Article Text

Download PDFPDF

Understanding IMS III: old data shed new light on a futile trial
  1. Joshua A Hirsch1,
  2. R Gilberto Gonzalez2
  1. 1NeuroEndovascular Program, Massachusetts General Hospital, Boston, Massachusetts, USA
  2. 2Department of Neuroradiology, Massachusetts General Hospital, Boston, Massachusetts, USA
  1. Correspondence to Dr Joshua A Hirsch, NeuroEndovascular Program, Massachusetts General Hospital, 55 Fruit Street, Gray 241B, Boston, MA 02114, USA; hirsch{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Neurointerventionalists and other believers in intra-arterial thrombolysis (IAT) were disappointed to learn that the Interventional Management of Stroke (IMS) III trial was halted at a prespecified endpoint. Already there are a number of articles citing rationales for failure of bridging therapy to demonstrate what would seem like an intuitive benefit over intravenous tissue plasminogen activator (tPA) monotherapy.1 There are genuine questions relating to patient selection and device availability as to the ability of IMS III to demonstrate a positive effect. In this brief comment we explore an area not necessarily well-known to the broader neurointerventional community that might confound any bridging trial.

At the most basic level, acute ischemic stroke is caused by the lack of cerebral blood flow. This leads to neurologic dysfunction and brain tissue injury. Hence, prompt restoration of blood flow is the most logical therapeutic approach. Reperfusion remains the only proven method to treat large vessel stroke, and the criticality of recanalization is well-known to neurointerventionalists and is an active part of clinical practice.2 ,3

IMS III was a ‘bridging’ trial; patients were first treated with intravenous tPA, then randomized to mechanical recanalization versus no additional therapy. ‘Bridging’ had been shown to benefit patients.4 ,5 The rationale was to maximize chances for rapid clot lysis, restoring blood flow through large occluded cerebral arteries in as timely a fashion as possible by ‘priming the pump’ with intravenous tPA and following through with mechanical therapy. While we wish to congratulate the stroke neurology and neurointerventional investigators in IMS III for completing such an important trial, we are vexed by its futility.

The basic idea seems reasonable—start the clot lysis process with intravenous tPA as soon as possible and then go in and mechanically remove any clot that still remained. However, the premature stoppage of IMS III suggests that IAT plus intravenous tPA is no better than the latter alone. It is not intuitive why this would occur.

Neurotoxicity of tPA and related effects in animal models

Neurointerventionalists may not be aware that the potential harmful effects of tPA in stroke have been studied for almost two decades. It became a topic of active investigation after the 1998 publication by Wang et al6 which showed that tPA worsened outcomes in models of ischemic stroke using wild-type and tPA knockout mice. A series of subsequent studies demonstrated that both tPA and plasmin are neurotoxic if they reach the extracellular space. Additionally, severe ischemic injury to the blood–brain barrier (BBB) produces injury due to the protease effects of tPA. A variety of effects in the brain were demonstrated including direct vasoactivity, cleavage of the N-methyl-D-aspartate (NMDA) NR1 subunit, amplification of intracellular Ca2+ conductance and activation of extracellular proteases from the matrix metalloproteinase (MMP) family. These effects may increase excitotoxicity and increase the damage to the BBB, thereby worsening edema and increasing the probability of hemorrhage. Moreover, in some cell and animal models of stroke and brain injury, tPA can also bind to NMDA receptors thus potentially amplifying excitotoxicity in neurons.7–10 Whether the higher concentrations of tPA used in intra-arterial approaches may activate some of these hypothesized MMP and NMDA mechanisms of reperfusion injury remains to be determined.

Clinical evidence of harmful effects of tPA

Are the experimentally verified negative effects of tPA in animal models relevant to clinical ischemic stroke? Perhaps a clue might be found in MRI studies that suggested differences in BBB responses after thrombolytic versus mechanical reperfusion. Kidwell et al11 used GdDTPA-enhanced MRI to detect contrast leakage into the subarachnoid space as a measure of BBB integrity after acute ischemic stroke. They reported data from 140 patients, including 38 who received intra-arterial tPA, 24 administered intravenous tPA, 18 who underwent only mechanical embolectomy and 60 patients who had no acute intervention. BBB breakdown was found in 66% of patients who received intra-arterial tPA and in 50% of those who received intravenous tPA compared with 28% of patients undergoing mechanical embolectomy and 30% of patients who received no acute treatment (p=0.002). A key observation was that thrombolysis with tPA was an independent predictor of BBB breakdown (p=0.001), and BBB breakdown was an independent predictor of hemorrhagic transformation (p=0.007).

In that paper, BBB leakage was less with mechanical recanalization than with tPA treatments, and BBB leakage was more prevalent with intra-arterial tPA than with intravenous tPA. Is it possible that, besides clot lysis, tPA may be directly affecting neurovascular function and integrity? This would be in line with experimental models that have shown that high concentrations of tPA upregulate MMPs, which can damage the BBB.6 ,12–14

Future directions

The apparent failure of bridging therapy involving a combination of intravenous tPA followed by intra-arterial recanalization in IMS III is surprising. Clearly, more research is essential. For example, how does large vessel recanalization correlate with microvessel reperfusion? Since intravenous tPA does not work particularly well in large vessel occlusion, is it possible that mechanical revascularization is serving the perverse role of bringing tPA with its possible neurotoxicity directly to the brain parenchyma?15 Are further experimental animal model studies (using the more directly relevant non-human primate model) that mimic and test clinical bridging therapy approaches important? Can advanced neuroimaging be used to identify the best responders in eligible patients? We wonder if various neurovascular biomarkers might potentially help us to exclude patients who are more vulnerable to MMP and/or NMDA-mediated reperfusion injury. Do we need clinical trials that randomize patients to only intravenous tPA or only mechanical IAT?16

IMS III remains a landmark trial for stroke neurologists and neurointerventionalists. Understanding the failure of the bridging arm is critical in considering the role of IAT in the future. We believe that restoring cerebral blood flow is essential to saving brain tissue from acute ischemia. The logic is reasonable and remains clear. With this brief note, we hope to make neurointerventionalists aware of these potential phenomena of tPA toxicity. Perhaps bridging leads to the perfect storm of recanalizing ischemically injured arteries, permitting tPA to cross a damaged BBB into the brain interstitial space where further damage ensues. In view of the possibility that improved mechanical recanalization devices may paradoxically exacerbate this effect, we believe it important to raise the awareness of these existing data.17 Clearly, one thing we can say for sure is that more work needs to be done.


The authors thank Dr Eng H Lo for helpful discussions on potential signaling mechanisms of tPA and Irina Badayan for assistance in preparation of the manuscript.



  • Funding None.

  • Competing interests None.

  • Provenance and peer review Commissioned; internally peer reviewed.