EFFECT OF CEREBRAL PERFUSION PRESSURE ON CEREBRAL CORTICAL MICROVASCULAR SHUNTING AT HIGH INTRACRANIAL PRESSURE IN RATS

Abstract

Background and Purpose

Recently we showed that decreasing cerebral perfusion pressure (CPP) from 70 to 50 and 30 mmHg by increasing intracranial pressure (ICP) with a fluid reservoir induces a transition from capillary (CAP) to microvascular shunt (MVS) flow in the uninjured rat brain. This transition was associated with tissue hypoxia, increased blood brain barrier (BBB) permeability and brain edema. Our aim here was to determine whether an increase in CPP would attenuate the transition to MVS flow at high ICP.

Methods

Rats were subjected to progressive, step-wise increases in ICP of up to 60 mmHg by an artificial cerebrospinal fluid reservoir connected to the cisterna magna. CPP was maintained at 50, 60, 70 or 80 mmHg by i.v. dopamine infusion. Microvascular red blood cell flow velocity, BBB integrity (fluorescein dye extravasation) and tissue oxygenation (NADH) were measured by in vivo 2-photon laser scanning microscopy. Doppler cortical flux, rectal and cranial temperatures, ICP, arterial blood pressure and gases were monitored.

Results

The CAP/MVS ratio increased (P<0.05) at high ICP as CPP was increased from 50 to 80 mmHg. At an ICP of 30 mmHg and CPP of 50 mmHg, the CAP/MVS ratio was 0.6±0.1. At CPP of 60, 70 and 80 mmHg, the ratio increased to 0.9±0.1, 1.4±0.1 and 1.9±0.1, respectively (mean±SEM, P<0.05). BBB opening and increased NADH occurred at higher ICP as CPP was increased.

Conclusions

Increasing CPP at high ICP attenuates the transition from CAP to MVS flow, development of tissue hypoxia, and increased BBB permeability.

Introduction

When cerebral perfusion pressure (CPP) is reduced by increasing intracranial pressure (ICP) instead of lowering arterial blood pressure, the critical CPP of cerebral blood flow (CBF) autoregulation falls from 60 mmHg to 30 mmHg.^1–4 The reason for this decrease in critical CPP was unknown but might be interpreted as improved CBF autoregulation at high ICP. We hypothesized instead, that the decrease in critical CPP at high ICP was due to a pathologically maintained CBF caused by microvascular shunting (MVS). We tested this hypothesis⁵ by decreasing CPP in rats by either increasing ICP or lowering arterial pressure stepwise from 70 to 50 and 30 mmHg while measuring microvascular flow in capillaries (3–7 µm diam), microvascular shunts or thoroughfare channels (TFC) (8–15 µm diam).^6–11 NADH for tissue hypoxia, fluorescein dye transcapillary extravasation for blood brain barrier (BBB) permeability and brain water content were also measured. Decreasing CPP by increasing ICP caused a transition from capillary to MVS flow that did not occur when CPP was decreased by lowering arterial pressure. This transition was clearly associated with the development of tissue hypoxia, brain edema and increased BBB permeability: hallmarks of non-nutritive shunt flow.

In the clinical management of patients with high ICP (≥20 mmHg) the “optimal” CPP remains a matter of dispute.¹² Nevertheless, recommendations are CPP values between 50 and 60 mmHg and even up to 70 mmHg without excessive use of hypertensive agents or fluids. There are two camps of thought on the management of CPP in patients with high ICP. One is a CPP-directed approach suggesting maintenance of CPP of up to 80 mmHg and even higher.^13–15 The other is an ICP directed therapy using pharmacologic agents^{16, 17} while maintaining CPP between 50 to 60 mmHg. Our aim in this study was to evaluate the transition from capillary to MVS flow at high ICP as a function of CPP.

Methods

Animals and surgical procedures

Protocol #100916 was approved by the IACUC of the University of New Mexico Health Sciences Center. Acclimated male Sprague-Dawley rats were used (300–350 g, n=28, Harlan Laboratories, Indianapolis, IN). Anesthesia was induced with 4% isoflurane/70% nitrous oxide and 30% oxygen. The rats were intubated with a 14GX1.9” catheter and mechanically ventilated (Harvard Apparatus, Holliston, MA) on 2% isoflurane/30% oxygen/70% nitrous oxide; tidal volume, 2.0–2.5mL at a rate of 55–65/min. Rectal temperature was kept at 37±0.5 °C by a heated water blanket. Atropine (0.2 mg, i.p.) was used to reduce mucous secretions. Femoral artery catheters (PE-50) were used to monitor arterial blood pressure and blood sampling (0.3 ml each). Double lumen femoral vein catheters (Braintree Scientific Inc., Braintree, MA) were used for dopamine infusion, fluid replacement (lactated Ringers, 1 ml/hour) and fluorescein dextran injection. The rats were placed in a sterotaxic head frame (Kopf Instruments, Inc.). A catheter (PE-50) was inserted into the cisterna magna and glued in place to monitor and manipulate ICP by a reservoir of artificial cerebrospinal fluid (ACSF). A craniotomy (5 mm diam) over the left parietal cortex was filled with 1.5% agarose in saline and a cover glass slip over the craniotomy was glued to the skull. A cranial temperature probe was used to monitor brain temperature.

Experimental paradigm

The animals were studied in four groups (n=7). CPP was maintained at: I) 50 mmHg; II) 60 mmHg; III) 70 mmHg; and IV) 80 mmHg (Table 1). In each group ICP was sequentially increased from 10 to 30, 40 and 60 mmHg by raising the ACSF reservoir while CPP was manipulated by titrated i.v. dopamine (1 mg/ml) infusion with a Syringe Pump (Harvard Apparatus, Holliston, MA). The dopamine dose infused ranged from 67±16 to 400±70 µg/kg/min (mean±SEM). The average time to achieve a stable elevation in CPP was 7.4±5.2 min. Thirty min at each CPP was sufficient for stabilization of physiological variables and completion of the measurements.

Table 1.

Experimental Paradigm for Four Study Groups

	Group I		Group II		Group III		Group IV
Time from the start, min	CPP=50 mmHg (51.4±3.3)		CPP=60 mmHg (62.2±3.5)		CPP=70 mmHg (70.6±4.4)		CPP=80 mmHg (83.4±5.1 )
	ICP	MAP	ICP	MAP	ICP	MAP	ICP	MAP
0	10.6±2.3	62.5±4.6	10.8±4.1	73.3±6.9	11.2±3.4	92.5±9.1	10.3±5.3	92.7±7.6
30	29.8±4.1	81.4±5.1	30.5±3.6	94.5±9.6	31.6±4.1	104.3±11.2	30.2±4.8	111.4±9.7
60	40.1±5.2	93.2±5.2	42.6±5.3	103.4±8.7	40.4±4.8	112.4±10.8	41.6±6.1	122.6±14.3
90	60.3±4.4	111.2±8.5	61.3±5.6	119.8±7.9	63.1±6.7	129.7±13.4	60.5±7.2	139.8±12.8

n=7 rats/group, all measurements in mmHg. mean±SEM, CPP, cerebral perfusion pressure; MAP, mean arterial pressure; ICP, intracranial pressure.

Dopamine was chosen as the vasopressor because norepinephrine and phenylephrine induced severe arterial acidosis that was difficult to control with i.v. sodium bicarbonate. Arterial acidosis with dopamine was minimal and easily controlled by i.v. sodium bicarbonate infusion (Table S1).

Measured variables were: microvascular red blood cell (RBC) flow velocity; NADH autofluorescence; BBB permeability and cortical Doppler flux. Rectal and temporal muscle temperatures were continuously monitored. Arterial pressure, ICP and CBF were continuously recorded using Biopac preamplifiers and software (Goleta, CA). Arterial blood gases, hemoglobin, hematocrit, pH, glucose and electrolytes were measured at each ICP using a CG8+ cartridge (iSTAT, ABAXIS, Union City, CA).

In Vivo Two-Photon Laser Scanning Microscopy (2PLSM) for RBC Flow Velocity

The procedures were as previously described.⁵ Briefly, fluorescein isothiocyanate-labeled dextran (20 kDa) in physiological saline (5% wt/vol) was injected¹⁸ and visualized using an Olympus BX51WI upright microscope and water-immersion LUMPlan FL/IR 20×/0.50W objective. Images (512×512 pixels, 0.15 µm/pixel) were acquired using Prairie View software. RBC motion was detected from line-scans, i.e., repetitive scans along the central axis of a microvessel at several depths (100–300 µm) from the pia mater.^{5, 18} Capillary selection was based on tortuosity, degree of branching and diameters ranging from 3 to 7 µm.^{7–9, 19, 20} Multiple records were taken from the same vessel and mean microvessel RBC velocity and SEM were calculated. A total of approximately 100 microvessels were scanned at each CPP and the diameter of each vessel measured.

Using NIH ImageJ in offline analysis, three-dimensional anatomy of the vasculature in a region of interest was reconstructed from planar images obtained at successive focal depths in the cortex (XYZ-stack). Volume flow (µl/s) was calculated by multiplying the cross sectional area of the microvessel in mm² by microvessel flow velocity in mm/s.

In Vivo 2PLSM of NADH

NADH autofluorescence was measured using 2PLSM as described above. Fluorescence was emitted by 740 nm center wavelength and filtered at 425–475 nm.^{5, 21} Twenty planar scans of fluorescence intensity were obtained in 10 µm steps starting at 100 µm from the pia matter at each CPP. In offline analyses, average intensity was calculated from the maximal intensity projection for each CPP.

Cortical Doppler Flux

Relative changes in cortical flow were measured continuously by Doppler flux using a single fiber 0.8 mm diameter surface Doppler probe (Moor Instruments, Axminster, UK) on the temporal bone (burr hole) below the optical window.

Statistical analysis

Statistical analyses were done by independent Student’s t-test or Kolgomorov-Smirnov test where appropriate. Differences between groups were determined using two-way analysis of variance (ANOVA) for multiple comparisons and posthoc testing using the Mann–Whitney U-test. Bonferroni’s multiple-comparison test was used for post hoc analysis, where the effects of different CPPs were compared against each other. Significance level was preset to P<0.05. Data are presented as mean±SEM.

Results

Physiological variables were within normal limits throughout the studies and were not significantly different (Table S2). Blood glucose levels were however, elevated in all groups likely due to the stress of surgery and anesthesia. Variations in blood gases were adjusted by manipulation of the rate and volume of the ventilator. Base deficits ≤−5.0 meq/L were corrected by slow i.v. injection of 8.4% sodium bicarbonate.

RBC flow velocities

Microvascular blood flow was measured by fluorescein-dextran labeling of plasma to observe RBCs as negatively stained stripes in a background of labeled plasma (Fig. 1a). Line scans were performed at each ICP on approximately 100 microvessels (3–15 µm diam) from the regions imaged by 2PLSM microscopy at several depths (100–300 µm) from the pia mater. A line-scan through a microvessel leads to a sequence of alternating bright and dark pixels corresponding to labeled plasma and unlabeled RBC. The result is diagonal bands in a space–time image as illustrated in Fig. 1b. The slope of the stripes inversely reflects RBC velocity.

Fig. 1.

Red blood cell flow velocities were affected by increasing intracranial pressure (ICP) as a function of cerebral perfusion pressure (CPP) maintained by dopamine infusion. (a) Micrograph showing a region from which microvascular flow was recorded (b) Line-scan data of red cell velocities in the two microvessels shown in a with lower velocity in the first microvessel than in the second. (c) Changes in capillary/microvascular shunt flow (CAP/MVS) ratio showing that increasing CPP as a function of ICP attenuates the transition to MVS flow (mean±SEM, *P<0.05, n=7).

As we previously reported,⁵ at a normal CPP of 70 mmHg the proportion of low velocity capillary (<1 mm/sec, 3–7 µm diam) was 67.2%, and significantly greater than the 32.8% of high velocity microvessels (>1 mm/sec, 8–15 µm diam). CPP reduction to 30 mmHg decreased the proportion of low flow microvessels to 48.8% and increased the proportion of high velocity microvessels to 51.2%. This redistribution was associated with hypoxia, brain edema and BBB leakage: a hallmark of non-nutritive MVS flow.

The CAP/MVS flow ratio we use here (Fig. S1) reflects the relative proportion of capillary to MVS flow (Fig 1c). In all groups at normal ICP of 10 mmHg, the CAP/MVS ratio was similar to control with an average value of 2.15±0.29 for all 24 animals. A progressive increase in ICP from 10 to 30, 40 and 60 mmHg at constant CPPs resulted in a progressive increase in the number of microvessels with flow velocities > 1.0 mm/s suggesting a shift from low velocity capillary to high velocity MVS flow. At ICP ranging from 30, 40, and 60 mmHg and CPP at 50, 60, 70 and 80 mmHg, the CAP/MVS ratio increased significantly indicating a reduction in microvascular shunting (Table S3). Thus, increasing CPP at a given ICP attenuated the transition from capillary to MVS flow.

The increase in Doppler flux with stepwise increase in ICP at a constant maintained CPP reflects the flow shift from low velocity capillary to high velocity MVS flow (Table S3). The increase in Doppler flux with ICP elevation was steeper at lower CPP than at higher CPP correlating with a higher proportion of MVS flow at any given increased ICP and lower CPP.

Tissue NADH

NADH is an indicator of the status of mitochondrial oxidation. NADH is fluorescent while oxidized NAD⁺ is not. Thus, increased NADH autofluorescence reflects tissue hypoxia.^{21, 22} At an ICP of 10 mmHg and CPP of 70 mmHg, NADH autofluorescence was evenly distributed in a rat parietal cortex (Fig. 2a). NADH fluorescence in a tissue more proximal to microvessels was less bright than distal, reflecting better oxygenation due to oxygen gradients.²¹ In controls, NADH autofluorescence was unchanged over 120 min. Progressively increasing ICP from resulted in a marked increase in NADH fluorescence suggesting tissue hypoxia in all four experimental groups (Fig. 2b, c, Table S4, *P<0.05 and **P<0.01). The rise in NADH was less at higher CPP.

Fig. 2.

Tissue oxygenation by reduced nicotinamide adenine dinucleotide (NADH) autofluorescence with increasing ICP and CPP maintained by i.v. dopamine infusion. (a) Micrograph of NADH fluorescence at normal ICP (10 mmHg) and CPP (70 mmHg). (b) Micrograph of the same area at an ICP of 60 mmHg showing increased NADH fluorescence and tissue hypoxia. (c) Graph showing progression of tissue hypoxia with increasing ICP and constant CPP. Data were normalized as a percent of an averaged initial levels at normal CPP (70 mmHg) and ICP (10 mmHg) (mean±SEM; n=7; *p < 0.05, **p < 0.01 between groups; arrowheads show first significant difference within the group.

BBB Permeability

In intact brain, bright vessels, filled with fluorescein-dextran, were clearly seen over the dark background of unstained tissue (Fig. 3a). Increased BBB permeability leaks fluorescein dextran out of microvessels into tissue (Fig. 3b). The table in Fig. 3c demonstrates the number of rats out of seven for each group showing dye extravasation. In a control group (ICP 10, CPP 70 mmHg, n=7) only one rat showed BBB leakage after 90 minutes of recording (not shown). In Groups I through IV with CPP maintained at 50, 60, 70 and 80 mmHg and ICP 10–60 mmHg, the incidence of BBB leakage decreased. Average dye fluorescence in each group showed a significant, gradual, increase of fluorescence in brain tissue with ICP increasing from 10 to 60 mmHg reflecting opening of the BBB (*P<0.05, **P<0.01). Higher CPPs attenuated BBB degradation compared to lower CPPs (Fig. 3d).

Fig. 3.

Fluorescein extravasation reflecting blood–brain barrier breakdown with increasing ICP and CPP maintained by dopamine infusion. The squares on the micrographs show tissue fluorescence at baseline ICP (a) and at ICP of 60 mmHg (b) with extravasation of fluorescein into the tissue. The table demonstrates the number of rats out of seven for each group showing dye extravasation. The graph illustrates the average of fluorescein isothiocyanate fluorescence in brain tissue in absolute units (a.u.) at CPP of 50, 60, 70 and 80 mmHg during gradual increase of ICP. Data normalized to normal ICP and presented as mean±SEM, n=7, *p < 0.05, **p < 0.01 between groups; arrowheads show first significant difference within the group.

Discussion

Important distinctions are to be made regarding the cerebral microvasculature. Capillaries are distinguished from shunts by their size (3–7 µm diam), branching and tortuosity.^{6–9, 11, 19} Larger diameter microvessels from 8 µm to 45 µm fall in the range of arterioles and venules and arterio-venous, arterio-arterio, veno-veno shunts that are pre-capillary shunts capable of shunting blood away from capillary beds.^{6, 9–11} Thoroughfare channels (TFC) as described by Hasegawa and Ravens^{6, 11} however, range in diameter from 5 to 12 µm and course through capillary beds. These distinctions are supported by observations of capillary rarefaction in infarcted tissue observed histologically while larger TFC shunts continue to perfuse the infracted tissue without nutrient and gas exchange.^{23, 24} Thus, whereas arterio-veno, arterio-arterio, and veno-veno shunts would result in capillary rarefaction, persistence of TFC shunts continues non-nutritive perfusion through the brain. The result is marked hyperemia through infarcted tissue without gas or nutrient exchange resulting in tissue hypoxia and increased BBB permeability. A pO₂ “gap” is noted with low tissue pO₂ and high cerebral venous pO₂.²⁵ The percentage of these shunts in the brain is small relative to capillaries but exactly what percentage they represent is unknown and has never been quantitated.¹⁹ Our data in the normal rat brain suggest a value of ~35% however, representing not only TFC shunts but all microvessels including arterioles and venules in the range of 8 to 15 µm diameter.

Increased BBB permeability as a result of the increase in capillary pressure would also lead to the initiation of inflammatory mediators such as cytokines and TNF-α from endothelial cells, pericytes, mast cells, and neurons and from the migration of neutrophils, macrophages and microglia into the brain parenchyma.^{26, 27} However, the specific sequence of events in these inflammatory processes requires further investigation.

The use of dopamine to raise CPP may affect MVS flow not only by the increase in CPP but also through vasoactive effects. Dopamine has both β2 adrenergic chronotropic and inotropic effects on the heart but also β2-adrenergic vasodilatory and α1-adrenergic vasoconstrictive effects on cerebral blood vessels.^{28, 29} At low doses, dopamine induces a vasoconstrictive α₁-adrenergic effect while at higher doses a vasodilatory β₂-adrenergic vasodilatory effect. In unanesthetized monkeys, we reported a 20–30% increase in CBF and cerebral metabolic rate for oxygen with dopamine infusion at 100 µg/kg/min.³⁰ The extent to which these effects of dopamine on the cerebrovasculature affected MVS flow remains to be determined.

This study is our first attempt to use the transition from capillary to MVS flow to study the interaction of CPP and ICP in the uninjured brain. The results show that increasing CPP at high ICP from 10 to 60 mmHg progressively attenuates the transition from capillary to MVS flow. The range of ICP studied is higher than that tolerated clinically but shows that at high ICP, the effects of increasing CPP on the CAP/MVS ratio is continuous and monophasic.

Determination of an optimal CPP in patients with elevated ICP is complex. It is multifactorial in causation and varies with magnitude of injury severity and the increase in edema and cerebral blood volume. This study is an extension of earlier studies^1–4 showing that increased ICP results in a decrease in the critical CPP of CBF autoregulation. The decrease in critical CPP was due to the appearance of “apparently” preserved autoregulation caused by a transition to pathological MVS flow and a falsely elevated CBF and autoregulation.⁵ This phenomenon in the normal brain is important because to the dynamics of the susceptibility of normal brain to increased ICP which also coexists in the injured brain and provides a basis for comparison with future studies in the traumatized or injured brain.

CBF autoregulation has been historically evaluated by the relationship between CBF and CPP but our studies show that at high ICP the relationship between CBF and CPP can be misleading. The conventional CBF autoregulation curve for determination of the critical CPP fails at high ICP due to microvascular shunting. In the injured brain, CBF autoregulation may be accurately assessed by transient increases in arterial pressure while recording the change in CBF or ICP. A change in CBF in response to a change in CPP reflects cerebrovascular reactivity (CVRx) and the change in ICP, pressure reactivity (PbRx).^31–33

The transition from capillary to MVS flow secondary to increased ICP is likely caused by an increase in cerebral venous back pressure due to the rise in CSF pressure.⁵ Increased venous pressure decreases the transcapillary pressure gradient resulting in decreased capillary flow and arteriolar dilation thereby increasing the transcapillary pressure gradient to restore capillary flow at a higher capillary hydrostatic pressure.^34–37 The higher capillary hydrostatic pressure promotes the development of brain edema and increased capillary resistance which ultimately redirects flow through lower resistance microvascular shunts resulting in capillary rarefaction and non-nutritive hyperemia.^{24, 38} Elevated CPP in the face of high ICP should promote flow through high resistance capillaries and reduce flow through MVS.

In summary, our studies show that increased CPP at high ICP attenuates the transition from capillary to MVS flow. The question remains as to when it is safe to increase CPP. CPP cannot be raised in a patient with high ICP and loss of CBF autoregulation but patients with high ICP and intact CBF autoregulation may benefit which remains to be determined.