By now, most interested people are well-aware of the MR CLEAN results; if not, you can read the article here. Following are just a few thoughts and observations:
One thing that caught my eye is that the absolute difference between endovascular therapy and conservative care with respect to the (secondary) outcome of achieving a modified Rankin scale (mRS) score of 0-2 was ~ 13%. This is the same effect size that IV tPA had on the (secondary) outcome of achieving a mRS of 0-1 in the pivotal NINDS study. Put another way, the number needed to treat (NNT) with IV tPA vs. placebo to save a subject from any disability is the same NNT with endovascular intervention vs. tPA alone to save a subject from slight disability (8 subjects). I thought it interesting how similar those effect sizes are.
Next, the risk of symptomatic intracerebral hemorrhage (sICH) in the control arm (91% of whom received IV tPA) was 6.4%. Again, this is exactly the sICH risk in the pivotal NINDS trial and is very similar to other trials. It’s remarkable how consistent that finding is.
Third, there are some parallels between the early failures of IV tPA and the early failures of endovascular therapy. Multiple early studies of IV thrombolysis for acute ischemic stroke were negative. They used inferior drugs (streptokinase) and long time windows (5-6 hours, with most subjects being randomized toward the later part of the windows). Eventually, the NINDS tPA study group hit upon an important innovation: They required that about half of the subjects be randomized within 1.5 hours. If a study site had a run of 1.5 – 3 hour randomizations, they had to halt enrollment until they got some subjects within the 1.5 hour window. This ensured that sites really worked hard to treat people quickly rather than dally “because we still have over an hour to go”. Turns out that tPA benefit is exquisitely time-sensitive.
Similarly, early endovascular trials used a drug not currently in use (pro-urokinase, although this was actually a positive trial) or substandard tPA dosing, enrolled subjects without an identified vascular occlusion, used less-efficacious embolectomy devices, and had long onset-to-treatment times. Moreover, recruitment was often very sluggish, possibly because the patients felt to be most likely to benefit from endovascular treatment were treated off-study. MR CLEAN rectified all of those issues.
Fourth, I think that MR CLEAN vindicates both the cautious and the enthusiastic schools of thought regarding endovascular treatment. It’s easy to see how strong endovascular proponents would feel vindicated–we finally have a really well-done randomized controlled trial showing the robust clinical benefit they always believed in. But looking back on the purported reasons for the earlier studies’ negative results also vindicates those of us who were more cautiously optimistic (as I have always considered myself to be). If one accepts the proposition that earlier failures were due to factors such as immature technology, then the current success doesn’t show that those technologies were beneficial all along–it shows that only in this newer era of stent retrievers is endovascular therapy truly beneficial.
Finally, a few words about the other major trials not yet published. Several ongoing endovascular studies have halted enrollment, with interim analyses occasioned by the favorable MR CLEAN results either pending or showing clear benefit. The ethical argument for this is that we should not be randomizing subjects to conservative care once we know that endovascular care is beneficial. However, there are important reasons to continue randomization. One is that our estimate of the overall effect size will be more precise if we have a larger sample. Another argument in favor of completing the trials is that it will allow for more robust subgroup analyses. While it’s great to finally have compelling data in support of a beneficial stroke treatment, it would be even better to have a finely-grained understanding of exactly who benefits and by how much.
Take carotid endarterectomy (CEA) for symptomatic disease, for example. I wasn’t around back then, but I was taught that strong supporters were vociferous in their insistence that CEA is beneficial and that randomizing subjects was unethical. Indeed, NASCET substantiated this belief, but also provided rich data sets that taught us the differential benefits (or lack thereof) according to degree of stenosis, time from stroke/TIA to randomization, patient gender, presence of contralateral occlusion, presence of tandem stenosis, etc. It’s very gratifying to know that we’re moving into an era of expanded therapeutic options for acute ischemic stroke, but many important questions remain to be answered about endovascular therapy: Is there a differential effect according to patient age? Gender? Location of occlusion? Is recanalization beneficial out to 8 hours? 12 hours? What role should perfusion imaging play? How about ASPECTS scores? DWI? Assessment of vascular collaterals? These questions will be much harder to address if we stop all trial enrollments and return to relying on lower-quality retrospective data.
Still, it’s a very exciting time in the stroke world–I’m sure this year’s conference in Nashville will be humming. Wish I could go, but someone’s got to mind the fort . . .
* OK, so the game pictured is Atari’s Breakout, not “Breakthrough”, but close enough. If you’re in the age range where you used to play such Atari games for hours on end, and you can afford to blow off the rest of your day, go to Google image search and type “Atari Breakout”. Or just click on the picture 🙂