Postresuscitation Cerebral Vasospasm and Capillary Failure After Experimental Asphyxial Cardiac Arrest Authors: Christian Staehr, Cecilie Munch Johannsen, Victoria Hinkley, Asger Granfeldt, Eugenio Gutiérrez‐Jiménez Published: 2026 ABSTRACT AbstractBackgroundDisrupted cerebral blood flow is suggested to contribute to secondary brain injury after cardiac arrest (CA). This study aimed to investigate cerebral microcirculation after return of spontaneous circulation (ROSC) following asphyxial CA. We hypothesized that contraction of pericytes after ROSC compromises cerebral capillary blood flow.MethodsTransgenic C57BL/6NRj mice (n=19) of both sexes were studied. Mice were expressing tdTomato as a fluorescent reporter under the control of the platelet‐derived growth factor receptor β promoter to label pericytes. Chronic cranial windows were implanted 3 weeks before the experiment. Four minutes of asphyxial CA was followed by cardiopulmonary resuscitation. Two‐photon microscopy assessed cerebral hemodynamics in the same cohort of mice at both 3 and 24 hours after ROSC.ResultsOf 13 mice in the CA group, 9 achieved ROSC; 6 and 5 mice survived to 3 and 24 hours after ROSC, respectively. Arterial blood pressure was similar between groups 3 and 24 hours after ROSC. At 3 hours after ROSC, pial arteries and penetrating arterioles were constricted in the CA group compared with sham (arteriole diameter, 12.2 μm [95% CI, 10.9–13.4] versus 15.6 μm [95% CI, 13.8–17.5] in CA and sham; P=0.003). Similarly, first‐ and second‐ to third‐order capillaries showed reduced diameters 3 hours after ROSC (first‐order diameter, 3.9 μm [95% CI, 3.5–4.2] versus 5.3 μm [95% CI, 4.8–5.9] in CA and sham; P=0.002). The vasoconstriction was associated with slower red blood cell velocities throughout the capillary network (for second‐ to third‐order capillaries upstream from venule; 0.78 mm/s [95% CI, 0.46–1.09] versus 2.14 mm/s [95% CI, 1.73–2.55]; P<0.001) and increased capillary flow stalling. Artery‐to‐vein mean transit time was increased and relative transit‐time heterogeneity was decreased 3 hours after ROSC. By 24 hours after ROSC, vessel diameters, blood flow velocity, transit time, and capillary stalling were not different compared with sham.ConclusionsCerebrovascular vasospasm 3 hours after ROSC was associated with impaired cerebral microcirculation and increased capillary flow stalling.Keywords: capillaries, cardiac arrest, microcirculation, pericytes, resuscitationSubject Categories: Cardiopulmonary Resuscitation and Emergency Cardiac Care, Ischemia, Vascular Biology, Vascular Disease, Cerebrovascular Disease/Stroke Nonstandard Abbreviations and Acronyms cardiac arrest capillary transit‐time heterogeneity mean transit time return of spontaneous circulation Clinical Perspective What Is New? Cerebral vasospasm occurs early (3 hours) after return of spontaneous circulation and is associated with significant microcirculatory impairment. Cerebral vasospasm occurs early (3 hours) after return of spontaneous circulation and is associated with significant microcirculatory impairment. What Are the Clinical Implications? Microvascular impairment normalized by 24 hours after return of spontaneous circulation, indicating that cerebral microcirculatory dysfunction peaks during the early post–return of spontaneous circulation period. Microvascular impairment normalized by 24 hours after return of spontaneous circulation, indicating that cerebral microcirculatory dysfunction peaks during the early post–return of spontaneous circulation period. These findings suggest a potential therapeutic window within the first hours after ROSC, during which interventions aimed at reducing cerebral vasospasm could be beneficial. These findings suggest a potential therapeutic window within the first hours after ROSC, during which interventions aimed at reducing cerebral vasospasm could be beneficial. Neurological injury is the leading cause of disability and death after successful resuscitation from cardiac arrest (CA), accounting for approximately two thirds of all deaths. 1 , 2 Following return of spontaneous circulation (ROSC), many patients experience progressive neurological deterioration, driven by secondary brain injury. Multiple mechanisms have been suggested to be implicated in secondary brain injury after ROSC, including ischemia–reperfusion injury, blood–brain barrier dysfunction, cerebral edema, and cerebrovascular dysregulation. 1 Impaired cerebrovascular autoregulation and fluctuations in cerebral blood flow have been reported after ROSC, resulting in regional hypo‐ or hyperperfusion. 3 , 4 The mechanisms underlying these heterogeneous perfusion patterns remain poorly understood, but may involve altered blood viscosity, 5 microthrombus formation, 6 , 7 and dysregulated cerebrovascular tone. 8 , 9 , 10 , 11 , 12 Cerebral capillary dysfunction has also been proposed as a contributing factor, 13 but in vivo characterization of the cerebral microcirculation after CA remains technically challenging. 4 , 8 , 10 , 12 , 14 , 15 , 16 Pericytes, which surround cerebral capillaries, play a critical role in maintaining blood–brain barrier integrity and regulating immune cell trafficking. Ensheathing pericytes, located on low‐order capillaries near penetrating arterioles, express contractile proteins that enable them to actively regulate capillary diameter. 17 , 18 , 19 , 20 In contrast, mesh and thin‐strand pericytes, which are surrounding higher‐order capillaries, are thought to have less contractile function. 21 , 22 Hypoxic exposure in acute brain slices has been shown to induce pericyte‐mediated capillary constriction, 23 , 24 a mechanism suggested to underlie microcirculatory dysfunction following ischemic stroke. 25 , 26 Another factor contributing to microcirculatory dysfunction after hypoxic brain injury is capillary stalling, which consist of transient episodes of capillary no‐flow associated with leukocyte plugging. 25 , 27 , 28 Both pericyte contraction and capillary stalling may contribute to heterogeneous capillary blood flow, known as capillary transit‐time heterogeneity (CTH), resulting in impaired oxygen extraction and clearance of metabolic waste from the brain. 29 We hypothesized that microvascular dysfunction following asphyxial CA is associated with capillary pericyte contraction and capillary stalling. To test this, we used high‐resolution in vivo imaging in mice implanted with chronic cranial windows to characterize vascular bed–specific hemodynamics in the brain at 3 and 24 hours after ROSC. Methods The data underlying this article are available in the article and in its online supplemental material. All experiments were conducted in accordance with a prespecified study protocol that complied with the Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes. The protocol was approved by the Animal Experiments Inspectorate of the Danish Ministry of Environment and Food (License No. 2024‐15‐0201‐01706) and reported in accordance with the Animal Research: Reporting of In Vivo Experiments guidelines. 30 Transgenic Mice C57BL/6NRj mice aged 18 to 20 weeks at the time of experimental CA were used in this study. A group of mice exposed to asphyxial CA were compared with a group of mice undergoing sham surgery. An equal number of females and males were included. Mice were genetically modified to express Cre recombinase under the control of the platelet‐derived growth factor receptor β promoter, a well‐characterized marker of pericytes and other mural cells. These mice were crossed with a Cre‐dependent Salsa6f reporter line, which encodes GCaMP6, a calcium‐sensitive green fluorescent protein and tdTomato, a calcium‐insensitive red fluorescent protein. However, analysis of calcium imaging was not performed for this study. Cre‐mediated recombination and reporter expression were induced by administering tamoxifen (0.2 mg/kg, oral gavage) once at least 4 weeks before experimental CA. Animals were housed under identical environmental conditions, including a 12:12 hour light/dark cycle and ad libitum access to food and water. Chronic Cranial Window Procedure At least 3 weeks before experimental CA, mice were implanted with a round chronic cranial window (Ø=3 mm) over the sensory barrel cortex of the left hemisphere, as previously described. 31 Briefly, following removal of the scalp, a craniotomy matching the window diameter was carefully drilled while preserving the dura mater. The glass coverslip was affixed to the skull using cyanoacrylate (Loctite, Düsseldorf, Germany). A custom‐made stainless steel headplate was then secured around the window using Scotchbond (3M, Maplewood, MN) and RelyX Universal Resin Cement (3M). For postoperative care, buprenorphine was administered subcutaneously (0.1 mg/kg) for 3 days. Mice with a >15% weight loss after surgery were excluded from further study (Figure S1). 32 Allocation Concealment and Blinding To ensure allocation concealment, a random number generator (Google) was used to randomly select which mouse to include on the day of the experiment. Outcome assessments, including image analyses and quantifications, were conducted by investigators blinded to group identity and timing (ie, 3‐hour versus 24‐hour follow‐up). For capillary flow stalling analysis, 2 investigators independently conducted blinded assessments, and the average of their results was used for statistical comparisons. While the analysis was conducted blinded, blinding was not feasible during the 2‐photon imaging phase, as the same investigator was responsible for imaging, resuscitation, and post‐ROSC care. CA Procedure Mice were fasted for 45 minutes before induction of anesthesia. Mice were anesthetized with 2% isoflurane (Scanvet Animal Health, Fredensborg, Denmark) in 21% O2 and maintained under mask ventilation (1.5% isoflurane in 21% O2). Two tail vein polyethylene tubing catheters (size 10, PE‐10) with a needle were placed, and a femoral arterial catheter (pulled PE‐10 catheter) was inserted via a left groin incision. The femoral artery was ligated distal to the insertion site, while the vein and nerve were preserved. Invasive arterial blood pressure was continuously monitored (BP‐1 system, World Precision Instruments, Sarasota, FL), and arterial blood was collected in 60 μL capillary tubes for blood gas analysis (ABL90 Flex; Radiometer Medical ApS, Brønshøj, Denmark). Blood gas results are reported only for mice that survived to the end of the 3‐hour follow‐up. Arterial and venous catheters were flushed with heparinized saline (50 IE heparin/mL; Panpharma, Le Clairay, France), and the groin incision was sutured. Mice were endotracheally intubated and mechanically ventilated (MiniVent 845; Harvard Apparatus, Holliston, MA) with end‐tidal CO2 monitored via micro‐capnograph (CI289; Columbus Instruments, Columbus, OH). Continuous monitoring and recording of 3‐lead ECG, blood pressure, and end‐tidal CO2 was performed using PowerLab (AD Instruments, Oxford, UK). Inspired O2 levels were measured with an R‐17MED oxygen sensor (Medical Scandinavia, Horsens, Denmark). For 24‐hour follow up, blood pressure was assessed using tail cuff plethysmography (CODA; Kent Scientific Corporation, Torrington, CT). Before CA, mice were ventilated with 21% O2 and 1% isoflurane, with a positive end‐expiratory pressure of 1.5 cm H2O and tidal volume set to 10 μL/g body weight. Respiratory rate was adjusted to an end‐tidal CO2 between 3 and 4% and to suppress spontaneous respiration, typically at a rate of 125/min for a 25‐g mouse. At 90 seconds before asphyxia, isoflurane was discontinued and rocuronium (10 mg/kg IV, 2.5 mg/mL, Hameln Pharma, Hameln, Germany) administered to prevent spontaneous respiration. Asphyxia was induced by turning off the ventilator. CA was defined by a mean arterial pressure <20 mm Hg and loss of arterial pulsations. After 4 minutes of CA, cardiopulmonary resuscitation was initiated, including manual chest compressions (330–360/min), mechanical ventilation with 100% O2, and epinephrine (0.1 mg/kg, 0.02 mg/mL IV; Bradex; FrostPharma, Danderyd, Sweden). ROSC was defined as mean arterial pressure >50 mm Hg. If ROSC was not achieved after 2 minutes, another bolus of epinephrine (0.1 mg/kg) was administered. If required the first 2 hours after ROSC, noradrenaline (0.2 mg/mL; Kalceks, Riga, Latvia) was continuously infused via syringe pump (model 11 Plus, Harvard Apparatus) and titrated to maintain mean arterial pressure >60 mm Hg. Ten minutes after ROSC, inspired oxygen was reduced from 100% to 50%. In the CA group, isoflurane was withheld from asphyxia onset until 3‐hour follow‐up imaging. Sham mice underwent the same surgical and pharmacological interventions, except for asphyxia, and remained on 1% isoflurane with matched gas concentrations. In Vivo 2‐Photon Microscopy Three hours after ROSC, mice were head‐fixed and maintained under mechanical ventilation with 50% O2 and 1% isoflurane. Two‐photon microscopy was performed using a Prairie Ultima IV In Vivo Laser Scanning Microscope (Bruker, Billerica, MA) and the PrairieView Software version 5.1 (Bruker), as previously described. 31 Excitation of fluorescent dyes was performed using a tunable Ti:Sapphire laser Chameleon (Coherent, Santa Clara, CA). Emission of fluorescence was detected using 2 Gallium Arsenide Phosphide photomultiplier modules (GaAsP‐PMT, H10770; Hamamatsu, Shizuoka, Japan). During imaging, a 20× water‐immersion objective (UMPLFLN; Olympus, Tokyo, Japan) was used, and a custom‐built heating system maintained the cranial window and immersion water at 37 °C. For imaging, a region of interest in the barrel cortex including both a pial artery branch from the middle cerebral artery and a pial vein was selected. For estimation of mean transit time (MTT) and CTH, we used an indicator‐dilution method as previously described. 33 Briefly, the concentration‐time curves of fluorescent dyes boluses were acquired through time‐series imaging using a spiral scanning mode at 12.6 frames/s for 60 seconds. Five seconds into each recording, an intravenous bolus of 10 μL albumin‐Alexa Fluor 680 (2.5 mg/mL in PBS, Carlsbad, CA) was administered. Boluses were injected at a rate of 20 μL/s using an infusion pump (GenieTouch; Kent Scientific Corporation). In each mouse, 50 μL albumin–Alexa Fluor was injected at each imaging session. Time‐series data were acquired at 160 × 160 pixel resolution (estimated pixel size, 3.68 μm; dwell time, 4 ms), using 850‐nm excitation for the albumin‐bound Alexa Fluor dye. The venous catheters and injected solutions were prewarmed to approximate physiological temperature using a heating circulation bath (ThermoScientific, Waltham, MA). Capillary diameter and red blood cell (RBC) velocity were assessed using line‐scan scanning mode at 4× digital zoom, with a dwell time of 3.5 to 4 ms, producing a scan period of 1.4 to 2.1 ms. Repetitive line‐scans were recorded over 30 to 40 seconds. The line‐scan path was defined using the free‐hand mode of the scanning software. Capillaries located 50–200 μm below the cortical surface were assessed. To systematically select capillaries for analysis, penetrating arterioles and venules were first identified, followed by stepwise imaging of capillary segments within their respective vascular trees. Dual‐channel fluorescence detection was performed to visualize both Alexa albumin‐bound intravascular dye and tdTomato‐labeled platelet‐derived growth factor receptor β–expressing pericytes. The tdTomato fluorescence was excited at 1100 nm using an optical parametric oscillator (Coherent Inc., Saxonburg, PA). Emission from fluorescent signals was collected using an et690/50 bandpass filter (Chroma Technology, Bellows Falls, VT) for the Alexa far‐red intravascular dye, and an et605/70 bandpass filter (Chroma Technology) for the tdTomato signal. Capillary stalling was assessed on time series, which consisted of 300 frames acquired at 0.5 frames/s using the Galvo scanning mode (field‐of‐view, 512 × 512; XY pixel size, 1.15 μm; dwell time, 6.5 ms). Rhodamine 6G (50 μL of 1 mg/mL; Thermo Scientific) was administered intravenously 5 minutes before acquisition to label leukocytes as previously described. 27 Using an excitation wavelength of 850 nm, fluorescence from rhodamine 6G was detected using an et525/50 bandpass filter (Chroma Technology). Structural angiograms were obtained using Z‐stacks with 3‐μm steps from 50‐ to 300‐μm cortical depth, in a 512 × 512 pixel XY plane with 1.15‐μm pixel size. Three fluorescent signals were recorded: the far‐red intravascular dye, green emission (rhodamine), and red emission for platelet‐derived growth factor receptor β–TdTomato‐expressing pericytes. A Z‐stack was recorded covering pial vessels and at a depth of ≈250 to 300 μm. The laser intensity and GaAsP‐ photomultiplier tube gain were gradually increased the deeper into the tissue the recording was made. The reason for choosing 3 and 24 hours after ROSC as follow‐up time points was that the 3‐hour time point represented the earliest imaging session permitting stable physiological conditions during head fixation. The 24‐hour follow‐up allowed evaluation of the persistence or resolution of early cerebrovascular alterations and potential delayed vascular responses, representing a clinically relevant subacute phase after CA when secondary brain injury typically evolves. The same regions of interest were scanned at the 3‐ and 24‐hour follow‐up scan with the same depth range. We aimed for the same capillary segments to be recorded at the 3‐ and 24‐hour imaging sessions by saving the X, Y, Z coordinates. However, in some cases, it was not possible to retrieve individual capillary segments. After the 24‐hour imaging session, mice were euthanized via transcardial perfusion. Under pentobarbital anesthesia (150 μL IP, 400 mg/mL; Alfasan, Utrecht, Netherlands), a cannula was inserted into the left ventricle, and an incision was made in the right atrium to allow for drainage. Mice were first perfused with 10 mL of heparinized PBS (50 IU heparin/mL), followed by perfusion with 50 mL 4% formaldehyde in PBS for fixation. Imaging Analysis MTT and CTH were estimated using the concentration‐time curves and modeling the tissue transport function (TF) through deconvolution, as previously described. 31 , 33 Briefly, the transport function was estimated from a pial artery and a draining pial vein, which served as the arterial input function and venous output function, respectively, for the scanned vascular network. Vessel types were confirmed by the sequential appearance of the intravascular dye following bolus injection, with arteries showing the earliest fluorescence peak and veins the latest. The transport function was modeled by a gamma probability density function with parameters α and β, and a scaling factor, f. MTT (α·β) was defined as the mean, and CTH as the SD (√α·β) of the function. For this study, an arterial input function and a venous output function, were analyzed over 3 scans. Line‐scan analysis was performed to estimate capillary RBC velocity and diameter using custom MATLAB scripts, as previously described. 31 , 33 RBC velocity was derived from axial line‐scan images using a Radon transform‐based algorithm, which estimates streak angles within 500‐ms windows, displaced by 250 ms per step. Capillary diameter was estimated from cross‐sectional line scans as the full width at half maximum of the intravascular dye intensity profile, averaged over 500‐ms intervals while excluding periods with RBCs. Median diameter and velocity values in each capillary segment were calculated. Pial and penetrating vessel diameters were assessed from Z‐stacks covering the depth range of the vessel of interest. The vessel diameter was assessed semiautomatically using VasoMetrics software. 34 Capillary stalling events were identified using a semiautomated approach based on a custom MATLAB graphical user interface previously described. 28 Automatically detected putative stalls were based on the absence of RBC movement across ≥2 consecutive frames within individual capillary segments. All candidate stalls were subsequently validated through manual inspection. Only capillaries with sufficient contrast to visualize RBC movement were included. A separate imaging channel simultaneously recorded rhodamine 6G‐labeled leukocytes, and this signal was incorporated into the capStall software to evaluate leukocyte involvement in stalling. Following the initial classification of capillary stalling events based on RBC movement, a secondary analysis was performed to assess leukocyte involvement using the corresponding signal. Stalls were classified as leukocyte associated if a leukocyte was visibly present within the same capillary segment for ≥2 consecutive frames. This classification required that the capillary segment had already been identified as stalled on the basis of the absence of RBC movement. Final outputs included mean stall duration, point prevalence, and stalling incidence. Incidence was defined as the percentage of capillaries exhibiting at least 1 stall during the 10‐minute time series and point prevalence as the proportion of capillaries exhibiting stalls among the total capillary analyzed per mouse at any given frame. The RBC stalling analysis was independently performed by 2 investigators, each conducting a complete and blinded assessment of the data. Sample Calculation Sample size was determined on the basis of power calculations performed in G*Power version 3.1.9.7 (Heinrich‐Heine‐Universität, Düsseldorf, Germany). Based on pilot data of first‐ to third‐order capillary diameters (mean±SD: sham, 6.07±1.31 μm; CA, 3.85±1.39 μm), a standardized effect size (Cohen's d) of 1.64 was calculated. To detect a statistically significant difference of P<0.05 with 80% power, the analysis indicated that 6 mice should be included per group. To accommodate an anticipated ROSC rate of ≈70% observed in pilot studies, additional mice were initially allocated to the CA group. Outcomes and Statistical Analysis The primary outcomes were capillary diameter and RBC velocity. Secondary outcomes included diameters of pial and penetrating vessel diameters, MTT, and CTH. Statistical analyses were performed using R Studio version 2024.12.1+563 and GraphPad Prism version 10 (GraphPad Software, La Jolla, CA). For data sets with multiple capillary measurements at each follow‐up scanning per mouse, group comparisons were performed using linear mixed models (Data S1 and S2). For pial artery diameter assessment, there was only 1 observation per mouse 3 and 24 hours after ROSC. Therefore, group differences were assessed using a 2‐sided permutation t test with 10 000 random permutations. For statistical comparison of the indicator‐dilution results, an average of the results from repeated bolus tracking in each experiment was calculated, and groups were compared using permutation t test. Arterial gas values and vital parameters were compared between groups using mixed‐effects analysis with Bonferroni's correction for multiple comparisons. Results are presented as means or medians, with error bars indicating the SEM. A P value <0.05 was considered statistically significant. Results Outcomes and Physiological Parameters Following Asphyxial CA Of the 13 mice subjected to CA, 9 achieved ROSC (Figure S1). Among those that survived to the 3‐hour follow up scan, the mean time from asphyxia to CA was 1 minute 39 seconds (95% CI, 1:13–2:04 min:sec). Pulseless electrical activity was consistently observed throughout the 4‐minute CA period (Table S1 and Figure S2). The mean duration from initiation of cardiopulmonary resuscitation to ROSC was 1 minute 10 seconds (95% CI, 0:22–2:17 min:s). Following ROSC, mice in the CA group that survived to the 3‐hour follow‐up required a cumulative noradrenaline dose of 106 μg/kg (95% CI, 47–165 μg/kg) to maintain a mean arterial pressure >60 mm Hg during the first 2 hours after ROSC. None of the mice required vasopressor support at the 3‐hour follow‐up, that is, during imaging of the cerebral circulation. The mean time from cessation of noradrenaline infusion to imaging was 114 (95% CI, 58–169) minutes. No correlation was found between arteriole or first‐order capillary diameters and either the cumulative noradrenaline dose or the time since cessation of noradrenaline infusion (Figure S3A,B). Arterial blood gas samples 2 minutes after ROSC revealed that the CA group exhibited lower pH, elevated Pco 2 and lactate, reduced base excess, and higher glucose levels compared with sham (Figure S4). These parameters were similar to those in the sham group 45 and 90 minutes after ROSC. Six of the 13 mice in the CA group survived to the 3‐hour follow‐up scan, and 5 mice survived to the 24‐hour follow‐up scan (Figure S1). In the sham group, 1 mouse died from major bleeding caused by the arterial catheter following the 3‐hour scan. Hence, only data from 5 mice in each group were included for the 24‐hour follow‐up scan. At 24 hours after ROSC, 3 mice exhibited paraplegia, a phenotype not observed in any of the sham‐operated mice. Vasospasm of First‐ to Third‐Order Capillaries and Reduced Capillary Blood Flow Velocity Were Observed 3 Hours After ROSC Capillary diameters were assessed in first‐ to third‐order capillaries with ensheathing pericytes, fourth‐ to 10th‐order capillaries, and capillary branches located first to third order before the venule, associated with mesh and thin‐strand pericytes (Figure 1A through 1F). At 3 hours after ROSC, the diameters of first‐order and second‐ to third‐order capillaries were reduced compared with that observed in the sham group (Figure 1G and 1H). There was no significant difference in diameters of fourth‐ to fifth‐ or sixth‐ to 10th‐order capillaries (Figure 1I and 1J). Prevenule capillaries located second‐ to third‐ and first‐order branch from the venule also showed similar diameters 3 hours after ROSC compared with sham (Figure 1K and 1L). Reduced diameter of first‐ to third‐order capillaries 3 hours after ROSC. Two‐photon microscopy representative images with platelet‐derived growth factor β in cyan color and intravascular Alexa‐conjugated albumin dye in red color. Images show a penetrating arteriole with the first‐ and second‐order capillaries with ensheathing pericytes (A), downstream capillaries with mesh pericytes (B), and prevenular capillaries with thin‐strand pericytes (C) 3 hours after ROSC and in sham (D–F; scale bars 25 μm). Platelet‐derived growth factor β in cyan color and intravascular Alexa‐conjugated albumin dye in red color. The diameter of first‐order capillaries (G) and second‐ to third‐order capillaries (H) was significantly reduced 3 hours after ROSC (n=6; 3 women and 3 men) compared with sham group (n=6; 4 women and 2 men). There were no differences between groups in diameters of downstream capillary orders (I–L). Diameters of capillaries were assessed using line scanning (see also Figure 2A–2D). The diameter of 157 vessels from mice undergoing sham surgery and 137 vessels from the CA group were analyzed (n=6 mice). For information about number of included capillaries in each branching order, please refer to Data S1. Groups were compared using linear mixed model. CA indicates cardiac arrest; Cap, capillary; and ROSC, return of spontaneous circulation. Capillary segments 3 hours after ROSC (Figure 2A–2D) showed reduced RBC velocity 35 of first‐order capillaries (Figure 2E), a group difference that did not obtain statistical significance in second‐ to third‐order capillaries (Figure 2F). No difference in RBC velocity was observed in fourth‐ to fifth‐order capillaries (Figure 2G), and the reduced velocity of sixth‐ to 10th‐order capillaries did not reach statistical significance (Figure 2H). Further downstream in prevenule first‐ to third‐order capillaries, RBC was reduced 3 hours after ROSC in the CA group compared with sham (Figure 2H through 2J). Reduced RBC velocity 3 hours after ROSC. The velocity of RBC passage throughout the capillary bed was assessed using 2‐photon microscopy line scanning. Representative images and line‐scans of capillaries located 3 branching orders upstream from the penetrating venule 3 hours after ROSC (A, B) and in sham (C, D; scale bars 15 μm). These capillaries were associated with thin‐strand pericytes (cyan; A, C). Representative line‐scans showing slower RBC velocity 3 hours after ROSC compared with sham with the steepness of lines (B, D) representing the distance–time correlation of the capillary blood flow. The RBC velocity was calculated using Radon transformation based on streak angles. 35 The RBC velocity was reduced in first‐order capillaries (E). No significant difference was detected in second‐ to 10th‐order capillaries (F–H). Downstream first‐ to third‐order branch capillaries upstream from the penetrating venule showed slow RBC velocity 3 hours after ROSC compared with sham (I, J). Measurements from 169 capillaries in the sham group (n=6; 4 woman and 2 men) and 146 capillaries in the CA group (n=6; 3 women and 2 men). Groups were compared using linear mixed model. CA indicates cardiac arrest; Cap, capillary; RBC, red blood cell; and ROSC, return of spontaneous circulation. At 24 hours after ROSC, first‐order capillaries remained constricted in the CA group; however, flow velocity had recovered to levels comparable with those observed in the sham group (Figure 3A–3C). Downstream high‐order capillary segments showed similar diameters and velocities in both groups at the 24‐hour follow‐up (Figure 3D–3H). Similar capillary blood flow velocity was observed between groups 24 hours after ROSC. Representative Z‐stack angiogram 3 hours (A) and 24 hours after ROSC (B; scale bars 50 μm). Red color, intravascular Alexa‐conjugated albumin; cyan color, platelet‐derived growth factor β. The diameter of first‐order capillaries remained reduced in the CA group (n=5; 3 women and 2 men) compared with sham (n=5; 3 women and 2 men) 24 hours after ROSC (C). There was no difference in capillary diameter of downstream capillaries and RBC velocity was also similar between groups throughout the capillary bed (C–H). Diameter measurements from 153 capillaries in the CA group and 153 capillaries from the sham group. Velocity measurements from 138 capillaries in the cardiac arrest group and 125 capillaries from the sham group. For information about number of included capillaries in each branching order, please refer to Data S1. Groups were compared using a linear mixed model. CA indicates cardiac arrest; RBC, red blood cell; and ROSC, return of spontaneous circulation. Increased Cerebral Capillary Stalling 3 Hours After ROSC Associated With Increased Leukocyte Adhesion Capillary flow stalling was assessed from 2‐photon microscopy time‐series scans with fluorescent labeling of leukocytes (Figure 4A through 4F). At 3 hours after ROSC, the stalling incidence (Figure 4G) and point prevalence were increased compared with sham (Figure 4H, Table S2). The difference between groups in mean stalling duration at 3 hours after ROSC was not statistically significant (Table S2). The incidence and point prevalence of capillary stalling caused by leukocytes was larger at 3 hours after ROSC compared with sham (Figure 4I, Table S2). The percentage of stalls that were associated with leukocyte plugging was increased 3 hours after ROSC compared with sham (Figure 4J, Table S2). No significant difference between groups in the mean duration of leukocyte stalls at 3 hours after ROSC was observed (Table S2). By the 24‐hour follow‐up scan, there was no difference between the groups in any of the stalling parameters (Figure 4K–4N, Table S2). Increased capillary flow stalling 3 hours after ROSC. Representative image of the chronic cranial window 24 hours after ROSC (A; scale bar 500 μm) and the region of interest for the 2‐photon stalling assessment (B; scale bar 100 μm). Example of a capillary segment with normal flow (C) and during flow stalling caused by plugging of rhodamine 6G‐labeled leukocyte (D; scale bars 25 μm). Representative stall‐o‐grams of the same mice as representative images 3 hours after ROSC (E), and a sham (F) showing the timeline of stalling capillaries through the 300 consecutive angiograms recorded over 600 seconds (capillary segments on y axis and time on x axis). Stalling incidence (G) and stalling point prevalence (H) were increased in the cardiac arrest group (n=6; 3 women and 3 men) compared with sham (n=6; 4 women and 2 men) at 3 hours after ROSC (G). The incidence of flow stalling caused by leukocytes were (I) and the percentage of stalls associated with leukocyte plugging were higher 3 hours after ROSC than in the sham group (J). No difference in stalling prevalence or incidence was observed between groups 24 hours after ROSC (K, L) and a similar incidence and fraction of leukocyte‐associated stalls was seen (M, N; n=5; 3 women and 2 men in each group). Stalling was assessed in 360 capillaries in the CA group and 333 capillaries in the sham group 3 hours after ROSC; and in 227 capillaries in CA and 305 capillaries in sham 24 hours after ROSC. One mouse from the CA group showed unsuccessful labeling with rhodamine and was therefore excluded for analysis (I, J, M, N). CA and sham were compared with the t test. For statistics including stalling durations, see also Table S2. CA indicates cardiac arrest; and ROSC, return of spontaneous circulation. Constriction of Cerebral Pial Arteries and Penetrating Arterioles 3 Hours but Not 24 Hours After ROSC Diameters of pial and penetrating vessels were assessed from 2‐photon microscopy angiograms and estimated using VasoMetrics plug‐in for ImageJ (Figure 5A through 5F). 34 By 3 hours after ROSC, the diameter of pial arteries downstream of the middle cerebral artery was reduced in the CA group compared with sham (Figure 5G). Downstream penetrating arterioles also showed a reduced diameter in the CA group 3 hours after ROSC (Figure 5H). No differences were observed in the diameters of penetrating venules and pial veins at the 3‐hour follow‐up (Figure 5I and 5J). By 24 hours after ROSC, the diameters of pial arteries and penetrating arterioles were no longer different between groups (Figure 5K and 5L), and the diameters of penetrating venules and pial veins remained similar in both groups (Figure 5M and 5N). Increased constriction of pial arteries and penetrating arterioles 3 hours after ROSC. Representative images of the chronic cranial window at the 3‐hour follow‐up in a sham mouse (A) and in a CA mouse 3 hours (B) and 24 hours after ROSC (C). Image showing the cranial window preparation of a mechanically ventilated head‐fixed sham mouse at 3‐hour follow‐up (D). Representative 2‐photon microscopy of pial vessels (E) and a penetrating arteriole (F) imaged 50 to 70 μm and 80 to 100 μm from the brain surface, respectively. Scale bars 500 μm (A–C), 5 mm (D), 75 μm (E), and 10 μm (F). Straight white lines show the semiautomatic diameter assessment in the Z‐stacks of the pial artery (E) and a penetrating arteriole (F) using VasoMetrics plug‐in for ImageJ software. 34 Pial arteries (G) and penetrating arterioles (H) were more constricted in the CA group (n=6; 3 women and 3 men) 3 hours after ROSC compared with sham (n=6; 4 women and 2 men). No difference in penetrating venule (I) and pial vein diameters (J) were observed between groups at the 3‐hour follow‐up. At the 24‐hour follow‐up, no significant differences were observed between the groups in pial artery (K), penetrating arteriole (L), penetrating venule (M), or pial vein diameters (N); n=5; 3 women and 2 men in both groups. Other vessel diameters were compared using a linear mixed model. For pial arteries, where only 1 vessel diameter was included for each mouse, groups were compared using the permutation t test without random effects. CA indicates cardiac arrest; and ROSC, return of spontaneous circulation. Prolonged Capillary Transit Time and Capillary Transit Time Heterogeneity After Cardiac Arrest The indicator‐dilution method using 2‐photon microscopy was applied to estimate MTT and CTH across the capillary network, from the pial artery to the pial vein (Figure 6A through 6C). The MTT from artery to vein was increased at 3 hours after ROSC in the CA group compared with sham (Figure 6D–6F). The difference in CTH did not reach statistical difference between groups (Figure 6G). No difference in MTT or CTH were observed 24 hours after ROSC (Figure 6F and 6G). The relative transit‐time heterogeneity (CTH/MTT– relative transit‐time heterogeneity) was reduced in the CA group at 3 hours after ROSC, a difference that was no longer observed at 24 hours after ROSC (Figure 6H). Prolonged MTT and reduced RTH 3 hours after ROSC. Capillary MTT) and CTH were assessed by tracking 10‐μL intravenous boluses of Alexa‐conjugated albumin dye (A–C; scale bars 85 μm). Representative traces show the intensity increase in pial artery and vein after bolus injection 3 hours after ROSC (D, E). The MTT (F) from pial artery to pial vein were increased 3 hours after ROSC (n=6; 3 women and 3 men) compared with sham (n=6; 4 women and 2 men). The difference in CTH did not reach statistical significance (G). There was no difference in MTT and CTH between groups 24 hours after ROSC compared with sham (n=5; 3 women and 2 men in both groups). The pial artery to vein RTH (RTH=CTH/MTT) was reduced 3 hours after ROSC compared with sham, while no difference between groups in RTH was seen at the 24‐hour follow‐up (H). CA and sham groups were compared at each follow‐up using the permutation t test. CA indicates cardiac arrest; CTH, capillary transit time heterogeneity; MTT, mean transit time; ROSC, return of spontaneous circulation; and RTH, relative transit time heterogeneity. Discussion In this study, the acute and subacute effects of asphyxial CA on the cerebral microcirculation were examined in mice. Reduced RBC velocity, vasoconstriction of first‐ to third‐order capillaries, and increased vascular tone in upstream pial arteries and parenchymal arterioles were observed 3 hours after ROSC. These changes were accompanied by increased capillary stalling, associated with leukocytes becoming lodged. Importantly, post‐ROSC cerebral vasospasm occurred despite normal arterial blood pressure and in the absence of vasopressor treatment. By 24 hours after ROSC, the cerebrovascular disturbances were no longer evident. Overall, these results demonstrate that cerebral microcirculatory blood flow is impaired 3 hours after ROSC and offer novel insight into cerebrovascular hypoperfusion after ROSC, identifying vasospasm as a potential therapeutic target to improve cerebral perfusion following CA. Increased Cerebrovascular Resistance in the Early Hours After ROSC Cerebral hypoperfusion has been described in the early post‐ROSC phase and associated with progressive neurological deterioration, driven by secondary brain injury. 1 A recent study using computed tomography in resuscitated patients demonstrated that region‐specific cerebral hypoperfusion within the first hours after ROSC was predictive of poor neurological outcomes. 36 Transcranial ultrasound studies suggest that impaired cerebral perfusion 4 hours after ROSC is associated with elevated middle cerebral artery pulsatility index, indicating increased cerebrovascular resistance. 8 , 15 In the current study, the observed constriction of pial arteries, penetrating arterioles, and first‐ to third‐order capillaries provides mechanistic insight into the increased cerebrovascular resistance and resulting cerebral hypoperfusion observed in the early hours after ROSC in clinical settings. As some mice required noradrenaline support in the early post‐ROSC period, similar to clinical post‐CA care, we carefully evaluated its potential contribution to the observed cerebrovascular constriction. No correlation was observed between cumulative noradrenaline dose and the degree of arteriolar or capillary constriction. Moreover, the mean time from cessation of noradrenaline infusion to imaging was 114 minutes, and given the short plasma half‐life of noradrenaline of 1 to 2 minutes, 37 it is unlikely that residual drug effects contributed meaningfully to the observed post‐ROSC cerebral vasospasm. The pial artery vasoconstriction observed in the current study during the early post‐ROSC period is consistent with findings in a preclinical neonatal rat model using multiphoton imaging, which demonstrated similar pial artery constriction as early as 30 minutes after ROSC. 11 , 12 However, findings related to the diameter of penetrating arterioles after ROSC varied. Li et al reported no change in diameter of penetrating arterioles, 12 whereas Oghifobibi et al reported arteriole constriction after ROSC in neonatal animals resuscitated with epinephrine. 11 Notably, the latter study found that arteriole vasoconstriction was absent after ROSC if cardiopulmonary resuscitation was performed without epinephrine, whereas pial artery vasoconstriction occurred irrespective of epinephrine administration. 11 The current study differs by using a chronic cranial window in adult mice and assessing longer‐term effects (3 and 24 hours after ROSC), extending the analysis estimating capillary transit‐time distribution, quantification of capillary stalls, and characterization of pericyte subtypes. We found widespread reduced RBC velocity throughout the capillary network at 3 hours after ROSC, in contrast with Li et al, who reported no change in capillary RBC velocity in the first hour after ROSC. This discrepancy likely reflects differences in experimental models and, importantly, post‐ROSC experimental timing. Thus, it is possible that the slowing of capillary blood flow occurs 3 hours after ROSC, but not within the first hour after ROSC. Another study in adult rats with CA induced by electrical fibrillation demonstrated cerebral hypoperfusion 45 minutes after ROSC. 9 Although vascular‐specific changes were not assessed, the study showed improved cerebral perfusion after ROSC following administration of an endothelin type‐A receptor antagonist, suggesting that elevated cerebrovascular tone contributes to post‐ROSC cerebral hypoperfusion. Notably, the same group demonstrated that treatment with the endothelin type‐A receptor antagonist administered 15 minutes after ROSC improves long‐term neurological recovery in rats. 38 Together with the results of the current study, these findings suggest that targeting post‐ROSC vasospasm pharmacologically may represent a promising strategy to enhance neurological recovery after CA. The cellular and mechanistic basis of the cerebral vasospasm observed 3 hours after ROSC remains to be elucidated. One possibility is that enhanced vasoconstriction arises from Ca2+ overload in cerebrovascular mural cells, a well‐described consequence of ischemia–reperfusion injury. 39 , 40 Other mechanisms that modulate mural‐cell contractility by altering the Ca2+ sensitivity of the contractile apparatus, without necessarily changing intracellular Ca2+ levels, may also contribute. 41 Indeed, exaggerated vasoconstriction of pial arteries in mice and penetrating arterioles in rats has been reported in the early hours after ischemic stroke, despite no detectable increase in smooth muscle Ca2+. 42 , 43 The poststroke vasoconstriction was attributed to increased Ca2+ sensitivity in cerebrovascular smooth muscle cells, potentially mediated by Src kinase–dependent signaling. 41 , 42 , 44 , 45 Other work suggests that hypoxia‐induced pericyte contraction may be driven by the release of 20‐HETE from astrocytes, a potent vasoconstrictor capable of initiating constriction both by elevating intracellular Ca2+ and by enhancing Ca2+ sensitivity. 24 , 46 Increased Capillary Stalling and Prolonged Capillary Transit Time 3 Hours After ROSC At 3 hours after ROSC, we observed a significant increase in capillary stalling, which was in part attributed to leukocyte adhesion. The pathophysiology of CA, including early endothelial injury and inflammation, supports our findings of leukocyte adhesion in the cerebral microvasculature. 47 , 48 A previous neonatal rat CA study 11 reported a stalling incidence of 20–30% within the first hour after ROSC, closely matching the 30% stalling incidence observed in the current study. However, the previous study did not evaluate stall dynamics or the underlying causes of these stalls. Comparisons with ischemia–reperfusion stroke models show similar stalling incidence values of ≈25% to 40% within 2 to 3 hours after reperfusion. 25 , 26 , 27 , 28 These ischemic stroke studies generally report higher stalling point prevalence of ≈5% to 15%, compared with the 2% point prevalence observed 3 hours after ROSC in the current study. Leukocyte‐mediated stalls accounted for ≈10% of all stalls in the current study, which is lower than the ≈15% and ≈20% leukocyte‐mediated stalling reported in the ischemic core and penumbra, respectively, 2 hours after stroke. 27 Differences in no‐flow duration and ischemic severity between experimental models of CA and focal stroke may partly explain variability in reported capillary stalling metrics. Nevertheless, experimental stroke studies targeting the neutrophil adhesion protein, Ly6G, with monoclonal antibodies effectively reduces capillary stalling after stroke and improves neurological outcomes. 27 , 28 Whether similar immunomodulatory approaches might improve outcomes following CA remains to be examined. Consistent with reduced capillary RBC velocity, the indicator‐dilution method used in the current study showed increased MTT and also a tendency toward increased CTH, with relative transit‐time heterogeneity being reduced at 3 hours after ROSC. By 24 hours after ROSC, all 3 parameters returned to sham levels. The observed increase in MTT in the early post‐ROSC phase is consistent with computed tomographic imaging data indicating evidence of delayed perfusion following ROSC. 36 The reduced relative transit‐time heterogeneity at 3 hours after ROSC indicates a relatively uniform flow despite slower velocities. This phenomenon may be a compensatory mechanism to homogenize capillary flows under low‐perfusion conditions to improve oxygen availability as previously estimated. 49 Notably, this profile in CTH has been associated with capillary no‐flow and infarct progression following ischemic stroke. 29 Pericyte Contraction Underlies Capillary Vasospasm 3 Hours After ROSC To our knowledge, this is the first study demonstrating low‐order capillary constriction after resuscitation from CA. However, capillary disturbances have been observed in other disorders also associated with hypoxic brain injury. Thus, hypoxia exposure triggers smooth muscle cell and pericyte contraction, a mechanism suggested to cause microcirculatory failure in the peri‐ischemic cortex following stroke. 25 Similarly, exaggerated vasoconstriction of pial arteries in mice and penetrating arterioles in rats has been observed in the early hours after ischemic stroke. 42 , 43 In acute brain slices, it is also suggested that hypoxia elicits irreversible contraction of pericytes, leading to persistent capillary failure. 24 , 46 The findings of the current study suggest that hypoxic injury during CA similarly triggers pericyte‐mediated sustained constriction of first‐ to third‐order capillary pericytes 3 hours after ROSC, whereas downstream capillaries did not constrict. The constriction of low‐order capillaries and arterioles was associated with reduced capillary RBC velocity. However, unchanged RBC velocities were seen in fourth‐ to fifth‐order capillaries. These capillary branches exhibited higher‐velocity variability, which may limit sensitivity to detect group differences. In addition, the distal capillary bed forms a densely interconnected network with multiple parallel flow paths, allowing local redistribution of flow within the plexus. Such network topology has been shown to redirect flow around perturbed capillary segments, 50 potentially preserving blood flow velocities in some capillary branches despite upstream microvascular impairment. Pial Artery and Arteriole Diameter Normalized at 24 Hours Post‐ROSC At 24 hours post‐ROSC, pial artery and penetrating arteriole diameters had normalized, suggesting a reversal of the hypercontractile state observed 3 hours after ROSC. This arterial relaxation at the 24‐hour follow‐up was associated with perfusion of the downstream capillary bed comparable with that of the sham group. Similarly, resuscitated patients with CA exhibit higher cerebrovascular perfusion associated with lower cerebrovascular resistance at 24 hours after ROSC compared with the early hours after ROSC. 8 , 15 This recovery might reflect compensatory vasodilatory mechanisms activated in the delayed postischemic phase, similar to pial artery dilation in peri‐ischemic tissue observed 24 hours after stroke. 25 Supporting these observations, increased vasodilation of isolated rat middle cerebral arteries has been reported following asphyxial CA. 51 Enhanced vasodilation was observed at 4 hours but not at 2 hours after ROSC, suggesting a delayed activation of compensatory mechanisms that may be driven by endothelial‐derived hyperpolarization. One potential adaptative mechanism could be the increased expression of endothelial large conductance calcium‐activated channels following hypoxia. 52 This upregulation have been shown to modulate cerebrovascular tone, suggesting that hyperpolarization of cerebral endothelial and smooth muscle cells may contribute to compensatory vasodilation after ROSC, although this mechanism has yet to be demonstrated. A limitation of this study is the lack of continuous cerebral perfusion monitoring during CA and immediately after ROSC, precluding assessment of early hemodynamic shifts, including possible transient hyperperfusion. 4 , 12 Using animal models to study CA has inherent limitations. While results from the current study appear to align with several clinical studies, 8 , 10 , 15 , 36 there may be species‐dependent differences in cerebrovascular changes after ROSC. Furthermore, CA model variations (eg, asphyxia versus fibrillatory) may limit direct translation. 53 Nevertheless, the cerebrovascular hypoperfusion phenotype seen in this study is similar to other rodent CA models, including electrical fibrillation–induced CA. 9 It cannot be excluded that cerebrovascular changes following shorter or longer CAs may differ from those observed after the 4‐minute arrest used in the current study. Finally, another limitation of this study is the lack of post‐ROSC behavioral assessment of neurological injury. In conclusion, this study demonstrates that at 3 hours post‐ROSC, cerebral arteries and low‐order capillaries were constricted, leading to impaired capillary blood flow and increased capillary stalling. At 24 hours after ROSC, cerebral hemodynamics had normalized, suggesting a reversible cerebrovascular dysfunction. The observed post‐ROSC cerebral vasospasm may contribute to secondary brain injury after CA, suggesting a novel treatment target to improve capillary perfusion and post‐ROSC outcomes. Sources of Funding This work was funded by Novo Nordisk Foundation, grant number 0086088. The study was also supported by the Danish Cardiovascular Academy, which is funded by the Novo Nordisk Foundation (Grant No. NNF20SA0067242) and the Danish Heart Foundation. Aarhus Universitets Forsknings fond covered part of the publication costs. Disclosures The authors declare no conflicts of interest. Supporting information Tables S1–S2 Figures S1–S4 Data S1–S2 Acknowledgments The authors gratefully acknowledge Dr Vitalii Dashkovskyi and Dr Dmitry Postnov for their expert assistance with the optical imaging experiments. The authors also thank Dr Frederik Boe Hansen for his valuable guidance and expertise on the experimental asphyxial CA model. We are grateful that Dr John T. Giblin, Dr Sefik Evren Erdener, and Prof David Boas have generously made their capillary stalling analysis code available. The authors thank Oliver Rasmus B. Christensen and Dr Irene Klærke Mikkelsen for their valuable assistance with programming. Finally, the authors are grateful to Susanne Smith Christensen and Anna Bay Nielsen for their excellent technical support in the laboratory and to Henriette Blæsild Vuust for her assistance in creating the graphical abstract. This manuscript was sent to Neel Singhal, MD, PhD, Associate Editor, for review by expert referees, editorial decision, and final disposition. Supplemental Material is available at https://www.ahajournals.org/doi/suppl/10.1161/JAHA.125.047582 For Sources of Funding and Disclosures, see page 15. References Associated Data Supplementary Materials Tables S1–S2 Figures S1–S4 Data S1–S2