombectomy failure in anterior and posterior circulation stroke: current results from a high-volume comprehensive center

Abstract

Background

Mechanical thrombectomy (MT) is an established therapy for acute ischemic stroke (AIS), but recanalization is not always achieved. Common reasons are inadequate removal at the thrombus site and difficulties with the access route. In order to identify risk factors for MT failure we conducted a retrospective study on a high-volume comprehensive stroke center.

Methods

Evaluation of 552 thrombectomies (2019-23; anterior and posterior circulation, direct aspiration +/- stent retriever [SR]). MT failures (= modified Thrombolysis in Cerebral Infarction score 0 or 1) were analyzed for age, sex, pre- and post-MT modified Rankin Scale, bridging intravenous thrombolysis (IVT), occlusion site (anterior / posterior circulation, proximal / distal), the Kaesmacher classification and time trend results.

Results

MT failure occurred in 56 patients (10.1%; median age 76; 53.6% female). Nineteen (33.9%) patients received IVT (p = 0.326). Logistic regression analysis did not show a significant association of age, sex or occlusion site with MT failure (p = 0.165, p = 0.738, p = 0.838). Distal MT generally demonstrated lower success rates (p < 0.01). According to the Kaesmacher classification SR failure was the most frequent cause of MT failure (category 2B: 48%, p < 0.001). Time trend analysis suggests improving recanalization rates in the further course (4 times in year-on-year comparison; p < 0.01).

Conclusion

MT failure occurs in AIS treatment, even in high-volume centers and occurs more frequently in distal occlusions. Improvements in device technology, particularly SR, and ongoing refinements in access route selection offer the prospect of better outcomes in the future.

Background

In acute ischemic stroke (AIS), mechanical thrombectomy (MT) using aspiration techniques in combination with stent retrievers is established for recanalization of large vessel occlusions (LVO) in both the anterior and posterior circulation [1–6]. The frequency of MT is increasing overall, particularly with regard to the recent positive randomized controlled trial results even in large infarct cores and later time windows [7–11]. In addition, the age range for this endovascular procedure has grown in recent years, with the additional and primarily self-evident inclusion of children and adolescents as well as old and very old patients despite recommended careful selection of them [12, 13]. The spectrum is also extended by increasing evidence for MT of distal arteries [14–17]. MT failure (= modified Thrombolysis in Cerebral Infarction score [mTICI] 0 or 1 respectively 0-2a) is generally a known caveat in the treatment of AIS and has already been reported in several studies with frequencies between 10.6 and 26.1% [18, 19, 20–26; Table 1]. Common reasons for MT failure are a missing device breakdown during the thrombus removal itself and inadequate cervical or intracranial access routes [19, 23, 25, 27]. These findings raise the question of more individualized planning and implementation of interventional stroke therapy to avoid MT failure, especially with a focus more on the specific thrombus optimized devices and potential bailout strategies to control anatomical problems [28–30]. While taking this into account, Kaesmacher et al. developed a classification for MT failure in particular with stent retrievers considering non-technical or technical reasons [19; Table 2].

Table 1.

Previous studies addressing mechanical thrombectomy failure [16, 17, 19–25]

Authors, year of publication	Number of patients	Study design	Study period	MT failure definition (mTICI)	MT failure rate	Technique	Anterior circulation	Posterior circulation	Inclusion of distal vessel occlusions (up to)
Our study	552	Single center	2019–2023	0/1	10.1%	DA + SR	+	+	M3, P2
Lajthia et al., 2023	1010	Single center	2013–2021	0-2a	11.9%	DA + SR	+	+	M4
Marnat et al., 2022	5076	Multicenter (n = 8)	2015–2020	0-2a	12.4%	DA + SR	+	-	no
Flottmann et al., 2021	2211	Multicenter (n = 25)	2015–2028	0-2a	16.8%	DA + SR	+	-	no
Weyland et al., 2021	218	Single center	2009–2019	0/1	13.8%	DA + SR	-	+	no
Heider et al., 2020	596	Single center	2014–2018	0-2a	16.8%	DA + SR	+	-	no
Goda et al., 2019	119	Single center	2015–2019	0-2a	26.1%	DA + SR	+	-	M3
Leischner et al., 2019	648	Single center	2010–2017	0/1	11.0%	DA + SR	+	-	no
Kaesmacher et al., 2018	592	Single center	2012–2017	0/1	10.6%	SR	+	+	no
Gascou et al., 2014	144	Single center	2009–2011	0-2a	13.9%	SR	+	+	n.a.

Legend: DA - direct aspiration, MT– mechanical thrombectomy, mTICI - modified thrombolysis in cerebral infarction scale, n.a. - not applicable, SR - stent retriever

Table 2.

Known classification of mechanical thrombectomy failure reasons as published by Kaesmacher et al. [19]. The changes we have added are highlighted in italics

Category	Classification system for mechanical thrombectomy failure
1	Technical reasons: Target occlusion not reached 1 A: Intracranial target occlusion was not reached due to marked cervical vessel tortuosity including twisted, looped, or kinked vessels. The proximal cervical vessels were successfully catheterized. 1B: Target occlusion was not reached owing to failed catheterization of proximal supra-aortic vessels. The proximal cervical vessels were not successfully catheterized owing to difficult aortic arch anatomy. 1 C: Target occlusion was not reached due to the inability to pass a cervical ICA occlusion (e.g. unpassable tandem lesion).
2	Technical reasons: Target occlusion reached 2 A: Target occlusion was reached, but the operator was unable to pass the thrombus with the microwire/microcatheter. In these cases, no stent retriever was deployed. 2B: Target occlusion was reached, the stent retriever was deployed, but no reperfusion occurred after multiple retrievals (no clot retrieval or dislocation), thus, stent retriever failure. 2 C: Initial reperfusion was achieved, followed by spontaneous or iatrogenic reocclusion (e.g. intracranial stenosis, intracranial dissection, or perforation with subsequent vessel sacrifice).
3	Non-technical reasons Presumed futility. Adverse non-neurological event with the need to stop mechanical thrombectomy. Signs of contrast extravasation without perforation (early hemorrhagic transformation). Puncture of a suitable vessel for access route not possible. Periinterventional arterial dissection. Classification is not feasible based on the available documents.

Legend: ICA - internal carotid artery

Given the urgency of more detailed understanding of MT failure, our large retrospective single center study from a German university hospital aims to comprehensively analyze demographic characteristics (age, sex, pre- and post-MT modified Rankin Scale [mRS] values, bridging intravenous thrombolysis [IVT] with plasminogen activator) and vessel occlusion site (anterior vs. posterior circulation respectively proximal vs. distal segments) as well as other risk factors for MT failure according to the classification of Kaesmacher et al. In addition, we carried out an analysis of MT failure over time.

Methods

Study design

In a retrospective analysis we evaluated 552 patients in total who underwent MT due to AIS in the anterior or posterior cerebral circulation between August 2019 and March 2023 in our high-volume comprehensive stroke center of a German university hospital. Our standard MT approach was as follows: After induction of general anesthesia, a long 6 french (F) sheath is placed in the internal carotid artery (ICA) or vertebral artery via femoral access with a 5 F guide catheter. An appropriately sized aspiration catheter is placed at the face of the thrombus in a triaxial approach and two attempts of direct aspiration are performed. If recanalization was unsuccessful, the decision to employ an additional stent retrieval technique, repeat the aspiration maneuver or to end the interventional procedure was made by the neuro-interventionalist.

Mechanical thrombectomy failure was defined as mTICI score of 0 or 1.

Patients with MT failure were analyzed for demographic characteristics age, sex, mRS values at admission (pre-MT) and at discharge from hospital (post-MT), bridging IVT and site of vessel occlusion (anterior vs. posterior circulation, proximal ICA, M1, M2, A1, A2, basilar artery [BA], P1 vs. distal [A3, M3, P2] segments) as well as reasons for MT failure according to the classification of Kaesmacher et al. [15, 19]. In addition, we analyzed MT failure of patients over time.

All angiographic results were evaluated by two experienced neuro-interventionalists with board certification.

Approval of our institutional review board was obtained, whereby research ethics and informed consent were considered (file reference: 2023 − 662).

Statistics

Statistical analyses were performed using R (version 4.3.1). Pearson’s Chi-square tests (parametric) were employed for categorical variables, logistic regression (parametric) for binary outcomes and Wilcoxon rank-sum and signed-rank tests (non-parametric) for ordinal data. Fisher’s exact test (non-parametric) was used for categorical data and ordinal logistic regression (parametric) for ordered categorical outcomes. The alpha level was set at 0.05, with p-values less than this threshold considered statistically significant.

Results

All vessel occlusions and mechanical thrombectomies

In all 552 patients receiving MT, 469 vessel occlusions (85%) occurred in the anterior circulation and 73 (13.2%) in the posterior circulation. The 10 remaining cases (1.8%) constituted combined occlusions in different territories, whereby two vessel segments alone were affected at the same time (each without simultaneous MT). The occlusion leading to thrombectomy was found in a proximal vessel segment in 527 cases (95.5%) and in a distal vessel segment in 25 cases (4.5%; A3 n = 1 [left], M3 n = 9 [8 left, 1 right], P2 n = 15 [9 left, 6 right]). 496/552 patients (89.9%) showed successful MT after use of direct aspiration and/or stent retriever, with n = 29 (5.8%) undergoing intracranial rescue stenting.

Mechanical thrombectomy failure patients

Demographic characteristics and site of vessel occlusion (anterior vs. posterior circulation, proximal vs. distal segments)

Mechanical thrombectomy failure occurred in 56 patients (10.1%; n = 47 anterior circulation, n = 9 posterior circulation). In these cases, 49 patients (87.5%) finally showed mTICI 0, while in 7 patients (12.5%) MT resulted in mTICI 1. There was no case with rescue stenting that resulted in MT failure. The median age of the patients with an unsuccessful MT was 76 (interquartile range: 69.75-82). Thirty patients (53.6%) with MT failure were female and n = 26 male (46.4%). The distribution of mRS values at admission for patients with MT failure showed a median of 4 (range 1–5), whereas the median at discharge from hospital was 5 (range 2–6). Nineteen patients (33.9%) with MT failure received bridging IVT, while n = 37 (66.1%) did not. There were 48 cases of failed thrombectomies in proximal (85.7%) and 8 (14.3%) in distal vessel occlusions, with M2 branch being the most frequent location of MT failure. Site distribution of MT failure in detail is presented in Table 3.

Table 3.

Site distribution of mechanical thrombectomy failure

Vessel occlusion	Left (n)	Right (n)
ICA	6	4
M1	7	4
M2	9	8
M3	3	0
A1	1	0
A2	1	3
A3	1	0
BA	1
P1	3	2
P2	1	2

Legend: BA - basilar artery, ICA - internal carotid artery, mTICI - modified thrombolysis in cerebral infarction scale

Review of MT failure, according to the Kaesmacher classification, shows that stent retriever failure was the most frequent cause in our study cohort (category 2B: 27/56 patients [48%]). The detailed distribution is listed in Table 4.

Table 4.

Mechanical thrombectomy failure reasons in our study using the classification system of Kaesmacher et al. as supplemented by us [19]

Specific statistical analyses including the classification of Kaesmacher et al. and trend over time

A logistic regression analysis showed a statistically non-significant trend towards a negative association of age with MT failure (p = 0.165). We found no significant association between sex and unsuccessful endovascular recanalization (p = 0.738). Bridging IVT was not significantly associated with successful MT (p = 0.326). An improvement in the clinical outcome on the basis of pre- and post-MT mRS was significantly associated with thrombectomy success (TICI ≥ 2a; p < 0.001). There was also no correlation of vessel occlusion localization in the anterior or posterior circulation on angiographic reperfusion after MT (p = 0.838). However, distal vessel occlusions (A3, M3, P2) generally demonstrated lower MT success rates than proximal occlusions (ICA, M1, M2, A1, A2, VA, BA, P1) (p < 0.01).

Our review of MT failure in detail, according to the Kaesmacher classification, showed that stent retriever failure was the most frequent cause (category 2a: 48%, p < 0.001). Furthermore, demographic factors such as age or sex showed no correlation across different failure types (p = 0.610, p = 0.805). Concerning the success rates of MT in different calendar years, the results indicated significant differences between 2020 (MT failure n = 22) and 2022 (n = 8, p < 0.01), 2020 (n = 22) and 2023 (n = 0, p < 0.01), 2021 (n = 23) and 2022 (n = 8, p < 0.01) as well as 2021 (n = 23) and 2023 (n = 8, p < 0.01; Fig. 1). Comparisons between other years did not yield statistically significant findings (p > 0.05).

Fig. 1.

Mechanical thrombectomy failure over time in our study. Legend: MT - mechanical thrombectomy

Discussion

Our aim was to analyze patients with MT failure. Considering this ongoing and dynamic issue, we present results from a large single center study conducted in a high-volume comprehensive center of a German university hospital, including both anterior and posterior circulation strokes. Mechanical thrombectomy failure patients were analyzed for baseline characteristics (age, sex, clinical conditions, bridging IVT) as well as vessel occlusion localization (anterior vs. posterior circulation, proximal [ICA, M1, M2, A1, A2, VA, BA, P1] vs. distal [A3, M3, P2] segments) and other reasons for MT failure according to the classification of Kaesmacher et al. In addition, we carried out an analysis of MT failure over time.

Unsuccessful endovascular recanalization is a well-recognized challenge in the treatment of acute ischemic stroke, with prior studies reporting failure rates ranging from 10.6 to 26.1%, based on the mTICI score [18–26]. Overall, our MT success rate is in line with the results of other work groups, although MT failure was not consistently defined across studies (mTICI 0/1 vs. 0-2a), making comparisons difficult in some cases (18, 19, 20–26; Table 1).

Our logistic regression analysis showed a trend towards a negative association of age with MT failure (p = 0.165) and we found no significant association between sex and unsuccessful endovascular recanalization (p = 0.738). Different from this finding, Goda et al. identified female sex as a predictor of MT failure [18]. Furthermore, in our results, bridging IVT was not significantly associated with MT success (p = 0.326), partly arguing against recent results from the SWIFT DIRECT trial that supported IVT additionally to MT for AIS treatment [31]. However, our data did not result from a randomized trial. Therefore, there is a clear selection bias between both groups that might influence results.

Our study further enhances understanding of MT results across different vascular territories. Several studies have already dealt with this; we refer to a meta-analysis of Adusumilli presenting comparable mechanical recanalization rates in anterior and posterior circulation [32]. Also, our results did not show a correlation of site of vessel occlusion in the different circulations on angiographic reperfusion after MT (p = 0.838). The lack of significant differences between the territories suggests that thrombus location does not compromise MT success, which is in a way in line with results of registry data [33].

We also compared the proximal and distal distribution in relation to thrombectomy success. In our study, the exploration of MT outcomes by the location of occlusion reveals that thrombectomy targeting distal vessels (in our cases A3, M3 and P2) have significantly lower recanalization rates compared to proximal thrombus locations (p < 0.01). In contrast, a systematic review indicates satisfactory MT outcomes for more distal cerebral arteries [34]. Nevertheless, the difference revealed in our study obviously confirms that distal occlusions of small vessels are more challenging, potentially due to increased tortuosity, resulting in less accessibility for MT devices, which may be more difficult to navigate, therefore increasing the complexity of the endovascular procedure. In the case of distal emboli, we see a higher number of left-brain findings in our results (18 of 25 cases). In this context, the literature predominantly describes more effective MT outcomes for LVO associated with left cerebral ischemia [35].

While previous work such as that by Heider et al. has already examined mechanisms and co-factors of MT failure in detail, we would also like to add new aspects to the topic [24]. The additional classification by Kaesmacher et al., which distinguishes between technical and non-technical reasons for MT failure, offers valuable insights into the factors contributing to MT outcomes [19]. We have also used this classification in our collective for patients treated with direct aspiration in connection with AIS. Our review of factors associated with unsuccessful thrombectomies according to the proposal of Kaesmacher et al. shows that stent retriever failure, despite correct positioning and potentially multiple retrieval attempts, is the primary cause (category 2B: 27/56 patients [48%]) and this demonstrated significant differences to the other categories (p < 0.001). Notably, our findings indicate a higher failure rate than reported by Kaesmacher et al., who noted stent retriever failure in 39.7% of their 63 patients [17]. This highlights the need for even greater emphasis on further development of stent retrievers, with a focus more on the specific thrombus optimized devices [30]. However, this kind of MT failure directly at the occlusion may be influenced by challenging vascular anatomy, such as tight curves or extensive atherosclerosis, which complicates the retrieval process itself [28, 30]. Furthermore, especially for the patients classified by us according to Kaesmacher et al., the demographic factors age and sex showed no correlation across different failure types (p = 0.610, p = 0.805). This suggests that they may influence overall stroke outcomes, but not directly impact MT success.

Lastly and as a further new aspect for literature, our time trend results suggest improving MT success rates in the further course. This is particularly noteworthy in the period from 2021 to 2023, with not a single failure documented in the last year in 43 MT patients examined (Fig. 4). It is worth mentioning that a relatively long period of time (almost 4 years) is included in this study, in which the development of new tools has likely played a role in decreasing the number of MT failures over time. The observed improvements in endovascular recanalization rates may be attributed in part to the increased use of thrombectomy especially for distal vessel occlusions with enhanced knowledge of material handling. As neuro-interventionalists become more familiar with the techniques of navigating and using these technologically advanced devices, they are able to perform MT more successfully and with better outcomes [36]. Especially for distal occlusions, therefore, advancements in devices, particularly stent retrievers, are crucial in shaping the future of MT [37]. Temporal trends in thrombectomy success have so far only been systematically investigated for LVO [38].

In general, additive MT procedures through other access routes (radial or direct ICA puncture) present a promising approach as an alternative strategy in cases of initial procedural failure via femoral artery access and certainly require further investigation [39–41]. Additionally, the utilization of simulators capable of replicating intricate anatomical scenarios could provide valuable training opportunities [42]. Furthermore, recent studies have shown that the use of intra-arterial tissue-type plasminogen activators after unsuccessful MT did not result in a higher risk of symptomatic intracranial hemorrhage or mortality, making it a low-risk bailout procedure [43, 44].

Limitations

This study’s retrospective design with single-center focus, different vessel occlusion durations especially due to “drip- and-ship”, use of varying procedural techniques for MT with changing experience levels of the neuro-interventionalists (learning curves) and lack of longer-term follow-up data limits its generalizability and introduces potential biases. As retrospective studies carry inherent limitations, more prospective studies on this topic are crucial.

The mTICI score was not originally intended for the description of therapy results of the posterior circulation, but is also widely used for this purpose [16]. In addition, there are different definitions for distal vessel occlusions and also overlaps with Medium Vessel Occlusions (MeVO), and we have referred to the descriptions of the multicenter initiative Stroke Thrombectomy and Aneurysm Registry (STAR) [15, 45, 46]. In the retrospective approach applied here, the exact reason for a persistent vessel occlusion after MT and possibly the lack of a change in interventional strategy with consideration of rescue stenting, among others, could not be clearly traced, although there are also novel developments in this area in particular [47–50]. Furthermore, in future the goal should be to better assess the additive effect of bridging IVT depending on localization of the vessel occlusion to be mechanically recanalized and achieve a better understanding of predictors for MT failure, particularly in very old patients [51, 52].

Beyond that, with regard to outcome, predictive clinical and radiological parameters [53] as well as other influences such as anticoagulation [54], tandem occlusions with simultaneous emergent ICA stenting [55], collaterals [56] and secondary intracerebral hemorrhages [57, 58] have not been analyzed by us. However, these will certainly also be the subject of future research. In addition, the question of whether the duration of the interventional procedure and the number of thrombectomy maneuvers have a decisive influence on the outcome in the case of unsuccessful endovascular recanalization was also discussed very recently, which was not considered in our survey [59].

Funding

None

Open Access funding enabled and organized by Projekt DEAL.

Data availability

The data that support the findings of this study are available from the authors Alexander Neumann and Paul Schildhauer, upon reasonable request.

Declarations

Ethical approval

Our study has been reviewed by the appropriate ethics committee (University of Lübeck, file reference: 2023 − 662) and has therefore been performed in accordance with the ethical standards laid down in an appropriate version of the 1965 Declaration of Helsinki.

Consent for publication

All authors consent to the publication of this manuscript.

Competing interests

HS: Lecture fee from Cercare Medical. GR: Compensation from Cardinal Health 200 LLC for consultant services; compensation from Novartis Pharma AG for other services; compensation from AstraZeneca for consultant services; travel support from Boehringer Ingelheim; compensation from Boehringer Ingelheim for consultant services; compensation from Ipsen Pharma SAS for consultant services; and compensation from Bristol-Myers Squibb for consultant services. UJK: Payment for attending a training course from Cerenovus. AN, PS, SMW, PS: None.