Quantification of Capillary Perfusion in an Animal Model of Acute Intracranial Hypertension

Abstract

Intracranial hypertension (IH) is a common feature of many pathologies, including brain edema. In the brain, the extended network of capillaries ensures blood flow to meet local metabolic demands. Capillary circulation may be severely affected by IH, but no studies have quantified the effect of intracranial pressure (ICP) and cerebral perfusion pressure (CPP) on capillary perfusion during the development of brain edema. We used optical coherence tomography angiography to quantify relative changes of fractional perfused volume (FPV) in cortical capillaries and simultaneously monitored ICP and blood pressure (BP) in anesthetized male C57Bl/6NTac mice during development of brain edema induced by water intoxication (WI) within 30 min. WI induced severe IH and brain herniation. ICP and CPP reached 90.2 mm Hg and 38.4 mm Hg, respectively. FPV was significantly affected already at normal ICP (ICP <15 mm Hg, slope ≈ −1.46, p < 0.001) and, at the onset of IH (ICP = 20-22 mm Hg), FPV was 17.9 ± 13.3% lower than baseline. A decreasing trend was observed until the ICP peak (Δ%FPV = -43.6 ± 19.2%). In the ICP range of 7-42 mm Hg, relative changes in FPV were significantly correlated with ICP, BP, and CPP (p < 0.001), with ICP and CPP being the best predictors. In conclusion, elevated ICP induces a gradual collapse of the cerebral microvasculature, which is initiated before the clinical threshold of IH. In summary, the estimate of capillary perfusion might be essential in patients with IH to assess the state of the brain microcirculation and to improve the availability of oxygen and nutrients to the brain.

Introduction

Acute elevation of intracranial pressure (ICP) and brain edema (BE) are life-threatening conditions that are commonly present in patients with traumatic brain injury (TBI), stroke, or infectious diseases.¹ An increase in ICP above 20–22 mm Hg is considered intracranial hypertension (IH)^2,3 and is managed as a clinical emergency since it restricts brain blood perfusion and indicates a risk for the further elevation of ICP which could result in the development of severe cerebral ischemia.^4,5 IH can produce uncal herniation and mechanical damages to the medulla oblongata, increasing the risk for respiratory arrest, coma, and death.^1,6

Cerebral perfusion pressure (CPP = mean arterial pressure [MAP] – intracranial pressure [ICP]) and cerebral blood flow (CBF) are the primary determinants of the cerebral perfusion,⁷ whose regulation is crucial to maintain proper delivery of oxygen and nutrients to the brain. Besides active regulation of CBF during brain activity, an efficient blood flow distribution and perfusion within brain capillaries is crucial for adequate oxygen diffusion to the brain tissue during increased metabolic demands.^8–10

During BE, increasing ICP restricts brain perfusion due to a reduction of CBF and CPP. This may occur through mechanisms at several levels of the brain vasculature. Capillary perfusion, which is also regulated,¹¹ is of particular importance since it is an essential part of the diffusion-based supply of oxygen and nutrients.^12,13 The evaluation of capillary perfusion as a function of ICP is therefore fundamental for a complete understanding of the pathological mechanisms of ICP-induced hypoperfusion and ischemia in cases of BE.

Even before the occurrence of life-threatening complications, increased ICP strongly affects CPP, which oscillates between 70–85 mm Hg in healthy adults.⁷ The current management of patients with BE and IH after a TBI is focused on lowering ICP and maintaining normal CPP. However, previous studies have shown important shortcomings of clinical interventions based on ICP monitoring alone, as maintaining ICP below 20 mm Hg did not show better outcome in patients compared with standard imaging assessment of BE.¹⁴ Since capillaries possess limited morphological resistance to elevation of external pressure,¹⁵ elevated ICP will be expected to decrease capillary perfusion. A better understanding of the relation between IH and capillary flow dynamics may benefit patients experiencing elevated ICP, since maintenance of proper capillary flow is extremely important to avoid brain hypoperfusion and ischemia. It is of particular interest to elucidate at which level of ICP first capillary perfusion is affected and if this happens before ICP reaches the threshold for treatment (20–22 mm Hg in TBI).^2,3

The main aim of this study is to quantify the effect of ICP and CPP on brain capillary perfusion during IH induced by water intoxication (WI) in mice. In addition, we also aim to evaluate whether capillary perfusion is affected before the current threshold for treatment. To address these aims, we used optical-coherence tomography (OCT) to quantify the total amount of capillary perfusion in anesthetized mice undergoing severe brain edema by WI. OCT is an imaging method that uses near-infrared and visible wavelengths for imaging of optically scattering tissues, using low-coherence interferometry.¹⁶ In the study of brain hemodynamics, OCT has showed several applications for accurate detection of erythrocytes movement without the need of exogenous contrast, which provides quantitative and qualitative information on brain vessels function and morphology.^17,18

Methods

Animal preparation

In vivo experiments were performed in adult male C57BL/6NTac mice (Taconics Bioscience Inc., Ejby, Denmark), between 9–18 weeks old and between 25–32 g of weight. All animals were housed at the Danish Neuroscience Center (DNC, Aarhus University) in group cages (between 3–5 mice/cage) with ad libitum access to water and standard diet and with a 12 h:12 h light-dark cycle at constant temperature and humidity (21°C ± 2 and 45% ± 5 relative humidity). All experiments were performed between 10:00–15:00. All housing, breeding, and experimental protocols were conducted according to the regulations of the Danish Ministry of Justice and Animal Protection Committees, with the Danish Animal Experiments Inspectorate permit 2015-15-0201-00509.

Two experimental groups were used: WI (n = 5) and Sham-operated mice (n = 5). We performed all surgical procedures and OCT scans under surgical anesthesia with isoflurane (3% in 30% medical air for induction and 1.5–1.75% for surgical procedure and scans) and local anesthesia with lidocaine (10 mg/mL⁻¹; Farmaplus) on all surgical incision areas. We placed a bladder catheter using a polyethylene tube, PE-50 (I.D. 0.58 mm, O.D. 0.96 mm; Intramedic Clay Adams, Becton Dickinson, NJ) and collected urines as previously described.¹⁹ Blood pressure (BP) was monitored using a blood pressure monitor (WPI Inc., Sarasota, FL) through a polyethylene tube, PE-10 (I.D. 0.28 mm, O.D. 0.61 mm) inserted and secured into the femoral artery. This catheter was also used to extract blood samples for arterial blood gas analysis. The mouse was mechanically ventilated through a tracheostomy using a polyethylene tube (PE-90, I.D. 0.86 mm, O.D. 1.52 mm) connected to a mechanical ventilator SAR-830/P (CWE Inc., Ardmore, PA). We also monitored end-tidal CO₂ (EtCO₂) using a micro-capnograph (Microcapstar, CWE Inc., Ardmore, PA). We maintained body temperature at 37 ± 0.5°C with a heating pad connected via feedback to a rectal thermometer (HB 101/2, Harvard Apparatus, Holliston, MA). To avoid body dehydration during the surgical procedure and before WI, we injected intraperitoneal (IP) 50 μL of isotonic saline (0.9% NaCl) every hour.

The ICP was measured using a solid-state transducer (Codman Microsensor, DePuy Synthes, Warsaw, IN) that was inserted through a cranial window of ∼2 mm drilled on the parietal and temporal bone, centered on the squamosal suture. For easy retraction after the experiment, we covered the probe with a layer of petroleum-based jelly and then secured with biocompatible dental cement for non-permanent applications (Carboxylatzement; Speiko, Münster, Germany).

For visualization of the cortical microvasculature by OCT, the head of the mouse was fixed to a custom-made head-holder through a metal bar glued to the frontal bone. We placed an open-sealed cranial window of ∼3 mm on the parietal bone over the somatosensory area S1, contralateral hemisphere to the ICP probe, as previously described.²⁰ Briefly, the bone was drilled and the skull removed. Dura mater was carefully removed using fine forceps to improve visualization of the cortical vasculature. Before closing the window with a 5-mm cover glass, we covered the exposed brain with a layer of a mixture of 1.5% agarose in artificial cerebrospinal fluid (EcoCyte Bioscience, Dortmund, Germany).

We collected all the physiological data using a data acquisition device (PowerLab 8/35, ADInstruments) and acquisition software (LabChart, ADInstruments Ltd, Oxford, UK). We estimated CPP as CPP = BP - ICP. Arterial blood samples (55–70 μL) were examined with ABL90 FLEX blood analyzer (Radiometer Medical ApS, Copenhagen, DK) to measure plasma pH, pCO₂, pO₂, Na+, K+, Cl-, Ca²⁺, HCO₃-, glucose, oxygen saturation (sO₂), hemoglobin concentration and hematocrit. If necessary, we adjusted base-deficit with I.V. administration of a sodium-bicarbonate solution (NaHCO₃; 75mg/mL in physiological solution, 0.9% NaCl). Arterial blood gases and urine production during baseline and after brain herniation. Arterial blood samples of 60 μL were withdrawn 10 min after induction of WI (baseline) and after signs of herniation (∼30 min. after WI). Urine production was measured over a 1-h timespan.

OCT scans

For imaging, we used a commercial spectral-domain OCT system (Telesto-II, Thorlabs, Germany). The light source of the system has a central wavelength of 1310 nm, and a bandwidth of 170 nm. The A-line sampling frequency of the system is 76 KHz. The estimated imaging range and resolution in the axial direction were 1.6 mm and 2.9 μm in biological tissue, respectively. The lateral resolution was estimated as 3.5 μm using a 10 × objective lens (Mitutoyo, Kawasaki, Japan). The cortical microvascular structure was imaged using an OCT angiographic technique.¹⁷

The volumetric angiographic imaging was performed over a 1 × 1 mm² field of view (FOV) in the cranial window, with 512 × 512 scan points. Each OCT-angiogram took ∼13.5 sec, with 5 sec between the end of the previous volume and the beginning of the next one. We consecutively acquired 120 such volumes (Fig. 1A). We initiated the OCT acquisition of the first angiogram volume simultaneously with the 2 min of water infusion.

FIG. 1.

(A) Maximum intensity projection of 100 pixels of the en face image of the brain microvasculature of a single experiment from the optical-coherence tomography (OCT) volumetric scan. A total of 120 volumes were acquired in each experiment (volume scan time ∼13.5 sec). (B) Vascular mask of the data in (A) generated by a three-dimensional enhancement filter and thresholding. The parameter used in this study to quantify capillary perfusion (fractional perfused volume) is the ratio of the number of capillary voxels to the total number of voxels, which is taken as a measure of the density of perfused capillaries. Color image is available online.

OCT post-processing

With OCT-angiography, the contrast originates from scattering of red blood cell (RBC) motion.¹⁷ Therefore, the intensity of OCT angiogram index the RBC flow, such that high (or low) intensity represents high (or low) RBC flow, and under the condition of diminished RBC flow in some vessels, the associated signal intensity within those vessels will drop to the level of static tissue.

We developed a MATLAB GUI to select the three-dimensional (3D) capillary region (100 μm) for which we wanted to perform the analysis. Pial and larger vessels were excluded from the analysis by selecting a region 100 μm in cortical depth. The MATLAB routine segmented the vessels from the background using a 3D enhancement filter²¹ and thresholding (Fig. 1B).

After the identification of perfused microvessels, we defined fractional perfused volume (FPV) as the ratio of perfused capillary voxels to the total number of voxels. We used FPV as an optical index of the degree of microvascular perfusion. In our quantification setup, a decrease in FPV reflects a decrease in perfused capillary density due to an increase in the number of capillaries without active perfusion (stagnant blood or vessel collapse). We calculated FPV for each volumetric angiogram under the same FOV, and changes in the density of perfused capillaries during WI are shown as the relative change (%) in FPV to volume #1.

Water intoxication

We induced WI as previously described.¹⁹ Briefly, a solution 0.4 μg·kg⁻¹ of desmopressin (dDAVP, Sigma-Aldrich) in sterile water was injected through a peritoneal catheter (PE-10) using an infusion pump (GenieTouch Syringe pump, Kent Scientific Corporation, Torrington, CT). The total volume injected per mouse was 20% of the body weight (BW), injected over a 2-min period. Mice were monitored for 40 min after WI initiation. Sham-operated mice underwent identical surgical, monitoring and postmortem analysis protocols without the injection of 20% BW distilled water with 0.4 μg·kg⁻¹ dDAVP.

Postmortem analysis

After the OCT scans, the brain was dissected to measure water-accumulation by the wet–dry method.²² The two hemispheres were dissected and individually weighed and dried at 110°C for 24 h. The water content was estimated using the following equation:

Brain Water content (%) = 100* [(Wet weight (g) – Dry weight (g))] / [(Wet weight (g))].

Statistical analysis

We used R (R Core Team, 2012) for statistical analysis. The variability of physiological parameters and FPV during normal ICP were analyzed with an analysis of variance test. Linear regression analyses were run by linear mixed models constructed with relative change in FPV as the predictor, the physiological measurement (ICP, MAP, CPP) as the fixed effects, and subjects as the random effect. We generated locally weighted scatterplot using running average smoothing (LOWESS smoother – span = 0.07) to describe the relations between the increase in ICP and FPV in each animal. Each value is expressed in the text as mean ± standard error of the mean unless specified elsewhere.

Results

Pathological kinetics of BE induced by WI

WI produced global BE in all experimental mice, resulting in brain water content postmortem of 82.4 ± 0.1% (Table 1). The water content was significantly different to the observed in the control group (78.0 ± 2.4; p = 0.04; CI: 0.3 - 8.4). During development of BE, dramatic changes in FPV, BP, ICP, and CPP were observed in all animals (Table 1 and Fig. 2).

Table 1.

Physiological Changes during WI and Development of Intracranial Hypertension

	Baseline		ICP peak	After herniation*
Parameter	Control	WI	WI	WI
Time from WI (min)	———	———	28.0 ± 2.6	227.0 – 36.6
BP (mm Hg)	74.9 ± 2.10	78.2 ± 1.4	128.5 ± 6.5	67.3 ± 8.5
ICP (mm Hg)	12.0 ± 1.46	11.3 ± 1.5	90.2 ± 7.4	58.9 ± 6.08
CPP (mm Hg)	62.89 ± 1.32	66.9 ± 0.4	38.31 ± 3.07	8.4 ± 2.9
HR (bpm)	387 ± 14	406 ± 19	473 ± 26	391 ± 22
BR (bpm)	106 ± 3	96 ± 2	92 ± 6	97 ± 2
EtCO2 (mm Hg)	30.0 ± 3.1	30.0 ± 3.1	29.8 ± 6.8	32.2 ± 6.4
Brain water (%) post mortem	78.0 ± 2.4	-	-	82.4 ± 0.1

Values expressed as mean ± standard error of the mean of the indicated experimental time period.

^*Calculated as the mean over the time after the sign of brain herniation (n = 5 per group).

WI, water intoxication; ICP, intracranial pressure; BP, blood pressure; CPP, cerebral perfusion pressure; HR, heart rate; bpm, beats per minute; BR, breathing rate; EtCO₂, end-tidal CO₂.

FIG. 2.

Time-courses of capillary perfusion and physiological parameters (n = 5 per group). (A–D) Maximum intensity projection of 100 pixels (axial resolution ∼2.9 μm) of a single experiment during normal intracranial pressure (ICP; 7.62 and 15.96 mm Hg) and during increased ICP (30.29 and 44.75 mm Hg). (E) Time-series plot depicting the fractional perfused volume (FPV) as a function of time from the first optical-coherence tomography (OCT) angiogram. Red: water intoxication (WI), Blue: Control (mean ± standard error of the mean). Individual dots that are the mean of the given parameter during each OCT-acquisition (volume scan time ∼13.5 sec). (F) blood pressure (BP), (G) ICP, and (H) cerebral perfusion pressure (CPP; BP-ICP) for each animal. Note the similarities between animals in the general course of the parameters and order of events, despite some individual variation in the absolute timing of events. Color image is available online.

During baseline, ICP was stable at 11.3 ± 1.1 mm Hg. Upon severe WI, changes in ICP were observed already 5 min after water IP infusion (Fig. 2G). The increase in ICP was steady and linear (increase rate = 2.9 ± 0.5 mm Hg/min) and the criterion of IH (20 mm Hg) was reached 11.0 ± 2.9 min from the induction of WI. The maximum peak in ICP (ICP-peak) was 90.1 ± 7.4 mm Hg, reached 28.0 ± 2.6 min from the beginning of WI and followed by a sudden drop to 54.0 ± 14.4 mm Hg. Severe IH (40–60 mm Hg) was observed during the rest of the scanning protocol.

BP also showed an increase during WI (Baseline BP: 76.8 ± 1.63 mm Hg; Fig. 2F). We observed a maximum hypertensive peak on BP of 128.5 ± 6.5 mm Hg and irregular breathing simultaneously with ICP-peak, and a complete failure of self-respiratory abilities with accumulation of CO₂ and reduction in sO₂ after ICP-peak (Supplementary Table 1). BP showed a similar drop as ICP after the maximum peak, reaching hypotensive levels (BP = 40 – 50 mm Hg) towards the end of the experiment. The rate of increase in BP was slower than observed in ICP. Consequently, CPP gradually decreased, reaching 38.3 ± 3.1 mm Hg at ICP-peak, which corresponds to a 42.61% reduction compared to baseline (Fig. 2H).

Arterial blood gases were also thoroughly affected consequentially to the severe WI (Supplementary Table 1). We observed a metabolic and respiratory acidosis as expected after the Cushing reflex. The Na+ dropped to a severely hyponatremic level (111.8 ± 4.0 mmol/L). WI halted urine production (82% decrease from baseline). Hematocrit was unchanged (baseline: 42.9 ± 2.3%; WI: 43.3 ± 2.4%).

Effect of IH on capillary perfusion

In the normal ICP range (ICP <15 mm Hg), FPV was already significantly affected (slope ≈ −1.46, p < 0.001; C.I. −2.22 – −0.70). This effect was dependent on the data points at the high end of this range (14–15 mm Hg), since if only data below 13.96 mm Hg were analyzed, the correlation was below the significance level (n = 8, p = 0.05; CI: −1.90 – 0.01). FPV started decreasing as ICP increased already 5 min. from WI (Fig. 2E and 2G). At the onset of IH (ICP = 20 mm Hg at 11.0 ± 1.4 min from WI), FPV was already significantly reduced by 17.9 ± 13.3% (mean decrease in FPV in ICP between 15 and 20 mm Hg: slope ≈ −3.65, p < 0.001, CI: −5.4– −1.8; Fig. 3A). In the same time range of ICP = 15–20 mm Hg, we did not observe any significant reduction on FPV as a function of CPP (p = 0 .84; Fig. 3B). A similar reduction in FPV of 30.8 ± 8.73% and 57.5 ± 9.9% was observed when ICP reached 30 mm Hg and 40 mm Hg, respectively (14.8 and 20.2 min from WI initiation). No further significant changes were observed after the ICP peak value.

FIG. 3.

Scatterplots of the observed fractional perfused volume (FPV) as a function of ICP (A), cerebral perfusion pressure (CPP; B) in the intracranial pressure (ICP) range between 15–20 mm Hg (n = 5 per group). A linear mixed-model was performed on the WI group to evaluate the effect of ICP and CPP on FPV. Data between 0-20 mm Hg from the control group was included as a reference. A model was also constructed for this data. Each colored slope represents the best individual fit for each mouse experiment. The best fit line from the linear mixed-models of the water intoxication (WI) group is plotted in black. Color image is available online.

We observed a sudden high variance in FPV measurement (Fig. 2E) in one experimental animal at ∼12 min. from water infusion. Similarly, at 25 min from WI, when ICP reached its maximum point for each animal, high variability in the FPV estimate suddenly increased. We also observed a significant variance in FPV estimates from scan estimates acquired at ICP >40 mm Hg. We attributed the production of this optical noise to brain movement secondary to the irregular breathing and mass effect induced by elevated ICP. Therefore, to investigate the correlations between BP, ICP, and CPP with the relative changes in FPV, we only analyzed data with ICP below 42 mm Hg.

We observed that ICP, BP, and CPP were all significant predictors of relative changes in FPV (Fig. 4A-C). ICP was a negative predictor of changes in FPV (r2: 0.56, p < 0.001; slope ≈ −2.16), whereas CPP showed a positive linear correlation with FPV (r2 = 0.33, p < 0.001; slope ≈1.9). BP was also a negative predictor of FPV, but the weakest compared to ICP and CPP (r2: 0.06, p < 0.001, slope ≈ −2.17).

FIG. 4.

Scatterplots of the observed fractional perfused volume (FPV) as a function of blood pressure (BP; A), intracranial pressure (ICP; B), and cerebral perfusion pressure (CPP; C; n = 5 per group). Colored circles represent FVP estimates after average smoothing during each individual optical-coherence tomography-angiogram, Red: water intoxication (WI), Blue: Controls. Each colored slope represents the best individual fit for each mouse experiment. The best fit line from the linear mixed-models of the WI group is plotted in black. Color image is available online.

Discussion

In this study, we aimed to quantify the effect of increased ICP on cerebral capillary perfusion. We employed an OCT-based scanning platform in combination with a severe model of BE induced by WI in mice. The kinetics of the WI model demonstrates the development of acute and severe cerebral edema as a consequence of plasma hypo-osmolality. BE was associated with sudden and drastic changes in MAP, CPP, and ICP, with ICP spanning from normal physiological values to a maximum of approximately 8-times baseline. The combination of Cushing reflex elements and vegetative coma after ICP-peak is in line with a diagnosis of brain herniation damaging the medulla oblongata of the brainstem, which is a typical characteristic of severe WI animal model.¹⁹

We selected OCT imaging to evaluate the changes in the perfusion of the brain capillary bed due to its advantage of performing high-speed volumetric scans with high spatial resolution without the usage of exogenous fluorescent contrast agents. These characteristics allowed us to identify the changes in single capillaries during the fast development of IH and brain edema during WI. OCT signal is limited when used in opaque tissues, such as retina and brain, and the stability of the underlying tissue's optical properties are an essential component for the optical signal and the interpretation of the results.^17,23 This limitation is a disadvantage for the evaluation of severe brain edema, since the accumulation of water in brain parenchyma may change the underlying optical properties of the brain and limit the penetration depth.²⁴

The density of perfused capillaries is reduced before the current clinical threshold for IH

Our data showed that both ICP and CPP were good predictors for capillary perfusion density during IH. CPP is currently considered the best clinical reference for determining neurovascular perfusion during IH.⁷ Capillary perfusion and flow distribution play an essential role in tissue oxygenation^8,9 and dysregulation of capillary flow distribution is linked to impaired oxygen availability and local ischemia.^13,25,26

The clinical threshold for the treatment initiation of IH after severe TBI has recently been set to 22 mm Hg^3,27 while defining a threshold for IH in previous guidelines has been challenging,²⁸ partly due to the cardiovascular heterogeneity of patients with elevated ICP.^29,30 In our study in mice, we observed that capillary perfusion was significantly affected by ICP levels down to 15 mm Hg, which is well below the current IH treatment threshold used in most clinical settings.³¹ This finding prompts the question, if slight reductions in capillary perfusion at ICP considered mildly elevated (ICP >15 mm Hg) may have unrecognized clinical effects. Such changes at the microvascular level may contribute to explain the difficulties in finding a consensus on a treatment threshold for elevated ICP.^31–34 Since the reference values for normal ICP may be lower than the 7–15 mm Hg range previously considered,³⁵ our study suggests that the effect of mild increases in ICP should be investigated, and that more studies with more controlled settings and higher time resolution in the ICP range 14–22 mm Hg are needed to fully describe changes in brain capillary hemodynamics during normal ICP and IH.

Possible mechanisms for the reduction of capillary perfusion with increasing ICP

During BE, several mechanisms can act together to cause the collapse of the cerebral microvasculature. Primarily, the increase in ICP may exert direct mechanical compression and constriction of the capillaries, which cannot be overcome by the arterial pressure and thus leads to capillary collapse. An in vitro study showed that increasing external pressure by 10-mm Hg is sufficient to induce a state of no-flow in up to 50% of capillaries.¹⁵ In this situation, we assume that CPP determines capillary perfusion and this is supported by our finding of CPP and ICP to be the best predictors of relative changes in FPV.

Second, swelling of astrocyte end-feet, which surround brain capillaries, has been observed both in animal models and in biopsies of TBI patients,^36,37 and could play a role in reducing capillary perfusion. Astrocytic end-feet are part of the brain–blood barrier and contain large amounts of the water channel aquaporin-4 (AQP4).³⁸ Therefore, astrocytic end-feet swelling might exacerbate BE by promoting water movement from plasma into astrocytes and brain parenchyma. As astrocytes end-feet completely wrap brain vasculature, their swelling may produce a robust mechanical strain on brain capillaries with the risk of compression of microvessels and hindered perfusion.³⁷ It is not known if a relationship exists between end-feet swelling and ICP elevations.

Third, failure of cerebral autoregulation could play a role in the reduced capillary perfusion. IH following BE leads to a decrease in CPP and CBF, which is counteracted by increasing systolic pressure as also observed in our study. The phenomenon of cerebral autoregulation implies that brain arterioles dilate and constrict in response to changes in systemic pressure to maintain a stable CBF.⁷ However, a certain degree of perfusion pressure is required for this mechanism to work.³⁹ In this study, even before brain herniation, CPP drops below the critical limit for CBF autoregulation (50 mm Hg when CPP is lowered by IH).⁴⁰ In this situation, cerebral arterioles reach their maximum dilation and therefore become unable to further increase CBF available for the microcirculation.³⁹

At normal ICP, 60–70% of the CBF in the microvasculature is distributed across slow-flow capillaries (< 1 mm/sec).²⁵ As BE develops, the increase in tissue pressure will induce CBF diversion towards vessels with lower resistance. In particular, this effect will favor the distribution of blood flow through high-speed capillaries, also called thoroughfare channels (TFC), which have been found to promote the development of non-nutritive CBF by venous shunting.²⁵ This phenomenon increases the heterogeneity of capillary flow distribution which in turn will reduce the oxygen availability to the brain tissue.^8,13 Therefore, the global reduction in the number of perfused capillaries as a function of ICP, which we observed in our study, provides essential evidence of profound changes in blood flow distribution across brain capillaries during BE, which strongly compromise tissue oxygenation even in situations of normal CBF. Overall, elevated ICP induced by WI can affect capillary flow microcirculation during IH, and the decreased efficiency of capillary flow may anticipate local ischemia in neurons and astrocytes.⁹ Energy depletion results in the disruption of the normal ionic homeostatic regulation in the brain, which is the priming mechanism of cytotoxic brain edema and anticipates other pathways of neuronal cell death.^41–44 More experiments are needed to study the relationship between loss of capillary perfusion and the death of neuronal cells.

Future perspectives and limitations

Hypoxic and ischemic insults lead to the influx of cerebrospinal fluid in the brain parenchyma, generating cytotoxic BE.⁴⁵ The overaccumulation of fluid disrupts the normal ionic gradient required for neuronal activity and survival,⁴¹ decreasing neuronal activity excitability.⁴⁶ Restoring ionic homeostasis in brain parenchyma is a primary function of all cells in the neurovascular unit,44 but it requires active use of energy and oxygen delivery which, at elevated ICP, is restricted by the collapse of microvessels. Since capillary perfusion and flow distribution play an essential role in local tissue oxygenation,^8,9 and dysregulation of capillary flow distribution is linked to impaired oxygen availability and local ischemia,^13,25,47 ensuring effective capillary perfusion at elevated ICP may improve oxygenation and reduce energy depletion of neurons and astrocytes. However, future studies with other imaging techniques⁴⁸ are needed to quantify the decrease in oxygenation following ICP-induced collapse of capillaries and how this reduced oxygenation influences neuronal activity and astrocytes homeostatic functions.

The translational value of our study might be limited by the anatomical differences between human and mice (e.g., the skull thickness and the lissencephalic structure of the rodent brain).⁴⁹ However, both species share similarities in their ICP and microvascular profile such as similar ICP levels at a horizontal position,^35,50–52 equivalent response on the sympathetic activation during mild ICP increase⁵³ and topological equivalence of the cortical microvascular network.⁵⁴

We did not investigate a significant correlation between ICP, CPP, and MAP with FPV for ICP above 42 mm Hg. Although values of FPV were still significantly lower than baseline after ICP of 42 mm Hg, we observed a very high variance in the FPV estimate in combination with very elevated ICP (Fig. 2E, 2G) and noticed an almost complete disappearance of perfused microvessels when ICP exceeded 40–45 mm Hg (Supplementary Video 1). Therefore, OCT-based estimation of FPV was not accurate during severely increased ICP (ICP >42 mm Hg) and we cannot provide quantitative measurements in this range.

The experimental noise observed during severely high ICP may be a consequence of the ICP mass effect and consequent misplacement of cerebral regions. This is a known component of BE pathophysiology and finds its most extreme manifestation in the incarceration of the brain stem, occurring in this animal model at ∼25 min. from WI. Such displacement phenomenon may modify the FOV under which the OCT scans were acquired. This consideration is in agreement with previous experiments in which, similarly to the setup used in this study, an OCT scanner was used in combination with a more severe WI.24 In that study, a significant 6–8% increase in average attenuation coefficient, which reflects changes in optical conformation of the brain, was observed between 20 and 25 min after WI. In our model, this range corresponds to the time of ICP-peak before fatal displacement of the brain stem due to herniation. Thus, in accordance with our suggestion, brain movement, induced by the ICP mass effect, can explain the increasing experimental noise we observed in the phase of severely high ICP.

Conclusion

Based on OCT-angiograms and a WI mouse-model of BE, we observed a significant reduction in capillary perfusion as a function of ICP in clinically relevant range of elevated ICP (20–42 mm Hg). This observation supports the hypothesis that dysregulation of perfusion in microcirculation, which in normal physiological conditions has a fundamental role in maintaining brain oxygenation, already emerges at levels of ICP that would not be clinically diagnosed as IH. Given the challenges in finding consensus on the threshold treatment for IH,³¹ these results might be relevant for future discussions and policy making. Further studies will quantitatively determine if capillary flow dynamics are affected by high ICP to provide a more comprehensive description of the effects of IH and brain edema on microvascular dynamics.

Funding Information

This study was supported by the following grant from the Danish Medical Research Council (DFF grant id: 4004-00504) and VELUX foundation (ARCADIA). Further financial support was obtained from Korningfonden and the Højmosegårdgrant (Lægeforeningen, 2017-1064/102 KBN).

Author Disclosure Statement

The authors declare that no competing financial interests exist.

Supplementary Material

Supplementary Table S1

Supplementary Video S1