Is there a time after symptom onset at which acute ischaemic stroke patients should no longer be eligible for mechanical thrombectomy? This has become one of the most important questions now that the safety and efficacy of mechanical thrombectomy has been established.
According to Tudor Jovin (University of Pittsburgh School of Medicine, Pittsburgh, USA) we should be asking the question, “Should there be a time window at all? And should we even be concerned about time in acute stroke interventions?” Jovin weighed in on the subject at the International Stroke Conference (ISC; 17–19 February, Los Angeles, USA). His main point was that there are two types of stroke patients to consider—“fast progressors” and “slow progressors”—and there is a need for fast workflows and fast recanalisation overall.
He pointed out that recent trials have shown that the purpose of intervention is to prevent infarct growth and to end up with a small infarct at the end of the procedure because there is a very strong correlation between infarct volume and good clinical outcomes. It was shown in IMS III and replicated in SWIFT PRIME that patients with small infarcts have a much higher chance of doing well.
“We know that when a vessel is occluded we are dealing with two compartments: core and penumbra (core being dead brain and penumbra tissue that is functionally impaired but structurally intact). We are all after penumbral salvage, but we need to understand that this is a time dependent process, and with time, the core grows and the penumbra shrinks until the entire territory that is supplied by the occluded vessel has infarcted and at that point, any attempts to recanalise the vessel are futile or even detrimental,” Jovin explained.
What is perhaps less well understood, according to Jovin, is that this process occurs at different speeds in different individuals. This was illustrated with data from DEFUSE 2 (a study that required an MRI scan prior to endovascular therapy) where there was a range of patients who had similar infarct growth and widely different time points—“fast progressors” and “slow progressors”. He explained that it is the collaterals that determine this rate of growth.
“It was worked out after IMS III by Khatri et al, that every 30 minute delay in reperfusion is associated with a 10% relative reduction in probability of good clinical outcome (mRS 0–2). If we look at the modern endovascular trials, we need to revise the Khatri curve. I think we are dealing perhaps with different patient populations in the most recent trials. In MR CLEAN there is still a curve and it is significant, but it is only 5% relative reduction in probability of good clinical outcome. Why is that? Perhaps in MR CLEAN even though there were no requirements for eliminating patients with large infarcts, based on the fact that median ASPECTS score was 9, maybe MR CLEAN enrolled patients who had better collaterals than IMS III,” he said. Similarly, in REVASCAT, the Khatri curve mirrors that seen in MR CLEAN—5% reduction for every 30-minute delay.
ESCAPE, Jovin went on, required inclusion only of patients who had good collaterals, and the Khatri curve in ESCAPE patients is almost flat, indicating that every 30-minute delay in reperfusion is associated with a 0.5% relative reduction in probability of good clinical outcome.
Jovin reported a deeper analysis of the REVASCAT data, detailing all patients that were subjected to endovascular therapy, not only the ones who reperfused. He said that it was found that from symptom onset to reperfusion there was a 26% reduction in probability of good outcome with every 30-minute delay. But from symptom onset to imaging, there is no association with time; the curve is in fact driven by the imaging to reperfusion interval. When it comes to patients who had poor collaterals, ie. those with low ASPECTS scores, there is a very steep curve—that is where there is a strong relationship between time to reperfusion and good outcome. On the other hand, in patients who have good ASPECTS scores, ie. good collaterals, this association does not exist. He added that in DEFUSE 2, which enrolled patients with good collaterals evidenced by a median baseline infarct volume of 16cc, the Khatri curve is actually in the opposite direction and there is absolutely no association between time to symptom onset and good outcome.
“How do I make sense of all these data? I think we are looking at different patient populations here. In ESCAPE and in DEFUSE 2 we are looking at an enriched patient population with respect to good collateral status, we are looking at ‘slow progressors’, that is why this curve is much flatter or the association between time of stroke onset to reperfusion and outcome is very attenuated or not detectable, whereas if we take a less strictly selected segment of the acute stroke population such as those enrolled in IMS III and (to a lesser extent) MR CLEAN and REVASCAT we are going to deal with a higher percentage of ‘fast progressors’ and that is where we are going to encounter a steep curve,” Jovin said.
As for what happens if these patients are treated, Jovin explained that data from DEFUSE 2 suggest that if there is mismatch and good collaterals, whether a patient is treated beyond six hours or within six hours what really counts is whether they reperfuse, and not the time window at which they are treated.
“What does this mean? Should we treat patients with large vessel occlusions and mismatch beyond six hours without any time limit in routine clinical practice? Not necessarily. Do we need any other proof? Yes, we do need proof because the problem is that we have absolutely no idea what the behaviour is of the patients who have mismatch and good collaterals but do not get treated with endovascular therapy. Because they have good collaterals it may be that their stroke will never grow and therefore they may do well without intervention. The challenge is to figure out who are those people with good collaterals who are going to experience growth of core with persistent vessel occlusion and devise strategies for the treatment of those patients,” he maintained.
To answer these questions, there are now two large randomised trials ongoing—DAWN and DEFUSE 3. DAWN is a single device trial (Trevo embolectomy device) sponsored by Stryker Neurovascular, with a time window of six to 24 hours, enrolling patients with proximal anterior circulation occlusions (M1, ICA T), with a core and clinical mismatch type selection paradigm where age is also a factor. Nearly 100 patients have been enrolled to date in this world-wide trial. DEFUSE 3 is an NIH-funded, prospective, randomised, multicentre, adaptive blinded endpoint trial that uses diffusion and perfusion mismatch as selection criteria. Mismatch on diffusion and perfusion is the same as the DEFUSE definition—core ˂70mL, mismatch ratio >1.8 and mismatch volume ≥15cc.
Concluding, Jovin stated, “Time dependency of favourable outcomes in patients with large vessel occlusion in the anterior circulation is most pronounced in those with poor collaterals—that does not mean that in patients who are ‘slow progressors’ (those with good collaterals) we have a justification to take our time. We need to have fast workflows and fast recanalisation even in the ‘slow progressors’ because there could be other benefits that current outcome measures are too insensitive to detect when we open up vessels very fast in these ‘slow progressors’. A non-insignificant proportion of patient with large vessel occlusion stroke in the anterior circulation (estimated at about 20–30%) presenting in the beyond six-hour time window have substantial mismatch (‘slow progressors’). Contrary to popular belief, it has been shown that whether they are wake-up or witnessed onset does not make a difference in terms of their response to endovascular therapy. In this patient population, endovascular treatment appears to be feasible and safe, but because we do not know anything about the natural history of these ‘slow progressor’ patients we need to complete a randomised trial to establish the clinical efficacy of this approach”.