Background Imaging-based patient selection for neurothrombectomy is reliant on the identification of irreversibly damaged brain tissue (core) and salvageable tissue (penumbra). The DAWN trial used the clinical-core mismatch (CCM) paradigm (clinical deficit out of proportion to infarct volume). We aim to determine the prevalence of CCM in large vessel occlusion (LVO) strokes and study the impact of time and the Alberta Stroke Program Early CT Score (ASPECTS) on the likelihood of mismatch.
Methods We performed a retrospective observational analysis of internal carotid artery/middle cerebral artery M1 occlusions with available advanced imaging (relative cerebral blood flow/MRI). We used automated software for infarct volume analysis and ASPECTS determination. The prevalence of CCM and the impact of time and ASPECTS were analyzed.
Result One hundred and eighty-five LVO strokes were included. Mean age was 71±15 years and median National Institutes of Health Stroke Scale score was 17 (range 12–21). Mean ischemic core volume was 50±69 mL. Within 0–24 hours, CCM was present in 53% and ranged from 63% in 0–3 hours to 25% at 21–24 hours (p=0.03). Prevalence of mismatch reduced 1.6% for every 1 hour increase in time to imaging. CCM prevalence by ASPECTS groups was: ASPECTS 9–10: 77%, ASPECTS 6–8: 65%, ASPECTS 0–5: 13% (p<0.01), with a 6.4% decrement for every 1 point decrease in ASPECTS. The prevalence of mismatch did not diminish over time among ASPECTS groups and higher ASPECTS was an independent predictor of CCM (OR 1.4 (95% CI 1.1 to 1.7), p<0.001).
Conclusions CCM is present in 57% and 50% of LVO strokes in the 0–6 and 6–24 hour window, respectively. The prevalence of mismatch declines with increasing time (1.6%/hour) and decreasing ASPECTS (6.4%/point). Among ASPECTS groups the prevalence of mismatch does not decline over time. These data support the use of an ASPECTS-based paradigm for late window patient selection.
- blood flow
Statistics from Altmetric.com
Endovascular thrombectomy (EVT) is the standard of care for acute ischemic stroke caused by large vessel occlusion (LVO).1 In 2015, early time window (0–6 hours) EVT trials established the superiority of EVT over best medical management, and a majority of these trials selected patients based on CT head-based Alberta Stroke Program Early CT Score (ASPECTS) (≥6).2–6 Pooled analysis of early EVT trials7 suggests that the treatment effect of EVT disappears at approximately 7.3 hours. This could be attributed to completion of stroke (absence of mismatch) and/or large stroke burden. Hence, late window (6–24 hours) trials were designed to select patients for EVT based on the presence of mismatch—small irreversibly infarcted brain tissue (core) surrounded by a large area of reversibly dysfunctional and potentially salvageable brain tissue (penumbra) at risk of infarction. The DAWN trial8 used the concept of clinical-core mismatch (DAWN-CCM), defined as the presence of severe and disproportionate clinical deficits in the setting of a small infarct volume up to 24 hours from time last known well (TLKW). Similarly, the DEFUSE-3 trial9 (6–16 hours) established the presence of mismatch using perfusion imaging (minimally hypoperfused tissue in the presence of a large volume of tissue with prolonged mean transit time). Consequently, the Americal Heart Association/American Stroke Association (AHA/ASA) thrombectomy guidelines recommend advanced imaging-based patient selection beyond 6 hours of TLKW.10
Increasingly, it has been recognized that there is variability in neuronal loss, suggesting that the prevalence of LVO strokes with mismatch may similarly vary with time and differential stroke pathophysiology.11 Simplifying patient selection for thrombectomy has the potential to reduce time to treatment, increase EVT eligibility, and facilitate adoption in resource-limited environments. This requires a better understanding of the characteristics of mismatch. In this study we aim to determine the overall prevalence of CCM in anterior circulation LVO strokes and study the impact of time and ASPECTS on the likelihood of mismatch.
After IRB approval we performed a retrospective analysis of prospectively screened consecutive acute ischemic strokes presenting to a comprehensive stroke center between November 2014 and February 2017. Strokes with anterior circulation large vessel occlusion (LVO internal carotid artery (ICA) and/or middle cerebral artery M1 (MCA-M1)) with available advanced imaging (CT perfusion or MRI) were included in the study. Administration of intravenous tissue plasminogen activator (IV-tPA) was per AHA criteria1 and EVT was offered based on provider discretion. Treated and untreated LVO strokes were included. Demographic (age, sex), clinical (National Institutes of Health Stroke Scale (NIHSS), TLKW, IV-tPA, EVT), and radiological (ischemic core volume measurement based on relative cerebral blood flow on CT perfusion (rCBF <30%) or diffusion-weighted MRI calculated using RAPID Version 4.9 (IschemaView) and ASPECTS on CT head determined using eASPECTS Version 7.1 (Brainomix)) information were collected. CCM was defined using criteria described in the DAWN trial (DAWN-CCM): NIHSS ≥10 and core infarct <31 mL (age <80) or NIHSS ≥20 and core infarct <51 mL (age <80), or NIHSS ≥10 and core infarct <21 mL (age ≥80). Prevalence of DAWN-CCM across time and ASPECTS groups was calculated and statistically tested using the χ2 test and logistic regression analyses. Statistical analyses were performed using SPSS 23 (IBM, Armonk, New York, USA).
A total of 185 patients with rCBF/MRI were included in the study for mismatch and time analysis, and 116 of them with CT head within 120 min of rCBF/MRI were analyzed for the relationship between mismatch, time, and ASPECTS. Mean age was 70.6±15 years and 40% (n=73) were men. Median NIHSS score was 17 (range 12–21), mean TLKW to core imaging was 9±5.8 hours, and 30% (n=55) of patients had an ICA occlusion. Mean and median ischemic core volumes were 50±69 mL and 17 (0–82) mL, respectively. Rates of IV-tPA and EVT were 23% (n=43) and 48% (n=89), respectively. One hundred and ten patients (59%) underwent core imaging at 6–24 hours of TLKW, and 85% (n=157) had an NIHSS score ≥10. Stroke onset was witnessed in 56% of patients (n=104).
Mismatch over time in NIHSS ≥10 (n=157)
In the 0–24 hour window, DAWN-CCM was present in 53% of patients (n=83). Rates of DAWN-CCM ranged from 63% (n=17) in the first 3 hours to 25% (n=1) at 21–24 hours (p=0.03). Assuming linearity, the prevalence of mismatch reduced 1.6% for every 1 hour increase in TLKW to imaging (table 1A, figure 1).
Mismatch across ASPECTS groups in NIHSS ≥10 (n=98)
In the 0–24 hour window, rates of DAWN-CCM by ASPECTS group were: ASPECTS 9–10: 77% (n=34), ASPECTS 6–8: 65% (n=20), and ASPECTS 0–5: 13% (n=3) (p<0.01). The prevalence of mismatch reduced 6.4% for every 1 point decrease in ASPECTS (table 1B).
Relationship between mismatch, time, and ASPECTS
Mismatch over ASPECTS in different time epochs
The prevalence of mismatch diminishes with decreasing ASPECTS across different time intervals. In the 0–6 hour window the prevalence of mismatch reduced from 65% in the ASPECTS 9–10 group to 13% in the ASPECTS 0–5 group (p=0.03). Similar reductions were observed across 6–12 hour and 12–24 hour time windows (table 1C, figure 2).
Mismatch over time in different ASPECTS groups
In the ASPECTS 9–10 group the prevalence of mismatch across 24 hours was 77%, and over time ranged from 65% in the 0–6 hour window to 81% in the 12–24 hour window (p=0.17). Similarly, in the ASPECTS 6–8 and ASPECTS 0–5 groups the prevalence of mismatch across 24 hours was 65% and 13%, respectively, and remained constant over time (ASPECTS 6–8 group: p=1, ASPECTS 0–5 group: p=1) (figure 2).
In multivariate logistic regression analysis including time and ASPECTS, higher ASPECTS score was an independent predictor of the presence of DAWN-CCM (OR 1.4, 95% CI 1.1 to 1.7; p<0.001) and time was not a predictor of DAWN-CCM (OR 0.99, 95% CI 0.93 to 1.07; p=0.96). In the entire cohort, a higher ASPECTS score was an independent predictor of DAWN-CCM (OR 1.6 95% CI 1.3-2.2, p<0.001) while time of presentation and NIHSS were not.
Mismatch and large core volume across different time epochs and ASPECTS ≥6
Overall, the prevalence of mismatch (56.9% vs 50%, p=0.42) and core volume >70 mL (29.3% vs 34.8%, p=0.49) were comparable in the 0–6 hour and 6–24 hour time windows. In LVO strokes with ASPECTS ≥6, the prevalence of mismatch (64.9% vs 78.9%, p=0.20) and core volume >70 mL (16.2% vs 10.5%, p=0.51) were comparable in the 0–6 hour and 6–24 hour time windows. In LVO stroke with ASPECTS ≥6 and available perfusion imaging, the prevalence of DEFUSE-3 trial-defined mismatch eligible patients was 69.2% in patients presenting 0–6 hours after symptoms onset versus 73.9% in patients presenting beyond 6 hours of symptoms onset (p=0.63).
Reasons for lack of DAWN-CCM
Seventy-four patients with NIHSS ≥10 in the 0–24 hour window did not have DAWN-CCM, 15% due to inadequate mismatch and 85% due to infarct volume ≥51 mL.
The main finding of this study is that CCM is present in 57%, 50%, and 53% of LVO strokes in the 0–6, 6–24, and 0–24 hour time windows, respectively. In the 0–24 hour time window, the prevalence of CCM reduces with increasing time (63% at 0–3 hours compared with 25% at 21–24 hours), translating to a mean rate of decay of 1.6%/hour. Additionally, the prevalence of CCM declines from 77% in the ASPECTS 9–10 group to 13% in the ASPECTS 0–5 group, translating to a mean rate of decay of 6.4%/ASPECTS point. However, among specific ASPECTS groups (9–10, 6–8, 0–5), the prevalence of mismatch does not diminish over time, and ASPECTS and not time independently predicts the presence of CCM.
To our knowledge, this is the first study systematically defining the prevalence of mismatch over time and ASPECTS among consecutive treated and untreated LVO strokes. The concept of CCM was first described by Davalos et al as clinical-DWI mismatch (NIHSS ≥8 and core ≤25 mL). They reported a prevalence of 52% within 12 hours of stroke onset, similar to our study.12 Previously, other perfusion studies have reported a prevalence of perfusion mismatch ranging from 50% to 70% in the early time window following symptoms onset.13 14 A recent study reported the existence of DEFUSE-3 trial-defined perfusion mismatch of approximately 20% beyond 24 hours of last known well.15 This is similar to the prevalence of mismatch in our study: 25% between 21 and 24 hours. Also, we have previously reported that CCM exists well beyond 24 hours and up to as long as 6 days in a minority of patients.16
The need for advanced imaging to select patients for neurothrombectomy in the late window to measure core volume and identify the presence of mismatch is a matter of debate among stroke experts. In addition, AHA/ASA guidelines necessitate identification of mismatch beyond 6 hours according to late window trial criteria. It is widely believed that advanced imaging is justified beyond 6 hours due to the increased likelihood of a large infarct burden and the decreased prevalence of mismatch. In this study we show that the proportion of patients with CCM gradually declines with increasing time. For the first 9 hours after stroke onset the prevalence of CCM is >50% and then gradually declines to 40% for the next 9 hours, finally reaching the lowest prevalence of 25% at 21–24 hours. However, the proportion of patients with mismatch (57% vs 50%, p=0.42) and core volume >70 mL (29% vs 35%, p=0.49) is comparable between the 0–6 and 6–24 hour time windows. Our findings suggest that, in the early time window, ASPECTS cut-off (≥6) may also be used for the late time window. Immediately after the onset of ICA/MCA-M1 occlusion, a large territory of MCA is hypoperfused to varying degrees depending on collateral blood flow. This causes variable infarct growth rate,11 leading to distinct phenotypes of infarct growth called slow and fast progressors (estimated to be approximately 55% and 25% in the LVO population).17 Hence, the combination of the prevalence of slow/fast progressors and the time from stroke onset together contribute to the prevalence of mismatch.
A novel analysis of our study includes the relationship between mismatch (adjudicated using advanced imaging) and ASPECTS (adjudicated using CT head), with and without the impact of time. In the 0–24 hour window the mean rate of mismatch decay per decrease in ASPECTS point was 6.4% (ASPECTS 9–10 group 77%, ASPECTS 6–8 group 65%, ASPECTS 0–5 group 13%; p<0.01). When including time and ASPECTS in our analysis, we found that patients with ASPECTS 9–10 have a high proportion of mismatch (65–80%) compared with a moderate proportion in those with ASPECTS 6–8 (60–65%) and a low proportion in those with ASPECTS 0–5 (10–20%). Importantly, among individual ASPECTS groups (9–10 vs 6–8 vs 0–5) the prevalence of mismatch did not decline over time. A slight increase in the prevalence of mismatch is observed in the 6–12 hour window compared with the 0–6 hour window, irrespective of the ASPECTS group. One possible explanation could be the relatively high prevalence of fast progressors in the 0–6 hour time window (specifically 3–4.5 hours), as previously reported by our group.17 We also report that ASPECTS score and not time is an independent predictor of the presence of mismatch.
A secondary analysis of the DAWN trial suggests that ASPECTS modifies the treatment effect in the trial.18 Also, early window thrombectomy trials selected patients without accurate delineation of mismatch, suggesting that a certain proportion of patients with less than adequate or no CCM were included in those trials and yet did not reduce treatment effect. Our findings provide a basis for further investigation as to what level of expected proportion of mismatch patients is required to necessitate advanced imaging for patient selection, and whether CT head alone is adequate. We report that the prevalence of mismatch (65% vs 79%, p=0.2) and core volume >70 mL (16% vs 11%, p=0.56) in LVO strokes with ASPECTS ≥6 is comparable in the early and late time windows, questioning the utility for advanced imaging in the beyond 6 hour window. Similarly, other studies have shown that ASPECTS-based late window patient selection yields equivalent outcomes compared with early window patient selection19 and compared with selection by advanced imaging.20 The current CCM definition is relatively strict, with the most common reason (85%) for CCM ineligibility being an infarct volume ≥51 mL (ASPECTS <6). Data from large core thrombectomy trials will provide more information on this population.
Our study has several strengths, including a robust sample of 185 consecutive treated and untreated patients with LVO stroke with automated infarct volume analysis. The study is limited by its retrospective design and potential referral bias by enriching for slow progressors (more mismatch) in the late window due to LVO stroke transfers. However, such bias is minimal because in our hub and spoke network it is routine practice to transfer all LVO strokes for advanced neurological monitoring. A limitation of our study is that, due to lack of perfusion imaging in all patients, we do not address the DEFUSE-3 trial-defined target mismatch criteria which include patients with NIHSS score 6–9 and ischemic core volume up to 70 mL. Additional weaknesses include the fact that the true time of stroke onset was unavailable for unwitnessed and wake-up strokes.
CCM is present in 57% and 50% of LVO strokes with NIHSS score ≥10 in the 0–6 hour and 6–24 hour time windows, respectively. The prevalence of CCM gradually declines with increasing time (1.6%/hour) and decreasing ASPECTS (6.4%/point). However, among ASPECTS groups, the prevalence of CCM does not decline over time. These data support the use of an ASPECTS-based paradigm for late window patient selection.
Twitter @shashvatdesaiMD, @ashupjadhav
Contributors Conception and design: SMD, APJ. Acquisition of data: SMD. Analysis and interpretation of data: All authors. Drafting the article: SMD, APJ. Critically revising the article: All authors. Administrative/technical/material support: All authors. Study supervision: APJ.
Funding Research reported in this publication was supported by the National Institute of Neurological Disorders and Stroke of the National Institutes of Health under Award Number U10NS086489 (NIH Stroke Trials Network-Regional Coordinating Center University of Pittsburgh).
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
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.