Response to: Basilar artery occlusion and unwarranted clinical trials

In their letter to the editor, the authors from the Karolinska Institute argue that, with the current available evidence of efficacy of mechanical thrombectomy for acute ischemic stroke, it is unethical to perform a trial in patients with basilar artery occlusion in which patients are at risk of being withheld this effective treatment. The performers of such trials are compared with representatives of the tobacco industry, breachers of the Nuremberg code and accused of being scientifically biased against, and unable to understand, observational studies. As the principal investigators of three such trials, we are more than happy to respond to these grave accusations.

In the heat of a discussion, arguments are often simplified or distorted. Several papers have been published on the so-called parachute trials – a type of trial that aims to prove the obvious.^1,2 Although such papers serve a purpose, the simplification of a discussion has the risk of being taken too seriously.

Current endovascular treatment (EVT) options in acute ischemic stroke date back to the first attempts of local thrombolysis in patients with basilar artery occlusion in the early 1980s by Herman Zeumer and his team in Hamburg.³ In those days, basilar artery occlusion was considered to be incompatible with independent survival – comparable to the outcome after jumping from an airplane without a parachute. Good outcomes of a few selected patients after intra-arterial thrombolysis (IAT) were enough evidence for the stroke community to obviate the performance of a randomized trial of patients with basilar artery occlusion. The idea of performing such a trial only surfaced after the failure of numerous trials to show the efficacy of IAT in the anterior circulation. In the meantime, case series had been published showing that patients with basilar artery occlusion could have a good outcome without being treated with IAT. Meta-analysis of retrospectively collected data of case series of selected patients showed no “obvious” difference in outcome between patients treated with intravenous thrombolysis (IVT) or IAT.⁴

The BASICS registry (2002–2007), a prospective multicenter observational registry of patients with BAO, was started in an attempt to show the efficacy of EVT without having to perform a randomized trial with similar arguments as used by the authors in their letter to the editor. Over a five-year period, over 600 patients with BAO were included in 50 dedicated stroke centers worldwide. Against all odds, this registry failed to show superiority of EVT over IVT.

Of course, there were many arguments for EVT believers to discard the results of this observational study. Probably the most valid argument was the presence of a potential selection bias, inherent to the observational, non-randomized nature of the registry. Another reason was the use of inferior outdated endovascular therapy; most EVT patients were treated with IAT alone or first-generation thrombectomy devices.

Based on available knowledge of observational studies in patients with BAO, the type of study favoured by the authors from the Karolinska Institute, a trial testing the efficacy of additional EVT, was therefore a logical next step. To date, no additional data on patients with BAO have emerged contradicting these observational studies.

The reason we started a trial of patients with BAO, and continued the trial after the publication of positive results of anterior circulation trials, is that we believe that there are major differences in vascular anatomy (collateral flow and perforating arteries) between the anterior and posterior circulation which may cause differences in treatment response. Although our original presumption was that this difference might benefit the efficacy of EVT, these differences may equally favour IVT.

We would like to counter argue the authors as they state; (1) that restoring blood flow is equally essential in both the anterior as the posterior circulation; we could argue that superior collateral flow in the posterior circulation may increase the resistance to ischaemia potentially prolonging the treatment window. (2) The efficacy of IVT in restoring blood flow in BAO is low (13%); the authors base this conclusion on five small studies including a total of 49 patients with BAO. In the BASICS registry, recanalization after IVT was 67% in 207 patients on follow-up imaging, only slightly less than at the end of the EVT procedure (72%). Superior collateral flow may improve the efficacy of IVT by a similar thrombolytic effect distal as proximal to the thrombus thanks to direct retrograde filling of the basilar artery from the anterior circulation through the posterior communicating arteries potentially shortening time to recanalization and improving recanalization rates. Although the timing of recanalization after IVT may have been delayed, the effect of recanalization on outcome was similar as in EVT. (3) There is a vast number of observational studies showing the superiority of mechanical thrombectomy compared to IVT alone in BAO; the authors refer to a meta-analysis limited to case series of patients treated with endovascular therapy (N = 491).⁵ A much more inclusive meta-analysis (N = 2056) including case series of patients treated with second-generation devices and with IVT alone (N = 341), found comparable death and dependency rates between EVT and IVT.⁶ Both meta-analyses concluded that randomized controlled trials in patients with BAO are warranted.

Both results from observational studies and randomized controlled trials may be influenced by selection bias. In observational studies, this bias is caused by the treating physician. The reasons for clinicians to select a specific treatment option in an individual patient are complex leading to confounding by indication. Multivariable analysis can never adjust completely for systematic differences between treatment groups. We agree with the authors that randomized trials may fail to show efficacy of a highly effective intervention if only a selected minority of eligible patients is randomized. The one and only randomized controlled trial of parachute use to prevent death and major trauma when jumping from an aircraft failed to show efficacy of parachute use because only subjects jumping from a stationary aircraft on the ground were randomized.¹ One way to make sure a trial is not harmed by this particular bias, is to require participating centers to enter data on all non-randomized patients in a screening log stating the reason why a patient was not eligible, or if eligible, why not randomized.

The authors could have used the same arguments against performing the anterior circulation trials, or any acute intervention trial. Trials are rarely started without some evidence of efficacy from observational studies. As a result, the randomizing physician often experiences a certain discomfort in randomizing a patient in a trial with a 50% chance of receiving a treatment which is perceived to be inferior. For many clinicians, this is a reason to limit their participation to observational studies.

There are multiple examples of failed randomized controlled trials despite robust observational data in cerebrovascular medicine. There was very strong non-randomized evidence to support the use of anticoagulation in intracranial atherosclerotic disease (ICAD), yet the WASID trial refuted its efficacy.⁷ Similarly, well-conducted observational studies and prospective registries supported the use of intracranial stenting for ICAD. However, SAMMPRIS and VISSIT subsequently showed that the intervention was not only inefficacious but even harmful.^8,9

The authors are surprised that there is still no consensus on how to treat acute BAO, despite overwhelming evidence of efficacy of EVT in stroke in the anterior circulation. We would argue that the acknowledgement of this lack of consensus on the best treatment option in patients with BAO based on the available observational data described above is the most valid reason to increase the level of evidence by performing a trial.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.