Two Paradigms for Endovascular Thrombectomy After Intravenous Thrombolysis for Acute Ischemic Stroke

Key Points

Question  Do patients benefit from bridging therapy under the drip-and-ship paradigm?

Findings  In a registry study of 159 patients who received intravenous thrombolysis for acute ischemic stroke with a large vessel occlusion, the proportion with a favorable neurologic outcome at 3 months was similar in the drip-and-ship and mothership groups, even after adjusting the analysis for the baseline National Institutes of Health Stroke Scale score, diffusion-weighted imaging Alberta Stroke Program Early Computed Tomography Score, and general anesthesia.

Meaning  This study found that the 3-month functional independence rates observed in the clinical trials can be reproduced in everyday practice, proving that patients treated under the drip-and-ship paradigm also benefit from bridging therapy.

Importance  Intravenous thrombolysis (IVT) followed by mechanical thrombectomy (MT) is recommended to treat acute ischemic stroke (AIS) with a large vessel occlusion (LVO). Most hospitals do not have on-site MT facilities, and most patients need to be transferred secondarily after IVT (drip and ship), which may have an effect on the neurologic outcome.

Objective  To compare the functional independence at 3 months between patients treated under the drip-and-ship paradigm and those treated on site (mothership).

Design, Setting, and Participants  This study used a prospectively gathered registry of patients with AIS to select patients admitted through the Saint-Antoine and Tenon (drip and ship) or the Fondation Rothschild (mothership) hospitals from January 1, 2013, through April 30, 2016. The study included patients older than 18 years treated with bridging therapy for AIS with LVO of the anterior circulation. Among the 159 patients who received MT at the mothership, 100 had been transferred after IVT from the drip-and-ship hospitals and 59 had received IVT on site.

Main Outcomes and Measures  The main outcome was 3-month functional independence (modified Rankin scale score ≤2). Both groups were compared using a multivariate linear model, including variables that were significantly different in the 2 groups.

Results  During the study period, 497 patients were hospitalized at the drip-and-ship and mothership hospitals for an AIS eligible to reperfusion therapy; 11 patients had a basilar artery occlusion and were excluded, leaving 100 patients in the drip-and-ship group (mean age, 73 years; age range, 60-81 years; 57 men [57.0%]) and 59 in the mothership group (mean age, 70 years; age range, 58-82 years; 29 men [49.2%]). The proportion of patients with a favorable neurologic outcome at 3 months was similar in both groups (drip and ship, 61 [61.0%]; mothership, 30 [50.8%]; P = .26), even after adjusting the analysis for the baseline National Institutes of Health Stroke Scale score, diffusion-weighted imaging Alberta Stroke Program Early Computed Tomography Score, and general anesthesia (P = .82). Patients had less severe conditions in the drip-and-ship group (median baseline National Institutes of Health Stroke Scale score, 15 vs 17 [P = .03]; median diffusion-weighted imaging Alberta Stroke Program Early Computed Tomography Score, 7.5 vs 7 [P = .05]). Process times were longer in the drip-and-ship group (onset-to-needle time, 150 vs 135 minutes; onset-to-puncture time, 248 vs 189 minutes; and onset-to-recanalization time, 297 vs 240 minutes; P < .001). Both groups were similar in terms of substantial recanalization (Thrombolysis in Cerebral Ischemia scores 2B to 3; drip and ship, 84 [84.0%]; mothership, 47 [79.7%]; P = .49) and symptomatic hemorrhagic transformation (drip and ship, 2 [2.0%]; mothership, 2 [3.4%]; P = .63).

Conclusions and Relevance  This study found that patients treated under the drip-and-ship paradigm also benefit from bridging therapy, with no statistically significant difference compared with those treated directly in a comprehensive stroke center.

Acute ischemic stroke (AIS) management has evolved significantly during the past 20 years, with the development of stroke units and advances in reperfusion therapies. Until recently, reperfusion mainly consisted of intravenous thrombolysis (IVT), which has proven efficacy when administered within 4.5 hours after symptom onset.^1-3 Mechanical thrombectomy (MT) was limited to basilar artery occlusions and situations in which IVT was contraindicated.^4-8 However, since December 2014, several clinical trials^9-14 have acknowledged the superiority of bridging therapy, which consists of IVT within 4.5 hours followed by MT within 6 hours of symptom onset, over IVT alone in AIS with a large vessel occlusion (LVO) of the anterior circulation.

Quiz Ref IDThis paradigm shift has proved to be difficult to implement given that fewer than one-third of all stroke centers in the United States (327 of 1148) and in France (37 of 132) are comprehensive stroke centers (CSCs) with on-site interventional neuroradiologic services.^15,16 In 2011, only 56% of the US population had 60-minute ground access to a CSC, 66% to a primary stroke center (PSC), and 81% to an IVT-capable hospital. Therefore, admitting patients directly to CSCs seems incompatible with the need to start reperfusion therapy as soon as possible. To solve this issue, it has been proposed that patients could receive IVT in a hospital before being transferred to a CSC for MT, a procedure referred to as the drip-and-ship (DS) paradigm, in opposition to direct admission to a CSC, referred to as the mothership (MS). The DS paradigm unavoidably leads to longer process times and could cause a significantly worse prognosis for patients because several meta-analyses^17-23 from clinical trials have found that delaying reperfusion decreases the benefit of MT. Most of the studies on the DS vs MS debate were performed before the generalization of bridging therapy. They focused on IVT or thrombectomy for basilar artery occlusions and often assessed patient outcome at discharge.^24-31 Three studies^32-34 have more recently been published on the effect of the DS paradigm on patients undergoing bridging therapy, but they have produced conflicting results. The aim of this study was to compare the neurologic outcome and the process times between patients admitted for an AIS with LVO who underwent bridging therapy in a DS vs MS paradigm.

Patients were identified from 2 prospectively gathered and previously described AIS registries (Saint-Antoine and Tenon PSC and Fondation Ophtalmologique Adolphe de Rothschild CSC).^35,36 Both hospitals, which are 3.2 miles apart (25 minutes by ambulance), cover and are located in the northeastern part of Paris, France. Quiz Ref IDThe patient delivery rule was to go to the nearest stroke center with available beds. The DS group included patients treated with IVT in the PSC, and the MS group included those directly admitted to the CSC. For the DS patients, the MS was contacted after the brain imaging while IVT was being administered to validate MT indication and coordinate for the transfer. In 2012, both stroke centers decided to systematically provide treatment with IVT and MT to patients presenting with AIS with LVO. Therefore, we decided to include the patients admitted from January 1, 2013, to April 30, 2016, who met the following criteria: 18 years or older, evidence of an AIS with an LVO of the anterior circulation (internal carotid artery [ICA], M1 or M2 middle cerebral artery [MCA] occlusion), treated with IVT, and eligible for MT within 6 hours after symptom onset. During the study period, MT techniques mostly consisted of stent retriever and direct aspiration. Until February 2015, the CSC participated in the Trial and Cost Effectiveness Evaluation of Intra-arterial Thrombectomy in Acute Ischemic Stroke (THRACE) clinical trial.¹⁴ The study protocol was approved by the Comité de Protection des Personnes III Nord Est Ethics Committee and the research boards of the CSCs. Patient consent was waived, and approval for the retrospective analyses of patients’ records and imaging data was obtained from the institutional review boards of both hospitals. Data were deidentified in the final database.

We collected the following information from the registries, reviewing medical records when data were missing or insufficient: age, sex, prestroke modified Rankin scale (mRS) score, vascular risk factors (hypertension, diabetes mellitus, hyperlipidemia, and smoking), prestroke medication (antiplatelet, anticoagulant), index stroke onset time and type (clear, wake-up, or unclear), severity measured by the National Institutes of Health Stroke Scale (NIHSS), IVT time, MT characteristics (anesthesia, groin puncture time, reperfusion time, and Thrombolysis in Cerebral Ischemia [TICI] score, which ranges from 0 [no reperfusion] to 3 [complete reperfusion]), neurologic outcome within 24 hours measured by the NIHSS, malignant infarct, discharge NIHSS score, 3-month mRS score (or, when missing, discharge mRS), and cause determined by the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification (large-artery atherosclerosis, cardioembolism, small vessel occlusion, undetermined, or other cause).^37,38 The primary outcome was the percentage of patients with favorable functional outcomes (mRS score ≤2) 3 months after the stroke. Secondary outcomes were substantial recanalization and symptomatic intracranial hemorrhage (sICH) rates. Substantial recanalization corresponded to the antegrade reperfusion of more than half (TICI 2B) or all (TICI 3) of the previously occluded target artery ischemic territory. We used the Safe Implementation of Thrombolysis in Stroke-Monitoring (SITS-MOST) study sICH definition: local or remote parenchymal hemorrhage type 2 associated with an increase of 4 points or more in the NIHSS score.³⁹

All patients underwent brain magnetic resonance imaging (MRI) and/or injected computed tomography (CT) during the stroke alert and 24 hours after reperfusion. All were assessed retrospectively by a trained neurologist for the following criteria: diffusion-weighted imaging Alberta Stroke Program Early Computed Tomography Score (DWI-ASPECTS) for infarct size (G.G.), axial fluid-attenuated inversion recovery (FLAIR) early vessel sign (G.G.), and occlusion site at time-of-flight magnetic resonance angiography or injected brain CT (I.-P.M.) and after 24 hours for ICH transformation according to the European Cooperative Acute Stroke Study (ECASS) radiologic classification (G.G.).³⁶

Baseline characteristics were compared between patients treated in the DS and MS groups using the 2-tailed, unpaired t test for continuous variables with a normal distribution and the χ² test or the Fisher exact test for categorical variables. We then compared 3-month functional independence, substantial recanalization, and sICH rates between the 2 groups by using a multivariate linear model. We included in this final model only variables that were significantly different in the 2 groups (P < .05) in the univariate analysis. We chose not to include process times in the model because they were expected to be longer in the DS group. All statistical analyses were performed using the computer software R-Studio, version 0.99.903 (RStudio). In all analyses, 2-sided P < .05 was considered statistically significant.

During the study period, 497 patients were hospitalized at the DS and MS hospitals for an AIS eligible for reperfusion therapy (Figure 1): 61 were treated with intra-arterial therapy alone (31 in the DS group and 30 in the MS group) because of contraindications to IVT, 266 were treated with IVT alone (144 in the DS group and 122 in the MS group), and 170 were treated with IVT followed by intra-arterial therapy. Among them, 11 patients had a basilar artery occlusion (10 in the DS group and 1 in the MS group) and were excluded, leaving 100 patients in the DS group (mean age, 73; age range, 60-81 years; 57 men [57.0%]) and 59 in the MS group (mean age, 70 years; age range, 58-82 years; 29 men [49.2%]). The DS and MS groups did not differ significantly in terms of demographic characteristics, such as age, sex, vascular risk factors, prestroke antithrombotic treatment, and stroke cause (Table 1).

Almost all patients in both groups (92 [92.0%] in the DS group and 56 [94.9%] in the MS group) underwent brain MRI during the stroke alert. Quiz Ref IDPatients in the DS group had significantly less severe stroke, with a median baseline NIHSS score of 15 (interquartile range [IQR], 8-20) vs 17 (IQR, 13-21; P = .03) and a median DWI-ASPECTS of 7.5 (IQR, 6-8) vs 7 (IQR, 5-8; P = .05). Both groups were similar in terms of left hemisphere stroke proportion, FLAIR early vessel sign, and occlusion site. Patients in the DS group had significantly less general anesthesia (9 [9.0%]) than those in the MS group (12 [20.3%]; P = .04). One patient in each group did not undergo arteriography because of a recanalization on a control MRI in the DS group or for having a stroke that was too severe in the MS group. Both groups did not differ significantly in terms of proportion of patients who received MT (74 [74.0%] in the DS group and 51 [86.4%] in the MS group; P = .06). Reasons for not receiving MT were clinical recovery (3 [3.0%] in the DS group and 1 [1.7%] in the MS group), recanalization on arteriography (18 [18.0%] in the DS group and 4 [6.8%] in the MS group), severe ICA stenosis (2 [2.0%] in the DS group), severe MCA stenosis (1 [1.0%] in the DS group), chronic MCA occlusion (1 [1.7%] in the MS group), good collateral circulation (1 [1.7%] in the MS group), and failed aortic catheterization (1 [1.7%] in the MS group). Thrombectomy devices used in both groups were also similar.

Median process times were significantly longer in the DS group: onset-to-needle time of 150 minutes (IQR, 120-190 minutes) vs 135 minutes (IQR, 114-155 minutes; P = .02), onset-to-puncture times of 248 minutes (IQR, 220-291 minutes) vs 189 minutes (IQR, 163-212 minutes; P < .001), and onset-to-recanalization time of 297 minutes (IQR, 255-357 minutes) vs 240 minutes (IQR, 202-285 minutes; P < .001). Needle-to-puncture time, an approximation of the transfer delay, was significantly longer in the DS group: 93 minutes (IQR, 79-110 minutes) vs 48 minutes (IQR, 30-67 minutes; P < .001). Four patients in the DS group received IVT after 270 minutes, and MT was started (groin puncture) after 360 minutes in 8 patients (7 in the DS group and 1 in the MS group).

Three-month mRS scores were available for 149 patients (95 in the DS group and 54 in the MS group), and discharge mRS scores were available for the remaining patients. In the univariate analysis (Table 2), there was no statistically significant difference between the 2 groups for the main outcome (61 [61.0%] in the DS group vs 30 [50.8%] in the MS group; P = .26). In the multivariate analysis, when adjusted on the baseline NIHSS score, the DWI-ASPECTS, and general anesthesia, there was still no significant difference between the 2 groups for the main outcome (P = .82). This was also the case when the onset-to-recanalization time was added in the multivariate linear model (47 of 80 [58.8%] in the DS group vs 28 of 55 [50.9%] in the MS group; P = .14). Substantial recanalization (TICI 2B-3) rates were similar in both groups (84 [84.0%] in the DS group and 47 in the MS group [79.7%]; P = .49), even after adjustment (P = .59). Both groups had a parenchymal hemorrhage type 2 hemorrhagic transformation rate of 8% and did not differ significantly in terms of sICH (2 [2.0%] in the DS group and 2 [3.4%] in the MS group; P = .63), even after adjustment (P = .65).

Among the study population, 125 patients effectively underwent MT. In a per protocol analysis (eFigure in the Supplement), there was still no statistically significant difference for the main outcome (40 [54.1%] in the DS group and 24 [47.1%] in the MS group; P = .44). Substantial recanalization (63 [85.1%] in the DS group and 43 [84.3%] in the MS group; P = .90) and sICH (2 [2.7%] in the DS group and 2 [3.9%] in the MS group; P > .99) rates did not differ significantly.

We performed a subgroup analysis excluding the patients with an M2 MCA occlusion (15 in the DS group and 5 in the MS group) for 2 reasons. First, we wanted to see whether the difference in terms of the baseline NIHSS score and DWI-ASPECTS was attributable to the more important proportion of M2 MCA occlusions in the DS group, even though it was not statistically significant. Second, we wanted to assess its effect on the main outcome because, in most of the published clinical trials, the proportion of M2 MCA occlusions were lower than in the DS group. The median baseline NIHSS score was still significantly lower in the DS group (16 vs 17 in the MS group; P = .03), but this was not the case for the DWI-ASPECTS. The subgroups did not differ significantly for the main outcome (49 of 85 [57.6%] in the DS group vs 28 of 54 [51.9%] in the MS group; P = .26) or in terms of substantial recanalization (70 [82.3%] in the DS group and 43 [79.7%] in the MS group; P = .69) or sICH (2 [2.4%] in the DS group and 1 [1.9%] in the MS group; P > .99) rates.

To assess whether early IVT administration had attenuated the effect of increased delays, we compared the main outcome among the patients who underwent MT alone (eTable in the Supplement). There were 54 patients (28 in the DS group and 26 in the MS group) with an occlusion of the anterior circulation. The median onset-to-recanalization times were higher than in the IVT plus MT groups at 301 minutes (IQR, 244-439 minutes) in the DS group vs 286 minutes (IQR, 218-340 minutes) in the MS group, but these times did not differ significantly (P = .10). The 3-month functional independence rate was lower in both groups (10 [35.7%] in the DS group and 7 [26.9%] in the MS group) compared with the IVT plus MT groups but did not significantly differ between them (P = .49). The substantial recanalization rates were slightly but not significantly lower in the DS group (19 [67.8%] in the DS group vs 23 [88.5%] in the MS group; P = .42).

Our study found that patients treated using the DS paradigm also benefit from bridging therapy, with no significant difference compared with those treated directly in a CSC and with substantial recanalization and 3-month functional independence rates similar to the results of the previously published clinical trials (Figure 2 and Figure 3). These results should be interpreted in the context of the 2 debates that have arisen regarding bridging therapy during the past year.

In the first debate, it was questioned whether patients with LVO should be treated as soon as possible with MT, without IVT if it took too long. Indeed, the reperfusion therapy paradigm is that “time is brain” and that the sooner the penumbra is reperfused the better the neurologic outcome will be. However, IVT recanalization rates are known to be low in LVO, as Seners and colleagues⁴⁰ reported in a meta-analysis, at 35% for M1 MCA and 13% for ICA occlusions. However, preclinical data suggest that the role of IVT is more complex. In a transient MCA occlusion model, IVT acts on the downstream microvascular thrombosis that starts immediately after the transient occlusion, limiting the infarct extension and yielding better functional results.⁴¹ Moreover, previously published registry studies^42,43 had conflicting results when comparing MT alone with bridging therapy. In the absence of a clinical trial designed to answer this question, bridging therapy remains the only proven treatment for patients presenting with AIS with LVO. Therefore, IVT is the first line of reperfusion treatment and should be administered as soon as possible within 4.5 hours of symptom onset.³

The second debate has focused on the organization required to maximize patient access to bridging therapy, opposing the DS and MS paradigms. With more PSCs than CSCs, DS allows patients to receive IVT as soon as possible at the nearest center and transfer them for MT secondarily. On the other hand, it has been questioned whether the increased process times would result in worse clinical outcomes and higher rates of sICH. Our study found that this supposition is not the case, even though the onset-to-needle time was significantly longer in the DS group (150 vs 135 minutes; P < .001). An important factor in this result was the transfer for MT, which lengthened needle-to-puncture times by only 45 minutes, from 48 minutes in the MS group to 93 minutes in the DS group (P < .001). Indeed, Saver and colleagues¹⁷ recently found in a pooled meta-analysis that the benefit of bridging therapy becomes nonsignificant when the onset-to-puncture time is longer than 438 minutes. Thanks to this limited increase, the median onset-to-puncture time was 248 minutes (Figure 3), and only 1 patient was over the limit (442 minutes). This finding was equally made possible by the geographic proximity of the 2 centers and the good upstream cooperation. Indeed, Sun and colleagues⁴⁴ found that the more often patients are transferred by the same hospital to a CSC, the more often the picture-to-puncture time decreases. Quiz Ref IDSimilarly, we observed a 21-minute decrease of the needle-to-puncture time during the study period in the DS group. We believe that this outcome was the combined result of a better consideration of the importance of the process times after the publication of the trials and of a learning curve effect at several levels (cooperation between both centers, internal organization, and neuroradiologists’ MT expertise). In addition, these results, similar to the ones from the Extending the Time for Thrombolysis in Emergency Neurological Deficits–Intra-arterial (EXTEND-IA) trial, were achieved using only the FLAIR early vessel sign, which both groups had in a very high proportion (93.5% in the DS group and 89.3% in the MS group), to assess perfusion and good collateral vascularization.

Another way to answer the DS vs MS question could be a prehospital triage that allows transfer of patients with stroke and suspected LVO directly to a CSC. Currently, this scenario remains impossible for several reasons. Quiz Ref IDIt is currently impossible to select patients with LVO purely on clinical criteria. Turc and colleagues⁴⁵ recently found that using the suggested NIHSS triage cutoffs would result in a false-negative rate of more than 20% and that reducing this rate to 10% would result in sending almost every patient to a CSC. The only way to perform prehospital LVO triage would be to use a specialized CT-equipped mobile stroke unit because it has been tested in Berlin for IVT.⁴⁶ However, this process has yet to be tested in a clinical trial and would also require a utility cost analysis given its high cost. Furthermore, even a better prehospital triage would not solve the issue of geographic access, which is particularly important in rural regions, where CSCs are even rarer than in urban areas. Given the available data, it is imperative to start reperfusion therapy with IVT as soon as possible, and delaying IVT to start it directly in a CSC could represent a worse prognosis for patients. Finally, at the current stage, all the CSCs could not admit all the patients requiring bridging therapy. There would have to be a massive development to increase their number and capacity to make it possible, and such a process would take time.

Our study has several strengths. Both centers had similar practices, and all the patients were treated by the same neuroradiologists. All the imaging criteria were interpreted by the same trained neurologist. Every patient had a baseline NIHSS evaluation, and only 6.3% did not have a 3-month mRS assessment. The study also has limitations. Both centers were in a metropolitan area; thus, the findings may not be relevant to rural hospitals. Its design was retrospective, with a relatively small number of patients. The patients in the MS group had slightly but significantly more severe conditions. This difference could be attributable to an implicit recruitment bias, favored by the presence in the MS of a neurosurgery department and a neurologic intensive care unit.

This study found that patients treated using the DS paradigm also benefit from bridging therapy, with no statistically significant difference compared with those treated directly in a CSC (MS). The delays and 3-month functional independence rates observed in the clinical trials can be reproduced in everyday practice for patients requiring a transfer for MT.

Corresponding Author: Sonia Alamowitch, MD, Service de Neurologie et d’Urgences Neurovasculaires, Hôpital Saint-Antoine, 184, rue du Faubourg Saint-Antoine, 75012 Paris, France (sonia.alamowitch@aphp.fr).

Accepted for Publication: December 6, 2016.

Published Online: March 20, 2017. doi:10.1001/jamaneurol.2016.5823

Author Contributions: Dr Alamowitch had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Gerschenfeld, Muresan, Piotin, Alamowitch.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Gerschenfeld, Alamowitch.

Critical revision of the manuscript for important intellectual content: Muresan, Blanc, Obadia, Abrivard, Piotin, Alamowitch.

Statistical analysis: Gerschenfeld, Alamowitch.

Administrative, technical, or material support: Obadia, Abrivard.

Study supervision: Muresan, Blanc, Obadia, Piotin, Alamowitch.

Conflict of Interest Disclosures: Drs Blanc and Piotin reported receiving a grant from Penumbra for a clinical study and personal fees from Medtronic for an educational presentation outside the submitted work. Dr Alamowitch reported receiving personal fees from the Revue Neurologique (editorial board), AstraZeneca (consultancy and lectures), and Bayer (lectures) and grants from AstraZeneca and Programme Hospitalier de Recherche Clinique for clinical trials outside the submitted work. No other disclosures were reported.

Additional Contributions: Marie Ecollan, MD, MSc, Department of General Medicine, Paris Descartes University, Paris, France, helped perform the statistical analyses without financial compensation.