Early post-thrombectomy MRI markers: temporal evolution and association with reperfusion and clinical outcome

Abstract

Background and Objectives

A significant proportion of endovascular therapy (EVT) treated ischemic stroke patients do not achieve favorable clinical outcomes despite successful recanalization. Identifying patients who may benefit from adjunctive therapies is a priority. We hypothesized that incomplete reperfusion is associated with evolution of imaging markers on early post-EVT MRI and clinical outcome.

Methods

Patients who underwent attempted EVT for anterior circulation large vessel occlusion (LVO) stroke and had early (<6 hour) and planned 24-hour MRI, were enrolled in the Guiding neUroprotection After Reperfusion to prevent Damage in acute ischemic Stroke (GUARDS) study. Imaging markers were evaluated by consensus, blinded to clinical data, except the LVO target. Frequencies of imaging markers, lesion patterns, and associations with reperfusion and clinical outcome were reported. Good clinical outcome was defined as modified Rankin Scale (mRS) 0–2 at 30-days.

Results

A total of 207 patients were included: median age 67 years, 53% female, 47% Black or African American, admission NIHSS 16 [10–21], pre-admission mRS 0 [0–1], and onset-to-triage time 192 minutes. Patients with incomplete reperfusion (54%) had larger ADC and perfusion volumes post-EVT, and by 24-hours, higher rates of lesion growth and edema. Thirty-five percent of those with early incomplete reperfusion had a Mixed Perfusion pattern at 24-hours. A Stable Lesion pattern was seen in only 12% of all patients at 24-hours. Early complete reperfusion was associated with lower NIHSS, 8 vs. 13 (p=0.011), and increased Early Neurological Improvement (ENI), 66% vs. 43% (p=0.002), at 24-hours. Absence of Mixed Signal on ADC (OR 3.01, p=0.015, 95% CI: [1.24–7.33]) remained a significant predictor of good clinical outcome at 30-days, even after adjusting for age, admission NIHSS, and ENI.

Discussion

This observational study systematically characterized MR imaging markers at serial follow-up early after EVT. The heterogeneous and dynamic evolution observed during the hours and days after EVT in most patients suggests further opportunity to intervene, with early incomplete reperfusion and reperfusion injury as prognostic differentiators. Our findings can serve as a roadmap for designing future clinical trials to develop adjunctive therapies targeting these imaging markers.

Introduction

Endovascular therapy (EVT) is effective and safe for treating patients with acute stroke due to a large vessel occlusion (LVO). ^1–4 The substantial effect size favoring mechanical recanalization over medical management observed in most trials has contributed to the rationale that EVT should not be withheld from patients with a LVO, and its use has been advocated for patients who did not meet the criteria in the clinical trials.⁵ This has led to the question: "What are the outer limits of safety and efficacy?".^{2, 3, 6, 7} While trials continue to explore these limits, it is reasonable to expect that the population treated with EVT as part of routine standard of care will grow.

A significant proportion of patients do not achieve favorable outcomes even when large vessel recanalization has been successful. Conversely, well intended adjunctive therapy may offer little prospect of benefit in patients who have a favorable response to EVT but could introduce unnecessary risk. Reports of recent meetings of STAIR XI and XII^8–10 highlight the need to understand and target secondary effects of endovascular therapy, including intra-ischemic¹¹ and reperfusion injury, to improve clinical outcomes in this patient population. Potential interventions include concurrent thrombolysis beyond the current indications (such as tenecteplase (TNK) trials), early cytoprotective strategies such as Efficacy and Safety of nerinetide [NA-1] (ESCAPE-NA1)^{12, 13}, and prophylactic treatments aimed at preventing later secondary injuries, such as the development of malignant edema (e.g., glyburide and glibenclamide).^{13, 14}

In contrast to most randomized controlled trials, where patients are selected through a pre-specified pathway, significant heterogeneity exists in the diagnostic pathways of routine clinical practice. Pre-treatment imaging, which may include CT or MRI with or without perfusion imaging, is conducted either at a referring hospital (often a primary stroke center) or at a thrombectomy-capable/comprehensive stroke center (eFigure 1). When transferred from another hospital, patients on arrival are increasingly brought directly to the interventional suite without stopping for interval additional diagnostic imaging.¹⁵ This variability leads to differences in pre-treatment imaging modalities, available image series, imaging parameters, image quality, and the time intervals from pre-treatment imaging to endovascular recanalization.¹⁶ However, the patients’ clinical paths converge in the interventional suite, where post-EVT imaging may be more easily homogenized compared to the emergent pre-EVT setting.

The patient emerging from the endovascular suite often exhibits a clinical and radiological profile that differs markedly from their pre-EVT state. Complete recanalization is not always achieved, and even when it is, reperfusion may not follow.¹⁷ Distal embolization can occur, and iatrogenic vessel injury may result in vasospasm or disruption of cerebrovascular autoregulation. The blood-brain barrier may remain intact or permeable¹⁸, edema formation can commence¹⁹, and hemorrhagic transformation may be present and evolving²⁰. These heterogeneous biological responses to EVT underscore the need for tailored adjunctive therapies. Therefore, after EVT, clinicians and researchers should anticipate variability in patient response and consider establishing a new post-thrombectomy baseline to guide further management and therapeutic decisions. In this context, non-invasive imaging for the diagnosis and stratification of injury is invaluable. This raises the question: what is the utility of early post-thrombectomy imaging?

We have explored early imaging markers post-EVT and have found a dynamic evolution in response to perfusion status.^18–23 With this study, we aimed to systematically evaluate our endovascular-treated LVO patient population, stratify by incomplete reperfusion on early <6 hours post-EVT MRI, compare the evolution of imaging markers between cohorts, and explore the association of these markers with clinical outcomes. The primary research question was how to characterize the heterogenous imaging response in patients post-EVT stratified by reperfusion status.

Methods

Patient Population

Patients presenting to MedStar Washington Hospital Center (Washington, DC) and Suburban Hospital (Bethesda, MD) between April 1, 2018, and August 30, 2024, who met the following criteria were prospectively screened and approached for enrollment: i) age ≥18 years, ii) confirmed LVO stroke of the anterior circulation, iii) no contraindication to 3T MRI, iv) EVT attempted, v) early post-EVT MRI, and vi) 24-hour MRI planned at the time of consent.

Standard treatment with IV thrombolysis or intra-arterial (IA) tPA was not an exclusion, and patients were not treated with any adjunctive neuroprotective therapies. Post-EVT care of study patients was performed according to American Heart Association (AHA) guidelines.²⁴ Multiple passes was defined as >1 pass of any device during EVT. Symptomatic intracranial hemorrhage (sICH) was defined using the Safe Implementation of Thrombolysis in Stroke - Monitoring Study (SITS-MOST) criteria as parenchymal hematoma grade 2 (PH-2) at 24-hours in combination with an increase in National Institutes of Health Stroke Scale (NIHSS) ≥4 points or death (NIHSS of 42). Early neurological improvement (ENI) was defined as a NIHSS score decrease ≥4 points, or total score of 0–1 at 24-hours. Good clinical outcome was defined as a modified Rankin Scale (mRS) of 0–2 at 30-days. Good clinical outcome and mortality rates using the 30- and 90-day assessments were reported separately. Outcome assessments were performed independent of reperfusion status reported in this study.

Imaging protocol

Study baseline MRI was defined as the early post-EVT (<6 hours) scan. Patients with insufficient imaging studies at the post-EVT including motion artifact or missing required sequences were excluded from this analysis. Patients underwent MRI on Siemens 3T or Phillips 3T early post-EVT, 24-hours, and 5-days. The MR imaging protocol included diffusion tensor imaging (DTI) with apparent diffusion coefficient (ADC) maps, gradient recalled echo (GRE), fluid attenuated inversion recovery (FLAIR), dynamic susceptibility contrast perfusion-weighted imaging (PWI) using gadolinium-based contrast agent (GBCA), MRA, and post-FLAIR. Perfusion-weighted imaging was performed using a weight adjusted single-dose of macrocyclic-GBCA injected via a power injector at 5ml/sec. The perfusion-weighted imaging parameters included TR/TE = 1,000–1,200/25 ms, number of excitations (NEX) = 1, flip angle (FA) =70–80, 96×96–128×128 matrix, 20–7 mm thick slices, and 80–120 dynamics and maps of relative cerebral blood flow and mean transit time were generated using AIF deconvolution. Time-to-peak perfusion maps were also generated. The imaging protocol was similar for patients with MRI available pre-EVT. However, the pre-EVT timepoints were not evaluated in this study.

The modified thrombolysis in cerebral infarction (mTICI)²⁵ score was reported by the neurointerventionalist at the end of the EVT procedure and was not a consideration for study inclusion; successful recanalization was defined as mTICI≥2b, complete recanalization as mTICI=3.

Qualitative Image Analysis

Raters included 2 trained vascular neurologists (AWH,SS) and 2 imaging scientists (LLL,ML) with experience in evaluating MRI of acute stroke patients. Consensus was reached on the imaging markers by at least 3 participating raters during joint image read sessions, conducted on a weekly basis, beginning January 2023 through February 2025. Image reads were not performed by individual raters. This consensus read approach with the same raters in this study has been previously published.²⁰ MRIs were interpreted once the 5-day follow-up time point was completed. The consensus panel reviewed the MRIs collectively using a commercially available Digital Imaging Communications in Medicine (DICOM) image viewer and reached agreement on the imaging markers across all available sequences including mean transit time, time-to-peak, and relative cerebral blood flow. The consensus panel was blinded to the clinical presentation and treatment data, including onset time (defined as last seen normal to triage time interval), admission NIHSS, IV thrombolysis treatment, and timing and procedural details of the EVT treatment including mTICI score, clinical outcomes, imaging complications, and radiology reports. The panel was provided the location of the imaging confirmed LVO target for EVT. eMethods provides the details on the definitions of the imaging markers read by consensus. These definitions were referenced, and each section of an electronic case report form was organized by imaging sequence and used during the image review sessions. Not all markers were available or interpretable, therefore the sample size “n” was reported for each imaging time point according to sections labeled as: i) ischemia, ii) perfusion, iii) clot, iv) hemorrhagic transformation (HT), v) lesion evolution, vi) hyperintense acute reperfusion marker (HARM)²⁶, and vii) re-occlusion (eTable 1). HARM was read as none, minor or severe by comparing the pre-Gad and post-Gad FLAIR at each follow-up timepoint (eTable 1). Partial reversal was defined as some normalization of the ADC, but with hypointense regions still present on ADC (Figure 1A).

Figure 1:

Lesion pattern definitions

A. Partial reversal was defined as some normalization of the apparent diffusion coefficient (ADC), magnified area on right panel, with corresponding diffusion-weighted imaging (DWI) slice (on left panel), but with hypointense regions still present. B. Vasogenic edema as evident by Mixed Signal was defined as heterogeneous signal intensity on ADC (magnified area). C. The heterogeneous signal intensity on ADC may have also included demarcated hyperintense “Rings” (magnified area) surrounding the ischemic ADC regions. D. Impaired microvascular reperfusion (IMR) was defined as incomplete reperfusion, shown on colorized mean transit time map with white arrow, in the absence of both clot on gradient-recalled echo (GRE) and angiographic re-occlusion on magnetic resonance angiography (MRA). E. Hyperemia was defined as an obvious increased signal intensity on relative cerebral blood flow (not shown) with corresponding areas of decreased mean transit time (shown on colorized map with dashed yellow oval). F. Mixed Perfusion was defined as the presence of both visually conspicuous hypoperfusion (with white arrow) and hyperemia, evident on colorized mean transit time map shown (with yellow arrow).

Incomplete reperfusion was defined as any visually conspicuous hypoperfusion consistent with the target vessel for EVT, independent of the presence of thrombus; complete reperfusion was defined as the absence of any hypoperfusion consistent with the target vessel for EVT.

Lesion Pattern Definitions

Clusters of congruent imaging markers at 24-hours were considered together to define i) incomplete reperfusion, ii) Stable Lesions, iii) impaired microvascular reperfusion (IMR)²⁷, iv) Mixed Perfusion, and iv) vasogenic edema (Figure 1, eFigures 2 and 3). Stable Lesion patterns were defined at 24-hours as: i) no lesion growth or edema on DWI or FLAIR, ii) no distal clot on GRE, iii) no HT or HT limited to hemorrhagic infarction grade 1 (HI-1) on GRE, iv) complete reperfusion on PWI, and v) complete recanalization on MRA. IMR at 24-hours was defined as the presence of visually confirmed hypoperfusion consistent with the target vessel for EVT, in the absence of both clot on GRE and angiographic re-occlusion on MRA. Obvious Mismatch was defined visually as PWI lesion > DWI lesion in size. Mixed Perfusion pattern was defined as having both visually conspicuous hypoperfusion and hyperemia evident on perfusion. Vasogenic edema signs were limited to heterogeneous “Mixed Signal” (both low and elevated) on ADC and formation of “Rings” (a pattern of contiguously elevated ADC adjacent to areas of normal or low ADC).

Quantitative Image Analysis

Ischemic “core” volumes were defined with ADC values based on two different thresholds, ≤620 μm²/s as the standard and ≤470 μm²/s given the recent publication²⁸, and measured using a fully automated novel NIH tool²¹ (“coretool”) in Mipav (v10, Center for Information Technology, NIH).¹⁹ The NIH tool was also used to calculate the T_max>6 sec delay perfusion volumes from the PWI. The mismatch volume, calculated by T_max>6 sec delay perfusion volume – ADC ≤620 μm²/s volume, was provided by the tool. This tool was programmed using heuristic approaches incorporated into plugins within Mipav that are available upon request. The heuristic steps including the thresholds are described in prior published studies.^{21, 23}

Statistical Analysis

Variables were reported as median (interquartile range 25–75) or percentage (n=number) as appropriate. eTable 1 specified the varying sample sizes associated with each time point and sequence. Frequencies of imaging markers were summarized across all timepoints. Univariate analyses were performed to compare patient cohorts using nonparametric tests (χ² or Fisher’s exact test) and stratification according to complete reperfusion post-EVT (present or absent), complete reperfusion at 24-hours (present or absent), and clinical outcome at 30-days. Binomial logistic regression including stepwise analyses was performed to identify post-EVT imaging markers that were predictors of clinical outcome. Multivariable logistic regression models were used incorporating covariates that were significant at the p<0.10 level in the univariate models. IBM SPSS Statistics v19.0 was used for all statistical analyses performed.

IRB Approval and Informed Consent

Guiding neUroprotection After Reperfusion to prevent Damage in acute ischemic Stroke (GUARDS) study was a prospective cohort of the Natural History of Stroke Study protocol (ClinicalTrials.gov Identifier: NCT00009243) approved by the National Institutes of Health institutional review board (IRB), the Georgetown University – MedStar Health Research Institute IRB, and the Johns Hopkins Medicine IRB, governed by the U.S. Department of Health and Human Services regulations 45 Code of Federal Regulations (CFR) 46 and in accordance with the ethical principles set forth by the Declaration of Helsinki. Written informed consent was obtained from all study participants and/or their legally authorized representatives.

Results

Patient Population

Between April 2018 and September 2024, a total of 772 patients underwent EVT across both hospital sites. Screening was inactive for 25% of the study period primarily due to pauses during the COVID-19 pandemic, MRI scanner replacement, and brief periods of staffing shortages. Of the patients treated, 74% (n=573) were screened for eligibility; 54% of those screened (n=311) met the eligibility criteria, and 67% of eligible patients (n=207) were enrolled and included in the present analysis (eFigure 4). In total, 572 MRI studies were reviewed by the consensus panel, comprising 207 post-EVT (2 hours [1–5]), 200 24-hour (23 hours [22–26]), and 165 5-day (5 days [4–6]) scans in 207 patients.

Table 1 contains the characteristics and clinical outcomes for the 207 patients overall, stratified by complete reperfusion status post-EVT. Median age was 67 years, 53% female, 47% Black or African American, 7% Hispanic ethnicity, admission NIHSS 16 [10–21], pre-admission mRS 0 [0–1], and onset time 192 minutes (last seen normal to triage). The number of patients via direct (68%) vs. interhospital transfer (32%) pathways are indicated in Table 1 with the additional details of the various imaging pathways provided in eTable 2.

Table 1.

Patient characteristics and clinical outcome data for GUARDS study population and cohorts, stratified by complete reperfusion post-EVT.

Variables n (%), median, [IQR]		All (n=207)		Complete reperfusion post-EVT (n=190)*		p-value
				Present (n=87,46%)	Absent (n=103,54%)
Age (years)		67 [57–76]		69 [57–76]	65 [57–75]	0.47
Sex (female)		109 (53%)		55 (63%)	47 (46%)	0.02
Mode of transfer for EVT	Direct	140 (68%)		50 (57%)	80 (78%)	0.002
	Interhospital transfer	67 (32%)		37 (43%)	23 (22%)
Hypertension		135 (65%)		59 (68%)	62 (60%)	0.25
Atrial fibrillation		52 (25%)		17 (20%)	28 (27%)	0.26
Diabetes		47 (23%)		24 (28%)	15 (15%)	0.029
Hyperlipidemia		69 (33%)		30 (34%)	33 (32%)	0.77
Congestive heart failure		20 (10%)		8 (9%)	8 (8%)	0.81
Onset (min)		192 [69–516]		215 [93–573]	167 [64–524]	0.30
Admission NIHSS		16 [10–21]		15 [10–20]	16 [10–23]	0.28
Pre-admission mRS		0[0–1]		0[0–1]	0 [0–0]	0.69
IV thrombolysis		86 (42%)		35 (40%)	44 (43%)	0.68
Hemisphere		Left	119 (57%)	49 (56%)	60 (58%)	0.55
		Right	84 (41%)	38 (44%)	39 (38%)
		Both	4 (2%)	0 (0%)	4 (4%)
Cardioembolic etiology		99 (48%)		41 (47%)	49 (48%)	0.89
Target vessel		eICA	2 (1%)	0 (0%)	2 (2%)	0.43
		iICA	30 (14%)	15 (17%)	14 (14%)
		M1-proximal	82 (40%)	31 (36%)	44 (43%)
		M1-distal	50 (24%)	26 (30%)	20 (19%)
		M2	41 (20%)	14 (16%)	22 (21%)
		M3	2 (1%)	1 (1%)	1 (1%)
Modified TICI score (mTICI)^		0	16 (8%)	2 (2%)	13 (13%)	<0.001
		1	2 (1%)	0 (0%)	1 (1%)
		2a	14 (7%)	2 (2%)	12 (12%)
		2b	88 (43%)	34 (39%)	48 (48%)
Variables n (%),		All (n=207)		Complete reperfusion post-EVT (n=190)*		p-value
		3	83 (41%)	49 (56%)	26 (26%)
Successful recanalization (mTICI 2b/3)^		171 (84%)		83 (95%)	74/100 (74%)	<0.001
Complete recanalization (mTICI 3)a		83 (41%)		49 (56%)	26/100 (26%)	<0.001
Multiple passes^		120 (59%)		45/84 (54%)	65/97 (67%)	0.07
Onset to groin puncture (min)		274 [177–648]		296 [183–690]	275 [172–660]	0.84
Onset to recanalization (min)		310 [211–680]		317 [208–665]	307 [215–752]	0.50
24-hour Clinical Outcomes
NIHSS at 24-hours		11 [4–18]		8 [3–17]	13 [5–21]	0.011
Change in NIHSS at 24-hours		−4 [−10–0]		−5 [−11–(−1)]	−3 [−7–2]	0.014
Early neurological improvement at 24-hours		116 (56%)		57 (66%)	44 (43%)	0.002
symptomatic ICH at 24-hours		1 (0.5%)		0 (0%)	0 (0%)	1.0
Discharge disposition (home)		67 (32%)		33 (38%)	31 (30%)	0.25
30-day Clinical Outcomes
mRS		3 [1–4]		2 [1–4]	3 [1–4]	0.05
Good outcome (mRS≤2)		92/184 (50%)		44/80 (55%)	43/91 (47%)	0.30
mortality rate		22/184 (12%)		7/80 (9%)	12/91 (13%)	0.41
90-day Clinical Outcomes
mRS		2 [1–4]		2 [0–4]	2 [1–5]	0.12
Good outcome (mRS≤2)		78/150(52%)		37/65 (57%)	37/73 (51%)	0.48
mortality rate		29/150 (19%)		11/65 (17%)	15/73 (21%)	0.55

Note – EVT: Endovascular Therapy; NIHSS: National Institutes of Health Stroke Scale; mRS: modified

Rankin Scale; mTICI: modified Treatment In Cerebral Infarction; ICH: Intracranial Hemorrhage.

^*17 patients excluded from cohorts due to missing or unevaluable Perfusion-Weighted Imaging (PWI).

^{^}7 patients had no device deployed.

Stratification by Early Reperfusion Status

Complete reperfusion post-EVT was achieved in 46% (n=87) of patients (Table 1). Female patients were more likely to achieve complete reperfusion, 63% (n=55) vs. 46% (n=47) in males, p=0.02. There were no other differences between demographic data, clinical severity on presentation, or onset times in reperfusion cohorts. Patients with complete reperfusion post-EVT did have higher rates of successful recanalization (95% vs. 74%, p<0.001) and complete recanalization (56% vs. 26%, p<0.001) than patients with incomplete reperfusion.

Patients with incomplete reperfusion had significantly larger ADC ≤470 mm²/sec volumes of 0.5mL vs. 0mL, ADC ≤620 mm²/sec volumes of 5.9mL vs. 2.4mL, T_max> 6 sec volumes (17.6 mL vs. 0.0 mL), infarct density (18% vs. 5%), and mismatch volumes of 0 mL vs. -1.8 mL, post-EVT compared to those with complete reperfusion (Table 2). Forty-eight percent (n=49) of those with incomplete reperfusion had persistent Obvious Mismatch with median T_max> 6 sec volume of 16.6mL (eTable 1). By 24-hours, those without complete reperfusion post-EVT had significantly higher rates of both lesion growth and edema, as seen on both DWI and FLAIR, compared to those with complete reperfusion (Table 2). Complete reperfusion at 24-hours was achieved in only an additional 20% (n=20) of patients of those who had not completely reperfused post-EVT, and 36% (n=27) of those still with incomplete reperfusion at 24-hours also had Obvious Mismatch. By 5-days, those with incomplete reperfusion post-EVT continued to have significantly higher rates of lesion growth on FLAIR (55% vs. 33%, p=0.007) and a trend towards higher rates of lesion growth on DWI and edema on DWI and FLAIR compared to those with complete reperfusion (Table 2). In line with the increased lesion growth and edema seen on DWI and FLAIR were other markers of edema with higher rates of Mixed Signal (eFigure 3) and mass effect on ADC at 5-days for those with incomplete reperfusion (Table 2).

Table 2.

Imaging marker data for GUARDS study population and patient cohorts, stratified by complete reperfusion post-EVT.

Variables n (%), median, [IQR]	All (n=207)		Complete reperfusion post-EVT (n=190)^		p-value
			Present (n=87,46%)	Absent (n=103,54%)
post-EVT Imaging Markers
onset to post-EVT MRI (hours)	9 [6–15]		10 [6–15]	9 [6–16]	0.69
EVT end to post-EVT MRI (hours)	2 [1–5]		3 [1–5]	2 [1–5]	0.12
ADC ≤470 mm2/sec volume (mL)	0.3 [0.0–3.6]		0.0 [0.0–1.1]	0.5 [0.0–7.0]	<0.001
ADC ≤620 mm2/sec volume (mL)	4.1 [0.4–18.0]		2.4 [0.3–10.0]	5.9 [1.2–28.3]	0.001
infarct density % (ADC ≤470 volume / ADC ≤620 volume)	12% [0–30]		5% [0–18]	18% [0–37]	0.003
Tmax >6 sec volume (mL)	1.2 [0.0–21.5]		0.0	17.6 [1.6–46.0]	n/a
mismatch volume (mL)	−0.6 [−9.4–6.0]		−1.8 [−10.4–0]	0 [−7.9–22.6]	<0.001
Obvious Mismatch	49 (26%)		0 (0%)	48 (47%)	<0.001
Mixed Signal on ADC	57 (28%)		24 (28%)	29/100 (29%)	0.88
Rings on ADC	44/204 (22%)		18 (21%)	24/100(24%)	0.63
Mass Effect	35 (17%)		13 (15%)	21/100 (21%)	0.30
Distal clot on GRE	56 (27%)		1 (1%)	51 (50%)	<0.001
Hyperemia^	90 (48%)		42 (48%)	48 (47%)	0.89
HT	63 (30%)		27 (31%)	33 (32%)	0.88
	HI-1 (HBC 1a)	40 (19%)	17 (20%)	20 (19%)	0.30
	HI-2 (HBC 1b)	17 (8%)	9 (10%)	8 (8%)
	PH-1 (HBC 1c)	5 (2%)	1 (1%)	4 (4%)
	PH-2 (HBC 2)	1 (0.5%)	0 (0%)	1 (1%)
HARM	145 (78%)		59 (70%)	86 (84%)	0.02
Severe HARM	101 (54%)		43 (51%)	58 (57%)	0.41
24-hour Imaging Markers
onset to 24-hour MRI (hours)	30 [26–37]		31 [26–37]	32 [27–38]	0.23
EVT end to 24-hour MRI (days)	23 [22–25]		23 [22–25]	23 [22–26]	0.57
Stable Lesion pattern*	24 (12%)		17 (20%)	4 (4%)	0.003
Lesion growth on DWI	114 (57%)		40 (47%)	68 (68%)	0.004
Lesion growth on FLAIR	142 (71%)		53 (62%)	80 (80%)	0.007
Edema on DWI	141 (71%)		51 (59%)	81 (81%)	0.001
Edema on FLAIR	135 (68%)		53 (62%)	76 (76%)	0.04
Rings on ADC	103 (52%)		45 (52%)	52 (52%)	1.0
Variables n (%),	All (n=207)		Complete reperfusion post-EVT (n=190)^		p-value
Mixed Signal on ADC	91 (45%)		40 (47%)	48 (48%)	0.89
Rings on ADC	103 (52%)		45 (52%)	52 (52%)	1.0
Mass Effect	60 (30%)		23 (27%)	36 (36%)	0.19
Distal clot on GRE	35 (19%)		1 (1%)	33 (33%)	<0.001
Complete reperfusion^^	109 (59%)		82 (80%)	20 (20%)	<0.001
Hyperemia^^	86 (47%)		41 (49%)	42 (42%)	0.35
Mixed Perfusion pattern^^	33 (18%)		0 (0%)	33 (35%)	<0.001
IMR	43/184 (23%)		0 (0%)	41 (40%)	<0.001
HT	105 (52%)		41 (48%)	57 (57%)	0.22
	HI-1 (HBC 1a)	46 (23%)	16 (19%)	25 (25%)	1.0
	HI-2 (HBC 1b)	49 (24%)	21 (24%)	26 (26%)
	PH-1 (HBC 1c)	5 (2%)	2 (2%)	3 (3%)
	PH-2 HT (HBC 2)	5 (3%)	2 (2%)	3 (3%)
HARM	164 (82%)		72 (84%)	86 (88%)	0.43
Severe HARM	121 (61%)		53 (62%)	64 (65%)	0.67
Re-occlusion on MRA	6 (3%)		0 (0%)	6 (6%)	0.02
5-day Imaging Outcomes
onset to 5-day MRI (days)	5 [4–6]		5 [4–6]	5 [4–6]	0.73
EVT end to 5-day MRI (days)	5 [4–6]		5 [4–5]	5 [4–6]	0.70
Lesion growth on DWI	44 (27%)		14 (20%)	28 (34%)	0.05
Lesion growth on FLAIR	73 (44%)		23 (33%)	45 (55%)	0.007
Edema on DWI	55 (33%)		18 (26%)	34 (41%)	0.05
Edema on FLAIR	92 (56%)		34 (49%)	52 (63%)	0.08
Mixed Signal on ADC	71 (43%)		23 (33%)	46 (55%)	0.007
Rings on ADC	102 (67%)		43 (61%)	59 (71%)	0.19
Mass Effect	53 (32%)		16 (23%)	35 (42%)	0.01
Distal clot on GRE	25 (17%)		2 (3%)	22 (27%)	<0.001
Hyperemia	46 (32%)		18 (28%)	25 (30%)	0.79
HT	85 (52%)		32 (46%)	47 (57%)	0.18
	HI-1 (HBC 1a)	38 (23%)	12 (17%)	21 (25%)	0.31
	HI-2 (HBC 1b)	35 (21%)	13 (19%)	21 (25%)
	PH-1 (HBC 1c)	5 (3%)	4 (6%)	1 (1%)
	PH-2 HT (HBC 2)	7 (4%)	3 (4%)	4 (5%)
Variables n (%),	All (n=207)		Complete reperfusion post-EVT (n=190)^		p-value
HARM	64 (31%)		22 (35%)	41 (52%)	0.04
Severe HARM	30 (14%)		9 (14%)	20 (25%)	0.07
Re-occlusion on MRA	6 (4%)		1 (1%)	4 (5%)	0.26

Note – EVT: Endovascular Therapy; ADC: Apparent Diffusion Coefficient; Tmax: time to peak of residue function following deconvolution of arterial input function; IMR: Impaired Microvascular Reperfusion; HT: Hemorrhagic Transformation; PH-2: Parenchymal Hematoma grade 2; HBC 2: Heidelberg Bleeding Classification grade 2; HARM: Hyperintense Acute Reperfusion Marker; DWI: Diffusion-Weighted Imaging; FLAIR: Fluid Attenuated Inversion Recovery; GRE: Gradient-Recalled Echo; MRA: Magnetic Resonance Angiography.

^*Stable Lesion pattern defined using 24-hour MRI as: i) no lesion growth or edema on DWI or FLAIR, ii) no distal clot on GRE, iii) none or HI-1 (1a) on GRE, iv) complete reperfusion, and v) complete recanalization.

^**Mixed Perfusion pattern defined as both incomplete reperfusion and hyperemia, visually confirmed on DSC Perfusion-Weighted Imaging (PWI) at 24-hours.

^{^}17 patients excluded from cohorts due to missing or unevaluable PWI.

^{^^}23 patients excluded due to missing or unevaluable PWI.

Among patients with incomplete reperfusion, 50% (n=52) may have had IMR, with no distal clot visible on GRE (Table 2). Incomplete reperfusion persisted in 41% (n=75) at 24-hours, and IMR was again suspected in 57% (n=43) of these patients (eTable 3). A Mixed Perfusion pattern at 24-hours was observed in 35% (n=33) of patients without early complete reperfusion, whereas it was absent in those with complete reperfusion (Table 2). Among Mixed Perfusion pattern cases, only one-third had visible distal clot, further supporting a role for IMR. Figure 2C displays a patient with IMR with lesion growth at 24-hours, and a separate patient with a Mixed Perfusion pattern (Figure 2D), with a region of hypoperfusion and an adjacent region of hyperemia with edema. The rates of hyperemia were comparable between patients with and without complete reperfusion after EVT and at 24 hours (Table 2 and eTable 3). HT occurred in 30% (n=63) of patients post-EVT, with only one PH-2 (0.5%). HARM was frequently seen post-EVT (78%) and at 24-hours (82%), more commonly in those with incomplete reperfusion (84% vs. 70%, p=0.02), and remained elevated at 5 days (52% vs. 35%, p=0.04) (Table 2). eFigure 2 provides examples of the most frequent blood-brain barrier disruption imaging markers seen across the timepoints: HT (hemorrhagic infarction grade 2 (HI-2), Heidelberg Bleeding Classification 1b) and Severe HARM. Serial imaging revealed dynamic progression in edema markers across the post-EVT to 24-hour to 5-day period: Rings (21% to 52% to 67%), Mixed Signal (28% to 46% to 43%), and Mass Effect (17% to 30% to 32%) (eTable 1). Meanwhile, complete reperfusion rates increased (46% to 59% to 72%) and distal clot prevalence declined (27% to 18% to 15%). Re-occlusion was rare, occurring in 3% of patients at 24-hours and 4% at 5-days (eTable 1).

Figure 2:

Recanalization patterns

A. A patient with post-EVT images shown from left to right: magnetic resonance imaging (MRA), gradient-recalled echo (GRE), colorized mean transit time map (dashed white oval indicating hypoperfusion), diffusion-weighted imaging (DWI), and 24-hr DWI (last image). The ischemic lesion was comparable to the patient illustrated in Figure 2B but with incomplete reperfusion due to unsuccessful recanalization, who demonstrated lesion growth at 24-hr DWI (last image in row). B. A patient with a Stable Lesion pattern with post-EVT images shown from left to right: MRA, GRE, mean transit time map, DWI, and 24-hr DWI (last image) after successful EVT and complete reperfusion. C. A patient with impaired microvascular reperfusion (IMR) and lesion growth at 24-hours, with post-EVT images shown from left to right: MRA, GRE, mean transit time map (indicated with yellow arrow), DWI, and 24-hr DWI (last image). D. A patient with a Mixed Perfusion pattern consisting of a region of hypoperfusion (indicated with yellow arrows) and an adjacent region of hyperemia (white arrow), both shown on colorized mean transit time map with post-EVT images shown from left to right: MRA, GRE, mean transit time, DWI, and 24-hr DWI (last image).

Clinical Outcomes

Complete reperfusion post-EVT was associated with greater ENI (66% vs. 43%, p=0.002) and greater NIHSS change (−5 vs. −3, p =0.014) compared to those with incomplete reperfusion, though 30-day outcomes only showed a trend toward benefit (mRS 2 vs. 3, p=0.05) (Table 1). Stable Lesion pattern at 24-hours was present in only 12% (n=24) of all patients (Table 2) and was associated with minimal vasogenic edema (Mixed Signal 13%, Rings 17%, Mass Effect 0%), favorable 24-hour NIHSS (median 4), high ENI (67%), discharged home (58%), and favorable 30-day mRS (median 0 [0–3], 74% good outcome) (data not included in table). Figure 2B illustrates a patient with a Stable Lesion pattern after successful EVT and complete reperfusion. In contrast, Figure 2A shows a patient with a comparable post-EVT ischemic lesion but incomplete reperfusion due to unsuccessful recanalization with distal clot on GRE, with lesion growth at 24-hours.

Patient characteristics and imaging markers between patient cohorts dichotomized by clinical outcome at 30-days are presented in Table 3, 50% (n=92) “good” with mRS≤2 vs. 50% (n=92) “poor” with mRS>2. Younger patients, female, lower admission NIHSS and pre-admit mRS, non-diabetic, non-congestive heart failure, non-cardioembolic etiology, and single-pass EVT were associated with good outcome at 30-days. Patients with good outcome at 30-days were more likely to have had ENI at 24-hours and to have been discharged home. Univariate analyses identified clinical and post-EVT imaging predictors of good outcome at 30-days. Model #1 included only clinical variables, with significant predictors being age, admission NIHSS, ENI, and discharge to home (Table 4). Model #2 adjusted for age, admission NIHSS, ENI, and discharge to home, and entered post-EVT imaging markers into the model including ADC ≤470 mm²/sec volume (mL), Absence of Mixed Signal on ADC, Absence of Mass Effect, Severe HARM, and Stable Lesion status at 24-hour. Absence of Mixed Signal on ADC (OR 3.01, p=0.015, 95% CI: [1.24–7.33]) remained a significant predictor of good clinical outcome at 30-days, even after adjusting for age, admission NIHSS, ENI, and Severe HARM in Model #3 (Table 4).

Table 3.

GUARDS patient cohorts stratified by good versus poor clinical outcome at 30-days.

Variables n (%), median, [IQR]		30-day Clinical Outcome (n=184)°		p-value
		Good (mRS≤2) (n=92,50%)	Poor (mRS>2) (n=92,50%)
Variables n (%),		30-day Clinical Outcome (n=184)°		p-value
Age (years)		63 [53–72]	72 [62–81]	<0.001
Sex (female)		42 (46%)	59 (64%)	0.012
Mode of transfer for EVT	Direct	62 (67%)	70 (76%)	0.18
	Interhospital transfer	30 (33%)	22 (24%)
Hypertension		54 (59%)	63 (68%)	0.16
Atrial fibrillation		20 (22%)	28 (30%)	0.15
Diabetes		16 (17%)	27 (29%)	0.054
Hyperlipidemia		28 (30%)	36 (39%)	0.20
Congestive heart failure		5 (5%)	15 (16%)	0.015
Carotid disease		5 (5%)	5 (5%)	1.0
Onset (min)		187 [69–458]	187 [67–538]	0.85
Admission NIHSS		13 [7–19]	19 [14–24]	<0.001
Pre-admission mRS		0 [0–0]	0 [0–1]	0.003
IV thrombolysis		41 (45%)	36 (39%)	0.46
Hemisphere	Left	51 (55%)	55 (60%)	0.49
	Right	41 (45%)	36 (39%)
	Both	0 (0%)	1 (1%)
Cardioembolic etiology		37 (40%)	53 (58%)	0.018
Target vessel	eICA	1 (1%)	1 (1%)	0.66
	iICA	13 (14%)	16 (17%)
	M1-proximal	33 (36%)	37 (40%)
	M1-distal	20 (22%)	23 (25%)
	M2	23 (25%)	15 (16%)
	M3	2 (2%)	0 (0%)
Modified TICI score (mTICI)^	0	7 (8%)	8 (9%)	0.72
	1	2 (2%)	0 (0%)
	2a	5 (5%)	5 (6%)
	2b	42 (46%)	35 (39%)
	3	35 (38%)	42 (47%)
Successful recanalization (mTICI 2b/3)^		77/91 (85%)	77/90 (86%)	0.86
Variables n (%),		30-day Clinical Outcome (n=184)°		p-value
Complete recanalization(mTICI 3)^		35/91 (38%)	42/90 (47%)	0.26
Multiple passes^		48/88 (55%)	63/89 (71%)	0.025
Onset to groin puncture (min)		261 [174–641]	272 [177–640]	0.92
Onset to recanalization (min)		299 [200–624]	319 [211–717]	0.50
post-EVT Imaging Markers
ADC ≤470 mm2/sec volume (mL)		0.0 [0.0–2.5]	0.4 [0.0–6.3]	0.048
ADC ≤620 mm2/sec volume (mL)		2.8 [0.3–13.2]	5.8 [1.0–32.7]	0.13
infarct density % (ADC ≤470 volume / ADC ≤620 volume)		12% [0–26]	14% [0–32]	0.41
Tmax >6 sec volume (mL)		0.4 [0.0–17.5]	5.7 [0.0–41.4]	0.07
mismatch volume (mL)		−0.8 [−8–2.0]	−0.4 [−10.6–18.0]	0.84
Complete reperfusion+		44/87 (51%)	36/84 (43%)	0.31
Hyperemia+		40/86 (47%)	40/87 (46%)	1.0
Distal clot on GRE		25 (27%)	24 (26%)	0.88
Mixed Signal on ADC		19 (21%)	29 (32%)	0.093
Rings on ADC		15 (16%)	21 (23%)	0.26
Mass Effect on ADC		12 (13%)	21 (23%)	0.083
HT	Any	25 (27%)	28 (30%)	0.66
	HI-1 (HBC 1a)	17 (18%)	16 (17%)
	HI-2 (HBC 1b)	6 (7%)	9 (10%)
	PH-1 (HBC 1c)	2 (2%)	2 (2%)
	PH-2 HT (HBC 2)	0 (0%)	1 (1%)
HARM		69 (77%)	71 (85%)	0.19
Severe HARM		54 (60%)	46 (55%)	0.49
24-hour Clinical Outcomes
NIHSS at 24-hours		5 [2–11]	17 [9–22]	<0.001
Change in NIHSS at 24-hours		−6 [−12–(−1)]	−3 [−7–2]	0.01
Early neurological improvement at 24-hours		59 (64%)	43 (47%)	0.025
Symptomatic ICH at 24-hours		0 (0%)	1 (1%)	0.34
Discharge disposition (home)		49 (53%)	11 (12%)	<0.001
24-hour Imaging Markers
Variables n (%),		30-day Clinical Outcome (n=184)°		p-value
Complete reperfusion+		53/87 (61%)	42/78 (54%)	0.36
Stable Lesion pattern*		14/91(15%)	5/88 (6%)	0.035
Mixed Perfusion pattern**		16 (18%)	13 (17%)	0.77
Hyperemia+		37 (43%)	37 (43%)	1.0
Distal clot on GRE		15/90 (17%)	18 (20%)	0.61
Lesion growth on DWI		48 (53%)	54 (61%)	0.24
Lesion growth on FLAIR		59 (65%)	68 (77%)	0.07
Edema on DWI		56 (62%)	69 (78%)	0.014
Edema on FLAIR		55 (60%)	65 (74%)	0.056
Mixed Signal on ADC		39/91 (43%)	41/88 (47%)	0.59
Rings on ADC		44/91 (48%)	47/88 (53%)	0.50
Mass Effect		22/91 (24%)	34/88 (39%)	0.031
HT	Any	41 (45%)	51 (58%)	0.67
	HI-1 (HBC 1a)	18 (20%)	22 (25%)
	HI-2 (HBC 1b)	21 (23%)	23 (26%)
	PH-1 (HBC 1c)	1 (1%)	2 (2%)
	PH-2 HT (HBC 2)	1 (1%)	4 (5%)
HARM		71 (79%)	78 (94%)	0.004
Severe HARM		50 (56%)	63 (76%)	0.005
5-day Imaging Outcomes
Lesion growth on DWI		18 (22%)	22 (31%)	0.19
Lesion growth on FLAIR		29 (35%)	35 (50%)	0.20
Edema on DWI		33 (40%)	21 (30%)	0.48
Edema on FLAIR		44 (53%)	38 (54%)	0.88
Mixed Signal on ADC		32 (39%)	34 (48%)	0.24
Rings on ADC		53 (64%)	50 (70%)	0.39
Mass Effect		21 (25%)	31 (44%)	0.016
Distal clot		12 (15%)	12 (17%)	0.74
Hyperemia+		22 (27%)	20 (30%)	0.69
HT	Any	38 (46%)	43 (61%)	0.72
	HI-1 (HBC 1a)	17 (20%)	19 (27%)
	HI-2 (HBC 1b)	16 (19%)	17 (24%)
	PH-1 (HBC 1c)	3 (4%)	2 (3%)
Variables n (%),		30-day Clinical Outcome (n=184)°		p-value
	PH-2 HT (HBC 2)	2 (2%)	5 (7%)
HARM		31 (39%)	30 (50%)	0.20
Severe HARM		10 (13%)	19 (32%)	0.007
Clinical Outcomes
30-day mRS		1 [0–2]	4 [3–5]	---
mortality 30-days		0 (0%)	22 (24%)	---
90-day mRS		1 [0–1]	4 [3–6]	---
90-day good outcome (mRS≤2)		67/68 (99%)	5/74 (7%)	---
mortality 90-days		0 (0%)	29/74 (39%)	---

Note – mRS: modified Rankin Scale; NIHSS: National Institutes of Health Stroke Scale; mTICI: modified Treatment In Cerebral Infarction; GRE: Gradient-Recalled Echo; ADC: Apparent Diffusion Coefficient; HT: Hemorrhagic Transformation; HBC: Heidelberg Bleeding Classification; HARM: Hyperintense Acute Reperfusion Marker; DWI: diffusion-weighted imaging; FLAIR: Fluid Attenuated Inversion Recovery.

^°23 patients did not have 30-day timepoint.

⁺13 patients excluded from cohorts due to missing or unevaluable Perfusion-Weighted Imaging (PWI) post-EVT; 17 patients excluded from cohorts due to missing or unevaluable PWI at 24-hours.

^{^}5 patients had no device deployed.

^*Stable Lesion pattern defined using 24-hour MRI as: i) no lesion growth or edema on DWI or FLAIR, ii) no distal clot on GRE, iii) none or HI-1 (1a) on GRE, iv) complete reperfusion, and v) complete recanalization.

^**Mixed Perfusion pattern defined as both incomplete reperfusion and hyperemia visually confirmed on PWI at 24-hours.

Table 4.

Predictors of good clinical outcome at 30-days.

Variable	Univariate	Multivariate – Model #1			Multivariate – Model #2			Multivariate – Model #3
	p-value	Adjusted odds ratio	95% CI	p-value	Adjusted odds ratio	95% CI	p-value	Adjusted odds ratio	95% CI	p-value
Age	<0.001	0.95	[0.92–0.99]	0.005	0.93	[0.89–0.96]	<0.001	0.92	[0.89–0.96]	<0.001
Sex (female)	0.012	0.56	[0.25–1.24]	0.151
Admission NIHSS	<0.001	0.92	[0.86–0.98]	0.006	0.90	[0.85–0.97]	0.002	0.89	[0.84–0.95]	<0.001
Preadmission mRS	0.003	0.75	[0.43–1.33]	0.327
Cardioembolic etiology	0.018	1.00	[0.44–2.31]	0.985
Diabetes	0.054	0.62	[0.25–1.55]	0.301
Congestive heart failure	0.015	0.52	[0.13–2.08]	0.352
Multiple passes	0.025	0.77	[0.33–1.81]	0.554
Early neurological improvement	0.025	2.95	[1.22–7.12]	0.016	3.78	[1.51–9.48]	0.005	4.19	[1.78–9.85]	0.001
Discharge disposition (home)	<0.001	3.61	[1.45–9.02]	0.006	2.01	[0.78–5.22]	0.151
ADC ≤470 mm2/sec volume (mL)	0.048				0.97	[0.93–1.01]	0.146
Absence of Mixed Signal on ADC post-EVT	0.093				2.35	[0.90–6.12]	0.082	3.01	[1.24–7.33]	0.015
Absence of Mass Effect post-EVT	0.083				1.19	[0.41–3.44]	0.746
Stable Lesion pattern at 24-hours	0.035				1.74	[0.37–8.16]	0.485
Severe HARM post-EVT	0.005				2.02	[0.90–4.63]	0.095	1.99	[0.90–4.42]	0.089

Note – mRS: modified Rankin Scale; NIHSS: National Institutes of Health Stroke Scale; ADC: Apparent Diffusion Coefficient; EVT: endovascular therapy.

Discussion

In this study, we have systematically characterized MR imaging markers at serial follow-up early after completion of endovascular therapy. We found a significant proportion of patients had incomplete reperfusion on post-EVT MRI with higher rates of lesion progression on subsequent timepoints and was associated with worse clinical outcome. Furthermore, even for those with early complete reperfusion, few had stable lesions. The dynamic progression of imaging findings over the first 24-hours to 5-days post-EVT suggests ongoing biological processes for which there may be opportunity to intervene, and post-EVT MRI may have utility in differentiating patients who have the greatest potential for benefit from those in whom additional therapy may not be warranted.

Ideally, early complete recanalization should achieve reperfusion, arrest lesion growth and development of edema, and not result in secondary injury such as blood-brain barrier disruption and hemorrhagic transformation. Despite half of our patients achieving complete reperfusion, only 1 out of 10 patients had a Stable Lesion pattern. It is not surprising that lesion progression, measured by lesion growth, edema, and hemorrhage, occurred more frequently in the cohort of patients with incomplete reperfusion on the post-EVT MRI, suggesting progression of primary ischemic injury. However, more than half of the population with complete reperfusion post-EVT also showed progression over the first 24-hours, suggesting an injury process secondary to reperfusion. This early, dynamic evolution on imaging may represent a combination of both primary and secondary injury (as a result of ischemia-reperfusion) that can aid in patient stratification and measure of future treatment response.

The ideal time to re-image a patient following EVT is not obvious and depends on the intended purpose.²⁹ For instance, in patients with early complete reperfusion, lesion growth beyond 24-hours occurred in only ~20% of patients, suggesting a 24-hour follow-up may be adequate. Conversely, in the 40% of patients who did not achieve complete reperfusion by 24-hours, subsequent imaging over the following days frequently revealed progression of multiple markers. As a result, later follow-up imaging is necessary to more accurately assess outcomes in this group. While these findings may not be surprising to the research community or vascular neurologists, this prospective, systematic analysis provides a comprehensive overview of the prevalence of these imaging markers during the first week following EVT. This roadmap can serve as a valuable resource for designing future clinical trials and has the potential to inform early clinical management and prognosis.

Our findings highlight that recanalization success does not guarantee microvascular tissue rescue, and for many patients post-EVT, there is a heterogenous tissue response.^{17, 28} Half of the patients who did not have complete reperfusion post-EVT may have exhibited IMR, with no distal clot visible on GRE. At 24-hours, a Mixed Perfusion pattern was seen in 1 in 3 patients who did not have complete reperfusion post-EVT, and IMR may have been contributing in one-third of these patients. In evaluating the association of post-EVT imaging markers with clinical outcomes, only absence of Mixed Signal on ADC, a marker of edema, was independently associated with good clinical outcome at 30-days.²⁸ Our findings of these dynamic changes following EVT have provided some insight to begin to understand why some patients may be left with persistent functional disability despite early recanalization. The findings from this GUARDS study further advance this understanding by demonstrating that, beginning with early post-EVT MRI, we can identify those who are at risk for further evolution and worsening of cerebral injury despite reperfusion. With early post-EVT MRI, we may be able to delineate appropriate candidates for adjunctive therapies versus those who will have a Stable Lesion pattern without indication for additional treatment.

Approximately 72% of all endovascular-treated acute ischemic stroke patients at our two centers had early post-EVT MRI, without complications, as part of their standard post-EVT clinical care. We found obtaining early post-EVT MRI to be feasible, safe, and, importantly, providing clinical information that can potentially help guide medical management during this dynamic period early after EVT. For example, early post-EVT MRI identified HT that may not have been appreciable on rotational flat-panel CT at the end of the EVT procedure or may have been attributed to contrast staining alone.^{20, 30} Given the evidence to date from observational studies and randomized trials, there is suggestion that moderate but not aggressive control of blood pressure after successful EVT may be beneficial, including within the first 24 hours.^{1, 31–37} Blood pressure parameters may have been lowered after the post-EVT MRI if extensive HT including parenchymal hematoma was demonstrated. Or, conversely, blood pressure parameters may have been raised after the post-EVT MRI if significant incomplete reperfusion was identified for a patient initially determined to have had mTICI 3 recanalization/reperfusion at the end of the EVT procedure.

We recognize that obtaining early post-EVT MRI is currently not part of the routine workflow for most and would require reallocation of clinical resources to support the safety of patients during transport and image acquisition. But, these are resources that one would expect to be available in the centers performing acute EVT, and the concept of early post-EVT MRI is not novel.^{28, 38–41} Two recently published neuroprotectant trials of patients with large vessel occlusion acute ischemic stroke illustrate this utility, with participating sites located in the United States and Canada (our sites/program were not involved in these trials).^{40, 41} In these trials, enrolled patients arrived either after transfer from another hospital or as direct admission. Per protocol, “baseline” imaging was an MRI obtained early post-EVT (2-hours and <5-hours, respectively) and again at 24- hours and 48-hours, respectively, to measure infarct growth.^{40, 41} Future studies could explore approaches to integrating MRI-based advanced imaging into standard post-EVT workflows.

Our study has several limitations. As with any observational study with prospective inclusion criteria, a potential bias may have been introduced with the requirement of a post-EVT MRI and 24-hour MRI. Therefore, more severe patients may have been less likely to be included. However, only 8% were unable to undergo MRI due to medical instability. Further potential for bias includes the requirement for consent for research by the patient or LAR. Although the demographic (e.g. age, sex) and some baseline characteristics of this study population (e.g. MRI screening and thrombolysis treatment rates, onset time, and successful recanalization) are in line with our overall EVT-treated patient population and with other published studies⁴², the patients included in this GUARDS study had lower admission NIHSS and better clinical outcomes based on 24-hour NIHSS, discharge disposition to home, and 30- and 90-day mRS 0–2. The mTICI scores were not centrally adjudicated.

The plethora of findings seen on post-EVT MRI that continue to persist or evolve through 24-hours and 5-days as evaluated in the GUARDS study, including lesion growth, edema, incomplete reperfusion, hyperemia, mixed perfusion pattern, blood-brain barrier disruption/HARM, and presence and extent of hemorrhage, have the potential to inform and better individualize the clinical management of patients after EVT. To successfully test for effective adjunctive therapies requires identifying patients who still have the potential to benefit. Our findings suggest that this can be accomplished with the use of multimodal MRI within just a few hours post-EVT when there may still be opportunity to improve outcomes with additional treatment. Our findings can serve as a foundation for understanding the frequency of these imaging markers in the setting of reperfusion and their association with clinical outcomes. This data can be used to inform the development and design of clinical trials of targeted adjunctive therapies to address reperfusion injury and other post-EVT changes that contribute to poor functional outcomes despite successful recanalization.

Data Sharing Statement

The data that support the findings of this study are available from the corresponding author on reasonable request.