An interview with Dr. Thabele (Bay) Leslie-Mazwi, MD, Director of Endovascular Stroke Services, Massachusetts General Hospital; Assistant Professor of Neurology, Harvard University; and Dr. Gregory W. Albers, MD, Director, Stanford Stroke Center; Professor of Neurology, Stanford University.
Interviewed by Kristina Shkirkova, BSc, Doctoral Student in Neuroscience, Department of Neurosurgery, Zilkha Neurogenetic Institute, University of Southern California.
They will be discussing the article “DEFUSE 3 Non-DAWN Patients: A Closer Look at Late Window Thrombectomy Selection,” published in the March 2019 issue of Stroke.
Ms. Shkirkova: Please briefly summarize the design and findings of your study.
Drs. Leslie-Mazwi and Albers: We evaluated DEFUSE 3 patients who would have been excluded from the DAWN trial based on DAWN eligibility criteria, with the goal of assessing treatment effect in that DEFUSE 3 subgroup (DEFUSE 3 Non-DAWN). The main reasons for DEFUSE 3 Non-DAWN were NIHSS 6-9, core too large (based on age and volume of established infarct), and mRS of 2. Patients with mRS 2 were included with the NIH stroke scale 6-9 group, as detailed in our paper, and so we analyzed the DEFUSE 3 Non-DAWN patients NIHSS 6-9 and core-too-large patients to assess treatment effect in that subgroup.
Patients with pretreatment core infarct volumes <70ml but too large for inclusion by DAWN criteria demonstrated robust benefit from endovascular therapy. Data supporting a beneficial treatment effect across the full range of NIHSS scores was documented in the entire DEFUSE 3 population. In our small subgroup of patients with NIHSS 6-9, we found a trend towards benefit.
Ms. Shkirkova: Were you surprised at the number of patients in the DEFUSE 3 trial who did not meet DAWN criteria? Is 39% (71/182) what you expected? Why do you think there was such a discrepancy?
Drs. Leslie-Mazwi and Albers: Trial design is always challenged by accurately reflecting the population at risk while maintaining enough homogeneity to produce a significant treatment effect. DEFUSE 3 was designed to be as inclusive as reasonable, based on work in the earlier DEFUSE trials. Patients were, therefore, included at more extreme ends of the spectrum of potential benefit. This drove the difference between DEFUSE 3 participants and DAWN participants. The almost 40% discrepancy underscores how small changes in inclusion criteria can lead to large changes in patient numbers for who can access the treatment.
Ms. Shkirkova: Considering differences between patient selection criteria in the DEFUSE 3 (perfusion imaging mismatch) and DAWN (discrepancy between clinical deficit severity and size of early infarct) trials, which approach do you think is more relevant to use in clinical practice? Are there practical issues that make one more relevant?
Drs. Leslie-Mazwi and Albers: Both trials utilized automated perfusion software. DAWN measured the core infarct and compared it to the patient’s clinical exam. DEFUSE 3 measured both the core infarct and the area of hypoperfused penumbra. Both approaches have clearly demonstrated their effectiveness in selecting patients in late windows. This is recognized by the recommendations in the AHA guidelines for 2018 that support either approach. The treatment effect in these 2 trials was huge (the largest in the history of stroke trials). Given the poor outcomes in the medical arm, it is highly likely that we can be more inclusive with endovascular approaches and still maintain a treatment benefit to undergoing thrombectomy. The implications of this for patient selection criteria are still being explored.
Practically, automated perfusion software simplifies and accelerates the interpretation of imaging, and facilitates distribution of results. However, the software is expensive, and this may represent a challenge for certain facilities.
Visualizing the volume of the ischemic penumbra may have important implications as we continue to expand treatment indications. Patients with large core infarcts (>70cc) or low NIHSS (<6) that also have large penumbral volumes are probably more likely to benefit from recanalization than patients who have small or absent penumbra volumes. Various options for expanding selection criteria will be evaluated in ongoing and future studies.
Ms. Shkirkova: In the study, the “core too large” DEFUSE 3, non-DAWN subgroup of patients has higher mortality in the treatment arm vs. medical arm (mRS 6 22% vs 13%) and no better mRS 5-6 outcomes (33% vs 33%), whereas there is benefit throughout in the “core not too large” subgroup. Do you think this suggests that potentially treating patients with even larger cores may not be beneficial?
Drs. Leslie-Mazwi and Albers: It is important to note that in these small subgroups, we are underpowered to assess every individual cut-point on the mRS, and the mortality trend you mentioned was not statically significant. The core-not-too large subgroup is much larger, so direct comparisons may be misleading. We certainly expected the “too large” group to have less favorable outcomes overall. As we embark on studies of patients with even larger core sizes, we will anticipate an even higher rate of poor outcomes because large core size is clearly a prognostic factor. However, our hope is to show overall treatment benefit. This benefit may be reflected primarily with a reduction in very poor outcomes (mRS 4-6) because functional independence (mRS 0-2 outcomes) may be uncommon in these patients.
Ms. Shkirkova: Based on the change of pattern in the benefit described in the question above, do you think there is still more penumbra threshold below the DEFUSE 3 cut-off, where patients would benefit, or is it beginning to hit the ceiling?
Drs. Leslie-Mazwi and Albers: This remains an open question. DEFUSE 3 allowed penumbra volumes as small as 15 ml; however, the vast majority of patients had very large penumbra volumes. It is possible that volumes below 15 ml will benefit but will likely result in dilution of the treatment effect, and, therefore, larger sample sizes may be required to establish efficacy. Since the benefit of this therapy is so dramatic in patients with favorable imaging profiles, it is important to remember that studies that include a subgroup of patients who do not have any salvageable tissue could still be “overall positive.” Therefore, it will be a challenge to sort out which patients actually do not benefit. Importantly, our metrics for outcome will also evolve. While we used functional outcome as the measure in this study, DEFUSE 3 used an mRS shift as the primary outcome. This allows a more discriminatory approach than a binary outcome of functional independence or not, and will pick up nuances of treatment effect. Finally, we do not have any clear evidence for harm in any of the endovascular trials to date. Therefore, it is important to attempt to further explore the limits of this therapy.
Ms. Shkirkova: Do you think that patients who present in the 16 to 24 hours window and meet DEFUSE 3 but not DAWN criteria should be treated based on the results of your analysis, even if they do not meet criteria set out in current guidelines?
Drs. Leslie-Mazwi and Albers: Yes. Results indicate that imaging selection can identify patients who will benefit for thrombectomy. We feel this is true between 16 and 24 hours, and is likely true after 24 hours (though, practically, patients presenting in that very late timeframe are limited in number). DEFUSE 3 and DAWN underscore the importance of the so-called “tissue clock” over the traditional “time clock.” This allows bespoke therapy for the individual patient, based upon the capacity of that patient’s collaterals to sustain the penumbra and minimize the size of the core.
Ms. Shkirkova: This study and the DEFUSE 3 and DAWN trials confirm the benefit of mechanical thrombectomy. Do you think that there are subgroups of patients not enrolled in these studies who may benefit from thrombectomy? Which are they? Are additional trials for these groups necessary?
Drs. Leslie-Mazwi and Albers: Without a doubt, we feel that additional groups benefit from thrombectomy that remain to be evaluated. This includes patients with more distal occlusions (M2, maybe M3), patients with NIHSS <6, patients with established core infarcts greater than 70 cc, and posterior circulation stroke. Importantly, trials are underway to assess treatment effect in all these groups. We anticipate continued expansion of thrombectomy in this current extremely dynamic and exciting era of stroke treatments.