After the recent randomized controlled trials demonstrating the superiority of mechanical thrombectomy (MT) over standard medical management alone, attention in the field is now focused on reducing time to revascularization, optimizing imaging methods for patient selection, and evaluating the best technical approach [1-4]. The Aspiration vs. Stent Retriever for Successful Revascularization (ASTER) trial aimed to compare the efficacy and safety of first-line endovascular treatment (EVT) using the contact aspiration (CA) technique versus the standard stent retriever (SR) technique and failed to show any superiority of CA on successful reperfusion [5]. In this context, thrombus subtype may play a key role in reperfusion success [6-8]. In the ASTER trial, a majority of patients were screened with magnetic resonance imaging (MRI) for acute stroke diagnosis. Thrombus composition could therefore be studied using the susceptibility vessel sign (SVS) on gradient recalled echo (GRE) sequences [9]. Intracranial clots are SVS-positive when red blood cells (RBCs) are predominantly present within the thrombus [8,10-12]. Retrospective and monocentric studies have shown that the qualitative evaluation of SVS could help predict recanalization and functional outcome after EVT [6,13-16].

We aim to assess the impact of SVS on ASTER results with a particular focus on a potential interaction with the first-line strategy (CA vs. SR).

The trial design has been published recently [17]. In brief, the ASTER trial was a prospective, randomized, multicenter, controlled, open-label, blinded endpoint (PROBE) clinical trial. It was an academic trial designed to address the question of which first-line strategy for MT (CA or SR) obtains higher revascularization rates at the end of the endovascular procedure. Patients were recruited at eight high-volume, comprehensive stroke centers in France, all of which regularly perform both types of technique. The study, which is registered with Clinical-Trials.gov (Identifier NCT02523261), was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice. Data collection during the study is reported in Table 1. The study protocol and the consent form were approved by the Comité de Protection des Personnes Ile de France VI (ID 2015-A00830-49). The details of the trial protocol were published previously [14].

Between October 2015 and October 2016, a total of 381 patients were randomized in the ASTER trial. Among these, 202 patients were eligible for inclusion in the present study (Figure 1). Reasons for exclusions were no EVT (n=45), internal carotid artery occlusion (n=65), and absence or poor MRI quality prior to MT for SVS diagnosis (n=69). Of the 202 included patients, 106 patients received SR in first-line and the remaining 96 received CA. Baselines characteristics and outcomes are provided for the overall study sample (Table 1) and according to the presence or absence of SVS. For the presence or the absence of the SVS, we found an excellent agreement between readers with an unweighted Cohen κ=0.88 (95% CI, 0.81 to 0.94). SVS was diagnosed in 70.8% of cases (n=143; 95% CI, 64.5% to 77.1%) with no difference according to the site of occlusion (M1, 69.0% vs. M2, 75.4%, P=0.36). As shown in Table 1, only the rate of IV tissue plasminogen activator (t-PA) prior to MT differed between the two groups with a higher rate in SVS (+) patients compared to SVS (-) patients (71.3% vs. 50.9%, P=0.005) (Table 1).

The main finding of our work is that the presence of SVS impacts the success rate of the thrombectomy strategy. Compared to CA, patients treated with SR achieve higher rates of complete reperfusion within fewer passes, which translates into a better functional outcome. In the absence of SVS, no differences were observed between the two techniques. Since the thrombus is the primary target of current reperfusion therapies, SVS evaluation on MRI before treatment appears to be relevant in order to determine the most effective MT strategy.

The SVS on a GRE sequence is defined as a hypointense signal, located at the site of the thrombus, exceeding the diameter of the contralateral artery [9]. SVS is observed in 50% to 85% of cases of hyperacute stroke, particularly in the case of a RBC-dominant thrombus, whereas a lack of SVS is indicative of a fibrin-dominant thrombus [6,7,13,14,16,25-28]. Of note, a linear variation of the RBC/fibrin ratio has been reported in samples of retrieved thrombi [29]. Hence, a cutoff ratio of erythrocyte components (>64% RBCs) has been related to successful recanalization using SR. Overall, SVS has been studied in the literature on ischemic stroke treated by IV t-PA or EVT with endpoints defined either on recanalization or functional outcome. The presence of SVS is associated with higher rates of recanalization and favourable outcome after EVT [6,7,13,15]. Conversely, the absence of SVS was related to better recanalization rate and clinical outcome after IV t-PA [14,16,30]. Interestingly, in vitro studies have shown that white thrombi, which are fibrin-rich, are not permanently embedded within the SR struts but are more prone to roll between the device and the vessel wall during retrieval [31]. On the other hand, red thrombi, are easily caught by SR [31].

We show in the present study that the rate of successful recanalization within three passes was higher in the first line SR arm than in the first line CA arm in the setting of SVS (+). Our hypothesis is that the SR struts are able to incorporate a soft RBC thrombus. This result could also be related to the use of the balloon-guiding catheter in the SR arm. Beyond the benefit of a mTICI 2b, 2c, or 3, time to reperfusion is a well-known independent prognostic factor for clinical outcome at 3 months, which underlines the importance of successful reperfusion within the first passes [1,32,33]. Additionally, numerous attempts and time consuming procedures, by increasing the risk of rethrombosis and per-procedural complications, compromise the probability of favourable clinical outcome [3,4,34]. This fact explains probably, at least in part, why in SVS (+) patients experience better clinical outcome with SR as a first line approach compared to CA.

Some limitations might be applied to our study. Since previous studies have shown that SVS (-) patients were better responders to IV t-PA than SVS (+) patients, our higher proportions of IV t-PA treatment among SVS (+) patients might be explained by a higher rate of recanalization after IV t-PA in SVS (-) patients, leading to their exclusion in the present study. This is probably why we found more patients who received IV t-PA before EVT in the SVS (+) group than in the SVS (-) group. However, our analysis was adjusted for previous IV t-PA. Although SVS identification is reliable and reproducible, with excellent interobserver and intraobserver agreements, SVS evaluation in the ASTER trial was performed on GRE sequences acquired on different MRI machines due to the multicentric design of the study. This is a significant limit, as we previously reported in an in vitro study that the accuracy of diagnosing SVS to identify RBC-dominant thrombi varies significantly among MRI machines [35]. Hence, a more reliable quantitative imaging biomarker able to overcome this limitation might be useful in order to adapt EVT strategy to the thrombus composition [7]. There is a need for biomarker comparison between different modalities including imaging and biological approaches to better define the significance of SVS. Advanced post-treatment software and acquisition sequences could be of a great interest in evaluating the composition of the thrombus, but one should keep in mind that time is still limited in this context and imaging must not delay treatment after the diagnostic confirmation. In addition, we have not been able to analyse thrombi retrieved after MT and therefore did not compare SVS and histology. However, published articles failed to find a reliable and reproducible association between the thrombi histology and recanalization results [10,36]. Indeed, there are inherent biases in assessing histologic characteristics of thrombi retrieved in vivo. IV t-PA is, when required, injected after imaging but before histological analysis and only the thrombi retrieved or parts of it are available for this analysis. Lastly, we have only included middle cerebral artery (M1 or M2) occlusions, since the skull base artefact in cases of more proximal occlusions can alter the evaluation of SVS. This certainly reduced the study sample but increased the homogeneity of the studied population.

Based on the ASTER trial population, we analysed the impact of SVS on MT strategy. With SR as a first line strategy, we found a higher rate of successful recanalization when SVS was present. Furthermore, for patients with SVS (+), and even after rescue therapy, the rate of favourable clinical outcome at 3 months was higher with SR compared to CA as the first line strategy. Thrombus MRI examination before MT thus appears relevant to adapt first line strategy in the setting of acute ischemic stroke due to large vessel occlusions. Large prospective studies using MRI clot evaluation are warranted in order to confirm this hypothesis.