See related article, p 3109

Preservation of the ischemic penumbra is the predominant mechanism of benefit from thrombectomy for large vessel occlusion stroke. Under guideline recommendations¹ endovascular therapy (EVT) is one of the most powerful treatments in medicine. As a result, expansion of thrombectomy to additional populations is the goal of multiple active trials. A common exclusion criterion for patients with large vessel occlusion stroke that do not undergo thrombectomy is an already large established infarct,² where reperfusion benefit is uncertain. In the SELECT trial (Optimizing Patient's Selection for Endovascular Treatment in Acute Ischemic Stroke), no benefit was seen in treating patients with an infarct volume >100 mL.³ The concern is 2-fold: first, futile recanalization, where the infarct is already complete and therefore the patient is beyond the opportunity for salvage. The second is more serious; reperfusion may cause harm by reperfusion injury, including hemorrhage. Tempering these concerns is the appreciation that if the large infarct population benefits from thrombectomy numbers of potential thrombectomy candidates may increase by as much as 35% to 40% and substantial disability in society prevented.⁴

On this backdrop, 4 active trials are randomizing patients to EVT versus best medical treatment (BMT) in the large infarct population. These trials are TESLA (REGISTRATION: URL: https://www.clinicaltrials.gov; Unique identifier: NCT03805308), IN EXTREMIS LASTE (REGISTRATION: URL: https://www.clinicaltrials.gov; Unique identifier: NCT03811769); SELECT 2 (REGISTRATION: URL: https://www.clinicaltrials.gov; Unique identifier: NCT03876457), and TENSION (REGISTRATION: URL: https://www.clinicaltrials.gov; Unique identifier: NCT03094715). A notable feature of these trial designs is that treatment selection is both pragmatic and inclusive. These studies represent a pendulum swinging away from the highly selective trials that helped establish the primacy of thrombectomy in both the early and late therapeutic window.

While we await large infarct core randomized trials to help guide our decision-making, Meyer et al⁵ have gathered data from EVT in patients with low Alberta Stroke Program Early CT Score (ASPECTS) in real clinical settings. They utilize the open label, prospective, multicenter German Stroke Registry, with 25 sites in Germany collecting data on consecutive patients undergoing thrombectomy. The comparison cohort is from 3 tertiary stroke centers where no EVT was offered, and patients received BMT only, including intravenous alteplase where indicated. ASPECTS and National Institutes of Health Stroke Scale differed (5 versus 4 and 18 versus 19 in the EVT group versus the BMT group, respectively). Given these differences and the difference in patient sources, the authors aimed to minimize the possibility of bias between the groups by using the statistical technique of propensity score matching.

Patients were included in their analysis after presenting with large vessel occlusion stroke and CT ASPECTS 0 to 5 (most patients were ASPECTS 4–5). Notably, median time from onset to imaging was ≈2 hours for both groups; these were patients in the early thrombectomy window. The authors apply a modified Rankin Scale (mRS) score dichotomized end point (mRS score ≤3, versus mRS score 5 and 6), with symptomatic intracerebral hemorrhage as their safety end point.

They find that outcomes were overall similar between groups, with 27.4% versus 25% (P=0.665) in the EVT group versus the BMT group respectively achieving mRS score 0 to 3 outcomes. Rates of very poor outcome (mRS score ≥5) were similar too (55.6% versus 58.1%, P=0.701). However, symptomatic intracerebral hemorrhage (16.1% versus 5.6%, P=0.008) and mortality (43.5% versus 28.9%, P=0.025) were both significantly higher in the EVT group after propensity score matching. For thrombectomy patients with incomplete modified Thrombolysis in Cerebral Infarction (mTICI) 2b and complete recanalization (mTICI 3), good functional outcome was observed in 28% and 35.7% (not statistically significant), respectively. For those with mTICI 3 reperfusion in <2 attempts, this trended toward significance (42.3%, P=0.074 compared with BMT). Interestingly, complete reperfusion trended toward protection against reperfusion hemorrhage; less cases were noted of symptomatic intracerebral hemorrhage (7.1%) in the subgroup of mTICI 3 compared with patients that did not receive complete reperfusion (26.2%; P=0.052).⁵

The authors are appropriately restrained in their conclusion: the absence of randomization and limited sample size means this study cannot be used to draw conclusions for decision-making in patients with stroke with low ASPECTS.⁵ However, the study highlights that equipoise regarding the best treatment for patients with large, established core certainly remains. Further, the risk of harm may limit the benefit of EVT.

How should these results inform our expectations about application of thrombectomy in the large infarct population? First, is this a representative population? The validity of ASPECTS may be debated in selecting patients in the early window, but this imaging approach is increasingly automated and likely to become increasingly reliable.⁶ Compared with DWI, for large, hemispheric infarcts human readers tend to underestimate CT infarct volume using ASPECTS,⁷ suggesting that the patients included in this study were indeed true large infarct stroke victims. Median time from onset to imaging was 131 minutes for the EVT group and 110 minutes for the BMT one; this study evaluates fast progressors,⁸ lacking adequate collateral support to sustain the penumbra. However, despite propensity matching the concern for selection bias is substantial. Consider the reported rates of intravenous alteplase administration. Why were tPA rates in the EVT arm not higher, closer to the 76.5% in the BMT arm and the 74% in patients in the randomized early window trials?⁹ Whether this reflects patient characteristics, operator or institutional bias or participation in direct to thrombectomy trials is unknown, as is the possible impact on results.

When treating large infarct patients, it is unrealistic to expect benefits of the magnitude seen in patients with higher ASPECTS; outcomes must be considered more broadly than merely functional independence. The authors utilize a modified dichotomized end point: broader good” outcomes (mRS score 0–3) compared with the extreme of severe disability (mRS score 5) or death. With restructured outcomes, the benefit of thrombectomy may be more visible. Can thrombectomy in this population, for example, reduce rates of hemicraniectomy¹⁰ and is that enough to recommend it? No hemicraniectomy data are available in this cohort the authors report on.⁵ Concerns for harm from thrombectomy in large infarct patients may be supported by the symptomatic intracerebral hemorrhage and mortality rates in this analysis. But results are more complex than that: EVT was associated with higher mortality than BMT but also shifted patients from mRS score 5 to lower mRS scores (a 10% reduction compared with BMT).⁵ If confirmed in larger trials, such findings will necessitate thoughtful discussion about implications of thrombectomy for this population at the level of a system of care.

Whichever outcome measures are applied, if benefit for thrombectomy in this vulnerable large infarct group exists, the implications are substantial. Prehospital, might this population be particularly sensitive to bypass protocols? They should be readily identifiable in the field, given the substantial clinical deficit that often accompanies these presentations. In hospital, variable rates of infarct progression between patients⁸ forms the basis of imaging selection for therapy. Will positive large infarct trials simplify imaging selection paradigms further, by reducing concerns about large established infarct cores?

Finally, the already firmly established importance of adequate and rapid reperfusion cannot be overstated. Results in those with mTICI 3 reperfusion compared with those patients undergoing BMT were the most indicative of benefit from reperfusion. The ability to predict reperfusion success may be an additional factor informing patient selection in large infarct populations. Improvements in device technology will also lead to more consistent complete reperfusion. Despite these expected advances, to anticipate that the pendulum will swing from current guidelines for imaging selection to broad thrombectomy access for all is too optimistic. Meyer and colleagues show that patient selection will need to continue to be refined if we are to identify those who manifest benefit or harm, and should serve to reinforce the equipoise that leads to enrollment of patients in the active large infarct thrombectomy trials. However, in this real-world registry data, there is a signal that a subgroup of large infarct patients benefit. We eagerly await the results of the pending randomized trials to tell us better who they are.

Disclosures Dr Simonsen reports grants from Novo Nordisk Foundation and Health Research Foundation of Central Denmark Region and is the Danish coordinator for the TENSION trial. Dr Leslie-Mazwi and Altschul have no relevant disclosures.

The opinions expressed in this article are not necessarily those of the editors or of the American Heart Association.

For Disclosures, see page 3120.

参见相关文章，第 3109 页

保留缺血半暗带是大血管闭塞性卒中血栓切除术获益的主要机制。根据指南建议^1，血管内治疗 (EVT) 是医学中最有效的治疗方法之一。因此，将血栓切除术扩大到更多人群是多项积极试验的目标。对于未接受血栓切除术的大血管闭塞性卒中患者，一个常见的排除标准是已经确定的大梗塞，²再灌注益处不确定。在 SELECT 试验（优化急性缺血性卒中血管内治疗的患者选择）中，未发现治疗梗死体积 >100 mL 的患者有任何益处。³担心有两个方面：首先，无效的再通，即梗塞已经完全，因此患者没有抢救的机会。第二种比较严重；再灌注可能会因再灌注损伤而造成伤害，包括出血。缓和这些担忧的是，如果大量梗塞人群受益于血栓切除术，潜在血栓切除术候选者的数量可能会增加多达 35% 至 40%，并且可以防止社会上的严重残疾。⁴

在此背景下，4 项正在进行的试验将患者随机分配到 EVT 与最佳药物治疗 (BMT) 的大梗死人群中。这些试验是 TESLA（注册：URL：https://www.clinicaltrials.gov；唯一标识符：NCT03805308）、IN EXTREMIS LASTE（注册：URL：https://www.clinicaltrials.gov；唯一标识符：NCT03811769）；选择 2（注册：URL：https://www.clinicaltrials.gov；唯一标识符：NCT03876457）和 TENSION（注册：URL：https://www.clinicaltrials.gov；唯一标识符：NCT03094715）。这些试验设计的一个显着特点是治疗选择既务实又包容。这些研究代表了一个钟摆远离高选择性试验，这些试验有助于在早期和晚期治疗窗口中确立血栓切除术的首要地位。

虽然我们正在等待大型梗死核心随机试验来帮助指导我们的决策，但 Meyer 等人⁵已经从真实临床环境中艾伯塔中风计划早期 CT 评分 (ASPECTS) 低的患者的 EVT 中收集了数据。他们利用开放标签、前瞻性、多中心的德国卒中登记处，在德国的 25 个站点收集连续接受血栓切除术患者的数据。比较队列来自 3 个未提供 EVT 的三级卒中中心，患者仅接受 BMT，包括静脉注射阿替普酶。ASPECTS 和美国国立卫生研究院卒中量表不同（EVT 组和 BMT 组分别为 5 对 4 和 18 对 19）。鉴于这些差异和患者来源的差异，作者旨在通过使用倾向评分匹配的统计技术来最大限度地减少组间偏差的可能性。

在出现大血管闭塞性卒中和 CT ASPECTS 0 至 5（大多数患者为 ASPECTS 4-5）后，患者被纳入他们的分析。值得注意的是，两组从发病到成像的中位时间约为 2 小时；这些是早期取栓窗口期的患者。作者应用改良的 Rankin 量表 (mRS) 评分二分终点（mRS 评分≤3，对比 mRS 评分 5 和 6），将有症状的脑出血作为其安全性终点。

他们发现两组之间的结果总体相似，EVT 组和 BMT 组分别有 27.4% 和 25% ( P = 0.665) 达到 mRS 评分 0 到 3 个结果。结果非常差（mRS 评分≥5）的比率也相似（55.6% 对 58.1%，P = 0.701）。然而，症状性脑出血（16.1% 对 5.6%，P = 0.008）和死亡率（43.5% 对 28.9%，P=0.025) 在倾向评分匹配后均显着高于 EVT 组。对于脑梗塞不完全改良溶栓 (mTICI) 2b 和完全再通 (mTICI 3) 的血栓切除术患者，分别在 28% 和 35.7%（无统计学意义）中观察到良好的功能结果。对于 mTICI 3 再灌注 <2 次尝试的患者，这趋于显着（42.3%，与 BMT 相比，P = 0.074）。有趣的是，完全再灌注倾向于防止再灌注出血；与未接受完全再灌注的患者相比，mTICI 3 亚组中出现症状性脑出血的病例较少 (7.1%)（26.2%；P = 0.052）。⁵

作者在他们的结论中受到了适当的约束：缺乏随机化和有限的样本量意味着该研究不能用于为低 ASPECTS 的卒中患者的决策得出结论。⁵然而，该研究强调，对于大的、已建立的核心患者的最佳治疗方案肯定仍然存在。此外，伤害风险可能会限制 EVT 的益处。

这些结果应该如何告知我们对在大量梗死人群中应用血栓切除术的期望？首先，这是一个有代表性的人群吗？在早期窗口选择患者时，ASPECTS 的有效性可能存在争议，但这种成像方法越来越自动化并且可能变得越来越可靠。⁶与 DWI 相比，对于大的半球梗塞，人类阅读者倾向于使用 ASPECTS 低估 CT 梗塞体积，⁷表明本研究中的患者确实是真正的大梗塞中风受害者。EVT 组从发病到成像的中位时间为 131 分钟，BMT 组为 110 分钟；这项研究评估了快速进展者，⁸缺乏足够的附属支持来维持半影。然而，尽管倾向匹配，但对选择偏差的担忧是巨大的。考虑报告的静脉阿替普酶给药率。为什么 EVT 组的 tPA 率没有更高，更接近 BMT 组的 76.5% 和随机早期窗口试验中患者的 74%？⁹这是否反映了患者特征、操作者或机构偏见或参与直接取栓试验尚不清楚，对结果的可能影响也是未知的。

在治疗大面积梗塞患者时，期望在 ASPECTS 较高的患者中获得如此大的益处是不现实的；必须更广泛地考虑结果，而不仅仅是职能独立。作者使用修改后的二分终点：与极端严重残疾（mRS 评分 5）或死亡相比，更广泛的“良好”结果（mRS 评分 0-3）。通过重组结果，血栓切除术的益处可能更加明显。例如，在该人群中进行血栓切除术能否降低偏侧颅骨切除术的发生率¹⁰并且足以推荐它？作者报告的这个队列中没有可用的偏侧颅骨切除术数据。⁵该分析中的症状性脑出血和死亡率可能支持对大面积梗死患者血栓切除术危害的担忧。但结果比这更复杂：EVT 与比 BMT 更高的死亡率相关，但也将患者从 mRS 评分 5 转移到较低的 mRS 评分（与 BMT 相比降低 10%）。⁵如果在更大规模的试验中得到证实，这些发现将需要在护理系统层面就血栓切除术对该人群的影响进行深思熟虑的讨论。

无论采用哪种结局指标，如果在这个易受伤害的大梗塞组中进行血栓切除术存在益处，则其影响是巨大的。院前，这个人群可能对旁路协议特别敏感吗？鉴于通常伴随这些表现的大量临床缺陷，它们在现场应该很容易识别。在医院中，患者⁸之间梗塞进展的可变速率构成了治疗影像学选择的基础。通过减少对已建立的大梗塞核心的担忧，阳性大梗塞试验是否会进一步简化成像选择范式？

最后，充分和快速再灌注的已经确立的重要性怎么强调都不为过。与接受 BMT 的患者相比，mTICI 3 再灌注患者的结果最能表明再灌注获益。预测再灌注成功的能力可能是告知大量梗死人群中患者选择的另一个因素。设备技术的改进也将导致更一致的完全再灌注。尽管取得了这些预期的进展，但预计钟摆将从当前的影像选择指南转向适用于所有人的广泛血栓切除术，还是过于乐观了。Meyer 及其同事表明，如果我们要确定那些表现出有益或有害的患者，就需要继续完善患者的选择，并且应该有助于加强导致患者参加活动性大梗塞血栓切除术试验的平衡。然而，在这个真实世界的注册数据中，有一个信号表明大面积梗死患者的亚组受益。我们热切地等待待定随机试验的结果，以更好地告诉我们他们是谁。

披露Simonsen 博士报告了诺和诺德基金会和丹麦中部地区健康研究基金会的资助，并且是 TENSION 试验的丹麦协调员。Leslie-Mazwi 博士和 Altschul 没有相关披露。

本文中表达的观点不一定是编辑或美国心脏协会的观点。

有关披露，请参阅第 3120 页。