The Quest to Prove Endovascular Stroke Therapy: Searching for the “Sweet Spot” in Patient Selection

Endovascular interventions targeted at improving outcomes in patients with acute ischemic stroke have been studied for more than 2 decades. During this period, intravenous (IV) recombinant tissue plasminogen activator (tPA) was established as an effective treatment for stroke, providing the impetus to build organized stroke systems of care and revolutionizing the treatment of stroke worldwide. Endovascular approaches to achieve revascularization in cerebral arteries have been proposed to have a number of advantages over IV tPA, including greater efficacy at reopening occluded vessels and fewer systemic bleeding risks. Although intra-arterial administration of thrombolytic agents has been shown to improve outcomes compared with placebo, proof of superior outcomes relative to IV tPA remains elusive.¹ In this issue of Mayo Clinic Proceedings, Singh et al² present a detailed meta-analysis of studies comparing endovascular therapy with IV tPA. Such an analysis is challenging, ie, pooling of heterogeneous studies with different inclusion criteria and different treatment approaches has the potential to create more confusion than clarity. The study design used by Singh et al included measures to work around these problems.

In their analysis, Singh et al systematically consolidated available evidence in the literature. Their literature search strategy initially identified 1252 articles, but their study inclusion criteria resulted in a total of only 1197 patients, 711 of whom were treated with endovascular therapy, reported in 5 randomized clinical trials to date. Their summary of results documented a resounding failure of endovascular interventions to provide a benefit superior to that achieved with IV tPA. In a paradoxically upbeat subgroup analysis of 271 patients with severe strokes (National Institutes of Health [NIH] Stroke Scale score ≥20) based on data in 3 studies, the authors suggested a potential benefit of endovascular therapy (risk ratio, 1.40; 95% CI, 0.86–2.28) for good outcome (modified Rankin Scale, 0–2 at day 90) when compared with an IV tPA–treated comparator group. Although greater efficacy of endovascular therapy in severe stroke, which is often due to major large-vessel occlusion unlikely to recanalize with IV tPA, is biologically plausible, the small sample size and well-recognized limitations of subgroup analysis limit the level of confidence in this finding. Furthermore, a relative benefit of endovascular therapy vs IV tPA may still be hampered by absolute poor patient outcomes in such a severely affected population.

Perhaps the most insightful perspective offered by the authors is their commentary on the effect of time in critically evaluating treatment interventions in acute stroke. Ischemic brain injury is a product of both the severity and the duration of decreased cerebral blood flow, often translated in current trials into imaging of brain tissue as a surrogate for severity and time from symptom onset to treatment for duration. These 2 key variables of severity and duration have motivated tria-lists to seek a “sweet spot” for selection of patients most likely to benefit from endovascular therapy, often using a simple paradigm based on findings on noncontrast computed tomography (CT) of the head and patient- or surrogate-reported time from symptom onset.

Not surprisingly, there is a price to be paid for simplicity. Although endovascular therapy can be implemented more swiftly if advanced imaging is avoided, doing so may also result in treating patients unlikely to benefit, either because irreversible, extensive brain injury has already occurred or the location and type of vessel occlusion indicates that IV tPA alone is likely to be similarly successful at achieving revascularization. The MR RESCUE (Mechanical Retrieval and Recanalization of Stroke Clots Using Embolectomy) trial utilized an advanced imaging model to categorize patients as having penumbral or nonpenumbral brain ischemia within an 8-hour treatment window that de-emphasized the need for ultraearly therapy.³ The MR RESCUE study was ultimately proclaimed a double failure, of both advanced imaging and endovascular therapy. Dismissing the concept of advanced imaging on the basis of the results of a single 120-patient study borders on reckless abandon, especially given the numerous advances in acute stroke imaging and device technology that have occurred since the MR RESCUE trial. Indeed, the relatively liberal definition of penumbral patterns and inclusion of patients with large ischemic core volumes on imaging may have accounted for the negative results of the MR RESCUE study.⁴ Exact definitions of penumbra, including specific perfusion parameters, thresholds, and volumes, have caused confusion in the acute stroke imaging field, yet simple patterns such as the “target mismatch” and “malignant” profiles employed in the recently completed DEFUSE 2 (Diffusion and Perfusion Imaging Evaluation for Understanding Stroke Evolution study 2) trial hold promise.⁵

A factor to which relatively little attention has been paid is the state of the collateral circulation in an individual patient.⁶ Recent studies have found that poor collateral circulation, even during the earliest epochs of the thrombolysis window, may be present in many patients and heralds poor outcome. If the ultimate goal is to improve patient outcomes after a particular treatment, then the ideal candidate for endovascular therapy must not be at the extreme range of known predictive outcome variables after stroke. For instance, age and NIH Stroke Scale severity at the time of the initial neurologic deficit are potent predictors of outcome, and these factors may limit the beneficial effect of any therapy. Focusing the next clinical trial of endovascular therapy on patients with NIH Stroke Scale scores of 20 or less may be predicated on an advantage over IV tPA as a comparator, yet many patients in such a trial will do poorly irrespective of treatment assignment. Furthermore, such an approach does not offer treatment to new populations and, in fact, competes with IV tPA for a minority of all stroke patients. Alternatively, treating very mild strokes may result in excellent outcomes, but such patients may have done well even without treatment.

Age and several other variables (including baseline neurologic status) similarly share a sweet spot or middle range at which the potential for benefit from any treatment is maximal. Because of these factors, a potential shift of individual patient selection for treatment based not as extensively on severity of neurologic status at presentation but instead on projected clinical outcomes at day 90 may optimize benefit from a new endovascular therapy. In developing such a selection strategy, we must remember that the baseline status of a patient with acute stroke still sways the expected course days and months later. However, the basis of this relationship is complex. Patients with poor collateral circulation tend to present early with severe neurologic deficits and rapid irreversible ischemic injury, whereas those with robust collaterals commonly present later with mild or fluctuating deficits. Collateral status may therefore also be a major determinant of baseline stroke severity and time to presentation. For the purposes of selecting patients for endovascular therapy, assessment of collateral flow may help avoid treating patients in whom poor outcomes are inevitable and, at the other extreme, those who will recover or compensate to a great extent with or without treatment. Evaluating collateral status in an endovascular study is straightforward because angiography of adjacent blood flow routes can be rapidly obtained. Ideally, a reliable noninvasive technique to establish markers of collateral flow on multimodal CT and magnetic resonance imaging studies with either angiography or perfusion images would facilitate triage before patients ever reach the angiography suite.

Keeping it simple has been the dogma of trialists striving to develop a practical treatment paradigm for stroke akin to the established model for IV tPA. Rapid triage with minimal imaging criteria beyond CT to rule out hemorrhage or major early signs of infarction has made the delivery of IV thrombolysis relatively simple. Such is not the case with endovascular therapies. The recently completed IMS (Interventional Management of Stroke) III trial followed this model, emphasizing the importance of time by focusing only on patients seen early after stroke, with minimal use of imaging.⁷ However, while only noncontrast head CT was required for the IMS III study, a subset of patients did have vascular imaging, typically CT angiography, before intervention. Although the IMS III trial did not find a benefit of endovascular therapy, a prospective evaluation of available angiographic collateral data throughout the trial revealed overt benefit of endovascular therapy to achieve superior reperfusion rates and good clinical outcomes in those with robust collaterals.⁸ Further analysis comparing outcomes between endovascular and IV tPA groups based on site and extent of vessel occlusion is clearly warranted.

Designing a simple endovascular stroke trial to prove superiority over IV tPA is not an easy task. Intravenous thrombolysis is relatively simple with respect to resources, cost, paucity of contraindications, and basic imaging requirements. Few stroke centers have a track record of implementing endovascular therapy with comparable speed. Rather than focusing on trying to change the paradigm for those patients already being treated with IV tPA, endovascular therapy should be developed to expand the number of patients we can treat by properly selecting those in the therapeutic sweet spot. Although imaging beyond noncontrast head CT will take extra time, even sophisticated CT and magnetic resonance imaging protocols can be rapidly processed and results returned to the bedside to help in patient selection. Striking a balance between added imaging information and simplicity of execution has already been reported in the DEFUSE 2 trial, in which practical imaging patterns successfully predicted ultimate clinical outcomes, the holy grail of endovascular therapies.⁵ The stark depiction offered in the meta-analysis by Singh et al² should not reinforce nihilism about endovascular stroke therapy but rather encourage us to dig deeper into the complex pathophysiology of stroke and develop efficient imaging protocols that will help identify patients in the therapeutic sweet spot when planning the next round of trials.