Exploring ischemic core growth rate and endovascular therapy benefit in large core patients

Abstract

After stroke onset, ischemic brain tissue will progress to infarction unless blood flow is restored. Core growth rate measures the infarction speed from stroke onset. This multicenter cohort study aimed to explore whether core growth rate influences benefit from the reperfusion treatment of endovascular thrombectomy in large ischemic core stroke patients. It identified 134 patients with large core volume >70 mL assessed on brain perfusion image within 9 hours of stroke onset. Of 134 patients, 71 received endovascular thrombectomy and 63 did not receive the treatment. Overall, poor outcomes were frequent, with 3-month severed disability or death rate at 56% in treatment group and 68% in no treatment group (p = 0.156). Patients were then stratified by core growth rate. For patients with ‘ultrafast core growth’ of >70 mL/hour, rates of poor outcome were especially high in patients without endovascular thrombectomy (n = 13/14, 93%) and relatively lower in patients received the treatment (n = 12/20, 60%, p = 0.033). In contrast, for patients with core growth rate <70 mL/hour, there was not a large difference in poor outcomes between patients with and without the treatment (55% vs. 61%, p = 0.522). Therefore, patients with ‘ultrafast core growth’ might stand to benefit the most from endovascular treatment.

Introduction

For patient with acute ischemic stroke caused by large vessel occlusion, perfusion imaging performed with computed tomography (CT) or magnetic resonance imaging (MRI) has been extensively used to identify cerebral areas of reversible (penumbra) and irreversible ischemia (core). ¹ The volume of ischemic core is considered one of the most important influences upon patient response to acute reperfusion treatment, including intravenous thrombolysis and endovascular thrombectomy (EVT). ² According to current stroke management guideline, ³ intravenous thrombolysis is not recommended beyond 4.5 hours from witnessed stroke onset, whereas EVT is recommended up to 24 hour in patients selected by perfusion imaging. The guideline is based on the previous clinical trials that incorporated ischemic core volume in their selection criteria as a means of specifically targeting patients who would have the greatest response to successful reperfusion.^{4 –7} Patients with large ischemic cores (typically defined as >70 mL) were excluded from those trials. Therefore, current stroke guideline does not recommend EVT in patients with large core. ³

New evidence indicates that patients with large core may still benefit from EVT. ⁸ Recently, three randomized controlled trials reported that patients with large cerebral infarctions resulted in better functional outcomes with endovascular therapy than with medical care alone (RESCUE-Japan LIMIT, ⁹ SELECT 2, ¹⁰ ANGLE-ASPECT ¹¹ ) However, the three trials also reported that EVT was associated with more intracranial hemorrhage or vascular complications. A secondary analysis of the RESCUE-Japan LIMIT trial suggested that moderately large cores appeared to benefit from EVT with less bleeding risk. ¹² In the trial, large core was not volumetrically quantified but defined by a score of ASPECTS 3–5. The HERMES collaboration pooled data from 6 trials, and it showed a benefit of EVT over medical care alone for patients with a CT perfusion (CTP) or MRI core volume of ≥70 mL. ² An observational cohort study also reported a positive association of EVT with 3-month outcome in patients with a baseline CT perfusion ischemic core volume 70–100 mL. ¹³ Moreover, in patients with large core, a recent prospective cohort study suggested that large penumbra was a predictor of good response to EVT compared to medical care alone. ¹⁴

In addition to core volume, core growth rate has been proposed as a modifier of the therapeutic effect of EVT. Albers postulated that slow core growth might be responsible for the greater benefits of EVT treatment (vs. No EVT) seen in late time window clinical trials, compared to early time window trials. ¹⁵ This has fueled a widely held belief that patients with ‘fast’ core growth rate are less likely to benefit from EVT treatment. However, the rates of ‘fast’ core growth in past studies have been described as >5–15 mL/hour.^{15 –17} Using the same linear infarct core growth model as in these studies, we recently observed faster core growth rates (>25 mL/hour) than described previously in early window patients with target mismatch and baseline core volume <70 mL. ¹⁸ Indeed, we found that these faster core growth patients actually had greater absolute benefit from EVT treatment compared to thrombolysis only treated patients. However, as with the previous studies, this was in relatively small core patients with a favorable imaging profile. There is virtually no data on core growth rate from patients with truly large cores. We hypothesized that large core patients would have even faster core growth rates than previously described, and, that this might have an influence on outcome and effect of EVT. Therefore, the aims of the current study were to explore in large core (>70 mL) patients: (1) the benefits of EVT; (2) whether the rate of core growth had an influence on these treatment benefits.

Material and methods

Patients

This retrospective cohort study used data from the International Stroke Perfusion Imaging REgistry (INSPIRE). Data in this study was collected between August 2011 and August 2019 and from 22 sites (9 Australian, 12 Chinese, and 1 Canadian). It identified anterior circulation stroke patients who underwent complete baseline multimodal CT, including non-contrast CT (NCCT), CTP, and CT angiography (CTA) within 9 hours of symptom onset. The following inclusion criteria were further applied: acute core volume >70 mL on CTP, large vessel occlusion (occlusion of internal carotid artery, middle cerebral artery M1 segment, or M2 segment), witnessed stroke with clearly established time of stroke onset, and a complete 3 month modified Rankin Scale score.

Ethics approvals and patient consents

The study had central ethical approval by Hunter New England Human Research Ethics Committee (HNEHREC Reference No: 11/08/17/4.01). Written informed consent was obtained for each patient for their data to be used in this study as part of INSPIRE research. The INSPIRE and all associated analyses are conducted in accordance with the ethical standards of the Helsinki Declaration of 1975 (and as revised in 1983).

EVT vs. No EVT group

Treatment decisions for endovascular thrombectomy were predominantly determined by respective national guidelines at the time when patients were recruited to the INSPIRE registry. Major changes to INSPIRE sites national guidelines occurred after the landmark EVT trials in 2015.^4,5 This resulted in two groups of large vessel occlusion patients within INSPIRE, an EVT group predominantly from 2015 onwards, and a historical, predominantly no EVT control group from 2012–2015. Intravenous (IV) thrombolysis was administered at the discretion of local clinicians, again following local guidelines.

Imaging acquisition and post-processing

Baseline CT imaging included brain NCCT, CTP, and CTA, obtained with different CT scanners (64, 128, 256, or 320 detectors, with Toshiba [Tokyo, Japan], Siemens [Munich, Germany], or GE [Cleveland, OH] scanners). Axial coverage ranged from 80 to 160 mm. Follow-up imaging included MRI and/or multimodal CT 24–72 hours after the baseline imaging, to verify recanalization status and assess for hemorrhagic transformation. MRI was performed on 1.5 T or 3.0 T scanner (Siemens Avanto or Verio), including diffusion weighted imaging and angiography. Follow-up multimodal CT was performed when MRI was not available or if the patient had contraindications to MRI. All imaging data were processed in a central neuroimaging lab.

The CTP raw data were centrally processed by commercial software MIStar (Apollo Medical Imaging Technology, Melbourne, VIC, Australia).^19,20 CTP parameters were generated by applying the mathematical algorithm of singular value decomposition with delay and dispersion correction. ²¹ The following four CTP parameters were generated: cerebral blood flow, cerebral blood volume, mean transit time, and delay time. The penumbra and core volume were measured on acute CTP with dual threshold setting: delay time at the threshold of 3 seconds for total ischemic lesion volume and cerebral blood flow at the threshold setting of 30% for acute core volume. ²² The mismatch ratio was defined as the total ischemic lesion volume divided by the acute core volume. Hypoperfusion intensity ratio was defined as the ratio of perfusion volume with delay time >6 seconds over delay time >2 seconds. ²⁰ High hypoperfusion intensity ratio indicated poor collateral status. ²⁰

Calculation of ischemic core growth rate

Clearly established time from stroke onset to CTP was required for this study. The core growth rate was calculated as the acute core volume on CTP divided by the time from stroke onset to CTP. We recently described and validated a method for estimating core growth based on stroke onset time and core volume on acute CTP. ²³ Essentially, it assumes that core volume was zero prior to symptom onset and that core volume grew in a linear pattern in the first 24 hours of stroke onset according to a previous study. ²⁴ In summary, this study calculated core growth by acute CTP with the following equation (Supplement Fig. I): Core growth rate = Acute core volume on CTP/Time from stroke onset to CTP

Patient outcomes

Given this was a large core cohort, we chose a primary outcome of 3-month modified Rankin Scale (mRS) of 5–6 at 3 months (defined as poor outcome). The secondary outcomes included 3-month death rate, good clinical outcome defined by mRS of 0–2, and fair clinical outcome defined by mRS of 0–3. The 3-month mRS was performed by trained personnel in each site.

The secondary outcomes also include hemorrhagic outcomes centrally assessed on follow-up MRI or CT. These were: 1) Hemorrhagic transformation (HT); 2) parenchymal hematoma type 2 (PH2); 3) symptomatic intracranial hemorrhage (sICH) based on European Cooperative Acute Stroke Study II criteria. ²⁵

Recanalization status was graded centrally at the INSPIRE core imaging laboratory (AB, MP, LL, CC) on digital subtraction angiography immediately after endovascular thrombectomy in the EVT group. A Modified Thrombolysis in Cerebral Infarction score of 2b-3 was classified as successful recanalization.

Statistical analysis

For patient characteristics and outcomes, continuous data were summarized by median and inter-quartile range (IQR). Between group differences were tested using Wilcoxon rank sum test. Categorical variables were summarized by proportion and compared with Pearson Chi-square test. Patients without 3-month follow-up clinical data were excluded in the statistical analyses.

Univariate logistic regression was performed to assess the predictive power of EVT treatment (vs. No EVT) on clinical outcomes. Multivariate logistic regression was then performed to adjust for different patient characteristics between the EVT and No EVT group.

Spearman’s correlation coefficient was performed to assess the relationship of core growth rate and core volume, as well as the relationship of core growth rate and onset to CTP scan time. Multivariate logistic regression was performed to assess the interaction between core growth rate and EVT treatment in predicting patient outcomes. The interaction of core volume and EVT and the interaction of time to CTP and EVT were assessed by multivariate logistic regression as well.

In the multivariate logistic regression, core growth rate was treated as a continuous variable. Then, core growth rate was divided into quartiles, and outcomes of EVT vs. No EVT patients were plotted across the quartiles. From the plot, a threshold of core growth was derived to dichotomize core growth rate into two groups: a newly described ‘ultrafast’ core growth vs. fast core growth. This dichotomy and new definition was used because all patients had core growth rates that would previously have been classified at least as fast growth (>5–25 mL/hour).^{15 –18} Within each core growth group, the predictive power of EVT (vs. No EVT) on patient outcomes was assessed by logistic regression respectively. Successful recanalization rate was compared between the ultrafast core growth vs. fast core growth group.

All statistical analysis was done using STATA 13.0 (Stata Corp, College Station, Texas, USA), with confidence interval (CI) set at 95% and significance level set at 0.05.

Results

Patients

Amongst 1879 patients in the INSPIRE registry, there were 155 patients with core volume >70 mL. All 155 patients had large vessel occlusion. Of the 155 patients, 14 were excluded for unknown stroke onset time and 7 patients were excluded for no 3-month follow-up outcome. Thus, 134 patients were included in final analyses.

Of the 134 patients, 71 received EVT and 63 did not. For the No EVT group, 83% (52/63) patients were enrolled before 2015 when endovascular thrombectomy was not routinely performed. For the EVT group, 96% (68/71) of EVT patients were enrolled after 2015 when endovascular thrombectomy was routinely performed, and 87% (61 out of 70) had successful recanalization from endovascular thrombectomy. Of the 134 patients, 131 patients received CTP scan within 6 hours of stroke onset, and only 3 patients had the scan between 6 and 9 hours after stroke onset. Fifty-two of the 134 patients did not receive IV thrombolysis.

Patient characteristics are summarized in Table 1. The median acute core volume of all 134 patients was 103.5 mL (IQR of 89–132), the median time from stroke onset to CTP was 2.36 hours (IQR of 1.68–3.45), and the median core growth rate was 44.0 mL/hour (IQR of 30.2–70.7). No difference was observed between the EVT and No EVT group in terms of acute core volume (p = 0.904), time to CTP (p = 0.784), or core growth rate (p = 0.913). The two groups also showed no difference in the hypoperfusion intensity ratio (0.43 vs. 0.42, p = 0.352). However, acute penumbra volume was significantly larger in the EVT group compared to the No EVT group (91 vs. 74 mL, p = 0.029). The EVT group had a lower IV thrombolysis rate compared to the No EVT group (51% vs. 73%, p = 0.008). In addition, the EVT group had a higher proportion of Chinese patients (79% vs. 43%, p < 0.001).

Table 1.

Patient characteristics of EVT and No EVT patients.

	No EVT (n = 63)	EVT (n = 71)	P
Chinese patients	43% (27/63)	79% (57/71)	<0.001†
Gender (male)	58% (37/63)	63% (45/71)	0.581
Age, Median (IQR)	75 (64–81)	71 (63–79)	0.277
Acute NIHSS, Median (IQR)	18 (16–20)	18 (14–22)	0.825
Systolic blood pressure (mmHg), Median (IQR)	155 (134–175)	155 (132–76)	0.908
Diastolic blood pressure (mmHg), Median (IQR)	86 (74, 100)	85 (74, 101)	0.872
Blood glucose (mmol/l), Median (IQR)	8.00 (6.10–9.80)	7.25 (6.50–9.30)	0.633
Smoking	31% (19/62)	35% (24/68)	0.574
Hypertension	71% (45/63)	66% (47/71)	0.515
Diabetes	21% (13/63)	13% (9/71)	0.214
Hyperlipidemia	18% (11/60)	14% (9/66)	0.471
Stroke history	14% (9/63)	23% (16/70)	0.206
Atrial fibrillation history	49% (31/63)	60% (42/70)	0.212
Large artery atherosclerosis	19% (12/63)	24% (17/70)	0.465
ICA occlusion	54% (34/63)	48% (34/71)	0.482
Acute ischemic lesion volume (mL), Median (IQR)	185 (145–236)	207 (162–249)	0.116
Acute core volume (mL), Median (IQR)	113 (85–132)	95.3 (90–134)	0.736
Acute penumbra volume (mL), Median (IQR)	74 (49–105)	91 (66–129.2)	0.029†
Mismatch ratio, Median (IQR)	1.65 (1.38–1.97)	1.78 (1.51–2.26)	0.056
Hypoperfusion intensity ratio, Median (IQR)	0.43 (0.26, 0.53)	0.42 (0.31, 0.55)	0.352
Core growth rate (mL/hour), Median (IQR)	43.3 (31.9–66.0)	44.8 (28.1–79.9)	0.913
Time from onset to CTP (hour), Median (IQR)	2.33 (1.71–3.20)	2.36 (1.67–3.82)	0.784
Intravenous thrombolysis	73% (46/63)	51% (36/71)	0.008†
Recanalization rate	31% (16/55)	87% (61/70)	<0.001†
Poor clinical outcome	68% (43/63)	56% (40/71)	0.156
3-month death	49% (31/63)	41% (29/71)	0.331
Good clinical outcome	10% (6/63)	20% (14/71)	0.098
Fair clinical outcome	19% (12/63)	30% (21/71)	0.158
HT	25% (15/59)	52% (35/68)	0.003†
PH2	7% (4/59)	21% (14/68)	0.026†
sICH	17% (10/59)	28% (19/68)	0.141

EVT: endovascular thrombectomy; IQR: interquartile range; NIHSS: National Institutes of Health Stroke Scale; ICA: internal carotid artery; HT: hemorrhagic transformation; PH2: parenchymal hematoma type 2; sICH: symptomatic intracranial hemorrhage; Good, fair, and poor clinical outcome are defined by 3-month modified Rankin Score of 0–2, 0–3 and 5–6 respectively.

^†p < 0.05.

EVT vs no EVT outcomes

The overall rates of poor clinical outcome and mortality were high in this study (62% and 37% respectively, Table 1). The EVT group tended to have a lower rate of 3-month mRS 5–6 compared to the no EVT group (56% vs. 68%, p = 0.156) and lower mortality rate (41% vs. 49%, p = 0.331). Regarding good clinical outcome, the EVT group had a trend towards increased 3-month mRS 0–2 compared to the No EVT group (20% vs. 10%, p = 0098, Table 1).

Compared to patients who did not receive EVT, the EVT-treated patients had an increased rate of any hemorrhagic transformation (HT, 52% vs. 25%, p = 0.003), as well as PH2 (21% vs. 7%, p = 0.026, Table 1). After adjusting for acute penumbra volume, IV thrombolysis rate, and proportion of Chinese patients in multivariable regression models, the EVT group still had an increased rate of HT (adjusted OR 4.32 [1.78–10.55], p = 0.001) and PH2 (adjusted OR 4.29 [1.14–16.23], p = 0.032). Regarding symptomatic ICH, the EVT group showed a trend towards higher rate than the No EVT group (28% vs. 17%, p = 0.141).

Core growth rate and EVT interaction

This cohort had extremely fast core growth overall, with a median core growth rate of 44.0 mL/hour (inter-quartile range of 30.2–70.7 mL/hour). There was no difference in core growth rate between the EVT and no EVT groups (median core growth rate 44.8 mL/hour vs. 43.3 mL/hour, p = 0.913). However, the treatment effect of EVT was mediated by core growth rate (Table 2), showing a beneficial interaction of EVT treatment and faster core growth rate on lower 3-month poor outcome (p = 0.043), as well as 3-month mortality (p = 0.023), and HT (p = 0.024). With increasing core growth rate, EVT (compared to No EVT treatment) decreased the odds of having poor clinical outcome (interaction OR = 0.97 [0.94, 0.99]), death (interaction OR = 0.97 [0.94, 0.99]), and HT (interaction OR = 0.97 [0.95, 0.99]). A beneficial interaction trend was also observed between EVT treatment and core growth for sICH [interaction OR = 0.98 [0.95, 1.00] p = 0.065]. After adjusting for acute penumbra volume, IV thrombolysis rate, and Chinese patient rate, the interaction of core growth and EVT remained significant in predicting poor outcome (adjusted interaction OR = 0.97 [0.94, 0.99], p = 0.032) and 3-month mortality (adjusted interaction OR = 0.97 [0.94, 0.99], p = 0.015, Table 2).

Table 2.

The interaction of EVT treatment and core growth in predicting clinical outcomes.

	Interaction OR	P	Adjusted interaction OR	P
Poor clinical outcome	0.97 [0.94, 1.00]	0.043†	0.97 [0.94, 0.99]	0.032†
3-month death	0.97 [0.94, 1.00]	0.023†	0.97 [0.94, 0.99]	0.015†
Good clinical outcome	1.01 [0.98, 1.05]	0.530	1.01 [0.98, 1.05]	0.524
Fair clinical outcome	1.01 [0.98, 1.05]	0.367	1.02 [0.99, 1.05]	0.309
HT	0.97 [0.95, 0.99]	0.024†	0.98 [0.96, 1.00]	0.061
PH2	0.99 [0.97, 1.01]	0.477	0.99 [0.97, 1.01]	0.405
sICH	0.98 [0.95, 1.00]	0.065	0.98 [0.96, 1.00]	0.117

Adjusted interaction OR is derived from adjusting for acute penumbra volume, IV thrombolysis rate, and Chinese patient rate.

^†p < 0.05.

Table 3 shows that the beneficial interaction of EVT treatment in patients with faster core growth was predominantly due to worse outcomes in the no EVT group with increasing core growth rate. Multiple regression, adjusting for acute penumbra volume, IV thrombolysis rate, and Chinese patient rate, showed that within the No EVT group, increasing core growth rate resulted in higher rate of poor clinical outcome (adjusted OR = 1.03 [1.00, 1.06], p = 0.027) and mortality (adjusted OR = 1.04 [1.00, 1.07], p = 0.004). An increase in HT (adjusted OR = 1.02 [1.00, 1.04], p = 0.046) and sICH (adjusted OR = 1.03 [1.00, 1.05], p = 0.020) was also seen in the no EVT group with increasing core growth rate. In contrast, increasing core growth rate did not have a significant influence either on 3-month clinical outcomes (p > 0.05) or on hemorrhagic outcomes (p > 0.05) in the EVT group.

Table 3.

Predicting clinical outcomes by core growth in logistic regression models.

	No EVT group (n = 63)
	Crude OR	P	Adjusted OR	P
Poor clinical outcome	1.03 [1.00, 1.05]	0.038†	1.03 [1.00, 1.06]	0.027†
3-month death	1.03 [1.01, 1.06]	0.009†	1.04 [1.00, 1.07]	0.004†
Good clinical outcome	0.99 [0.96, 1.02]	0.540	0.99 [0.95, 1.02]	0.405
Fair clinical outcome	0.98 [0.96, 1.01]	0.233	0.98 [0.95, 1.01]	0.161
HT	1.02 [1.00, 1.04]	0.033†	1.02 [1.00, 1.04]	0.046†
PH2	1.01 [0.99, 1.03]	0.180	1.02 [0.99, 1.04]	0.178
sICH	1.03 [1.00, 1.05]	0.018†	1.03 [1.00, 1.05]	0.020†
	EVT group (n = 71)
	Crude OR	P	Adjusted OR	P
Poor clinical outcome	1.00 [0.99, 1.01]	0.821	1.00 [0.99, 1.01]	0.853
3-month death	1.00 [0.99, 1.01]	0.662	1.00 [0.99, 1.01]	0.583
Good clinical outcome	1.00 [0.99, 1.01]	0.880	1.00 [0.99, 1.01]	0.993
Fair clinical outcome	1.00 [0.99, 1.01]	0.584	1.00 [0.98, 1.01]	0.534
HT	1.00 [0.99, 1.01]	0.421	1.00 [0.99, 1.01]	0.702
PH2	1.00 [0.99, 1.02]	0.330	1.00 [0.99, 1.02]	0.398
sICH	1.00 [0.99, 1.01]	0.505	1.01 [0.99, 1.02]	0.295

^†p < 0.05; Adjusted OR is derived from adjusting for acute penumbra volume, IV thrombolysis rate, and Chinese patient rate.

Core growth rate >70 mL/hour vs. Core growth rate <70 mL/hour

The interaction of EVT treatment and core growth rate on outcomes is further illustrated in Figure 1. For EVT treatment vs. No EVT, the rate of poor outcome was 38% vs. 43% in the first quartile of core growth (p = 0.848), 68% vs 65% in the second quartile of core growth (p = 0.813), 53% vs 64% in the third quartile of core growth (p = 0.447), and 60% vs. 93% in the fourth quartile of core growth (p = 0.033). A similar trend was observed with mortality for EVT vs. No EVT (p = 0.948, 0.492, 0.632, 0.032 respectively across the first to the fourth quartile of core growth rates). Only in the fourth quartile of core growth rate (>70 mL/hour) were poor outcome and 3-month mortality after EVT significantly less than that of the No EVT group. Therefore, we dichotomized core growth rate by the threshold of 70 mL/hour: <70 mL/hour (N = 100) as the fast core growth group, and >70 mL/hour (N = 34) as an ‘ultrafast’ core growth group. In summary, there was a clear increase in poor clinical outcome and death for the no EVT group in the highest core growth quartile (ultrafast growth >70 mL/hour), and this was not seen comparing core growth rate quartiles in the EVT group.

Figure 1.

Poor clinical outcome distribution across core growth quartiles. (a) 3-month poor clinical outcome measured by modified Rankin score of 5–6; (b) 3-month death. Core growth quartile 1: <30 mL/hour (n = 33), 2: 30–44 mL/hour (n = 34), 3: 44–70 mL/hour (n = 33), 4: >70 mL/hour (n = 34); SE refers to standard error.

Outcomes of core growth >70 mL/hour vs. <70 mL/hour are summarized on Table 4, Figure 2 and Figure 3. Regarding poor clinical outcome, there was not a difference between the EVT and no EVT groups in patients with core growth rate <70 mL/hour (p > 0.05). However, in patients with core growth rate >70 mL/hour, EVT substantially reduced the rate of 3-month poor outcome (60% in EVT group vs 93% in No EVT group, p = 0.033), and decreased 3-month mortality rate (50% in EVT group vs. 86% in No EVT group, p = 0.032).

Table 4.

Patient outcomes of EVT vs. no EVT patients across core growth groups.

	Core growth <70 mL/hour (n = 100)
	No EVT (n = 49)	EVT (n = 51)	P
Poor clinical outcome	61% (30/49)	55% (28/51)	0.522
3-month death	39% (19/49)	37% (19/51)	0.876
Good clinical outcome	10% (5/49)	22% (11/51)	0.121
Fair clinical outcome	22% (11/49)	33% (17/34)	0.226
HT	20% (9/46)	52% (26/50)	0.001†
PH2	2% (1/46)	14% (7/50)	0.036†
sICH	11% (5/46)	24% (12/50)	0.092
	Core growth >70 mL/hour (N = 34)
	No EVT (n = 14)	EVT (n = 20)	P
Poor clinical outcome	93% (13/14)	60% (12/20)	0.033†
3-month death	86% (12/14)	50% (10/20)	0.032†
Good clinical outcome	7% (1/14)	15% (3/20)	0.484
Fair clinical outcome	7% (1/14)	20% (4/20)	0.298
HT	46% (6/13)	50% (9/18)	0.833
PH2	23% (3/13)	39% (7/18)	0.353
sICH	38% (5/13)	39% (7/18)	0.981

^†p < 0.05.

Figure 2.

Clinical outcome comparison between core growth <70 mL/hour and >70 mL/hour. (a) 3-month poor clinical outcome; (b) 3-month death; (c) hemorrhagic transformation (HT) and (d) symptomatic Intracranial hemorrhage (sICH).

Figure 3.

The distribution of 3-month modified Rankin Score across core growth groups. (a) core growth <70 mL/hour and (b) core growth >70 mL/hour.

Regarding hemorrhagic outcomes, for patients with core growth rate <70 ml/hour, EVT (compared to no-EVT) hemorrhagic outcomes were increased (52% vs. 20% for HT, p = 0.001; 14% vs. 2% for PH2, p = 0.036). In contrast, for core growth rate >70 mL/hour, EVT did not cause an increase in hemorrhagic outcomes, although overall rates were quite high.

Core growth rate and time to CTP

The core growth rate showed a strong correlation with onset to CTP scan time (correlation coefficient = −0.84, p < 0.001) and a moderate correlation with acute core volume (correlation coefficient = 0.47, p < 0.001). When Core growth rate showed a significant interaction with EVT group (EVT vs. No EVT) in predicting 3-month poor outcome (interaction odds ratio = 0.97 [0.94, 0.99], p = 0.042), acute core volume did not show significant interaction with EVT group in predicting poor outcome (interaction odds ratio = 0.99 [0.97, 1.01], p = 0.277), nor did onset to CTP scan time (interaction odds ratio = 1.16 [0.67, 1.95], p = 0.585, Supplement Fig. II, Supplement Table I).

In addition, hypoperfusion intensity ratio showed no significant interaction with EVT treatment in predicting 3-month poor outcome (interaction odds ratio = 23.9 [0.34, 1659], p = 0.142, Supplement Table I). In this study, the median hypoperfusion intensity ratio was 0.43 (IQR 0.30–0.55), with 75% patients had the ratio >0.3. Adjusting for hypoperfusion intensity ratio did not change the interaction of core growth and EVT in predicting 3-month poor outcome (adjusted interaction OR = 0.97 [0.94, 0.99], p = 0.049). In patients with no EVT treatment, increasing core growth rate still resulted in increased odds of poor clinical outcome after adjusting for hypoperfusion intensity ratio (adjusted OR = 1.03 [1.00, 1.06], p = 0.028).

Discussion

This study reports on a cohort of patients with a large ischemic core and large vessel occlusion who underwent endovascular treatment compared to a predominantly historical control group not undergoing endovascular treatment. We have found that rates of good outcome are low in large core patients, but with trends towards better outcome in patients receiving endovascular treatment. We have also explored, for the first time, the relationship between core growth rate and outcomes in a large ischemic core population. Most notably, the therapeutic benefits seen between EVT and no EVT group was greatest in those with ‘ultrafast’ core growth of >70 mL/hour. Additionally, and reflecting the unique study population, such rapid core growth rates have not previously been described.

The rates of core growth seen in our study are much faster than seen in other reports (which contained mostly patients with smaller cores).^{15 –18} Thus, we coin the phrase ‘ultrafast core growth’. Limited studies have been conducted on core growth so far, with none being focused purely on large core patients. In a previous report, ¹⁶ the median core growth rate was 3–7 mL/hour, and ‘Fast’ core growth was defined by a threshold of >5 mL/hour. A threshold of >10–15 mL/hour was suggested to define fast core growth by Albers et al. ¹⁵ and Sarraj et al. ¹⁷ A recent study by our group, also in small core patients, indicates these rates are not truly ‘fast’ with a higher rate of >25 mL/hour defining the upper quartile of core growth. ¹⁸ Clearly, these rates are not particularly ‘fast’ in patients presenting with large core volumes. Our data suggests >70 mL/hour is truly fast, and, in fact, the vast majority of large core patients have growth rates much faster than 15 mL/hour. The ultrafast core growth might be explained by poor collateral flow. In this study, patients had a high proportion of severe hypoperfusion tissues within the ischemic region, which indicates very poor collateral status. ²⁰ Poor collateral flow has been reported as one of the determining factors for fast core growth. ²³ However, this study indicates that the prediction of fast core growth on patient outcome may be independent of collateral. A recent study suggests poor venous outflow may play an important role. ²⁶ Venous flow presents a novel approach to reflect the dynamics of arterial and venous microcirculation. The in-flow of arterial blood to the brain tissue, together with its outflow into the cerebral veins, modulates cerebral microcirculatory perfusion.^{26 –28} Further studies are required to explore the relationship between core growth rate, collateral flow, venous outflow, and microcirculation.

Our data does suggest that the overall rates of good outcome are rather low with large core patients, and it is quite clear that outcomes are very poor without EVT. Strong trends were seen towards better outcomes overall in the large core group, and it is particularly apparent that the patients with the fastest core growth rates did very poorly without EVT. Our study suggests that patients with large core and ultrafast progression (core growth rate >70 mL/h) have virtually no chance of a favorable outcome without EVT. This is probably due to the late recanalization or no recanalization by medical care alone. Although a high proportion of patients received intravenous thrombolysis as the best medical management in this study, thrombolysis compared to EVT would lead to a lower rate of recanalization rate, and in addition, if recanalization from thrombolysis occurred it might be delayed. According to a previous study, ²⁹ the early recanalization rate was around 20% from IVT, whereas this study showed the early recanalization rate was above 80% for patients with EVT. The delay from intravenous thrombolysis to recanalization would then lead to significant core expansion for patients with ultrafast faster core growth. In comparison, EVT leads to faster recanalization, thus, in patients with fast core growth, core expansion is halted, resulting in less infarct growth and hence improved clinical outcome. In patients with slightly ‘slower’ (<70 mL/hour) core growth rates, they may be less affected by the delayed recanalization, which explains the absolute difference in outcomes between the groups was not quite as strong.

Findings of this study may also be explained by the “no-reflow” phenomenon resulting in futile recanalization of intravenous thrombolysis. The no-reflow refers to the absence of tissue reperfusion despite complete recanalization. ³⁰ No-reflow is believed to reflect abnormal constrictions and vasodynamics of cerebral vessels in previously ischemic areas.^31,32 Previous studies showed a high no-reflow rate (21–40%) for patients receiving intravenous thrombolysis or no reperfusion, and a recent study showed very rare existence of no-reflow with EVT (3%). ³³ The study proposed a hypothesis that no-reflow may exist mainly with longer occlusion times, supported by animal studies indicated that no-reflow was prominent only after 3–4-hour occlusion, and it was absent following 2-hour occlusion. ³³ Patients receiving intravenous thrombolysis, compared to EVT treatment, would have delayed recanalization and thus, much longer occlusion times and higher rate of no-reflow. Moreover, microcirculatory dysfunction with microemboli has been suggested as a common mechanism for the no-reflow phenomenon. The microemboli can originate from debris caused by thrombolysis. ³⁰ This may further explain the poor outcome of large core patients treated with intravenous thrombolysis in this study. The microcirculation dysfunction can also lead to profound blood brain barrier disruption, ³⁴ which may explain the high bleeding rate of large core patients treated with thrombolysis. ³⁵ In this study, the differential effect of ultrafast core growth in the EVT vs No EVT groups on mRS seems at least in part explained by the increased rate of PH2 or sICH in the No EVT group who received medical care alone, which in turn may reflect inappropriate use of thrombolysis in very large core, particularly after 4.5 hours from onset.

Limitations of the study design include the following. First, this is not a randomized controlled study. The No EVT group was a historical control arm for the EVT group, which might explain the relative lack of baseline imbalance between the groups. However, there might be potential selection bias, such as patient selection for EVT based on large penumbra. Although penumbra volume was adjusted in the multivariate regression analysis, residual confounding effect might exist. In addition, this study might not correct for other unmeasured confounders, including differences over time in rates of hemicraniectomy (which were not recorded routinely in this registry). Second, this study is limited to a relatively small sample size. Third, a small number of 5% patients (7/155) were lost to follow up for 3-month outcome and these might have had some effect on the overall outcomes of the study. Nonetheless, our data provides strong impetus for randomized controlled trials of EVT versus standard care in patients with truly large core, and suggests that core growth rate might be a useful measure in which to stratify the treatment allocation.

When applying the findings of this study, the following should be noted. First, the calculation of core growth rate is based on the assumption of linear growth from stroke onset to imaging. This methodology has been validated ^23,24 and applied in previous studies.^{15 –18} However, the linear assumption may require investigation. Second, the calculation of core growth relies on accurate detection of ischemic core on CT perfusion. Cerebral blood flow with the threshold of <30% has been well validated to define the ischemic core. ²² However, it potentially overestimates core volume in patients scanned within 90 min from onset. ³⁶ For those patients, recent studies have suggested that a stricter cerebral blood flow threshold (<20%) provides a better estimate of ischemic core. ³⁷ New CT perfusion protocols with integrating regional perfusion information have the potential to further improve the identification of ischemic core. ³⁸ Third, since core growth rate in this study is based on the qualification of core volume on CT perfusion, findings of this study are not applicable to more qualitative estimates of core on NCCT, such as the ASPECTS score Moreover, to improve the accuracy of core growth calculation, we did not include patients with unknown onset time, so the data cannot be extrapolated to this group. Finally, most patients in this study presented to hospital in the early 6-hour time window (because they were truly ‘fast growers’). Thus, caution should be interpreted to applying these results in the extended window beyond 6 hours. Findings of this study support the use of CTP to select patients for EVT in the early time window, although current stroke guideline does not recommend the routine use of CTP in the early 6-hour window because most early-window trials did not use CTP for patient selection.

Conclusions

In conclusion, patients with a large ischemic core appear to benefit from endovascular thrombectomy, and, paradoxically, patients with the fastest core growth rates might stand to benefit the most, due to the dismal outcomes seen with standard care, including thrombolysis.

Name	Sites
Ferdinand Miteff, MD	John Hunter Hospital, Newcastle, Australia
Philip M. C. Choi, MD	Eastern Health, Box Hill, Melbourne, Australia
Timothy Kleining, PhD, MD	Royal Adelaide Hospital, Adelaide, Australia
Billy O’Brien, MD	Gosford Hospital, Gosford, Australia
Jianhong Yang, MD	Ningbo First Hospital, Ningbo, China
Congguo Yin, MD	Hangzhou First Hospital, Zhejiang University School of Medicine, Hangzhou, China
Peng Wang, MD	Zhejiang Provincial People’s Hospital, Hangzhou, China
Yu Geng, MD	Zhejiang Provincial People’s Hospital, Hangzhou, China
Weiwen Qiu, MD	Lishui People’s Hospital, Lishui, China
Qi Fang, MD	The First Affiliated Hospital of Soochow University, Soochow, China
Yi Sui, PhD, MD	The First People’s Hospital of Shenyang, Shenyang, China
Wenhuo Chen, MD	Zhangzhou Municipal Hospital, Zhangzhou, China
Gang Li, PhD, MD	Shanghai East Hospital, Tongji University School of Medicine, Shanghai, China