Improving reperfusion therapy for acute ischaemic stroke


Jeffrey L. Saver, Stroke Center and Department of Neurology, David Geffen School of Medicine at the University of California, Los Angeles, CA, USA.
Tel.: +1 310 794 6379; fax: +1 310 267 2063.


Summary. Background: The first generation of clinical reperfusion treatment, intravenous (IV) fibrinolysis with tissue plasminogen activator (tPA), was a transformative breakthrough in stroke care, but is far from ideal. Objectives: To survey emerging strategies to increase the efficacy and safety of cerebral reperfusion therapy. Methods: Narrative review. Results and Conclusions: Innovative IV pharmacologic reperfusion strategies include: extending IV tPA use to patients with mild deficits; developing novel fibrinolytic agents (tenecteplase, desmetolplase, plasmin); using ultrasound to enhance enzymatic fibrinolysis; combination clot lysis therapies (fibrinolytics with GPIIb/IIIa agents or direct thrombin inhibitors); co-administration of MMP-9 inhibitors to deter haemorrhagic transformation; and prehospital neuroprotection to support threatened tissues until reperfusion. Endovascular recanalisation strategies are rapidly evolving, and include intra-arterial fibrinolysis, mechanical clot retrieval, suction thrombectomy, and primary stenting. Combined approaches appear especially promising, using IV fibrinolysis to rapidly initiate reperfusion, mechanical endovascular treatment to debulk large, proximal thrombi, and intra-arterial (IA) fibrinolysis to clear residual distal thrombus elements and emboli.

Reperfusion therapy for acute ischaemic stroke is greatly in need of improvement

It is true that cerebral reperfusion therapy has been a tremendous success in its first decade and a half of application, since the publication of the two pivotal NINDS-TPA trials in 1995 [1]. Stroke is the leading combined cause of death and disability worldwide [2]. Each year, more than 15 million people suffer a stroke. IV fibrinolysis is now an accepted staple of acute stroke care throughout the world. Treatment within the first 3 h of onset improves outcome in c. 290 of every 1000 patients treated, and treatment between 3 and 4.5 h improves outcome in c. 140 of every 1000 patients treated [3,4]. Across the world, regional systems of acute care are being built out to route patients preferentially to recanalisation-therapy capable hospitals, increasing the frequency of treatment delivery [5–8]. In countries and regions with more advanced networks, IV fibrinolytic use within the first 3 h of onset has increased from 1–2% to 5–20% of all ischaemic stroke patients [9–11]. Endovascular recanalisation techniques that achieve recanalisation by direct delivery of fibrinolytic to target thrombus, mechanical clot retrieval, clot aspiration, and vascular lesion stenting are now also approved in many countries and complement IV pharmacotherapy.

However, the cerebral reperfusion therapy of 2011 has many limitations. Recanalisation efficacy is far from optimal. IV fibrinolysis achieves partial or better recanalisation in 20–40% of patients [12,13]. Endovascular techniques achieve partial or better recanalisation in 50–90% and complete recanalisation in only 20–50%. Haemorrhagic transformation is too frequent a complication. Symptomatic haemorrhages occur in 4–7% of patients treated with IV fibrinolysis and 8–12% of patients undergoing endovascular recanalisation therapy [1,13,24,25,52,85,87,88]. Too few patients are treated, with many arriving at centres incapable of rapid and safe therapy administration.

Also, reperfusion must occur early to be beneficial. Every minute that goes by without reperfusion, more tissue in the ischaemic zone progresses from ailing but salvageable penumbral tissue to irreversibly infarcted core. In the first 3 h after a vascular occlusion, virtually every patient still harbours at least some salvageable penumbral tissue. Thereafter, a steadily increasing proportion of patients will have fully completed their initial infarct. By 12–24 h, virtually no patient still has tissue at risk that has not already progressed to infarction. Techniques to identify the subset of late-arriving patients who still harbour salvageable tissues and will benefit from delayed recanalisation remain incompletely validated.

Fortunately, several strategies to improve cerebral reperfusion therapy are progressing in preclinical studies and human clinical trials. This review will highlight several advances that presage another coming paradigm shift, from an era of moderately effective to an era of highly effective reperfusion treatment for acute ischaemic stroke.

Implementing standard IV fibrinolysis more effectively

Systematic studies have shown that it takes an average of 17 years for new knowledge generated by randomised controlled trials to be incorporated into practice, and even then the application is highly uneven [14]. IV fibrinolysis for acute cerebral ischaemia, which will reach the 17-year translation milestone in 2012, well illustrates the challenges to rapid dissemination of a novel, efficacious therapy. When thrombolytic stroke therapy was introduced in the 1990s, it was a disruptive medical innovation that could not easily be incorporated into existing patterns of care. Neurologists resisted the changes in practice that lytic therapy required, including the need to respond emergently to the Emergency Department [15]. Radiologists resisted reading CT or MR scans emergently, rather than the following morning during routine reading sessions [16]. Emergency Physicians resisted having to give a therapy with risks as well as benefits, for a complex brain disease, if they were not supported by neurologist, radiologist, and institutional backing [17]. Delivering fibrinolytic therapy effectively required multiple system changes in acute stroke, including: cultural and generational change among neurologists to develop a critical mass of stroke specialists schooled in emergent response [18,19]; cultural and generational change among emergency physicians to develop a critical mass of emergency specialists ready to participate as vital members of stroke response teams [20]; hospital-level commitments to create stroke units and stroke centres [21]; development of regional and national certifying authorities to designate and regulate stroke centres [22]; changes in reimbursement to defray the added up front costs of delivering fibrinolytic therapy (more than recouped at the system level by reduced long-term nursing home costs from implementation of an efficacious treatment) [23]; supportive data confirming benefit from additional controlled trials and large-scale practice registries [24–27]; and studies indicating safety and benefit across race-ethnic groups with distinctive stroke aetiologies and haemorrhage propensities (including Asians and Blacks) [28–30].

A major challenge in the coming decade is to complete the build out worldwide of regional acute stroke systems of care. In a regional system, prehospital Emergency Medical Services providers preferentially route most acute stroke patients directly to Primary Stroke Centres capable of delivering IV fibrinolytic therapy reliably and safely, and select acute stroke patients directly or by interfacility transfer to Comprehensive Stroke Centres capable of endovascular recanalisation interventions, more sophisticated, multimodal imaging, and neurocritical care [31]. These systems to date have achieved wide, but uneven distribution. In the United States, at the end of 2010, 53% of the population lived in a state or a county with a regional acute stroke care system, but 47% did not [8]. In Europe, penetration of regional acute stroke systems varies markedly across countries. In much of Asia, individual hospitals are well-organised, but rationalised ambulance routing is fragmentary. In low and middle income nations of the world, access to recanalisation therapy is limited [5]. Fortunately, with increasing recognition that acute stroke systems pay for themselves by reducing long-term health system costs associated with post stroke disability, more and more national healthcare systems are devoting appropriate resources to build out of regional systems of care. The demonstration that telemedicine can cost-effectively extend imaging interpretation and bedside stroke specialist expertise to remote sites is accelerating the geographic development of stroke-ready hospitals [32].

Within established stroke centres, further improvements in outcome from standard IV fibrinolysis can be achieved by systematic efforts to reduce the time interval from patient arrival to start of drug infusion – the door to needle (DTN) time. The therapeutic benefit of tPA is greatest when given early after ischaemic stroke onset and declines exponentially with time, to no significant benefit after 4.5 h [24,33]. With every 10 min delay in the start of thrombolytic infusion within the 1–4.5 h treatment time period, one fewer patient out of 100 patients has an improved disability outcome [34]. Because of the importance of rapid treatment, treatment guidelines recommend that hospitals complete the clinical and imaging evaluation of acute ischaemic stroke patients and initiate IV tPA therapy within 60 min of patient arrival in those without contraindications [35,36]. However, this target is more often missed than achieved. In the United States, among over 25 500 tPA treated patients in the national Get With The Guidelines – Stroke registry, only a little over one quarter had DTN times within 60 min and the overall median DTN time was 78 min [27]. In Europe, among nearly 24 000 tPA patients in the SITS-ISTR registry, fewer than half had DTN times under 60 min, and median DTN time was 65 min for those treated within 3 h of onset and 77 min for those treated between 3 and 4.5 h [37]. To improve DTN times in the United States, the American Heart Association/American Stroke Association in 2010 launched the Target: Stroke campaign, disseminating to stroke centre hospitals key best practice strategies that are associated with achieving faster DTN times in acute ischaemic stroke. Among the 10 key strategies chosen by Target: Stroke are: Emergency Medical Service prearrival-notification, activating the stroke team with a single call, rapid acquisition and interpretation of brain imaging, use of specific protocols and tools, pre-mixing tPA, a team-based approach, and rapid data feedback.

With build out of regional stroke systems of care and acceleration of in hospital treatment delivery processes, it should be possible to deliver standard IV fibrinolytic therapy effectively to 8–15% of all patients with ischaemic stroke in any geographic region, including 30–50% of all patients with moderate to severe ischaemic strokes [38].

Extending standard intravenous fibrinolysis to milder stroke patients

It is a little appreciated, but well-demonstrated, fact that the majority of ischaemic strokes are mild. In one population-based sample, 67% of all ischaemic stroke patients present with an NIH Stroke Scale score of 5 or less, one common definition of mild deficits [39]. Among consecutive, hospital-presenting patients, the median NIH Stroke Scale deficit severity score among consecutive, unselected ischaemic stroke patients is 5–6 [40,41].

Mild or rapidly improving initial stroke deficits are a frequent reason for non-use of tPA treatment. Patients with nondisabling deficits at presentation were excluded from the pivotal trials because mild impairments were felt insufficient to justify exposure to the risk of haemorrhagic side effects. However, multiple studies have now shown that 25–40% of patients in whom IV tPA is withheld due to mild initial deficits have poor final functional outcomes [42–44]. Frequently, these patients have flow through collateral routes that is adequate to support the neural parenchyma during the first few hours after onset, but fails over the next day or two. A persisting large artery occlusion is a strong predictor of delayed deterioration among mild deficit patients [44]. When these patients deteriorate beyond the 3–4.5 h window, the opportunity to use thrombolytic therapy has been lost.

The risks of thrombolytic therapy in patients with mild or rapidly improving deficits are likely substantially lower than in moderate to severe deficit patients. Haemorrhagic transformation risk is directly related to the volume of irreversibly infarcted tissue present at the time of fibrinolytic exposure [45,46]. Since established infarct cores are quite small in patients with mild stroke deficits, haemorrhagic risk should also be much reduced.

However, the degree to which thrombolytic therapy may improve outcome in mild stroke patients is uncertain. Since many of these patients do well without recanalisation treatment, the scope for improving outcomes by intervention is constrained. Clinical trials are now being designed and launched to formally test the extension of thrombolytic therapy to the mild deficit population. In the United States, the Potential of rtPA for Ischaemic Strokes with Mild Symptoms (PRISMS) Trial proposes to randomise 1500 patients with nondisabling deficits within 4.5 h of onset to IV tPA or placebo. Population-based and cohort analyses suggest that expanding the indications for thrombolytic treatment to include patients with initial nondisabling deficits could double the number of patients treated [39,47,48]. Up to 20–30% of all ischaemic stroke patients in a region and would be eligible for thrombolytic therapy, with late presentation times being the only remaining frequent barrier to therapy.

Improving fibrinolytic therapy

IV tPA is far from an ideal acute reperfusion therapy for cerebral ischaemia. The recanalisation efficiency of IV tPA in acute ischaemic stroke is low. Early, sustained recanalisation is achieved in 20–45% of patients. (Initial recanalisation is achieved in 25–55% of treated patients, among whom early reocclusion occurs in up to one-third.) [49] tPA has several adverse effects. Cerebral haemorrhage temporally associated with early deficit worsening occurs in 6–7% of patients, and alters final outcome in 2–3% [26,50]. Angioedema occurs in 1% of patients. In animal models of cerebral ischaemia, the tPA molecule itself is neurotoxic [51].

Pharmacologic strategies to enhance recanalisation efficacy and minimise size effects include (i) new fibrinolytic molecular entities with more favourable effects than tPA, (ii) ultrasound enhancement of tPA fibrinolysis, and (iii) co-administration of various classes of antiplatelet and anticoagulant agents with tPA.

Novel fibrinolytic agents

Since the approval of tPA for acute ischaemic stroke, serious human clinical trial development programs have been pursued for at least seven novel fibrinolytic agents for a stroke indication. Four have been terminated. Alfimeprase, V10153, and microplasmin all entered early human trials but failed to demonstrate sufficient advantages over tPA to progress to pivotal studies. Pro-urokinase was found beneficial in a small phase 3 trial of IA delivery, but further development was curtailed by the sponsor [52]. Agents continuing in currently active programs include tenecteplase, desmoteplase, and plasmin.

Tenecteplase (TNK) is a genetically modified form of tPA that has 14-fold greater fibrin specificity, longer half-life, and 80-fold greater resistance to inhibition by plasminogen activator inhibitor type 1 (PAI-1) [53]. The long lifetime of TNK allows the use of a single-bolus administration. High fibrin specificity should confer the ability to induce faster and more complete clot lysis, with less bleeding complications. TNK administration has been demonstrated to avoid the systemic plasminogen activation and plasmin generation commonly seen after tPA therapy. Further, the lack of a procoagulant effect exhibited by TNK may reduce early re-occlusion. In comparative trials in myocardial infarction patients, TNK showed equivalent efficacy to tPA, a similar rate of intracranial haemorrhage (ICH), fewer non-cerebral bleeding complications, and less need for blood transfusion [53].

Tenecteplase is the only fibrinolytic agent whose development program has been pursued through direct, head to head testing against tPA in under 3 h patients. In pilot dose-escalation study of 75 patients, IV TNK < 3 h showed no haemorrhagic safety concerns at all three dose tiers studied, 01, 0.2, and 0.4 mg kg−1 [54]. However, in a subsequent adaptive sequential design, phase 2B trial enrolling 112 patients, the 0.4 mg kg−1 dose was associated with a high rate of symptomatic haemorrhage and discarded. The trial was prematurely stopped before the 0.1 mg kg−1 or the 0.25 mg kg−1 could be distinguished as the most propitious dose to advance to a phase 3 trial, due to slow enrolment [55]. An on-going Australasian phase 2B study, Tenecteplase in Stroke (TIS) Trial, is comparing tenecteplase at 0.1 and 0.25 mg kg−1 dose tiers with IV tPA, using reperfusion at 24 h as a technical success endpoint.

Desmetoplase is one of four distinct proteases found in the saliva of the blood-feeding vampire bat Desmodus rotundus, collectively referred to as D. rotundus salivary plasminogen activators (DSPAs). Desmetoplase is the α-1 variant among the DSPAs and exhibits more than 72% amino acid sequence identity with human tPA. Unlike human tPA, DSPA α-1 exists as a single-chain molecule and its catalytic activity is exquisitely dependent on the presence of fibrin as cofactor. In models of arterial thrombosis, DSPA α-1 induces faster and more sustained recanalisation than tPA and produces less antiplasmin consumption and fibrinogenolysis. Moreover, unlike tPA, DSPA α-1 does not enhance NMDA-mediated neurodegeneration [56]. Desmoteplase showed promise in two phase 2 ischaemic stroke trials enrolling patients 3–9 h after onset when a magnetic resonance imaging diffusion–perfusion mismatch pattern was present. An initial phase 3 trial was nonpositive, with CT mismatch-selected patients not benefitting while MR mismatch-selected patients trended beneficial [57]. Two further phase 3 trials are currently underway, treating patients in the 3–9 h window if evidence of occlusion in a proximal cerebral artery is evident on CT or MR angiography.

Human Tissue Urokinase type Plasminogen Activator (HTUPA): HTUPA is a genetically engineered hybrid molecule in which the urokinase kringle region has been inserted before the double-kringle region of tPA. In preclinical models, this hybrid molecule at low doses induces prompt, clot selective thrombolysis and sustained patency with reduced reocclusion [58]. In a pilot, dose escalation trial in an Asian population, among 32 patients treated at lower dose tiers up to 5 h after onset, symptomatic haemorrhages were observed in 1/32 (3.1%) and asymptomatic haemorrhages in 6/32 (18.8%) [59]. At 3 months, 34% of patients in the low dose tier achieved a final normal or near normal neurologic outcome (NIHSS 0–1).

Plasmin is being developed as a therapeutic fibrinolytic. Standard plasminogen activating drugs depend upon the local availability of plasminogen to generate active, fibrin-digesting, plasmin. Their efficacy is constrained by the availability of endogenous plasmin in the thrombus region. This limitation is overcome when exogenous plasmin is administered, and can act directly upon fibrin. Because human plasmin is rapidly inactivated by circulating antiplasmin, plasmin is not suitable for use as an IV treatment but is potentially very useful as a local, IA applied therapeutic agent [60]. In animal models, in contrast to plasminogen activators, which provoke some bleeding at any effective thrombolytic dose, plasmin is tolerated without bleeding at several-fold higher amounts than needed for thrombolysis [61]. An early phase, dose escalation, human clinical trial is now underway, enrolling 75 patients with middle cerebral artery occlusions within 8.5 h of onset, testing IA delivery of plasmin at 20, 40, and 60 mg dose tiers.


Experimental and clinical studies have consistently demonstrated the capability of ultrasound (US) to enhance enzymatic thrombolysis [62]. US application increases the transport of tPA into the thrombus, promotes the opening and cleaving of the fibrin polymers, and improves the binding affinity of tPA to fibrin. While low frequency ultrasound in tandem with tPA has been found to increase the risk of brain haemorrhage [63], high frequency ultrasound of the type used in standard diagnostic studies has appeared safe and potentially beneficial. CLOTBUST, a phase 2 multicentre randomised trial, demonstrated that 2-h continuous application of 2 MHz transcranial Doppler (TCD) ultrasound in tandem with tPA is safe and may improve outcome [64]. Among 126 patients randomised to tPA plus 2-h TCD monitoring (target group) or tPA alone (control group), symptomatic ICH occurred in 4.8% of target and 4.8% of control patients. Complete recanalisation or dramatic clinical recovery at 2 h after tPA bolus was observed in 49% of target and 29% of control patients (P = 0.02). Moreover, trends towards better clinical outcomes at 24 h and long-term were noted in sonothrombolysis patients.

Using standard diagnostic TCD ultrasound to enhance thrombolysis has a substantial practical drawback – the pulse wave Doppler technique requires the immediate availability of a skilled sonographer to position the ultrasound window over the target clot. A pivotal phase 3 trial, CLOTBUSTER, will instead test a novel, continuous wave Doppler device that can be placed by any health professional. CLOTBUSTER is planned to enrol 900 patients randomised to systemic tPA alone or combined ultrasound and tPA.

The clot disrupting effects of ultrasound energy can be potentiated by the addition of gaseous microsphere ultrasound contrast agents, which resonate, expand, oscillate, and detonate near and within the thrombus when subjected to externally applied ultrasound. The phase 2, TUCSON trial investigated perfluten microspheres co-administered with tPA and TCD and found symptomatic haemorrhages in 0/12 patient in the low dose tier and 3/11 in the moderate dose tier [65].

Combination pharmacotherapies to enhance intravenous fibrinolysis

A variety of combination pharmacotherapy strategies to improve IV fibrinolysis are under active investigation.

Combining fibrinolytics with additional agents active in clot lysis and deterrence of clot formation may yield higher rates of arterial recanalisation, lower rates of re-occlusion, reductions in the dose of fibrinolytic agent required, less no-reflow phenomenon in the microcirculation, and reduced frequency of haemorrhage transformation. Several studies have explored combining fibrinolytics with GPIIb/IIIa platelet disaggregating agents or with direct thrombin inhibitors.

Fibrinolytics plus tirofiban

In a series of studies, the Dusseldorf group treated up to 37 patients within 3 h of onset with reduced doses of IV tPA (typically 20 mg) and a 24 h infusion of tirofiban. Combined therapy resulted in a high rate (68%) of MCA recanalisation on MR angiography, greater salvage of perfusion MR defined tissue at risk, and better clinical outcome than standard IV tPA [66,67]. Low rates of symptomatic ICH were observed.

Fibrinolytics plus abciximab

A pilot study in 27 patients found combining abciximab with low-dose tPA (0.45 mg kg−1) appeared safe and resulted in higher rates of MCA recanalisation compared with full-dose tPA alone [68]. In a 34 patient dose-ranging trial, the combination of reteplase and abciximab, appeared safe and efficacious in achieving recanalisation in 3–24 h post onset window [69].

Fibrinolytics plus eptifibatide

In the NIH CLEAR phase 2, dose escalation, good safety was noted among 94 patients randomised to standard dose IV tPA or eptifibatide plus lower dose IV tPA at two dose tiers (0.3 and 0.45 mg kg−1) [70]. A follow-up phase 2 trial is exploring the safety of a higher combination dose tier (with 0.6 mg kg−1).

Fibrinolytics plus argatroban

The NIH-sponsored Argatroban tPA Stroke Study (ARTTS) is enrolling 65 patients in a prospective, single arm trial of combined standard dose IV tPA plus argatroban infused for 48 h adjusted to a target partial thromboplastin time of 1.75 times baseline. Among the first 20 patients enrolled, intracerebral haemorrhage rates were low [71]. Partial or complete recanalisaton at 2 h was achieved in 70% of patients. Complete recanalisation at 2 h trended higher in the combined argatroban plus tPA group than in historical controls (35% vs. 13%).

Neuroprotective agents

Neuroprotective agents block the molecular elaboration of cell injury in hypoxic–ischaemic tissues [72]. Often safe in haemorrhagic as well as ischaemic stroke (unlike fibrinolytics, which are absolutely contraindicated in acute haemorrhage), many neuroprotective agents can be given prior to brain imaging, including in the prehospital setting. By stabilising threatened brain tissue, early neuroprotective therapy may increase the volume of salvageable tissue that is still present at the time that reperfusion therapy can be started, after hospital arrival and initial brain imaging. Magnesium sulphate is reliably neuroprotective in animal models and neither inhibits or potentiates tPA activity [73]. In a pilot trial, paramedic initiation of magnesium sulphate neuroprotective therapy in the field was shown to be feasible and appeared safe in 20 stroke patients [74]. In the NIH-funded FAST-MAG Phase 3 randomised trial, paramedics are initiating magnesium sulphate or placebo in 1700 patients, within 2 h of stroke onset. Among the first 1208 patients, the median time from last known well to start of study agent was 46 min. Among patients with a final diagnosis of cerebral infarction (rather than intracerebral haemorrhage or TIA), 37% received IV tPA on hospital arrival after their field initiation of neuroprotective study agent, ensuring the trial will have substantial statistical power to explore whether hyperacute neuroprotection potentiates the benefits of subsequent reperfusion therapy.

Cerebral ischaemia damages the cerebral vessels as well as the neuronal parenchyma, disrupting vascular integrity and predisposing to intracerebral haemorrhage. Fibrinolytic agents exacerbate this haemorrhagic risk. Administering agents that are vasoprotective along with reperfusion interventions may reduce haemorrhagic transformation rates and improve the benefit-risk ratio, and increase the permissible time window for reperfusion therapy. The greatest degree of preclinical investigation has focused on modulators of matrix metalloproteinase-9 (MMP-9), a member of the matrix metalloproteinases family that normally remodel the extracellular matrix. MMP-9 is elevated after cerebral ischaemia and is involved in accelerating matrix degradation, disrupting the blood–brain barrier, and fostering haemorrhagic transformation. Many drugs that attenuate elevated expression levels of MMP-9 after ischaemia have been found to reduce the extent of haemorrhagic transformation in animal models, including tetracycline derivatives, cyclooxygenase inhibitors, angiotensin-converting enzyme inhibitors and angiotensin receptor blockers, and normobaric hyperoxia [75,76].

Endovascular approaches

Like coronary and limb revascularisation in the past, cerebral revascularisation is steadily moving from a strategy of systemic fibrinolysis alone, with its attendant modest recanalisation rates and extra-target organ risks, to a combined systemic-endovascular or pure endovascular approach, with higher recanalisation efficacy and reduced systemic complications. Endovascular recanalisation therapies for acute ischaemic stroke have rapidly evolved over the past decade and now comprise a wide range of pharmacologic and mechanical techniques.

Intra-arterial fibrinolysis

In local IA fibrinolysis, fibrinolytic agents are infused distal to, proximal to, and/or directly within thrombotic occlusions using a microcatheter delivery system. Compared with standard IV administration, the IA route offers several theoretical advantages, including: higher concentrations of fibrinolytic agent at the clot site; reduced systemic exposure to thrombolytics; an opportunity to carry out gentle mechanical disruption of the clot with the delivery catheter and wire; precise imaging of case-specific vascular anatomy, pathology and collateral patterns; and exact knowledge of the timing and degree of recanalisation achieved. In open clinical series, IA cerebral thrombolysis has yielded higher early recanalisation rates than IV therapy.

IA fibrinolysis also has a number of potential disadvantages, including: manipulation of a catheter within cerebral vessels, potentially increasing vulnerability to haemorrhage; the requirement for heparin administration intra-procedurally to deter catheter-induced thrombosis (potentially increasing haemorrhage risk); delay in initiation of fibrinolysis while the diagnostic angiogram is carried out and the delivery microcatheter positioned (start of IA lytic infusion typically occurs 50–90 min later than start of IV lytic infusion); the procedure is labour- and capital-intensive; and the intervention can only be performed at tertiary and secondary hospitals capable of acute endovascular therapy.

Five randomised trials of IA fibrinolysis vs. control have been performed, including the two major studies of PROACT 2 testing pro-urokinase in the United States and MELT testing urokinase in Japan [52,77]. A meta-analysis of these five trials, collectively analysing 395 patients, found that IA fibrinolysis increased the likelihood of both good outcome (OR = 2.05; 95% CI, 1.33–3.14; P = 0.001) and excellent outcome (OR = 2.14; 95% CI, 1.31–3.51; P = 0.003) [78]. Partial or better recanalisation was increased with IA fibrinolysis (64.6% vs. 17.8%, OR 6.42; 95% CI 3.67–11.24), as was complete recanalisation, (19.0% vs. 1.4%, OR 4.62; 95% CI 2.02–10.56). While symptomatic haemorrhage was increased with IA treatment, (8.9% vs. 2.3%, OR 2.87; 95% CI, 1.21–6.83), mortality was not elevated, (20.5% vs. 24.0% OR 0.82; 95% CI, 0.48–1.39).

Endovascular mechanical therapies

Endovascular mechanical therapies offer several distinct advantages over endovascular delivery of pharmacologic fibrinolytics. Mechanical therapies typically work more rapidly, achieving recanalisation within a few minutes, rather than the up to 120 min required with IA fibrinolytic administration; may have lower intracerebral and systemic haemorrhage risk, due to the avoidance of pharmacologic lysis; are more effective in disposing of large clot burdens in proximal vessels, such as carotid T occlusions, where the sheer volume of clot to be digested retards pharmacologic lysis; and may in general be more efficacious at achieving full recanalisation [79]. Reflecting the generally more rapid development cycle for medical devices compared with pharmacologic agents, new endovascular devices for cerebral revascularisation have been appearing every 12–18 months or so since their first clinical clearance in 2004, with a marked resulting expansion in the endovascular armamentarium for acute ischaemic stroke.

The heterogeneity of target vascular lesions in cerebrovascular disease mandates a diversity of mechanical treatment options for deployment by interventionalists. In many patients, the intracranial occlusion is an embolus that has arisen from the heart or a proximal aortocervical arterial source and landed in a relatively normal recipient artery. Such target thrombi respond well to retrieval and aspiration strategies. In other patients, the occlusive lesion is comprised of an in situ intracranial atherosclerotic plaque with supervening thrombosis. These target lesions will not respond well to retrieval devices, which catch on the plaque, or to aspiration devices, which are effective only for the thrombus component. However, they do respond well to angioplasty and stenting, which accomplish controlled cracking and dissection of the underlying atherosclerotic lesions. There is a notable race–ethnic variation in the composition of intracranial occlusions. Among whites, emboli from the heart or extracranial arterial sources are common; among Asians and Blacks, in situ intracranial atherosclerosis with supervening thrombosis is more frequent [80]. As a result, the most effective and commonly employed mechanical treatment strategies vary regionally across the globe.

The currently widely employed endovascular mechanical interventions may be classified into the categories of devices to augment fibrinolysis, angioplasty/stent devices, suction thrombectomy devices, and thrombus retrieval devices.

Augmented fibrinolysis

Several mechanical techniques may enhance pharmacologic fibrinolysis. Passage of a micro-wire through an occlusion during IA fibrinolytic procedures is a form of augmented fibrinolysis, not only directly disrupting the clot, but also increasing penetration of fibrinolytic agent throughout the target thrombus [81]. Endovascular ultrasound techniques to enhance enzymatic, IA delivered fibrinolytic agents are being developed, in a manner complementary to external ultrasound techniques to enhance intravenously administered fibrinolytics.

Angioplasty and stenting

Angioplasty and stenting are highly effective acute recanalisation strategies when target occlusions are in situ atherosclerotic plaques with supervening thrombosis. Across several early series, the recanalisation rate of acute intracranial angioplasty, largely without stenting, was 84% [82]. Stents flexible enough to be navigated through the tortuous cervical circulation to reach intracranial vessels have only recently been developed. Intracranial application of balloon-mounted stents may pose greater risks in the cerebral than coronary beds because of differences in angioarchitecture. Cerebral arteries lack an extensive external elastic lamina and are relatively fixed in position because of small branching and perforating arteries. Balloon-mounted stents are relatively inflexible and may be difficult to navigate to anterior circulation vessels. Intracranial arteries are more prone to vessel dissection and a ‘snow plough’ effect in which plaque displaced by the stent piles up and occludes the ostium of perforator vessels. Self-expanding stents are better suited to cerebral use: they can be delivered to the target vessel with reduced barotrauma and they adapt more fully to the shape and anatomy of the affected artery [83]. In a 20 patients, single centre, uncontrolled trial of the Wingspan self expanding stent, partial or better recanalisation was achieved in 100% and complete recanalisation in 60% of acute cerebral ischaemia patients [84]. Symptomatic haemorrhage rates were low (5%).

Suction thrombectomy

Suction thrombectomy devices employ vacuum aspiration to remove occlusive clot in acute ischaemic stroke. An advantage of aspiration devices is that the continuous suction forces reduce the risk of uncontrolled thrombus fragmentation and distal embolisation. Progress in developing aspiration devices required a technical solution to the problem of clogging of aspiration tips, a common occurrence when applying suction through a bore small enough to fit within intracranial arteries. The Penumbra System successfully overcame this obstacle by adding an in bore separator wire with a bulbous tip that the operator continually advances and retracts, disrupting attached clot and pulling in thrombus ahead of the catheter. In a multicentre, single arm cohort trial, the Penumbra System yielded some degree of recanalisation in 82% of patients [85,86]. Symptomatic intracranial haemorrhage (intracerebral and/or subarachnoid) occurred in 11.2% of patients. Functionally independent final outcomes tended to be more frequent among patients in whom recanalisation was achieved, 29% vs. 9%, P = 0.06.

Endovascular retrieval devices

Clot retrieval devices were first developed to capture errant coils and other foreign bodies that had embolised within the cerebral circulation during endovascular procedures. A natural next step was to apply these devices to capture and remove naturally arising thromboemboli. These devices ensnare a thrombus and then withdraw it out of the body, via the guide catheter. For cerebral revascularisation, the first approved, and most extensively studied, family of retrieval devices are the Merci Retrievers, helical nitinol coils that entrap the clot, like a corkscrew removing a cork from a wine bottle. In an initial, single arm, multicentre, technical efficacy trial, among 151 patients enrolled, moderate or complete recanalisation by use of the device alone was achieved in 48% [87]. Successful recanalisation was associated with markedly improved clinical outcomes (90 day modified Rankin Scale (mRS) 0–2 in 53% of recanalisers vs. 6% of non-recanalisers, P < 0.0001). Symptomatic haemorrhage occurred in 5% of patients treated with the device alone and 24% treated with the device plus an additional, rescue reperfusion intervention because of incomplete recanalisation response to the device (most commonly IA fibrinolysis). In a subsequent trial employing a later generation of devices, among 164 patients enrolled, moderate or complete recanalisation was achieved in 57% by use of the device alone and in 70% with the addition of adjunctive therapies (IA lytics or angioplasty) [88].

A recently emerging, and particularly promising, class of retrieval devices are the stent retrievers. A self expanding stent is deployed in the occluded vessel within the thrombus, pushing it aside and entangling it within the stent struts. The stent is then withdrawn in its unfolded state, often bringing out with it the enmeshed thrombus. Stent retrievers offer three potential advantages over other retrieval devices: (i) restoration of flow immediately upon deployment within the target artery, rather than only upon successful clot extraction, (ii) substantially higher successful recanalisation rates than other embolectomy devices, and (iii) the option to detach the stent and leave it in place in the vessel if the thrombus is unable to be withdrawn. In single centre series, retrievable stents have achieved rapid and frequent flow restoration [89,90]. Several multicentre trials comparing stent retrievers with earlier generation devices are under way.

The development of thrombus retrieval devices has allowed investigation of the histopathologic properties of actual human stroke-causing thrombi. While a minority of retrieved clots conform to classic descriptions of ‘red’ or ‘white’ thrombi, the preponderance have mixed architectural features, with fibrin/platelet deposits interspersed with linear collections of nucleated cells (monocytes and neutrophils) and confined erythrocyte-rich regions [91]. Radiopathologic correlation studies analysing retrieved thrombi have demonstrated that clot conspicuity on CT or MR scans is proportional to the red blood cell content of the thrombus – red blood cell and mixed clots frequently produce hyperdense artery signs on CT and hypointense artery signs on MRI, fibrin-dominant clots do not [92]. Better understanding of the molecular composition and physical–mechanical properties of retrieved thrombi promises to enable improved design of drug and device therapies aimed at clot removal.

Need for formal controlled trials and promise of combined IV and endovascular strategies

The regulatory approval framework for devices is markedly different from drugs, requiring that they demonstrate technical efficacy as task tools, e.g., clearing clot from the neurovasculature, not that they demonstrate improved final patient outcome. As a result, device approvals come from single arm, technical efficacy studies, not randomised controlled trials. Randomised trials with final clinical outcome primary efficacy endpoints are urgently needed to provide definitive evidence of treatment benefit. Among several such studies now underway worldwide are two NIH-funded trials: the MR RESCUE Trial comparing endovascular recanalisation vs. supportive care up to 8 h after stroke onset and the IMS 3 Trial comparing IV tPA alone vs. combined IV tPA and endovascular recanalisation therapy among patients presenting within 3 h of onset [93,94].

Preliminary studies suggest that ultimately combined IV and endovascular treatment will be a preferred strategy for cerebral revascascularisation, rather than either a pure IV or a pure endovascular approach [95,96]. Combined therapy offers the advantages of a speedy start of reperfusion treatment via the IV route (crucial for ischaemic stroke in which the optimal time window for intervention is brief) and high final recanalisation rates via endovascular options (recanalisation rates close to 100% are desirable, and are unlikely to ever be obtained by an IV strategy alone). In contrast, the disadvantages of the treatments do not appear to be additive – several pilot trials and cohort studies have suggested that the haemorrhagic risks of combined therapy are not substantially higher than the haemorrhage risk of endovascular treatment alone. In the future, it may be envisioned that tailored, patient-specific combinations of recanalisation therapies will be deployed to achieve optimum recanalisation rates, always bearing in mind that the cerebral vasculature is fragile and the amplitude of mechanical energies and intensity of pharmacologic therapies delivered to break up thrombi are constrained by the need to protect vessel wall integrity.


The ideal toward which reperfusion therapies for stroke strive is to achieve rapid, complete and sustained vessel patency in all patients harbouring salvageable tissue, with no risk of haemorrhagic transformation. Areas of recent advance include: wider implementation of standard IV fibrinolysis in regional acute stroke care networks; testing of newer fibrinolytic agents, investigation of combinations of fibrinolytics with thromboactive, neuroprotective, and vasoprotective agents; and availability of several classes of mechanical endovascular recanalisation devices, including stent retrievers. As these advances progress over the next decade, cerebral recanalisation therapy is poised to evolve from a treatment of modest efficacy and scattered availability to an intervention of remarkable power and scope for rescuing patients from the scourge of acute ischaemic stroke.

Disclosure of Conflicts of Interests

The University of California, Regents receive funding for Dr. Saver's services as a scientific consultant regarding trial design and conduct to BrainsGate, CoAxia, ev3, and Talecris (all modest). Dr. Saver is an investigator in the NIH FAST-MAG, MR RESCUE, CUFFS, CLEAR-ER and IMS 3 multicenter clinical trials for which the UC Regents receive payments based clinical trial performance; serves/served as an unpaid site investigator in a multicenter studies run by Lundbeck and Concentric for which the UC Regents received payments based on the clinical trial contracts for the number of subjects enrolled; is an employee of the University of California, which holds a patent on retriever devices for stroke; and is funded by NIH-NINDS Awards P50 NS044378 and U01 NS 44364.