Microvascular Occlusion and Postrecanalization Hyperperfusion in a Swine Model of Acute Ischemic Stroke

Abstract

Background

Despite successful recanalization after endovascular thrombectomy, more than half of patients with acute ischemic stroke with large‐vessel occlusions experience an unsatisfactory outcome. Incomplete microvascular reperfusion may contribute to it, but its occurrence remains debated, partly due to clinical observations of hyperperfusion after recanalization. This study investigates the relationship between ischemia duration, infarct development, microclot presence, and cerebral perfusion in a swine model of focal cerebral ischemia and reperfusion.

Methods

Twenty‐three swine underwent craniectomy and were randomized into 5 groups: 1‐hour, 2‐hour, or 4‐hour occlusion followed by 4‐hour recanalization, 8‐hour occlusion without recanalization (positive control), or without occlusion (sham). Middle cerebral artery occlusion was induced using aneurysm clips, with 3‐dimensional digital subtraction angiography confirming occlusion and recanalization. Three‐dimensional digital subtraction angiography was used for quantification of area at risk and tissue perfusion. Infarct size was measured using 2,3,5‐triphenyl‐2H‐tetrazolium chloride staining. The presence of microclots was quantified using CD61+ platelet immunostaining (number/mm²). Plasma markers of coagulation activation were measured before and during middle cerebral artery occlusion and after recanalization.

Results

Area at risk–normalized infarct size increased significantly with ischemia duration (r=0.8, P<0.001). Compared with the noninfarcted hemisphere, microclots in the infarcted hemisphere increased in middle cerebral artery occlusion groups (0.07 [0.04–0.13] versus 0.33 [0.16–0.71], P=0.003). Microclot density in area at risk region showed a positive correlation with ischemia duration (r=0.7, P=0.007), while no such correlation was observed in remote region. Simultaneously, higher perfusion levels were observed at both 2 hours (1.27±0.19, P<0.001) and 4 hours (1.23±0.16, P<0.001) following recanalization, irrespective of the duration of ischemia. No activation of coagulation was detected in systemic plasma.

Conclusions

Focal cerebral ischemia and reperfusion result in a substantial presence of microclots, which occurs alongside hyperperfusion in this swine model of recanalized acute ischemic stroke. These platelet microclots extend beyond the ischemic area. The density of microclots within the ischemic region increased with prolonged ischemia.

Nonstandard Abbreviations and Acronyms

AAR

area at risk

CHOICE

Chemical Optimization of Cerebral Embolectomy

DSA

digital subtraction angiography

EVT

endovascular thrombectomy

IMR

incomplete microvascular reperfusion

IS/AAR

percentage of infarct size to area‐at‐risk size

LH

left hemisphere

MCA

middle cerebral artery

MR CLEAN

Multicenter Randomized Clinical Trial of Endovascular Treatment for Acute Ischemic Stroke in the Netherlands

RH

right hemisphere

ROI

region of interest

TTC

2,3,5‐triphenyl‐2H‐tetrazolium chloride

Research Perspective.

What Is New?

• This study demonstrates that microvascular occlusions, such as platelet aggregates, can persist after successful recanalization, increasing with prolonged ischemia duration, and occur even outside the tissue at risk.

• This study also reveals that hyperperfusion can coexist with microvascular occlusions, providing new evidence for the phenomenon of impaired microvascular reperfusion despite apparent hyperperfusion after endovascular thrombectomy.

What Question Should Be Addressed Next?

• Future research is needed to determine in which vascular bed these microclots reside and how they impact long‐term neurological functional outcome.

In acute ischemic stroke caused by large‐vessel occlusion, endovascular thrombectomy (EVT) has significantly improved recanalization rates and outcomes. ¹ , ² However, despite successful recanalization, >50% of patients still experience unsatisfactory clinical outcomes. ¹ This suggests that successful recanalization of large arteries does not always guarantee successful tissue reperfusion, as the latter relies more on restoration of microcirculatory blood flow, which might be a stronger predictor of clinical recovery. ³ , ⁴ An observational study from MR CLEAN (Multicenter Randomized Clinical Trial of Endovascular Treatment for Acute Ischemic Stroke in the Netherlands) registry investigated predictors of poor functional outcomes in patients undergoing EVT, identifying nonmodifiable patient factors as key contributors to poor outcome. ⁵ However, a significant portion of poor outcomes remains unexplained. This highlights the importance of identifying new predictors, such as early microvascular patency, not only to improve outcome prediction but also to identify potential novel therapeutic targets. ⁵

Incomplete microvascular reperfusion (IMR) refers to the incomplete restoration of microcirculatory flow in certain areas of ischemic tissue after reopening an occluded blood vessel. Although IMR has been reported in preclinical studies, ⁴ , ⁶ its existence as a cause of poor outcome in patients with stroke who underwent thrombectomy remains debated, ⁷ partly due to frequent observations of hyperperfusion after EVT recanalization. ⁸ , ⁹ Additionally, clinical imaging techniques used to assess tissue reperfusion are not yet sensitive enough to detect and quantify microperfusion defects. ¹⁰ Given these challenges, animal stroke models can play a role in elucidating these phenomena by overcoming clinical limitations, enabling detailed pathological exploration of the tissue to gain a deeper understanding of IMR.

In light of these considerations, we used a swine model of focal cerebral ischemia–reperfusion, which allows extensive tissue and blood assessments and the use of clinical‐grade imaging techniques to characterize microclot formation and cerebral perfusion dynamics acutely after recanalization. Specifically, we studied the relation between duration of ischemia and infarct development, formation of microclots and their density, and cerebral perfusion, to characterize the occurrence of thrombotic microvascular occlusions.

METHODS

Data Availability

Data are available upon reasonable request to the corresponding author.

Animal Experimental Protocol

Twenty‐three female Landrace‐Yorkshire swine (farm‐bred, 3–4 months, 51.6±3.9 kg) were used in this study. A frontotemporal craniectomy was performed to gain access to the right middle cerebral arteries (MCAs). Animals were randomized to each of the following groups: 3 experimental groups with 1‐hour (n=5; 1hO/4hR); 2‐hour (n=5; 2hO/4hR); or 4‐hour (n=5; 4hO/4hR) occlusion followed by 4‐hour recanalization (reperfusion); or a positive control group with 8‐hour occlusion without recanalization (n=5; 8hO/0hR); or a sham group with 8 hours without occlusion (n=3; 0hO/8hR). Sample size calculation is described in Data S1. Group assignment was blinded to the surgeons and was revealed only after the targeted MCAs were dissected and completely prepared for occlusion with vascular clips. Animals remained under general anesthesia for the entire experimental procedure. The protocol was designed to ensure equal anesthesia and protocol duration across all groups by adjusting the timing of MCA occlusion to align with the recanalization phase (Figure S1). At the end of the study protocol, animals were euthanized, and the brain was collected for further analyses. The primary outcome was infarct size determined by 2,3,5‐Triphenyl‐2H‐tetrazolium chloride (TTC) staining.

All procedures were conducted in accordance with the Animal Research: Reporting of In Vivo Experiments guidelines ¹¹ and adhered to the Guiding Principles in the Care and Use of Animals set by the American Physiological Society. The study was approved by the Animal Care Committee of Erasmus University Rotterdam (AVD1010020198546).

Animal Preparation and Surgical Procedures

Overnight fasted animals were sedated with zoletil (tilatemine plus zolazepam [6 mg/kg]) plus xylazine [2.25 mg/kg] plus atropine (30 μg/kg) IM. Anesthesia was induced with propofol 1% [2 mg/kg, IV]. After orotracheal intubation, volume‐controlled mechanical ventilation was started (minute ventilation=6 mL/kg; respiratory rate=20 resp/min; fraction of inspired oxygen=0.30; and positive end‐expiratory pressure=4 mm Hg) and adjusted, if needed, to maintain normoxia and normocapnia on the basis of arterial and venous blood gas measurements. Anesthesia was maintained with propofol 2% (2.5–16 mg/kg per h IV] and sufentanil (25 μg/mL) (1–2 mg/kg per h IV). Catheters were inserted through the femoral artery and femoral vein to establish arterial and central venous access. Additionally, a catheter with temperature sensor was placed in the urinary bladder. Throughout the whole surgical procedure, animals were continuously monitored by ECG, peripheral oxygen saturation (pulse oximetry), end‐tidal CO₂, invasive arterial blood pressure and central venous pressure, and periodic urine production to maintain vital measurements within the physiological range. Physiological parameters including body weight, heart rate, systolic and diastolic blood pressure and core temperature were collected at baseline before craniectomy.

Once instrumented, animals underwent a right craniectomy. The animals were placed in a left lateral position and a semilunar incision of the skin posterior to the eye was created, from the zygomatic process of the temporal bone to the frontal bone (above the supraorbital process). The temporal muscle was then excised to expose the skull. A surgical window in the temporal bone was opened using a surgical drill (Saber drill, Core console, Stryker). Using a surgical microscope (Zeiss), the bone opening was carefully extended toward the skull base to gain access to the circle of Willis and the origin of the MCAs to allow occlusion using aneurysm clips (4/5/7/9 mm, B. Braun), which were removed at specific time points to allow timed recanalization of the MCAs.

Cerebral Digital Subtraction Angiography and Area at Risk Assessment

Digital subtraction angiography (DSA) (Artis Zee, Siemens Healthengineers, Netherlands) of the cerebral vasculature was performed at different time points and for different purposes. The contrast injection protocol to obtain 3‐dimensional angiography is described in detail in a previous study. ¹² Three‐dimensional rotational DSA images were postprocessed and reconstructed in situ using Leonardo DynaCT and InSpace 3‐dimensional software (Siemens Medical Solutions) to evaluate the anatomy of the MCAs and confirm their correct occlusion.

Coronal 3‐dimensional DSA angiography performed right after the onset of the MCA occlusion was used to obtain the area at risk (AAR) of infarction as a region of interest (ROI). AAR was quantified using the following method: The coronal 3‐dimensional DSA angiography was imported into ImageJ (National Institutes of Health) as a stack, consisting of 507 frames for each animal, with each frame having an original thickness of 0.2 mm. To enhance the ease and accuracy of brain tissue identification, every 20 frames were compressed into a single DSA image (thickness, 4 mm) to achieve a higher resolution. A single researcher, blinded to group allocation, individually reviewed the DSA images in random order, comparing the grayscale of the left and right hemispheres to identify areas without contrast flow. The AAR and the entire brain border were manually traced using the polygon selection tool. The AARs were multiplied by 4 mm (thickness) and summed to calculate the total AAR volume. The AAR results were expressed as a percentage of the right brain hemisphere volume also measured by DSA.

Perfusion Analysis

Coronal 3‐dimensional DSA angiography at baseline (before craniectomy), occlusion, 2 hours, and 4 hours of reperfusion were imported into ImageJ as stacks, and processed into 4‐mm DSA images using the same method as described above for the AAR. The AAR ROIs from the occlusion time points were transferred into corresponding baseline and reperfusion DSA images at the same position. Then the mean gray value inside AAR ROIs at the different time points was measured. In this method, the mean gray value reflects the total amount of contrast agent present in arteries, veins, and parenchymal tissue over a specific time, serving as a proxy for cerebral blood volume. The parameter primarily reflects the distribution and relative intensity of contrast agent perfusion in the ROI within a specific time. Therefore, for the sake of narration, we define this parameter as “perfusion level” in this article, represented by the mean gray value within the ROIs. More detailed descriptions of image adjustments can be found in Data S1.

Infarct Size Assessment

All animals were euthanized under deep anesthesia induced by propofol (intravenous bolus), followed by the administration of potassium chloride (14.9%, 20 mEq IV bolus; B. Braun, Netherlands) intravenously. Before euthanasia, unfractionated heparin (10 000 IU IV bolus; Leo Pharma, Netherlands) was administered to reduce postmortem clotting. Immediately after removal of the skull, the brains were placed in ice‐cold PBS and sliced into 5‐mm thick coronal sections using a pig brain slicer/matrix (Zivic Instruments, Pittsburgh, PA). The slices were incubated in a solution of 1% TTC in PBS1x at 37 °C for 30 minutes and photographed on both sides (Nikon D5300, Japan) as previously reported. ¹³ A single researcher, blinded to group allocation, individually reviewed the TTC images in random order. ImageJ was used to outline the entire brain border and TTC‐negative areas. The infarct volume of each slice was calculated as the average of the TTC‐negative areas on both sides, multiplied by the slice thickness (5 mm). Infarct size was expressed as a percentage of the total infarct volume relative to the volume of the infarcted hemisphere. The normalized infarct size, which was adjusted by the corresponding AAR size, was expressed as the percentage of infarct size to AAR size (IS/AAR [%]).

Immunostaining of Platelet Aggregations

Formalin‐fixed coronal brain slices were divided into 4 equal parts, processed for histology, and embedded in paraffin. Platelet aggregates were examined only in the superior regions of the brain slices, at the level of the caudate nucleus, where most infarcted tissue was located and where manipulation during surgery was minimal. Embedded tissues were sectioned into 5‐μm‐thick slices and stained for the platelet marker CD61 (MAB4190; Cell Marque, Merck, Germany) using 3,3′‐diaminobenzidine as a chromogen (DAKO K500711‐2, Netherlands) with nuclei counterstained briefly with hematoxylin).

The CD61‐stained sections were digitized using a Nanozoomer version 2.9.25 (Hamamatsu Photonics K. K., Japan) and analyzed with an open‐source image analysis tool (Orbit image analysis software, version 3.64; www.orbit.bio). A segmentation model was trained to identify platelet aggregates, defined as clusters of at least three 3,3′‐diaminobenzidine–positive cells. A researcher blinded to group allocation manually traced ROIs for both the right (infarcted) and left (noninfarcted) hemispheres, as well as for AAR, on the basis of corresponding 3‐dimensional DSA images. The region outside AAR ROI was defined as remote ROI. The optimized segmentation model was then applied to each ROI to quantify platelet aggregates and given as number/mm² of brain tissue.

Coagulation Markers in Serial Plasma Samples

Central venous blood samples were drawn from the sheath placed in the femoral vein and collected in 3.2% (w/v) sodium citrate and EDTA tubes at different time points described in Figure S1. After collection, plasma was prepared by double centrifugation at 2500g for 10 minutes each, at 21 °C (break off), with plasma stored (−80 °C) for contact activation marker analyses. Thrombin:antithrombin (common pathway), factor Xa:antithrombin (common pathway), factor IXa:antithrombin (intrinsic pathway), factor XIa:antithrombin (intrinsic pathway), factor VIIa:antithrombin complexes (extrinsic pathway) were measured in citrated plasma and factor XIIa:antithrombin complexes (intrinsic pathway) were measured in EDTA plasma using in‐house developed and validated ELISAs as described elsewhere. ¹⁴ , ¹⁵ , ¹⁶ Standard curves consisting of known concentrations of each complex were used to assess levels in individual samples.

Statistical Analysis

To study the effect of ischemia–reperfusion, comparisons were made only among the 3 recanalization groups (1hO/4hR, 2hO/4hR, and 4hO/4hR), while the 0hO/8hR and 8hO/0hR groups were used as references but were not included in the statistical analysis. Data normality was assessed using the Shapiro–Wilk test. Data are presented as either mean±SD or median (interquartile range). For multiple group comparisons, either 1‐way ANOVA or Kruskal–Wallis was used on the basis of data distribution. All statistical analyses were performed using SPSS version 28.0.1.0 (IBM, Armonk, NY). Statistical significance was set at P<0.05 (2‐tailed).

Infarct sizes were compared between 1hO/4hR, 2hO/4hR, and 4hO/4hR groups using the Kruskal–Wallis test. Spearman's rank correlation analysis was used to assess the correlation between infarct size and ischemia duration. Linear regression was performed to determine the association between multiple factors (ischemia duration and AAR size) and infarct size.

Perfusion levels at occlusion and postrecanalization were expressed as relative ratio to baseline. Changes in perfusion levels at occlusion and 2‐hour and 4‐hour recanalization were studied using a 1‐sample t test with a reference perfusion level of 1.0. For this analysis, data from the 1hO/4hR, 2hO/4hR, and 4hO/4hR groups were pooled to increase statistical power. Spearman's rank correlation analysis was used to evaluate the relationship between changes in perfusion levels and ischemia duration.

The microclot quantification comparisons were performed between the infarcted hemisphere (right) and the noninfarcted hemisphere (left), as well as across different ROIs, using the Wilcoxon signed‐rank test. Effects of ischemia and reperfusion on microclots were studied by pooling 1hO/4hR, 2hO/4hR, and 4hO/4hR groups. The normalized microclot quantification is presented as the difference between microclot density in the ROIs and left hemisphere (LH). Differences due to ischemia duration were studied by comparing the normalized microclot data among the 3 ischemia–reperfusion groups using 1‐way ANOVA. Correlations between normalized microclot quantification and ischemia duration were assessed using Spearman's rank correlation analysis.

Plasma levels of each coagulation marker were normalized by the relative ratio to baseline value after craniectomy. The coagulation marker levels were compared between 1hO/4hR, 2hO/4hR, and 4hO/4hR groups at time points after recanalization, using a generalized linear mixed model. The model set different time points as repeated measures and included time and group as fixed factors and animal ID as random effect to account for individual variability.

RESULTS

All 23 animals completed the experiment without major complications. Vitals, including heart rate, blood pressure, and fluid balance, were successfully maintained within the targeted ranges. All animals were included in the baseline comparison and infarct size analysis. However, 2 animals (from the 1hO/4hR and 4hO/4hR groups, respectively) showed no AAR ROI in the superior portion of the brain slice that was used for microclot density and were therefore excluded from the platelet aggregate analysis.

Baseline Physiological Parameters

The 5 groups did not significantly differ in body weight, heart rate, systolic and diastolic blood pressure, or temperature at baseline (Table S1).

Infarct Size, AAR, and Duration of Ischemia

The MCA clipping and unclipping methodology successfully created strokelike infarctions, as it resulted in well‐contrasted and extensive TTC‐negative areas in all the animals of the 2hO/4hR and 4hO/4hR groups. The 1hO/4hR group showed small to nonexistent TTC‐negative areas within the AAR. As for the control groups, TTC‐negative areas were extensively observed in the right hemisphere (RH) of the animals that underwent 8‐hour occlusion without recanalization but were not observed in the sham group.

Representative illustrations of infarction by TTC staining and AAR at the level of the caudate nucleus of each group are shown in Figure 1. The quantitative data of TTC‐determined infarct size, AAR size, and IS/AAR are summarized in Table 1. Briefly speaking, the AAR size was not significantly different among 1hO/4hR, 2hO/4hR, and 4hO/4hR groups (P=0.6). The TTC‐determined infarct size and IS/AAR all significantly differed among the three groups (resp. P=0.03, P=0.01). Differences of IS/AAR between 1hO/4hR, 2hO/4hR, and 4hO/4hR groups are also shown in Figure 2A. Additionally, infarct size seemed to reach a maximum after 4‐hour occlusion. A significant positive relationship was observed between IS/AAR ratio and ischemia duration (r=0.8, P<0.001; Figure 2B). Both ischemia duration and AAR size emerged as independent predictors of infarct size in multivariate linear regression analysis (adjusted R ²=0.6, P=0.001; Table 2).

Figure 1. Examples of TTC‐stained infarct and DSA‐defined AAR in 5 groups.

A, AAR, indicated by yellow line, appears as a dark region lacking contrast in 3‐dimensional DSA images. B, TTC staining. Red: live tissue; white: dead tissue/infarction. AAR indicates area at risk; DSA, digital subtraction angiography; L, left (noninfarcted) hemisphere; R, right (infarcted) hemisphere; and TTC, 2,3,5‐triphenyl‐2H‐tetrazolium chloride.

Table 1.

Infarct Size, AAR Size, and AAR Normalized Infarct Size in 1hO/4hR, 2hO/4hR, 4hO/4Hr, and 8hO/0hR Groups.

	1hO/4hR (n=5)	2hO/4hR (n=5)	4hO/4hR (n=5)	8hO/0hR (n=5)	P value*
AAR size (% hemisphere)
Median (IQR)	26.51 (14.65, 32.09)	28.47 (25.02–33.29)	30.49 (17.67–32.37)	26.57 (22.29–30.95)	0.6
Range	9.82–33.39	22.76–37.29	11.44–34.10	22.21–32.41
Infarct size (% hemisphere)
Median (IQR)	0.12 (0.03–5.54)	6.74 (6.48–17.96)	20.35 (7.29–29.15)	18.60 (12.48–24.40)	0.03
Range	0.00–10.00	6.27–20.75	4.35–33.50	10.55–27.06
Infarct size/AAR (%)
Median (IQR)	0.66 (0.32–20.59)	27.54 (24.12–55.77)	66.44 (40.41–89.77)	64.63 (55.11–82.65)	0.01
Range	0.00–37.70	23.69–70.85	38.01–98.23	47.17–83.45

AAR indicates area at risk; and IQR, interquartile range.

^*Statistical analysis was conducted only among 1hO/4hR, 2hO/4hR and 4hO/4hR groups.

Figure 2. AAR normalized infarct size increases with ischemia duration.

A, Normalized infarct size (% infarct relative to AAR) differs between 1hO/4hR, 2hO/4hR, and 4hO/4hR groups. B, Positive correlation between normalized infarct size and ischemia duration. **P<0.01. AAR indicates area at risk; IS, infarct size; O, occlusion; and R, recanalization.

Table 2.

Association Between Ischemia Duration, AAR Size, and Infarct Size

	Multivariate analysis*
	B (SE)	β	P value
Ischemia duration (h)	4.97 (1.32)	0.63	0.003
AAR size (% half brain)	0.63 (0.22)	0.48	0.01

This table describes the contributions of both ischemia duration and AAR to the infarct size as determined by TTC. AAR indicates area at risk; and TTC, 2,3,5‐triphenyl‐2H‐tetrazolium chloride.

^*Adjusted R ²=0.6 and P=0.001 for the linear regression equation. The multivariate model incorporated the following variables: ischemia duration, and AAR size.

Higher Perfusion Levels After Recanalization

Perfusion levels in AAR ROIs (representing the potentially infarcted tissue) were measured at baseline, during occlusion, and at the 2‐hour and 4‐hour recanalization time points (Figure 3A). The quantitative data are summarized in Table S2, Figure 3, and Figure S2. In summary, a significantly decreased perfusion level was observed during occlusion (P<0.001; Figure 3B), confirming the occlusion of the MCAs. Significantly higher perfusion levels were observed in all groups both after 2‐hour recanalization (P<0.001; Figure 3B), and 4‐hour recanalization (P<0.001; Figure 3B). Perfusion levels in sham and positive control group demonstrated stability across all time points (Figure S2). No significant relationship was found between perfusion intensity and ischemia duration at either 2‐hour (P=0.1; Figure 3C) or 4‐hour recanalization (P=0.2; Figure 3D).

Figure 3. Hyperperfusion after recanalization within the AAR.

A, Perfusion analysis example at baseline, occlusion, 2‐hour, and 4‐hour recanalization. Red circle marks the AAR region. B, Average perfusion density in AAR during occlusion, 2‐hour, and 4‐hour recanalization, shown as a ratio to baseline. Dashed line=hypothetical normal level (1.0). C and D, No correlation between perfusion intensity and ischemia duration at either 2‐hour or 4‐hour recanalization. ***P<0.001. AAR indicates area at risk.

Presence of Microclots

Platelet microclots in the cerebral vasculature are shown in Figure 4A. Their presence was observed with varying frequencies in both the infarcted RH and noninfarcted LH in all animals. Compared with LH, the infarcted RH showed a significant increase in microclot presence (P=0.003; Figure 4B). Interestingly, we also observed increased microclot presence in the positive control group, which underwent 8‐hour occlusion without recanalization. The quantitative data are summarized in Table S3.

Figure 4. Presence of microclots after occlusion.

A, Typical example of microclots (black arrows). B, Comparison of microclot density between nonischemic hemisphere (LH) and ischemic hemisphere (RH). Ischemic hemispheres show significantly more microclot presence than the nonischemic side. **P<0.01. Scale bar = 50 μm. AAR indicates area at risk; LH, left hemisphere; and RH, right hemisphere.

Regional Differences in Microclot Presence

Both the tissue within and outside the AAR showed a significant increase in microclot presence compared with the LH (LH versus AAR: P=0.003; LH versus remote: P=0.006; Figure 5A and 5B). Additionally, the microclot density in the AAR was higher than in the remote ROI (P=0.04; Figure 5C). The quantitative data of each ROI are summarized in Table S3.

Figure 5. Microclot density across ROIs.

A, Microclot density is significantly higher in the AAR region than in the noninfarcted hemisphere (LH). B, Remote region shows significantly higher microclot density than the LH region. C, AAR region has significantly more microclots than the remote region. *P<0.05, **P<0.01. AAR indicates area at risk; and LH, left hemisphere.

Microclot Presence and Ischemia Duration

Some microclots were also observed in the sham group and the LH of the rest of the groups, presumably due to the surgical procedure and brain collection. To minimize the impact of protocol‐derived microaggregates, microclot density in the LH was subtracted from all analyses of the RH. When considering the entire RH, a significant difference was observed between the 1hO/4hR, 2hO/4hR, and 4hO/4hR groups (P=0.04; Figure 6A). In the AAR, comparisons among the 3 groups also showed a significant difference in the presence of microclots (P=0.04; Figure 6C). In the remote ROI, no statistical differences were observed among the groups (P>0.05; Figure 6E).

Figure 6. Normalized microclot density and ischemia duration.

A, In the RH, 4hO/4hR group shows significantly higher normalized microclot density (RH–LH) than 1hO/4hR and 2hO/4hR groups. B, Positive correlation between ischemia duration and normalized microclot density is observed in RH. C, In the AAR region, the 4‐hour occlusion group has significantly increased normalized microclot density (AAR–LH) compared with other groups. D, Positive correlation between ischemia duration and normalized microclot density is observed in AAR regions. E, No difference in normalized microclot density (remote–LH) among occlusion groups in the remote region. F, No correlation between ischemia duration and normalized microclot density is found in the remote region. *P<0.05. AAR indicates area at risk; LH, right hemisphere; and RH, right hemisphere.

In the entire RH (ie, AAR plus remote), a significant positive correlation was found between normalized microclot density and ischemia duration, with more microclots with longer ischemia duration (r=0.7, P=0.007; Figure 6B). In the AAR, a similar significant positive correlation was observed (r=0.6, P=0.02; Figure 6D), while in the remote ROI, no significant correlation was found (P=0.1; Figure 6F).

Coagulation Markers

The plasma levels of all 5 coagulation markers showed no statistical differences among 1hO/4hR, 2hO/4hR, and 4hO/4hR groups at multiple time points after recanalization. The data are summarized in Table S4. Factor Xa:antithrombin and factor XIIa:antithrombin complexes were not analyzed due to insufficient data points, with >50% of samples having undetectable levels of these coagulation‐inhibitor complexes.

DISCUSSION

In this study, we investigated the effects of ischemia duration on IMR using a swine MCA occlusion model with an incremental occlusion duration. As expected, cerebral infarct size after ischemia and 4 hours of recanalization increased with ischemia duration and was influenced by the size of the AAR. The key findings were as follows: First, an increasing presence of microclots was observed within AAR, and there was a strong correlation between microclot density and ischemia duration. Second, microclots were found throughout the entire ischemic hemisphere, including tissue not directly impacted by ischemia. Third, within AAR, hyperperfusion was observed concomitant with an increasing presence of microclots. Finally, no coagulation activation was detected in systemic plasma. Importantly, our results were not influenced by surgery or anesthesia, as all groups were fully randomized and underwent identical surgical procedures and anesthesia durations, including the controls.

Ischemia Duration and Infarct Size

The infarct size of our model increased with ischemia duration, which aligns with findings of previous studies in rodent models. ¹⁷ , ¹⁸ Compared with the 8hO/0hR group, the 4hO/4hR showed larger infarct sizes and greater variability.

Area at Risk

In our model, we observed variants with 2 to 4 MCAs, which were also reported in previous studies. ¹⁹ We used DSA to define the AAR territory, which reflects the MCA branching and collateral patterns. Measuring the AAR is a standard practice in cardiology infarction studies, as it is a crucial determinant for normalization of infarct size. ¹³ , ²⁰ The results indicate that in our study AAR size also significantly affects infarct size. The high variability of MCAs and individual differences in collateral flow may be possible reasons for infarct size variations observed in previous swine MCA occlusion models. ²¹ , ²² , ²³ Moreover, based on our data, 11 animals per group are estimated to demonstrate a 50% reduction in AAR‐normalized infarct size (power of 0.80, α=0.05), while 26 animals per group would be required to detect the same reduction without AAR normalization. This approach of IS/AAR ratio therefore effectively reduces sample size while maintaining statistical power.

Ischemia Duration and Microclot Presence

Our findings demonstrated a positive correlation between ischemia duration and microclot density within the ischemic tissue. Evidence from previous research suggests 2 key factors that contribute to the progression of endothelial damage and inflammation over time. First, a longer duration of ischemia compromises endothelial integrity, increasing vascular permeability and exacerbating damage to endothelial cells. ²⁴ As a critical regulator of homeostasis, the endothelial surface within the microvasculature shifts during ischemia from an anticoagulant to a procoagulant state, promoting the aggregation of platelets, together with fibrin, and leukocytes. ²⁵ Second, as ischemia duration extends, platelets become progressively activated, releasing proinflammatory mediators like P‐selectin, which interact with endothelial cells to amplify tissue damage by recruiting more immune cells to the injury site. ²⁶ , ²⁷ The ischemic time‐related microvascular impairment is supported by findings from rodent models showing that prolonged ischemia significantly increases microcirculatory perfusion deficits. ²⁸ , ²⁹

Microclots were observed in the positive control group with 8 hours of occlusion without recanalization, suggesting that microclot formation occurs during prolonged ischemia even in the absence of reperfusion. This finding aligns with a previous rodent study showing microthrombi formation early after occlusion. ³⁰ However, contrary to the rodents where microclots dissolved after reperfusion, our swine model demonstrated persistent microclots even after recanalization. Notably, the 4hO/4hR exhibited even higher microclot density with considerable variability, suggesting a complex interplay of ischemia–reperfusion effects. On one hand, blood flow restoration may flush away nonadherent cells, potentially improving microcirculatory patency. ³⁰ On the other hand, it has been described that reperfusion triggers the production of reactive oxygen species and other metabolites, which activate platelet aggregation and enhance interactions with injured endothelial cells and leukocytes. ³¹ A dog model reported that the extent of microvascular obstruction significantly increased over the first 48 hours following reperfusion after myocardial infarction. ³² This dual impact of ischemia–reperfusion highlights that reperfusion influences microclot formation in a multifaceted manner. Furthermore, the 2 potential phases of microclot formation, during occlusion and after recanalization, highlight the potential benefit of combining thrombectomy with thrombolysis. The CHOICE (Chemical Optimization of Cerebral Embolectomy) trial suggested that administering intra‐arterial alteplase at the end of an endovascular procedure improved clinical outcome, potentially through enhancing microcirculatory reperfusion after thrombectomy supporting better functional recovery. ³³ However, a recent meta‐analysis challenges the effect size of the CHOICE trial. ³⁴ Given the current clinical evidence, it remains unclear to what extent microvascular‐level impairments exist and how it contributes to poststroke recovery. ³⁴ , ³⁵ , ³⁶ In this context, the discovery of microclot formation after recanalization in our swine model provides valuable insights and avenues for further investigation.

Microclots Outside the AAR ROI

The detection of a relevant number of microclots in remote ROIs of the infarcted hemisphere is particularly intriguing. It indicates that ischemia may provoke a clotting response that extends beyond the directly affected area. It cannot be ruled out that ischemia‐derived clotting mechanisms are having an effect outside the ischemic area via passage of reactive oxygen species or other metabolites. Nonetheless, other mechanisms, such as signaling via spreading depolarizations, could also be involved in the presence of microclots in remote cerebral tissue. Spreading depolarization originates from the ischemic core and has been shown to invade adjacent healthy zones with intact blood supply in both human and mouse models. ³⁷ , ³⁸ The vasomotor and hemodynamic responses to spreading depolarization, observed in microvessels, ³⁹ could potentially lead to microvascular stall events. Supporting this, Törteli et al demonstrated in a mouse model that inhibiting spreading depolarization significantly improved reperfusion outcome. ⁴⁰ Another explanation could involve the intricate pattern of the brain microvascular network. A primate study found couplings between arterioles and venules not restricted in the anatomic nearest neighbors; instead, their locations do not follow any identified spatial pattern. ⁴¹ These long‐distance couplings may enable distant blood supply or drainage. After MCA occlusion, ischemia triggers the release of cytokines, damage‐associated molecular patterns, and other proinflammatory mediators within the ischemic territory. ⁴² These prothrombotic factors may propagate through the extensive microvascular network, extending their effects to nonischemic regions. Given these complexities, future studies should aim to distinguish the vascular origin of microclots by using vessel‐specific markers or advanced imaging techniques. This will provide a clearer understanding of microclot development across different vascular beds and their role in microvascular reperfusion failure. Our findings underscore the complexity of IMR and its dependency on ischemia duration and regional factors. The finding also suggests that postrecanalization management could benefit from strategies that not only focus on the ischemic core but also on the prevention of distant microvascular occlusions.

Hyperperfusion and Microclots

One of the most relevant findings of this study is that concomitant with the presence of microclots, we observed persistent hyperperfusion in the ischemic tissue until at least, 4 hours after recanalization. This was independent of occlusion duration. Hyperperfusion after cerebral ischemia–reperfusion has been observed in several studies in rodent models. Premilovac et al compared cortical vascular reperfusion following 45 minutes, 60 minutes, or 90 minutes of MCA occlusion, finding the increase in cortical reperfusion after MCA occlusion was not significantly different among the 3 groups. ⁴³ Franx et al also observed early development of persistent hyperperfusion in the postischemic lesion, regardless of occlusion duration. ⁴⁴ Hyperperfusion is of complex clinical significance. Once regarded as a symbol of successful reperfusion indicating favorable clinical outcome, more recent evidence revealed it could also reflect impaired autoregulation, predicting poor outcome. In animal models, early hyperperfusion after recanalization usually suggests poor tissue outcome, as hyperperfusion injury might contribute to secondary blood–brain barrier disruption and cell damage to cause postischemic tissue injury. ⁴⁵ Notably, the simultaneous presence of microclots and hyperperfusion could reflect the complex hemodynamic state within the ischemic region. A previous clinical systematic review reported that IMR upon complete recanalization appears relatively rare and reported that the incidence highly depends on the definition used and confounding factors. ⁴⁶ Clinical IMR is usually studied in the scope of hypoperfusion. However, microvascular compromise in hyperperfused tissue could easily be missed in a clinical setting where hypoperfusion is the primary focus. Additionally, current clinical tools usually lack the sensitivity to detect microvascular occlusions, further limiting their identification. Our animal model provides a critical insight, as it allows detailed pathological assessment of the ischemic regions. The simultaneous observation of hyperperfusion and microclot presence after recanalization in our model underscores the paradoxical nature of IMR. The coexistence of both infarcted and salvageable tissues within the area at risk may yield regions with microvascular obstruction interspersed with perfused areas, creating a mixed physiological environment within the AAR. These findings align with the former understanding that the responses of cerebral microvasculature to focal ischemia is highly heterogeneous, both spatially and temporally. ⁴⁷

Coagulation Activation

The presence of microclots and their pattern of association with ischemia duration did not align with systematic coagulation activation. This suggests that plasmatic coagulation does not contribute to the time‐dependent increase of microclot formation in this model. In a previous clinical study, a significant temporary increase of factor Xa:antithrombin and thrombin–antithrombin complex was observed at 1 hour after EVT. ⁴⁸ In another study on hemostasis biomarkers in patients with acute ischemic stroke undergoing EVT, coagulation biomarkers such as factor VIII, endogenous thrombin potential, and thrombospondin 1 repeats 13, showed higher levels at 24 hours after EVT compared with before EVT. ⁴⁹ Additionally, our previous swine model of thromboembolic occlusion and EVT showed that EVT with stent‐retriever or direct aspiration thrombectomy triggers activation of the coagulation. ⁵⁰ However, the activation was not observed in the current model, suggesting that EVT, rather than ischemia–reperfusion alone, may be responsible for coagulation activation. Furthermore, in our study, factor Xa:antithrombin and factor XIIa:antithrombin complex concentrations were below detectable levels in most of the animals. This discrepancy could be due to differences in assay sensitivity across species, the use of healthy swine without atherosclerosis, or the lack of an embolized clot responsible for the stroke, which may not fully reflect the coagulation conditions observed in patients with clinical embolic stroke with atherosclerosis. As a result, strong conclusions cannot be drawn from these findings.

Another possible mechanism could involve other hemostasis biomarkers related to platelet aggregation. A previous clinical study found that lower levels of soluble glycoprotein VI are associated with poorer outcomes in patients undergoing EVT, suggesting that increased expression of platelet glycoprotein VI might contribute to platelet‐dependent thrombus formation. ⁴⁹ This presents a potential direction for future investigations into the mechanisms underlying microclot formation.

Limitations

Our study has several limitations that should be noted. First, we did not differentiate between the vascular origins and location of the platelet aggregates, so it remains unclear whether microclots developed in arterioles or venules. Additionally, we did not include other components, such as fibrinogen/fibrin, which may potentially underestimate the abundance of fibrin‐rich microclots. Second, our study concluded observations at 4 hours after recanalization, which means we lack data on potential infarct growth and long‐term functional outcomes. It is possible that infarct size and microclot presence continues to change over a longer period. Future studies will aim to extend the observation period to assess the long‐term evolution of microclots, infarct progression, and functional outcome. Third, our method of perfusion intensity assessment relied on the gray value detection in ImageJ, with brightness and contrast adjustments set to a reference unit. Importantly, statistical stability was observed across sham groups at different time points, suggesting this method's reliability. We did not compare the accuracy and reproducibility of our method with other imaging, hampered by metallic microclip presence, as it was outside the scope of our study. In future studies, we plan to use multiple imaging techniques to cross‐validate our findings. Our swine model of recanalized acute ischemic stroke based on MCA clipping does not incorporate the presence of an occluding thrombus, which could be a source of microclots and coagulation activation when pharmacologically or mechanically removed and further increase the presence of microclots in the distal cerebral vasculature. Further studies could be conducted using a thromboembolic stroke model to evaluate whether the presence of an occluding thrombus and its removal influences downstream microclot formation. Finally, we used healthy swine without comorbidities and included only prepubescent female animals (to allow urinary bladder catheterization without surgery and increase physiological homogeneity), which does not fully represent the complex human stroke population. Future studies should include both male and female animals with a larger sample size to explore potential sex differences on microclot formation and improve generalizability. Of note, the Highly Effective Reperfusion Evaluated in Multiple Endovascular Stroke Trials (HERMES) study showed no significant difference in treatment effect between men and women for endovascular treatment in acute ischemic stroke. ⁵¹

CONCLUSIONS

Our study demonstrates that ischemia impacts both infarct size and microclot presence within the ischemic region, both increasing with ischemia duration. Microclots were found throughout the ischemic hemisphere, even beyond the infarct core. Interestingly, microclot formation coincided with hyperperfusion in the area at risk. These findings highlight the need for further research into the mechanisms of incomplete microvascular reperfusion and adjunctive therapies to mitigate infarct progression and improve outcome.

Sources of Funding

The authors acknowledge the support of the Netherlands Cardiovascular Research Initiative, which is supported by the Dutch Heart Foundation (CVON2015‐01: CONTRAST) and the Brain Foundation Netherlands (HA2015.01.06), and the support of Health~Holland, Top Sector Life Sciences & Health (LSHM17016), Medtronic, Cerenovus, and Stryker European Operations BV. The collaboration project is additionally financed by the Dutch Ministry of Economic Affairs by means of the public–private partnership allowance made available by the Top Sector Life Sciences & Health to stimulate public–private partnerships. M.N. was supported by the Dutch Heart Foundation (03‐006‐2022‐0052).

Disclosures

Prof Dippel reports funding from the Dutch Heart Foundation, Brain Foundation Netherlands, The Netherlands Organization for Health Research and Development, and Health Holland Top Sector Life Sciences and Health and unrestricted grants from Penumbra Inc., Stryker European Operations BV, Medtronic, Thrombolytic Science, LLC, and Cerenovus for research, all paid to the institution. Dr van Beusekom reports funding from the Dutch Heart Foundation and consultancy fees from Thuja healthcare investors paid to Erasmus MC. Prof ten Cate reports research support from Bayer and Synapse; and consultancy fees from Astra Zeneca, Viatris, Novostia, Galapagos, and Alveron. All revenues are deposited at the CARIM Institute to support investigator‐driven research. The other authors report no conflicts of interest.

Supporting information

Data S1

Tables S1–S4

Figures S1–S2