Introduction

Since 2015, treatment guidelines have recommended endovascular therapy (EVT) for patients with acute ischemic stroke due to large vessel occlusions (LVO) involving the intracranial carotid artery or the middle cerebral artery (MCA)-M1 segment under level IA evidence.1 However, growing evidence suggests that EVT may also be safe and effective in the treatment of MCA-M2 segment occlusions.2–8 Some of these studies have compared the use of a stent-retriever (SR) in cohorts of patients with MCA-M2 versus MCA-M1 occlusions.

Improved devices for distal vessel navigation as well as new techniques for faster and more efficient clot removal have evolved including front-line aspiration approaches9 10 or the combined use of SR and contact aspiration (CA).11–14 Concerning the latter, several potential advantages have been described including the mechanical pinching of the thrombus between the SR and the aspiration catheter, allowing the clot to be held more effectively during removal.

Compared with more proximal intracranial occlusions, MCA-M2 occlusions represent a more heterogenous group with higher variability in terms of anatomy, clinical severity, and tissue-at-risk on presentation. The use of SR alone versus CA as a front-line EVT has been previously addressed in M2 occlusions15–17; however, the performance of combined SR+CA approaches remains unexplored.

We sought to compare whether reperfusion rates, procedure times, procedure-related complications, and clinical outcome differed between M2 occlusions treated with SR alone versus SR+CA in a large cohort of patients presenting with an isolated MCA-M2 occlusion.

MethodsStudy design and patient selection

The ROSSETTI registry is an ongoing investigator-initiated prospective study launched in June 2019, which includes consecutive patients with acute ischemic stroke due to anterior circulation occlusion treated with the more recent EVT device technology across 10 comprehensive stroke centers in Spain. The inclusion criteria for the ROSSETTI registry are as follows: age ≥18; confirmed occlusion in the intracranial internal carotid artery, MCA-M1 or MCA-M2 segments (from the MCA-M1 bifurcation to the proximal third of the vertical segment in the Sylvian fissure); time from last seen well (TLSW) to treatment ≤24 hours; National Institutes of Health Stroke Scale (NIHSS) score ≥2 and modified premorbid Rankin Scale (mRS) score ≤2. Tandem occlusions were excluded in the ROSSETTI registry. At each center, the EVT technique used was at the discretion of the operator. A stroke neurologist decided whether or not to perform intravenous thrombolysis based on conventional inclusion and exclusion criteria for intravenous thrombolysis, according to guidelines.1 Successful and near-complete/complete reperfusion were defined as scores of 2b–3 and 2c–3 on the expanded Treatment in Cerebral Ischemia (eTICI) scale, respectively.

The current study consists of a retrospective analysis of patients with isolated MCA-M2 occlusions who underwent EVT with SR alone versus SR+CA from the ROSSETTI registry between June 2019 and December 2020.

Device description

The choice of the type and size of both the SR and CA catheter in each case was primarily driven by availability in the department and ultimately to the criterion of the neurointerventionist.

The SR catheters used in both groups were Aperio (Acandis, Pforzheim, Germany), Catch (Balt, Montmorency, France), Embotrap (Cerenovus, Galway, Ireland), NeVa (Vesalio, Nashville, Tennessee, USA), pREset (Phenox, Bochum, Germany), Solitaire (Medtronic, Irvine, California, USA) and Trevo (Stryker, Kalamazoo, Michigan, USA).

The CA catheters used in the SR+CA group were 3MAX, Ace 64 and Ace 68 (Penumbra, Alameda, California, USA); AXS Catalyst 5, Catalyst 6 and Catalyst 7 (Stryker, Kalamazoo, Michigan, USA); Navien 058, React 68, React 71 (Medtronic, Irvine, California, USA); Q3, Q4, Q5 and Q6 (MIVI Neuroscience, Eden Prairie, Minnesota, USA); and Sofia 5F, Sofia Plus (MicroVention Terumo, Tustin, California, USA).

Data collection

The study data were collected prospectively through an online questionnaire. Data included the patient demographics, baseline characteristics (NIHSS score and Alberta Stroke Program Early CT Score (ASPECTS), time from clinical onset (defined as TLSW) to arterial puncture, use of IV thrombolysis), procedure details (technique employed in each pass, devices used, rescue therapy), procedure-related complications, angiographic outcomes (first-pass and final eTICI scores), clinical outcome (24-hour NIHSS and mRS at 90 days) and 90-day mortality rate.

Outcomes

The primary technical outcome was the rate of first-pass effect (FPE), defined as achieving near-complete/complete reperfusion (eTICI 2c–3) after the single-pass approach. Secondary efficacy endpoints included the final eTICI reperfusion rates, number of thrombectomy attempts, and the prevalence of rescue therapy. The latter is defined as the use of any EVT technique other than the initial approach; in the case of the SR group this was the use of SR+CA, CA alone, or rescue intracranial angioplasty±stent, and for the SR+CA group it was the use of SR alone, CA alone, or rescue intracranial angioplasty±stent.

The safety outcome endpoints were the presence of arterial dissection, arterial perforation, distal embolization in the same territory or new vascular territory, subarachnoid hemorrhage or symptomatic intracranial hemorrhage (sICH, according to ECASS II definition)18 at the 24-hour control head CT.

The clinical outcomes included the rates of early clinical improvement (difference in the NIHSS score from baseline to 24 hours) as well as good functional outcome (mRS 0–2) and mortality at 90 days. Reperfusion scores, sICH evaluation, and clinical outcome were self-adjudicated by experienced stroke neurologists.

Statistical analysis

All analyses were conducted using Stata version 15.0 (StataCorp, College Station, Texas, USA). Categorical variables were presented as number (frequency) and quantitative variables were presented as median (IQR) or mean (SD). Demographic, clinical data, procedure details, and angiographic and clinical outcomes were compared according to the EVT groups: SR versus SR+CA. χ2 or Fisher exact tests were used for categorical variables and the Mann–Whitney U test was used for continuous variables.

Subsequently, a propensity score matching analysis was performed, matching (1:1) according to the nearest-neighbor matching algorithm with a caliper of 0.05. The matched covariates included age, baseline ASPECTS, NIHSS score at admission, use of IV thrombolysis, and time from clinical onset to arterial puncture. Comparisons between categorical variables were made using McNemar’s exact test, the paired samples t-test was used in normal quantitative variables (age) and for the rest of non-normal distribution quantitative variables the Wilcoxon test was employed.

In addition, multivariable logistic regression analysis was performed in order to identify independent variables associated with FPE and good clinical outcome at 90 days. First, we performed a univariable logistic regression analysis with the variables potentially associated with reperfusion (age, baseline ASPECTS, NIHSS at admission, time from clinical onset to arterial puncture, administration of IV thrombolysis, use of balloon-guide catheter (BGC) and EVT technique used) or good clinical outcome at 90 days (age, baseline ASPECTS, NIHSS on admission, time from clinical onset to arterial puncture, administration of IV thrombolysis, use of BGC, EVT technique used, procedure time and FPE). To maximize sensitivity, those variables with a univariable association of p<0.20 were included in a multivariable logistic regression model.19 Quantitative variables were not dichotomized, but adjusted per 1-year increase (age), 1-point increase (baseline ASPECTS and NIHSS at admission), and 1-min increase (time from clinical onset to groin and procedure time). In addition, to ensure that multicollinearity did not impact the results, variance inflation factor was tested on all independent variables. Associations are presented as ORs with corresponding 95% CIs.

ResultsPatient inclusion

Of 834 patients included in the ROSSETTI registry from June 2019 to December 2020, 606 (72.7%) patients were excluded due to occlusion involving the intracranial ICA or MCA-M1 segment and 14 (1.7%) patients with MCA-M2 occlusion who were treated with front-line aspiration technique. Thus, 214 patients (mean±SD age 74±12.9 years; 51.4% women) were included for analysis (SR alone, 125 patients; SR+CA, 89 patients) (figure 1).

Figure 1

Flowchart of patient inclusion in the study. ASPECTS, Alberta Stroke Program Early CT score; EVT, endovascular treatment; IV, intravenous; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; tPA, tissue plasminogen activator.

Baseline characteristics and outcomes for the overall population

Online supplemental table 1 outlines the characteristics of the study cohort.

Groups were similar regarding age, sex, and baseline ASPECTS score. However, the SR group had a lower NIHSS score at baseline (106–15 vs 139–18; p<0.001), lower use of IV thrombolysis (16.8% vs 42.7%; p<0.001), higher use of BGC (98.4% vs 40.4%; p<0.001), and longer time from clinical onset to arterial puncture (7.3±5.3 vs 5.7±3.9 min; p=0.042). The SR group had a shorter procedural time (33.6±25.9 vs 50±30 min; p<0.001) and lower utilization of rescue therapy (4% vs 15.7%; p=0.003). However, the rates of FPE (44.8% for SR group vs 44.9% for SR+CA group) and final eTICI 2c/3 (76.8% vs 74.2%) were similar in both cohorts. In addition, both groups had comparable rates of clinical improvement at 24 hours as well as good clinical outcome and mortality at 90 days. Notably, the SR group had a lower rate of procedure-related complications (8% vs 22.5%; p=0.004) as well as a lower incidence of sICH (4% vs 12.4%, p=0.022).

Devices used and EVT technique

Online supplemental tables 2 and 3 provide descriptions of the SR and CA catheters used in both groups.

The most common type of SR used overall was Catch (44.4%). In the SR group, Catch was used in almost half of the cases; however, in the SR+CA group, the SR most frequently used was Trevo (47.2%). Regarding diameter, 4 mm SRs were used in 52.3% of cases and, in terms of length, a 20 mm SR was used in 55.6% of the patients. Similarly, in both groups, 4 mm diameter and 20 mm length SRs were used more frequently. In the SR+CA group, a total of 15 CA catheters were used, with Catalyst 5, Catalyst 6, Sofia 5F, and Sofia Plus comprising 68.5% of the CA catheters employed.

Concerning the technique used, in the SR alone group the combination of SR with BCG was used in nearly all cases (98.4%) but in the SR+CA group this decreased to 40.4%. In the SR+CA group the sequence of device use was the deployment of the SR over the occlusion site followed by the advancement of the CA catheter to the proximal clot and finally the combined removal of both devices.

Baseline characteristics and outcomes for the propensity score-matched cohort

The 1:1 propensity score matching included 67 patients in the SR group and 67 patients in the SR+CA group (table 1). Both groups had similar rates of angiographic outcomes. However, the SR group had significantly shorter procedural times (33.6±25.9 min vs 51.4±31.4 min, p=0.001) and lower utilization of rescue therapy (1.5% vs 16.4%, p=0.006). The incidence of distal emboli to the same territory after the thrombectomy pass was comparable between groups. However, there was a lower incidence of procedure-related complications with the use of SR (10.4% vs 25.4%, p=0.031). Both groups had similar rates of early clinical improvement as well as good clinical outcome and mortality at 90 days (table 1).

Table 1

Baseline characteristics according to matched groups

Predictors of first-pass effect

Multivariate analysis showed that baseline ASPECTS was the only independent factor associated with FPE (OR 1.494 (95% CI 1.162 to 1.922); p=0.002) (table 2).

Table 2

Multicollinearity test and univariable and multivariable logistic regression analysis for first-pass effect

Predictors of good clinical outcome

Age (OR 0.964 (95% CI 0.940 to 0.989); p=0.005), NIHSS at admission (OR 0.903 (95% CI 0.854 to 0.955); p<0.001), and procedure time (OR 0.985 (95% CI 0.973 to 0.997); p=0.015) were significant independent factors associated with a good clinical outcome (table 3).

Table 3

Multicollinearity test and univariable and multivariable logistic regression analysis for modified Rankin Scale 0–2 at 3 months

Discussion

Mechanical thrombectomy has become the standard of care for acute LVOs involving the ICA and in the MCA-M1 segment.1 However, the safety and efficacy of EVT in more distal intracranial branches remain unclear. This propensity score matching study showed that, after an adjustment for baseline covariates, there were similar rates of FPE, final reperfusion, and good clinical outcome at 3 months in patients with isolated MCA-M2 occlusions who were treated with SR alone versus SR+CA. The use of SR was associated with shorter procedural times, lower utilization of rescue therapy, and lower incidence of procedure-related complications. We performed propensity score matching to overcome the patients’ clinical heterogeneity, therefore, after adjusting for baseline characteristics, both groups were more homogeneous and comparable.

ASTER 2 was the first randomized clinical trial comparing SR alone versus combined EVT as front-line techniques in patients with distal ICA, MCA-M1, or MCA-M2 occlusions.20 This trial did not demonstrate any significant differences regarding the rate of final eTICI 2c/3 score between techniques. Moreover, a recent meta-analysis, focused on the efficacy of EVT techniques for isolated MCA-M2 occlusions, did not demonstrate differences in the rate of final eTICI 2b/2c/3 (OR 1.05 (95% CI 0.91 to 1.21); p=0.70) across SR (87%, n=200 patients) versus CA (80%, n=157 patients).8 We found that similar higher final eTICI 2b/2c/3 reperfusion rates were achieved in both groups (94% for SR alone and 95.5% for SR+CA) compared with previous published works.2–8 21 This might be attributable in part to the advent of new generation devices and a greater experience of the operators in treating M2 occlusions. The only independent variable associated with FPE was baseline ASPECTS score. Specifically, EVT technique was not a predictor of FPE.

The BGC technique has been associated with higher successful reperfusion rates, shorter procedure times, better clinical outcomes, and lower risk of mortality.22 23 In our study, the use of BGC was significantly unbalanced (98% in the SR group and 39% in the SR+CA group). Most large-bore aspiration catheters are not fully compatible with the commonly used BGCs, therefore operators more accustomed to using aspiration techniques tend to use guide catheters other than BGCs. Nevertheless, our analysis showed that BGC use was neither a predictor of near-complete/complete reperfusion nor good clinical outcome at 3 months.

Concerning the safety outcomes, several advantages were speculated in the use of a combined technique compared with single device techniques. First, the SR allows a distal trapping of the clot while the vacuum-assisted aspiration reaches the proximal side allowing a capture of the thrombus by both sides,24 25 thus lowering the risk of clot fragmentation during the thrombectomy pass and the risk of clinical worsening. Moreover, by capturing part of the SR by the distal aspiration catheter (pinning), there is less contact surface between the SR and the artery wall reducing the radial and tractional force, hence reducing the risk of vessel injury.26 However, we found a higher number of procedure-related complications in the combined EVT group, in addition to a higher incidence of sICH among the overall study population perhaps due to the combination of SR with large-bore aspiration catheters (the same devices used as in proximal LVOs), with the consequent higher risk of hemorrhagic complications. In this regard, Pérez-García et al reported a lower incidence of sICH in patients with distal medium vessel occlusions treated with combined EVT through the blind exchange with mini-pinning technique in comparison with SR alone,11 probably due to the use of lower profile devices.

Recent evidence shows that improving the rate of FPE and improving the procedure time will translate into clinical improvement and patient benefit.27 28 However, the larger number of studies focus on ICA or MCA-M1 occlusions, so the predictors of good clinical outcome after EVT in M2 occlusions are not fully defined. We found that age, NIHSS at baseline, and procedure duration were independently associated with a good clinical outcome at 3 months. In fact, recent data indicate that longer procedures may have a negative impact on clinical outcome,29 since prolonged procedures are associated with a higher risk of microembolism and hemorrhagic complications. The use of SR alone was associated with significantly shorter procedural times and a lower incidence of procedure-related complications. This could explain the trend towards a better clinical outcome in the SR alone group.

Our study has some limitations. The nature of the registry was non-randomized and there was no independent adjudication of the clinical and angiographic outcomes. This is a retrospective observational study that suffers from inherent methodological restrictions; nevertheless, the study was performed with data collected prospectively from several comprehensive stroke centers. The main results arise from cohorts homogenized by means of propensity score matching, thus we believe that the selection bias might be reduced. In addition, the election of any specific EVT technique was at the discretion of the operator in each case. The fact that different types and sizes of SRs and aspiration catheters were used across different anatomies should be mentioned as a potential bias. Notably, all thrombectomy devices used were latest generation.