Persistent Target Mismatch Profile >24-hours after Stroke Onset in DEFUSE 3

Abstract

Background and Purpose

Efficacy of endovascular thrombectomy has been demonstrated up to 24h after stroke onset in patients selected with perfusion imaging. We hypothesized that a persistent favorable perfusion profile exists in some patients beyond 24h from onset and can be predicted by a lower baseline hypoperfusion intensity ratio (HIR), which indicates favorable collaterals.

Methods

We identified control arm patients from the DEFUSE 3 trial with a DWI and perfusion MRI performed 24h following randomization and compared imaging and clinical variables between patients with persistent mismatch versus patients who no longer had a mismatch 24h after randomization.

Results

18% of the control arm patients had a persistent favorable profile >38h after last known well time. These patients had similar baseline DWI and Tmax >6s volumes as patients whose initially favorable perfusion profile became unfavorable (DWI 7mL vs. 17mL, p=0.17, Tmax>6s 98 vs. 110, p=0.48) yet experienced less infarct growth (15mL vs 59mL, p<0.001) and had 3-fold smaller infarct volumes (15mL vs 59mL, p<0.001) 24h after randomization. Patients with a persistent favorable perfusion profile had a significantly lower HIR on baseline imaging (0.2 vs 0.4, p<0.01). Favorable clinical outcome at 90 days occurred in only 10% of the persistent mismatch patients.

Conclusions

About 20% of patients with an MCA or ICA occlusion who present in an extended time window and are not treated with thrombectomy have a persistent mismatch for at least an additional 24 hours. These patients have a favorable HIR at presentation, may experience delayed infarct expansion, and have poor clinical outcomes. Clinical trials are needed to determine if patients with a favorable perfusion profile benefit from reperfusion beyond 24 h.

Introduction:

The DAWN and DEFUSE 3 trials demonstrated efficacy of endovascular thrombectomy in patients selected using perfusion imaging or MR Diffusion Weighted Imaging (DWI) and treated up to 24-hours after stroke onset.^{1, 2} We estimated the incidence of a persistent favorable perfusion profile from the control arm (standard medical therapy only) of DEFUSE 3, which typically obtained follow-up imaging studies about 40 hours after the patient was last known to be normal.

We hypothesized that patients with a persistent favorable profile would have a lower hypoperfusion intensity ratio (HIR), which indicates favorable collaterals, on their baseline imaging studies. The HIR reflects the percentage of the time to peak of the residue function >6 seconds (Tmax>6s) perfusion lesion that has severe contrast arrival delays (>10 seconds). Finally, we looked for an association between clinical characteristics and outcomes and the presence of a persistent favorable profile.

Methods:

DEFUSE 3 data have been made publicly available and can be accessed at https://clinicaltrials.gov/ct2/show/results/NCT02586415?term=DEFUSE+3&rank=1. All patients in DEFUSE 3 were required to have a favorable perfusion profile (target mismatch) and a middle cerebral artery (MCA) or internal carotid artery (ICA) occlusion at the time of enrollment. The target mismatch profile was defined as an ischemic core volume of less than 70 ml, a ratio of the volume of hypoperfused tissue (Tmax>6 seconds) to infarct core volume of 1.8 or more, and a mismatch volume between core and hypoperfusion of 15 ml or more. The volume of the ischemic core and hypoperfused tissue from CT perfusion or MRI diffusion and perfusion scans was automatically calculated with RAPID software (iSchemaView, Menlo Park, CA), an automated image post-processing system. The size of the ischemic core was established using an Apparent Diffusion Coefficient (ADC) threshold of <620×10⁻⁶ mm²/s on MRI-DWI or a relative cerebral blood flow of <30% on CT perfusion. Ethics approval was obtained from the local institutional review board at each center, and written informed consent was obtained from patients.

We identified patients with a persistent target mismatch profile on the follow-up MRI, obtained 24h after randomization, using the same definition of target mismatch. The RAPID software was used to quantify the hypoperfused tissue and the infarct core was traced manually on the DWI (automated core volumes based on ADC thresholds can be unreliable beyond 24-hours due to reperfusion-related increases in ADC).

Based on the perfusion pattern on the follow-up MRI, we identified 3 groups: 1) patients with spontaneous reperfusion (based on the DEFUSE 3 definition of >90% reperfusion between baseline and 24-hour follow-up hypoperfusion lesions); 2) persistent hypoperfusion lesion, but loss of the target mismatch perfusion profile at follow-up imaging; 3) persistent target mismatch profile. We compared group 2 (no reperfusion and mismatch not maintained) to group 3 )the group that maintained the target mismatch profile) using Wilcoxon Rank Sum test for continuous variables and Fisher’s exact test for proportions.

Results:

Out of the 90 patients in the DEFUSE 3 medical arm, 55 had MR DWI and perfusion follow-up imaging and were included in this analysis. Thirteen had no follow-up perfusion imaging and 22 had non-contrast CT/CT perfusion follow-up which does not provide accurate assessment of subacute infarct core volumes. Out of the 55 eligible patients, 10 (18%) patients had spontaneous reperfusion (group 1), 35 (64%) had a persisting Tmax lesion but no target mismatch (group 2), and 10 (18%) had a persistent target mismatch profile (group 3). The Table shows the characteristics of the patients who maintained a target mismatch profile (group 3) versus the patients that lost the target mismatch profile between baseline and the 24-hour post-randomization scan (group 2). Patients with a persistent target mismatch profile had a significantly lower HIR on baseline imaging. There was no association between baseline clinical characteristics or clinical outcomes and a persistent favorable profile (Table). Favorable clinical outcome (mRS 0–2) at 90 days occurred in 10% with a persistent target mismatch vs. 17% without a persistent mismatch (p=1.0)

Table.

Clinical and Imaging Characteristics

Characteristics	Persistent Target Mismatch Profile (n=10)	Lost Target Mismatch Profile (n=35)	p-value
Clinical Characteristics
Median Age (IQR)	77 (63–84)	64 (56–74)	0.08
Female Gender (%)	50%	51%	1.0
Median Baseline NIHSS (IQR)	13 (8.5–22.5)	19 (12.5–22.0)	0.23
Glucose	111.5 (108.0–124.8)	128.0 (112.0–151.5)	0.14
Systolic Blood Pressure	152.5 (132.5–160.0)	142.0 (131.0–159.0)	0.49
Diastolic Blood Pressure	82.0 (62.5–99.8)	80.0 (69.5–89.0)	0.96
Previous Myocardial Infarction	0%	9%	1.0
Hypertension	80%	80%	1.0
Atrial Fibrillation	30%	23%	0.69
Hypercholesterolemia	30%	34%	1.0
Diabetes	20%	34%	0.47
Prior Stroke	20%	9%	0.31
Antiplatelet therapy	40%	23%	0.42
Anticoagulant	10%	11%	1.0
Statin therapy	20%	29%	0.71
Diabetic Medication	0%	26%	0.17
Anti-hypertensive therapy	70%	57%	0.72
Median NIHSS at 24h (IQR)	14.5 (11.2–24.0)	17.0 (11.0–20.0)	0.83
Median mRS at day 90 (IQR)	5.0 (3.2–6.0)	4.0 (3.0–4.5)	0.14
mRS 0–2 at day 90 (%)	10%	17%	1.0
mRS 5–6 at day 90 (%)	60%	26%	0.06
Baseline Imaging Characteristics
Core volume, ml (IQR)	7 (3–12)	17 (3–34)	0.17
Tmax>6s volume, ml (IQR)	98 (66–137)	110 (76–157)	0.48
Tmax>10s volume, ml (IQR)	18 (4–50)	42 (17–80)	0.07
Mismatch volume, ml (IQR)	92 (62–131)	88 (59–129)	0.83
Hypoperfusion intensity ratio* (IQR)	0.20 (0.04–0.35)	0.38 (0.24–0.56)	<0.01
Percent MCA occlusions	60%	57%	1.0
Percent ICA occlusions	40%	43%	1.0
24-hour imaging characteristics
Last well to 24h post-randomization imaging, hrs (IQR)	38.4 (37.8–39.1)	36.5 (32.7–39.3)	0.25
DWI volume, 24h post-randomization, ml (IQR)	23 (11–31)	88 (34–134)	<0.001
Tmax>6s volume, 24 h post-randomization ml (IQR)	77 (60–91)	47 (24–75)	<0.05
Infarct growth baseline to 24h, ml (IQR)	15 (8–21)	59 (22–110)	<0.001
Change in Tmax>6s volume, baseline to 24h, ml (IQR)	-34 (−60–5)	-49 (−94- −17)	0.33
Percent reperfusion (IQR)	30% (10%−−40%)	50% (20%- 80%)	0.02
Mismatch volume ml (IQR)	50 (38–80)	-19 (−82–0)	<0.001

Two of the 10 patients with a persistent target mismatch profile had an additional MRI follow-up scan four days after randomization, allowing assessment of the infarct volume at an even later time point. One of these patients (Figure 1) experienced nearly a three-fold expansion of the infarct between the 24-hour follow-up study and day four and had a poor clinical outcome (90-day mRS=5). The other patient’s infarct grew slightly from 10mL at the 24-hour follow-up to 17mL 3 days later with maintenance of a large hypoperfusion deficit. The patient deteriorated and died from the stroke, but no additional imaging was obtained to clarify the final infarct volume.

Figure 1:

73-year-old female randomized into the DEFUSE 3 medical control arm following a wake-up stroke. Panels 1,2 and 3 show baseline, 24h post-randomization and 4-day imaging. Continued lesion growth is documented with expansion of the infarct into tissue that had no DWI abnormality 24h-post randomization.

Discussion:

Eighteen percent of the DEFUSE 3 medical arm patients had a persistent target mismatch profile when imaged a median of 38-hours [33–39 IQR] after last know well time. The patients with a favorable vs. unfavorable perfusion profile at follow-up presented with similar baseline core and hypoperfusion lesion sizes, but there was more than a 3-fold difference in infarct growth (15mL vs 49ml, p<0.001) and more partial reperfusion in the patients who lost the favorable profile at 24-hours. These findings demonstrate that patients can “lose” the favorable profile over time due to either infarct expansion or partial reperfusion leading to less tissue at risk.

The significantly smaller (23mL vs 88mL, p<0.001) DWI volumes in the persistent favorable profile group at the 24-hour follow-up did not translate into better day 90 mRS outcomes. 60% had mRS 5/6 and only 10% mRS 0–2. We suspect that these patients are at high risk for further infarct expansion and often, if not always, eventually develop substantial infarct growth. This concept is supported by the two patients who had day-4 scans, both showing continued infarct growth.

Excellent collateral flow, as measured by the HIR, was a strong predictor of the persistence of a target mismatch profile. This is consistent with previous findings that a high HIR heralds rapid infarct growth and faster disappearance of a favorable pattern.^{3, 4} Our findings suggest a low HIR, in patients presenting late after symptom onset, predicts slow infarct growth over the next 24-hours. Although the HIR has not been shown to improve patient selection for thrombectomy, it can help identify patients who are likely to maintain a target mismatch during lengthy transfers. In one recent study, patients with a favorable HIR had a median infarct core growth rate of only 1 ml/hr during transfer vs 10 ml/hr for patients with a high HIR.⁴ The Tmax>10s volume at baseline also trended lower (p<0.07) in the persistent target mismatch group, but our results suggest that HIR is a more robust predictor of persistent mismatch.

The notion that penumbral tissue can exist for more than 24-hours after stroke onset is supported by prior PET studies and MRI studies indicating presence of penumbral patterns up to 48-hours after onset.^{5, 6} Other studies have shown that infarct expansion appears to continue for several days in some patients who do not reperfuse.^{7, 8}

Our study suggests that approximately 20% of patients with a large vessel anterior circulation occlusion who have a persistent target mismatch imaging profile 6–16 hours after symptom onset will continue to have the target mismatch profile for at least 24 additional hours. In our patients, the persistent target mismatch profile did not predict a favorable outcome in the absence of spontaneous or mechanical reperfusion. We speculate that these patients continue to experience infarct growth over a prolonged period and may be candidates for reperfusion therapy even beyond 24-hours.

The incidence of target mismatch declines over time. Based on screening logs, approximately half the patients who met all other eligibility criteria for DEFUSE 3 had a target mismatch. Of these, it appears that about 20% will have a persistent mismatch if not treated with thrombectomy. Therefore, the incidence of persistent mismatch in patients who present 24–40 hours after onset with a large vessel anterior circulation occlusion may be as high as 10%, which could make a clinical trial in this population feasible. Trial feasibility is further enabled by the 2018 AHA recommendations that are expected to lead to a substantial increase in the number of stroke centers that perform perfusion imaging.⁹ Limitations of our study include the small sample size and the fact that nearly 40% of the DEFUSE 3 patients did not have follow-up imaging with MR perfusion, therefore our conclusions should be considered preliminary.

Conclusion:

Approximately 20% of medical arm patients in DEFUSE 3 had a persistent mismatch for 38 hours after stroke onset. These patients have slow expanding infarcts but eventually suffer very poor clinical outcomes. Clinical trials in patients treated beyond 24-hours after symptom onset are needed to test the hypothesis that late presenting mismatch patients can benefit from reperfusion therapies.