2015 American Heart Association/American Stroke Association Focused Update of the 2013 Guidelines for the Early Management of Patients With Acute Ischemic Stroke Regarding Endovascular Treatment: A Guideline for Healthcare Professionals From the American Heart Association/American Stroke Association

Since the publication of the most recent “Guidelines for the Early Management of Patients With Acute Ischemic Stroke” in 2013,¹ substantial new high-quality evidence on the clinical efficacy of endovascular treatments of acute ischemic stroke has become available. This focused update on endovascular treatment of acute ischemic stroke analyzes results from 8 randomized, clinical trials of endovascular treatment and other relevant data published since 2013 while taking into account the previous evidence summarized in the 2013 guidelines. This focused update is not intended to be based on a complete literature review from the date of the previous guideline publication but rather to include pivotal new evidence that justifies changes in current recommendations. When there is overlap, the recommendations made here supersede those of previous guidelines.

Members of the writing committee were appointed by the American Heart Association (AHA)/American Stroke Association Stroke Council’s Scientific Statement Oversight Committee and the AHA/American Stroke Association Manuscript Oversight Committee, representing various areas of medical expertise. Strict adherence to the AHA conflict of interest policy was maintained throughout the consensus process. Panel members were assigned topics relevant to their areas of expertise, reviewed the stroke literature with emphasis on publications since the prior guidelines, and drafted recommendations in accordance with the American College of Cardiology/AHA’s Level of Evidence grading algorithm (Table 1). All recommendations were unanimously approved by the members of the writing group.

Rapid administration of intravenous recombinant tissue-type plasminogen activator (r-tPA) to appropriate patients remains the mainstay of early treatment of acute ischemic stroke.¹ Timely restoration of blood flow in ischemic stroke patients is effective in reducing long-term morbidity. For patients who meet national and international eligibility guidelines, intravenous r-tPA improves functional outcomes at 3 to 6 months when given within 4.5 hours of ischemic stroke onset and should be administered. Every effort should be made to shorten any delays in the initiation of treatment because earlier treatments are associated with increased benefits. If patients who are eligible for intravenous r-tPA do not have intracranial vascular imaging as part of their initial evaluation, they should begin receiving intravenous r-tPA before being transported for additional imaging and before being transferred for endovascular treatment. This approach will help minimize onset-to-treatment times, a key driver of efficacy for r-tPA.^1–6

None of the 3 earlier studies carried out with primarily intra-arterial fibrinolysis or first-generation mechanical embolectomy devices showed a benefit of endovascular treatment over intravenous r-tPA in intravenous r-tPA–eligible patients either as a substitute for initial treatment (SYNTHESIS Expansion) or as subsequent intervention in those with persistent large-artery occlusion after intravenous r-tPA (IMS III and MR RESCUE). MR RESCUE also showed no benefit for other patients treated within 8 hours even if selected by multimodal neuroimaging criteria. These studies, using almost exclusively intra-arterial r-tPA and first-generation endovascular devices alone or in combination, achieved recanalization rates of 27% to 41%. The subsequent trials using stent retrievers almost exclusively demonstrated improved results for both recanalization rates and outcome. Studies have shown that clinical outcome improved with increasing effectiveness of recanalization. Those with partial recanalization (TICI grade 2a) did not do as well as those with nearly complete or complete recanalization (TICI grade 2b/3) reflected as both differences in discharge disposition (41.0% of the TICI grade 2b/3 group discharged home versus 17.4% of the TICI grade 2a group) and functional outcome (34% with a TICI grade of 2a had an mRS score of 0 to 2 at 90 days versus 49% with a TICI grade of 2b/3).^12,23 TICI grade 2b/3 recanalization was achieved in 59% to 88% of endovascularly treated subjects in the 5 stent retriever trials, whereas in the previous 3 studies, the rate had been 25% to 41%, as mentioned above. All 5 stent retriever studies showed clinical benefit in the endovascular group.

Of the 5 stent retriever trials, MR CLEAN, ESCAPE, and SWIFT PRIME permitted use of salvage intra-arterial fibrinolytic drugs, whereas EXTEND-IA and REVASCAT did not. These data do not establish the benefit of intra-arterial fibrinolytic salvage, nor can they establish lack of benefit. Such salvage techniques may be reasonable to use in some clinical circumstances.

The MR RESCUE trial enrolled patients up to 8 hours from symptom onset and showed no benefit from endovascular therapy with first-generation devices regardless of penumbral imaging pattern. Three of the 5 stent retriever studies specified a 6-hour window after stroke onset (2 specified 6 hours to groin puncture; the third specified 6 hours to start treatment). Aggregate data from REVASCAT and ESCAPE with treatment permitted out to 8 and 12 hours show a benefit, but ESCAPE enrolled too few patients after 6 hours to provide useful data, and REVASCAT provides no data about patients who underwent groin puncture between 6 and 8 hours. How much the overall positivity in these 2 trials was completely driven by those treated at shorter times is unknown at this time. The only time-dependent data are from the MR CLEAN presentation, which are not consistent with a benefit of treatment beginning after 6 hours. It will take patient-level meta-analyses to sort this out.

Every or nearly every patient in the 5 stent retriever studies first received intravenous r-tPA. Only REVASCAT stipulated the specific guidelines to be used to determine intravenous r-tPA eligibility (“guidelines provided by the European Stroke Organization”). EXTEND-IA refers to “standard criteria,” and the 3 other trials used “national guidelines.” Because it is not the purpose of this update is to address eligibility criteria for intravenous r-tPA, we have used the phrase “guidelines from professional medical societies” to address this issue in our recommendations. Too few data are available from the small number of those who did not receive intravenous r-tPA, for either time-based or non–time-based exclusion criteria, to determine with certainty whether there are characteristics that identify those who benefited from endovascular treatment. Two trials (MR CLEAN and REVASCAT) stipulated waiting for a period of time after beginning the administration of intravenous r-tPA before proceeding to endovascular therapy, whereas 3 trials (ESCAPE, SWIFT PRIME, and EXTEND-IA) did not. On the basis of these data, a waiting period is not necessary to achieve beneficial outcome in these patients.

All of these studies enrolled participants ≥18 years of age. There are no randomized trials of endovascular therapy in patients <18 years of age. Ischemic stroke resulting from large-vessel occlusion is rare in children and young adults relative to older individuals, posing challenges to rigorous study of this clinical scenario. Case reports and case series have documented that high rates of recanalization and favorable outcomes in young patients can be achieved with endovascular therapy.^24–26 Ideally, appropriate trials would be done to test the efficacy of endovascular therapy in young patients. Studies in the United States, the United Kingdom, Australia, and Canada have shown median times from onset of symptoms to initial brain imaging for pediatric stroke of 8.8 to 16 hours.²⁷ This problem of diagnostic delay will need to be addressed if trials of endovascular treatment for acute ischemic stroke are to be conducted successfully in this population.

Four stent retriever trials used NIHSS scores as eligibility criteria (>2, >5, 8–29, and >5), and the fifth enrolled patients with a similar distribution of NIHSS scores. From these trials, there are insufficient data in patients with NIHSS scores <6 to determine whether there is an overall net benefit from endovascular therapy in this population. Further randomized trials in patients with low NIHSS scores may be warranted. An NIHSS score of ≥6 was the minimum score used in 2 trials, thus fulfilling the AHA’s Level of Evidence grading algorithm for Level A evidence.

Four of the 5 stent retriever trials used a prestroke function eligibility criterion. REVASCAT and SWIFT PRIME used a prestroke mRS score of 0 to 1; EXTEND-IA used mRS scores of 0 to 2; and ESCAPE used Barthel scores of ≥90 to 100. MR CLEAN did not set a threshold and did not provide data on prestroke function. Thus, there are good data from 4 trials for patients with good baseline function (including 2 that required an mRS score of 0 to 1) and very few data for those without good baseline function.

All 5 stent retriever studies required baseline nonenhanced CT or MRI. MR CLEAN did not use a specific ASPECTS criterion for eligibility; it was the only positive trial that permitted enrollment of patients with ASPECTS <6. Although the treatment effect in that trial favored intervention in all 3 ASPECTS subgroups of 0 to 4 (28 patients), 5 to 7 (92 patients), and 8 to 10 (376 patients), the point estimate in the subgroup with an ASPECTS of 0 to 4 was close to unity with wide CIs (adjusted common OR, 1.09; 95% CI, 0.14–8.46). In the ESCAPE trial secondary analyses based on ASPECTS, the risk ratio favoring intervention was 2.6 (95% CI, 1.7–4.1) for patients with an ASPECTS of 8 to 10 and 2.7 (95% CI, 1.0–7.2) for those with a score of 6 to 8. EXTEND-IA did not report secondary analyses based on ASPECTS. SWIFT PRIME reported similar benefit for those with ASPECTS of 8 to 10 (OR, 1.62; 95% CI, 1.17–2.24) and 6 to 7 (OR, 1.98; 95% CI, 0.73–5.33), although the small number of 43 patients in the latter group produced wide confidence bounds. REVASCAT reported greater benefit for those with ASPECTS ≥8 (OR, 2.2; 95% CI, 1.1–4.4) than for those with ASPECTS <8 (OR, 1.4; 95% CI, 0.7–2.9). On the basis of these data, the benefit from endovascular therapy in patients with ASPECTS <6 is uncertain, and further randomized, controlled trials are warranted. An ASPECTS ≥6 was the minimum score used in 2 trials, thus fulfilling the AHA’s Level of Evidence grading algorithm for Level A evidence.

Each of the 5 stent retriever trials used different strategies of imaging-based selection criterion in addition to nonenhanced CT or MRI. Common to all was required demonstration, usually with a noninvasive vessel imaging study (CTA or MRA), of a large-vessel occlusion before randomization. MR CLEAN and REVASCAT also allowed digital subtraction angiography screening to identify a target occlusion. Two trials required noninvasive imaging to be performed at initial evaluation before intravenous r-tPA was started (combined occurrence of no clot at endovascular intervention in 12 of 200 [6.0%]); a third recommended the same (no clot at endovascular intervention in 8 of 233 [3.4%]); and a fourth stipulated that it be done at all centers for which this was part of local standard of care but otherwise after consent was obtained (no clot at endovascular intervention in 7 of 98 [7.1%]). REVASCAT stipulated that the imaging study must be completed no more than 90 minutes but ideally within 60 minutes before groin puncture, and for patients who received intravenous tPA, an imaging study assessing vessel patency must be obtained at a minimum of 30 minutes after that start of intravenous r-tPA infusion (no clot at endovascular intervention in 5 of 103 [4.9%]). The REVASCAT strategy did not result in a decrease in the number who failed to have a clot present at the time of endovascular intervention compared with the other studies. The goal of intravenous r-tPA and of endovascular therapy is to recanalize the occluded vessel as soon as possible. After the initiation of intravenous r-tPA, some patients will experience successful recanalization, obviating the need to pursue follow-on endovascular therapy.²⁸ However, because recanalization occurs in only a minority of patients with large-vessel occlusion receiving intravenous r-tPA alone (eg, 37.3% in the ESCAPE trial), noninvasive intracranial vascular imaging should proceed without delay before or immediately after initiation of r-tPA to identify the majority of patients who will benefit from follow-on endovascular therapy and to expedite its performance. This approach was explicitly taken by investigators in the ESCAPE trial, helping them achieve a median CT–to–groin puncture time of only 51 minutes.

The ESCAPE, EXTEND-IA, and SWIFT PRIME trials were initially designed with the intent to select and enroll only patients with small regions of ischemic cores and the presence of salvageable brain tissue (SWIFT PRIME and EXTEND-IA) and/or adequate collateral flow (ESCAPE). In ESCAPE, nonenhanced CT and CTA (preferably multiphase) were used to select patients with a target occlusion, small infarct core (ASPECTS 6–10), and moderate to good collateral circulation (filling of ≥50% pial arterial circulation visualized on CTA). EXTEND-IA required demonstration of potentially salvageable brain tissue on perfusion CT (mismatch ratio of >1.2, absolute mismatch volume of >10 mL) and ischemic core <70 mL (relative cerebral blood flow <30% of normal). All images were processed on site with a specialized software package.²⁹ Penumbral tissue was defined as regions with time-to-maximum (Tmax) perfusion values >6 seconds that were not included in the ischemic core. SWIFT PRIME excluded patients with evidence of frank ischemia in greater than one third of the MCA territory or involving >100 mL of tissue. For the first 71 patients enrolled, an additional inclusion criterion was the presence of target mismatch defined as infarct core ≤50 mL (as assessed by specialized software¹⁹) and ischemic penumbra ≥15 mL with a mismatch ratio >1.8. After the enrollment of the first 71 patients, the investigators switched to the criterion to ASPECTS of ≥6 for sites that did not have CT perfusion capability. To date, subgroup analyses with the various imaging criteria have not been published. In these trials, the use of advanced imaging selection criteria had the potential advantage of increasing the likelihood of showing treatment benefit by enhancing the study population with patients most likely to respond to therapy. However, the inherent disadvantage of this study design is the possibility that patients who may have responded to therapy were excluded. In contrast, the MR RESCUE trial was designed specifically to validate imaging biomarkers as a selection tool for endovascular therapy. However, the trial was unable to demonstrate an overall benefit from endovascular therapy with first-generation devices or in the subgroup with a favorable penumbral pattern. None of the 5 stent retriever studies was designed to validate the utility of the advanced imaging selection criteria themselves in either the early or late time windows. Thus, the role of these techniques for patient selection requires further study.

The overwhelming majority of patients in the stent retriever trials had ICA or proximal MCA (M1) occlusion. The number of patients with isolated M2 lesions was small; ESCAPE, REVASCAT, and SWIFT PRIME excluded patients with isolated M2 occlusions, although small numbers of these patients were enrolled in these trials. The distinction of M1 from M2 can be difficult in some patients because of early branches of the M1 such as the anterior temporal branch. Inadequate numbers of patients with occlusion of other vessels, including M3 and anterior cerebral artery occlusions and those in the vertebrobasilar circulation, also were enrolled to allow assessment of clinical efficacy in these territories.

The usefulness of mechanical thrombectomy devices other than stent retrievers is not well established, either for technical efficacy or for clinical benefit. Most of the patients in MR CLEAN and ESCAPE and all of the patients in EXTEND-IA, SWIFT-PRIME, and REVASCAT who underwent an endovascular procedure were treated with a stent retriever (81.5% in MR CLEAN, 86.1% in ESCAPE). These trials were not designed to demonstrate the superiority of stent retrievers over other devices such as snares or suction aspiration systems. Therefore, the recommendation that stent retrievers are preferred over MERCI is unchanged from the previous guidelines based on the SWIFT and TREVO 2 (Trevo Versus Merci Retrievers for Thrombectomy Revascularisation of Large Vessel Occlusions in Acute Ischaemic Stroke) studies.^30,31 At the time the present guidelines were written, there were no published randomized, clinical trials demonstrating clinical benefit or comparing the relative effectiveness of other devices versus stent retrievers.

None of these studies specified requirements for the use of a proximal balloon guide catheter, large-bore distal-access catheter, or cervical guide catheter alone or in conjunction with stent retrievers. The concomitant use of distal-access suction catheters during stent retriever mechanical thrombectomy has been described in retrospective case series.^32–34 The advantages of the combined stent-aspiration technique include a flexible large-bore catheter in a triaxial technique, which provides stability for the stent-retriever; flow reversal to prevent distal embolization during stent retrieval of the thrombus; and the potential synergistic effect of both techniques of suction aspiration and stent retrieval used simultaneously.^32,34 Clinical experience has shown the combination of balloon guide catheters or distal-access/aspiration catheters and stent retrievers to provide rapid, effective, and safe recanalization.^35,36

All the stent retriever trials allowed the inclusion of patients with proximal cervical carotid stenosis, and all but 1 trial allowed the inclusion of patients with complete atherosclerotic cervical carotid occlusion (SWIFT PRIME). One difficulty with this exclusion is that differentiating complete cervical carotid occlusion from a distal ICA occlusion is often not possible on CTA or MRA.³⁷ The number of patients with cervical carotid occlusion or stenosis was not consistently reported but was substantial, ranging from 18.6% (REVASCAT) to 32.2% (MR CLEAN). Stenting of the underlying stenosis or occlusion was discouraged in the ESCAPE protocol. Thirty of the 75 patients with carotid stenosis or occlusion in the intervention arm were stented during the thrombectomy procedure in MR CLEAN. Nine of the 19 patients with carotid occlusion in REVASCATS were stented at the time of thrombectomy. The management of the underlying lesion was not reported in the other trials. Outcomes for the subgroup of patients with cervical carotid occlusion were reported in ESCAPE (OR, 8.7; 95% CI, 1.9–39.4) and MR CLEAN (adjusted OR, 1.43; 95% CI, 0.78–2.64). Although thrombectomy for patients with cervical ICA occlusion is clearly indicated by these data, the optimal management of the underlying stenosis is not clear. There are several potential advantages and disadvantages for angioplasty and stenting at the time of thrombectomy. Although immediate revascularization may reduce the risk of recurrent stroke, urgent stenting generally requires antiplatelet prophylaxis, which has been associated with intracranial hemorrhage in this setting. Carotid stenting and intracranial thrombectomy for the treatment of acute stroke resulting from tandem occlusions with aggressive antiplatelet therapy may be associated with a high incidence of intracranial hemorrhage.^38,39 In addition, there is some risk for thromboembolic stroke at the time of stenting. Further studies are indicated.

General anesthesia with intubation and conscious sedation are the 2 most frequently used anesthetic approaches for patients with an acute ischemic stroke receiving endovascular therapy.⁴⁰ No dedicated randomized, controlled, clinical trials have addressed this issue. The MR CLEAN investigators have reported that the outcomes of the 79 patients in the endovascular group who received general anesthesia were not different from the outcomes of the 267 nonendovascular control patients (adjusted OR, 1.09; 95% CI, 0.69–1.71.), whereas the outcomes for the 137 endovascular patients who did not receive general anesthesia were better than the outcomes for the 267 control patients (adjusted OR, 2.13; 95% CI, 1.46–3.11).⁴¹ Similar data showing worse outcomes in those undergoing general anesthesia compared with conscious sedation for endovascular were reported in a recent meta-analysis of 9 nonrandomized studies comprising 1956 patients (814 received general anesthesia, 1142 received conscious sedation), with the largest study having 1079 patients and the smallest study having 66 patients.⁴² In this meta-analysis, compared with conscious sedation, general anesthesia was linked to lower odds of a favorable functional outcome (OR, 0.43; 95% CI, 0.35–0.53; P<0.01), higher odds of mortality (OR, 2.59; 95% CI, 1.87–3.58; P<0.01), and more adverse respiratory events (OR, 2.09; 95% CI, 1.36–3.23; P<0.01). No significant differences in the rates of asymptomatic ICH, sICH, or other vascular complications were seen between the groups. Furthermore, mean time to groin puncture, mean procedure time, and mean time from symptom onset to revascularization were not significantly different between the 2 techniques. There was substantial heterogeneity (I²>50%) across the included studies for the outcomes of functional status (I²=55%), time to revascularization (I²=60%), time to groin puncture (I²=83%), and procedure time (I²=91%). In most of the included studies, patients who received general anesthesia were typically in worse clinical condition at baseline, as reflected by their comparatively higher NIHSS scores. Only 6 of the 9 studies included information on baseline NIHSS score. Adjusting for NIHSS score by the use of meta-regression for the odds of having good functional outcomes yielded an OR of 0.38, which was similar to the unadjusted estimate of 0.43; however, the 95% CI became statistically insignificant (0.12–1.22). Thus, even after adjustment for initial stroke severity, the possibility of selection bias cannot be completely excluded. Patients with more severe strokes or poorer baseline conditioning may have received general anesthesia or may have been intubated before the procedure because of an actual or expected inability to maintain airway patency. Moreover, it is possible that the lower recanalization rates observed with general anesthesia in some studies were attributable to greater numbers of more technically difficult vascular occlusions in those who received general anesthesia. On balance, published data broadly indicate that conscious sedation might be safer and more effective than general anesthesia in the setting of endovascular therapy for acute ischemic stroke. However, specific randomized, controlled trial data are warranted to definitively establish conscious sedation as the preferred anesthetic technique in patients receiving endovascular treatment for acute ischemic stroke. Clinical trials are ongoing (http://www.clinicaltrials.gov; NCT01872884, NCT02317237).

The AHA’s Level of Evidence grading algorithm requires high-quality evidence from >1 randomized, controlled trial for Level of Evidence A. In accordance with this algorithm and the results from the 5 recent studies with stent retrievers summarized above, we concluded that the data supported Class I, Level of Evidence A recommendations but only for a carefully defined group of patients (see Recommendation 2). Subsequent meta-analysis of patient-level data may allow these recommendations to be expanded.