This was a retrospective analysis of prospectively maintained databases from three comprehensive stroke centers in the United States. We included consecutive patients who presented with intracranial internal carotid artery (ICA), middle cerebral artery (MCA) M1/M2, or basilar/vertebrobasilar artery occlusion strokes and underwent EVT using either SR, catheter aspiration (contact aspiration (CA)) or a combined technique (CT: SR plus CA) for the first pass strategy. Passes employing other techniques (eg, remote balloon guide catheter aspiration, stenting, angioplasty, microwire maceration, intra-arterial thrombolysis, or a combination of those) or targeting at any other vessel (eg, anterior cerebral artery, MCA M3) were excluded. The study period spanned January 2015 (or the date on which each center started prospectively collecting thrombectomy data) to August 2023.

Patients were included for the per-pass analysis if they did not respond to the previous pass. For the third pass, patients who had undergone two passes with repeated strategies were included. Reperfusion rating was based on the expanded thrombolysis in cerebral ischemia (eTICI) scoring system.2 For analyses evaluating eTICI 2c–3 reperfusion as outcome, the lack of response for each pass was defined as eTICI < 2c or eTICI 2c–3 with subsequent reocclusion. For analyses involving eTICI 2b–3, the lack of response was defined as eTICI < 2b. The choice of device and technique was at the treating neurointerventionists’ discretion.

The primary endpoint was complete or near-complete recanalization (eTICI 2c–3) per pass for both the second and third passes. The secondary outcomes were successful recanalization (eTICI 2b–3) for the second and third passes, the cumulative incidences of recanalization at pass 3 according to different technique sequences, as well as the first pass effect (FPE, first pass eTICI 2c–3), functional independence (modified Rankin Scale (mRS) score ≤2) at day 90, subarachnoid hemorrhage, and parenchymal hematoma according to European Cooperative Acute Stroke Study III criteria.10

Since no specific reperfusion strategy has been demonstrated to outperform others, the decision to initiate or switch techniques was not expected to be significantly influenced by clinical or demographic characteristics. Rather, the decision on which strategy to employ is probably to be explained by the previous reperfusion grade achieved, occlusion location, resource availability, and operator preference.

Variables of interest were described as median (quartiles) or means (SD), when numeric, or absolute and relative frequencies, when discrete or ordinal. Logistic regression models with prespecified adjustments for age, occlusion location, use of balloon-guide catheter, pre-attempt eTICI, and center were used in the strategy modification analysis. Missing covariates data were handled using multiple imputation. Complications were compared via Fisher’s exact test, the Χ2 test, or the Kruskal-Wallis test as appropriate. Aspiration catheter sizes were compared between CT and CA using Mann-Whitney’s test.

We estimated the cumulative incidences of reperfusion across different sequences of techniques using the Kaplan-Meier method. All participants with a sequence compatible with the one of interest were considered (online supplemental figure S2). We compared groups according to the technique for the three first passes for patients who underwent passes using the same technique (SR vs CA vs CT) using Cox regression adjusting for age, occlusion location, use of balloon guide catheter, and site. A two-tailed α of 0.05 was considered. All analyses were conducted using R Software v4.3.1.

In this cohort of patients undergoing mechanical thrombectomy in three comprehensive stroke centers in the United States, switching to the combined technique after as early as one failed contact aspiration pass or two failed stent retriever passes was associated with higher full/near-full reperfusion rates, whereas changing to contact aspiration after two failed combined technique passes was found to be associated with lower chances of eTICI 2c–3 reperfusion.

The extent and speed of reperfusion has a substantial impact on clinical outcomes, and first pass eTICI 2c–3 is the primary angiographic goal of mechanical thrombectomy since it has been unequivocally associated with the highest rates of functional independence. The HERMES meta-analysis demonstrated that, compared with eTICI 0, the adjusted odds of mRS 0-2 at 90 days increased along with the degree of reperfusion (eTICI 2a: OR=1.4; eTICI 2b50: OR=2.4; eTICI 2b67: OR=5.1; eTICI 2c: OR=5.2; eTICI 3: OR=7.3).2 Additionally, as the number of required passes increases, the odds of a favorable clinical outcome decrease while the risk of intracranial hemorrhage rises.1 3 11–14 Although final eTICI 2c–3 rates as high as 65% have been reported in clinical trials, the FPE was only achieved in 23–41% of patients with proximal large vessel occlusions in contemporary series and randomized trials.1 6 14–20 Incorporating practice patterns that lead to faster and more complete reperfusion is therefore essential for improving clinical outcomes.

Randomized studies comparing MT strategies include the ASTER trial,4 which showed no difference in angiographic outcomes between up to three contact aspiration versus stent retriever passes in 189 patients, with a cumulative eTICI 2c–3 proportion of 56.6% for SR versus 56.3% for CA in the intention-to-treat analysis (P=0.82) and cumulative eTICI 2b–3 proportions of 83.1% (SR) and 85.4% (CA). A secondary analysis showed similar rates of first pass effect between SR and CA (31.3% vs 26.3%, P=0.44).20 Our findings compare favorably, as our hazard ratios for complete or near-complete reperfusion were approximately 1 when comparing stand-alone CA against SR after adjusting for age, occlusion location, balloon-guide catheter use and center. In addition, the present data suggest a potential negative impact of transitioning to CA after one failed SR pass for eTICI 2b–3 when compared with repeating the initial strategy. The COMPASS trial randomized 270 patients and reinforced the hypothesis of similar performance between SR and CA for eTICI 2c–3 within 45 min of access and first pass eTICI 2b–3, although CA outperformed SR for eTICI 3 within 45 min and time to eTICI 2b–3.5 The study compared time to outcome rather than number of passes and did not disclose first pass eTICI 2c-3 data, limiting comparisons. Of note, 85% of the patients randomized to SR had distal aspiration combined with SR thrombectomy at some point during their procedure.

The ASTER 2 study6 did not observe significant differences between the CT and SR alone among 405 patients for the prespecified primary outcome of eTICI 2c–3 with up to three passes (65% vs 58%, OR=1.33, 95% CI 0.88 to 1.99; P=0.17). The present findings are consistent with ASTER 2, as the comparison between up to three successive CT passes and up to three isolated SR also showed a higher, but not significantly different proportion of eTICI 2c–3 in the CT group (67.9% vs 63.2%, aHR=1.1, 95% CI 1.0 to 1.3; P=0.103) (figure 3). Estimates for first pass effect were also comparable between the ASTER 2 trial and this study. Whereas the trial estimated an OR for FPE of 1.4 (95% CI 0.9 to 2.1; P=0.12), the present study found an adjusted OR of 1.1 (95% CI 0.9 to 1.3; P=0.296).

Studies directly comparing CA with CT are more scarce. The Penumbra Separator 3D trial21 supported the similarity of both techniques for cumulative mTICI 2–3 and 2b–3, although the 3D Separator device was not designed for, and has not been demonstrated to function by itself as, a reperfusion device. Preliminary data presented at the European Stroke Organization Conference 2023 on the Adaptive endovascular strategy to the CloT MRI in large intracranial vessel occlusion (VECTOR) trial demonstrated earlier reperfusion was accomplished with the CT than with CA, with an absolute difference of 11% in FPE favoring the combined technique, although the final rates or eTICI 2c–3, defined as the primary endpoint, were not significantly different between groups.22 23 Our analyses also showed a benefit for earlier reperfusion with the combined technique when compared to contact aspiration alone (aHR 1.3, 95% CI 1.0 to 1.7, P=0.014) which may be attributable to first pass effect, for which a difference in favor of CT was also noted (aOR=1.3, 95% CI 1.0 to 1.6; P=0.027). Contact aspiration techniques have had a remarkable evolution and increasing use in the past years, which was also observed in our study (online supplemental figure S4), and a part of the lower reperfusion rates in our study could be explained by successive changes in device characteristics over time as well as operator experience with the technique. The discrepancy in cumulative reperfusion with three passes between the studies may reside in the fact that VECTOR included solely subjects with a positive susceptibility vessel sign (indicating red-cell-rich thrombi) on MRI, while the present series did not take clot characteristics into consideration.

The responsiveness of patients to given strategies might vary due to several factors. Clot composition may play a significant role, with red-cell-rich thrombi potentially being more responsive to SR than to CA24 25. However, biomarkers for determining clot composition and data supporting device/technique-specific benefits related to preprocedural thrombus profiling are scarce. A failed pass might also contribute to responsiveness to further passes and to specific techniques through its impact on the vessel wall and clot structure.26 The responsiveness to a specific technique might influence the chances of reperfusion in subsequent passes,3 supporting the concept that very early modification of the thrombectomy technique may be desirable, and that the lack of difference in cumulative reperfusion observed in trials might relate to the less substantial benefit with repeated passes using the same technique. Vessel angulation at the site of occlusion has been demonstrated to affect the performance of the Merci retriever,27 as well as SR28 and CA29 thrombectomy. It is possible that the combined technique might allow for a synergistic effect by (1) straightening the artery and allowing for the aspiration catheter to engage the clot following its main longitudinal axis and (2) attenuating the effect of clot adhesiveness by minimizing the area of thrombus exposed to the endothelium and decreasing the magnitude of breakaway force required for clot dislodgement.

Our study has several limitations, including weaknesses inherent to the study design such as unmeasured confounders. A propensity for certain first pass techniques in different centers was noted, which led to a between-center exploratory analysis that ultimately showed no significant interactions. Sample size limitations precluded the comparison of effects between anterior and posterior circulation occlusions. The selection of techniques based on the neurointerventionist’s discretion and experience with diverse techniques might have introduced bias. This study was not designed to evaluate the impact of changes in reperfusion strategies on clinical outcomes. There was no central adjudication for reperfusion scoring. There was substantial heterogeneity in device types and sizes; device stratification would lead the analysis to limited subgroup sample sizes. Although SR sizing and length were not controlled for, the size of aspiration catheter inner diameter was comparable between CA and CT for the second pass and minimally larger for CA as compared to CT for the first pass, therefore not explaining the observed benefit of CT. The analysis related to procedural times is exploratory and did not consider potential confounders.