Main

IV/PV hemorrhage is a common complication in premature infants and is associated with a high incidence of neurologic morbidity. Intracranial hemorrhage may result in cerebral vasospasm with consequent ischemic insult, contributing to neurologic deficit(1–3). Bada(4) reported decreased end-diastolic flow velocity in infants within 48 h from onset of IV/PV hemorrhage. Consistent with this abnormal finding was the report of Ment et al.(5) that neonates with IV/PV hemorrhage had decreased CBF measured by xenon inhalation. Also, Volpe et al.(1), using positron emission tomography, observed a significant decrease in CBF in premature infants who sustained major IV hemorrhage.

Aneurysmal subarachnoid hemorrhage (SAH) in adults and SAH in animal models are associated with severe cerebral vasospasm(6, 7). No information exists as to the effects of IV/PV blood on cerebrovascular responses to physiologic or pathologic stimuli in newborns. Previous studies have shown alterations of vascular reactivity of vessels directly exposed to hematoma(8, 9). Extravascular blood produces high levels of activated oxygen that could directly affect blood vessels in the vicinity of the clot(10). However, more recent studies suggest that a hematoma may lead to production of factors;e.g. oxyhemoglobin, thromboxane, peptidoleukotrienes, ET-1, LPA, and PGF2α(2, 11–14) that can circulate in CSF, thus having the potential to modify reactivity of vessels not directly exposed to the hematoma(13, 14). The principal site of bleeding in premature newborn babies is ventricular rather than subarachnoid. However, because circulating factors in CSF appear to contribute to altered vascular reactivity, it is possible that clotted blood in the CSF reservoirs could result in release into the CSF of factors that compromise normal vascular function in areas far distant from the actual hematoma, producing the potential for injury to remote brain areas. Therefore, we wondered if IV/PV blood had the potential to alter vascular reactivity in the cerebral cortex. We hypothesized that IV/PV blood will alter the cerebrovascular reactivity in cortical areas distal from the site of hematoma.

METHODS

The animal protocol was reviewed and approved by the Animal Care and Use Committee of The University of Tennessee, Memphis. Sixteen newborn pigs(1.0-1.8 kg) of either sex, 1-2 d of age, were randomly assigned to this study to have either autologous blood (blood group) or aCSF (control group) injected into their left caudate nucleus.

Injection of blood/aCSF into the caudate nucleus. Newborn piglets were anesthetized with halothane and nitrous oxide. Using aseptic technique, a small longitudinal scalp incision was made over the sagittal suture. The scalp and the underlying periosteum were retracted to expose a 2-cm2 area of the calvarium. A 1-mm burr hole was drilled through the calvarium over the left parietal area, using the stereotaxic coordinates of 2.0 mm posterior to bregma and 4.0 mm lateral to the sagittal suture. These coordinates were derived from our previous experiments wherein methylene blue was injected into the caudate nucleus followed by pathologic confirmation to allow stereotaxic localization of the site for injection. A stainless steel guide cannula (20 g, 25.0 mm long), together with a stopper (26 gauge, 25.0 mm long), was lowered into the caudate nucleus (15.0 mm below the skull) at the site of the burr hole. The cannula was aspirated, and the absence of CSF withdrawal was assured at the time of placement of cannula. The cannula was fixed to the skull by applying dental acrylic cement around the cannula and over the anchor screws; then the scalp was sutured closed.

To inject the blood or aCSF, the stopper was removed, and an injection needle (26 gauge, 25 mm long) that was attached to a syringe via polyethylene tubing was inserted into the guide cannula. Three milliliters of fresh, sterile, nonheparinized autologous arterial blood (removed via direct cardiac puncture) or sterile aCSF were injected over 2 min into the left caudate nucleus; the volume injected was adequate to result in the extension of the injected blood into the lateral ventricles. Because the cannula had to be in place for at least a few hours to prevent back flow of blood over the cerebral cortex and because the removal of the cannula required anesthesia, the experiment was designed to remove the cannula 24 h later, when the animals had to be anesthetized for placement of a cranial window. After cannula placement, the piglets were treated with penicillin and gentamicin and observed closely until the effects of anesthesia had disappeared, then were caged individually with free access to pig milk and water for 24 h.

Cerebrovascular reactivity and CSF cAMP were determined 24 h after injection of blood or aCSF. This was based on our pilot studies in a selected number of piglets that showed a significant fall in global CBF as measured by the microsphere technique at 24 h after injection of blood, followed by a gradual rise of CBF to baseline value by 96 h.

Catheters and cranial window placements. The piglets were anesthetized with intramuscular ketamine hydrochloride and acepromazine (33 and 3.3 mg/kg, respectively). A catheter was inserted into a femoral vein for injection of α-chloralose at an initial dose of 30-50 mg/kg followed by 3 mg/kg every 2-3 h to maintain the desired level of anesthesia; this catheter was also used for maintenance i.v. fluid infusion (D5W, 100 mL/kg day). Another catheter was inserted into the abdominal aorta via a femoral artery for recording the BP and for sampling blood for pH and gases. The trachea was intubated, and the lungs were ventilated with room air using a newborn positive pressure respirator, BP-200 (Bourns Bear Cub infant ventilator, Bear Medical, Allied Healthcare Products, Inc., Riverside, CA). The systemic arterial BP and arterial blood pH and gases were maintained within the normal range during the experiment. Body temperature was maintained between 37 and 38°C using a heating pad with a servo-control system.

A closed cranial window was implanted over the left parietal cortex for direct measurements of pial arteriolar diameters, topical application of different compounds, and collection of CSF. A cranial window was implanted only on one side; our previous experiments in piglets with dual cranial windows revealed that the cerebrovascular responses to vasogenic stimuli were similar between the right and left cerebral cortex after injection of blood into the left caudate nucleus even up to 4 d after cannula placement. To implant the cranial window, the scalp was cut and retracted from the skull. The preimplanted cannula and the screws were removed from the skull. A hole approximately 2 cm in diameter was made in the skull (posterior to the cannula position). An incision was made through the dura and arachnoid membranes, and these membranes were retracted over the edge of the bone. A stainless steel ring with a premounted glass coverslip was inserted in the hole. The window was cemented in place with dental acrylic. Three needles that are pierced through the ring allow injection of aCSF under the window and sampling of CSF. The space under the window was filled with aCSF [Na+ 150 mEq/L, K+ 3 mEq/L, Ca2+ 2.5 mEq/L, Mg2+ 1.2 mEq/L, Cl- 132 mEq/L, glucose 3.7 mM, urea 6 mM, HCO3- 25 mEq/L, pH = 7.33, Pco2 = 46 mm Hg (6.13 kPa), Po2 = 43 mm Hg (5.73 kPa)]; aCSF temperature was kept between 37 and 38 °C. The volume directly below the window was approximately 500 μL.

Pial arterioles were observed with a Wild dissecting microscope. Pial arteriolar diameters were measured using a television camera mounted on the microscope, a videomonitor, and a video micrometer (model VPA-1000, FOR-A, Los Angeles, CA). Briefly, the images of arterioles (47-213 mm in diameter) were displayed on the television monitor, and the sides of selected vessels were bracketed by parallel lines projected by the micrometer. As the arteriolar diameter changed, the lines were moved manually to correspond with the vessel walls. Precalibration of the distance between the lines allowed determination of arteriolar diameters. After implantation of the window, a 30-min stabilization period was allowed before experimentation.

Experimental procedures. Experiments were started when the systemic BP, blood pH, and blood gases were within the normal ranges. Before topical application of vasoactive agents, 300 μL of CSF were collected from under the window by slowly infusing aCSF into the inlet port of the cranial window and allowing the CSF to drip freely into a collection tube from the outlet port. The collection tubes were prechilled and contained EDTA (final concentration in samples, 5 mM) buffered with Tris base to pH 7.4. Immediately after collection, the CSF samples were frozen and stored at -60 °C before assay for cAMP.

Two pial arterioles (one <100 μm and the other ≥100 μm in diameter) in each animal were selected for observation. We measured the pial arterioles every minute over a 10-min period during baseline conditions. Cerebrovascular reactivity to topically applied vasodilators PGE2(10-6 and 10-5 M) and histamine (10-6 and 10-5 M) and vasoconstrictors ET-1 (10-9 and 10-8 M) and LTC4(10-10 and 10-9 M) were tested in both groups of animals. These vasoactive agents (PGE2, histamine, ET-1, and LTC4) were purchased from Sigma Chemical Co.(St. Louis, MO). These agents were applied in random order but with the intention of not applying two dilators or two constrictors back to back. The vasoactive agents were dissolved in aCSF and freshly prepared on the day of experimentation. The lower concentration of each agent was first instilled under the cranial window, and pial arteriolar responses were measured every minute over 10 min. Then a higher concentration of the same agent was topically applied. Before the application of a different chemical agent, the space under the cranial window was flushed several times with aCSF over 30 min, until the pial arteriolar diameters resumed baseline values. A minimum of 10-min control arteriolar diameter measurements was always made between the application of different vasoactive agents. At the end of the experiment each anesthetized piglet was killed with i.v. injection of a KCl solution. The brain was removed for documentation of the presence or absence of the blood clot in the parenchyma/lateral ventricles.

cAMP assay. cAMP was measured in CSF samples using RIA procedures and commercial cAMP 125I scintillation proximity assay systems (Amersham International plc, Amersham, UK). The results are reproducible, and we have extensive experience with this assay(12, 15). Acetylation of CSF samples with a 2:1 mixture of triethylamine and acetic anhydride was performed immediately before assay to increase the sensitivity of the method (the analysis range was 2-128 fmol cAMP/sample). This RIA used the simultaneous addition of sample,125 I-cAMP, rabbit cAMP antibody, and an anti-rabbit second antibody bound to scintillant-incorporated microspheres. Samples were mixed overnight in an orbital shaker (200 rpm) at room temperature. To determine the amount of125 I-cAMP bound to the light-producing microspheres, the vials were counted using a β-scintillation counter. All unknowns were assayed at two dilutions. The cAMP concentration in the CSF sample was calculated from the standard curve.

Statistical analysis. All values were presented as mean± SEM. Separate statistical analysis was done for changes in the small pial arterioles (<100 μm in diameter) and the large pial arterioles(≥100 μm in diameter). The maximum response of the pial arterioles to vasoactive agents was chosen, and the percent change of diameter from the baseline value was calculated. Comparison of variables between the aCSF and the blood groups was made using t test for planned comparison(unpaired). First-level Winsorization(16) was used for statistical analysis of CSF cAMP in the two groups. A p value<0.05 was considered significant.

RESULTS

Two piglets from the blood-injected group were excluded from study. These piglets developed status epilepticus after the injection of blood and thus were anesthetized and killed with i.v. KCl. At autopsy we found extensive intraparenchymal hemorrhage in both piglets. In other blood-injected piglets at autopsy, the blood clot was found in the lateral ventricles and left caudate nucleus (Fig. 1).

Figure 1
figure 1

Autopsy finding in one piglet that received blood injection; blood clot is seen in left caudate nucleus and ventricles.

Full size image

Baseline arterial pH, Pco2, Po2, and systemic arterial BP were comparable in both groups (Table 1). Values of each of these parameters were kept in the normal range during the experiment. The baseline pial arteriolar diameters in the aCSF (control) group ranged from 60 to 213 μm (117 ± 10) and in the blood group ranged from 47 to 196μm (114 ± 12) in diameter.

Table 1 Arterial pH, blood gases, and mean arterial blood pressure (MAP) of piglets in control (aCSF) and blood groups*
Full size table

Cerebrovascular responses to vasoconstrictors. Both the saline and the blood groups responded similarly to vasoconstrictors ET-1 and LTC4. Figure 2 illustrates the large and small pial arteriolar responses in both groups.

Figure 2
figure 2

Percent change in pial arteriolar diameters in response to vasoconstrictor stimuli. Control (aCSF) group, n = 9 and blood group, n = 6. (A, upper panel) Pial arterioles ≥100μm in diameter. (B, lower panel) Pial arterioles <100 μm in diameter. *p < 0.05 blood group compared with control group.

Full size image

Cerebrovascular responses to vasodilators. Pial arterioles ≥100 μm in diameter (Fig. 3A). Topical application of PGE2 resulted in pial arteriolar dilation. However, the percent dilation was significantly lower in the blood group when compared with the control group (p < 0.0003). Topical application of histamine also resulted in pial arteriolar dilation. The percent change was significantly lower in the blood group compared with the control group(p < 0.0001).

Figure 3
figure 3

Percent change in pial arteriolar diameters in response to vasodilator stimuli. Control (aCSF) group, n = 9 and blood group,n = 8. (A, upper panel) Pial arterioles ≥100 μm in diameter. (B, lower panel) Pial arterioles <100 μm in diameter. *p < 0.05 blood group compared with the control group.

Full size image

Pial arterioles <100 μm in diameter (Fig. 3B). Topical application of PGE2 and histamine resulted in pial arteriolar dilation. The vasodilatory responses to PGE2 in the blood group appeared to be attenuated compared with the control group (p = 0.08). Pial arteriolar response to topical application of histamine was significantly attenuated in the blood group compared with the control group(p < 0.01).

The cAMP concentrations in cortical periarachnoid CSF collected before administration of the vasoactive agents were markedly reduced in the blood group compared with the control group (Fig. 4). The CSF cAMP levels in the blood group and the control group were 199 ± 31 and 1092 ± 238 fmol/mL, respectively (p = 0.0006).

Figure 4
figure 4

The cortical periarachnoid CSF cAMP levels 24 h after injection of aCSF (n = 6) or blood (n = 5).*p = 0.0006 blood group compared with the control group.

Full size image

DISCUSSION

Injection of autologous blood into the caudate nucleus resulted in extrusion of blood into lateral ventricles. Twenty-four hours after injection, the main findings in the cortical areas distant from the site of IV/PV blood included: 1) preservation of pial arteriolar responses to vasoconstrictive stimuli (ET-1 and LTC4), 2) a marked attenuation of pial arteriolar responses to cAMP-mediated vasodilators(PGE2 and histamine), and 3) a significant reduction in periarachnoid CSF cAMP concentration.

Under normal physiologic conditions, a balance exists between dilator and constrictor influences on the cerebral vasculature(14, 17). Such balance may be disrupted by IV/PV hemorrhage. To date, the effects of IV/PV blood on cerebrovascular reactivity have not been studied. Related studies have been performed on piglet models of subarachnoid hematoma with autologous blood injected in the subarachnoid area and cerebrovascular reactivity evaluated 1-4 d later(2, 8, 9, 18, 19). Findings from these studies have indicated preservation of pial arteriolar constrictive response to acetylcholine, attenuation of response to norepinephrine, and exaggeration of constriction in response to ET-1 and LTC4(9, 19). In our piglets, the vasoconstrictive responses to ET-1 and LTC4 were not altered.

There is a difference in the duration of blood contact with brain tissue between the subarachnoid hematoma model and our IV/PV blood experiment. In our piglets, this duration was only 1 d, whereas in the subarachnoid hematoma model(9) it was 4 d. Studies in adult humans and experimental animal models showed severe cerebral vasospasm only when exposure to extravascular blood was prolonged over 24 h(6, 7, 18, 19). Prolonged exposure to extravascular blood may increase CSF synthesis of vasoconstrictors, as reported in newborn and adult animal models of SAH, in human adults with SAH, and in neonates with posthemorrhagic hydrocephalus weeks after the detection of IV hemorrhage(2, 11, 13, 20).

Actual direct contact of a blood clot with the cerebral vessels may also affect vascular responsiveness differently from just contact with bloody or xanthochromic CSF. Patients with severe parenchymal or ventricular hemorrhages from ruptured aneurysms may not develop severe vasospasm, when the large subarachnoid blood clot coating the affected blood vessels is removed(3).

It has been reported that cerebral arterioles may develop tolerance to vasodilator agents after SAH(13). Of interest, we found a marked reduction of pial arteriolar dilatory responses to PGE2 and histamine, similar to the findings of Yakubu et al.(8), in piglets with subarachnoid hematoma. Findings from other studies on subarachnoid hematoma showed marked reduction in pial arteriolar dilation response to iloprost (PGI2 analog), and virtual elimination of responses to hypercapnia and hemorrhagic-hypotension(8, 18). The prostanoid system has an important role during the perinatal period, especially for regulation of cerebral circulation in both human infants and newborn piglets(21). Indomethacin, which is a PG-H synthase inhibitor, is commonly administered in premature infants for closure of patent ductus arteriosus and recently for IV hemorrhage prevention. The decreased influence of dilator prostanoid on the cerebral vessels due to IV/PV blood may result in prolonged and uninhibited constriction of pial arterioles in the presence of indomethacin, which could be a potential mechanism for neurologic damage.

The conceivable mechanisms for decreased pial arteriolar dilatory responses in the piglets with intracranial blood, either cortical or ventricular, include: 1) reduced synthesis of dilator prostanoids. This has been reported in experimental hemorrhagic models, in adults with SAH, and babies with IV hemorrhage(11, 22, 23);2) impaired dilatory ability of cerebral vessels due to decreased sensitivity to prostanoids(10, 13). In piglets with subarachnoid hematoma, the vasodilator responses to dilator prostanoids were attenuated(8); 3) loss or reduced affinity of receptors to specific dilator agents. Oxygen free radicals generated from blood clots have been shown to have deleterious effects on endothelial and smooth muscle cells, resulting in their altered receptor characteristics to prostanoids(24–26); and 4) selective alteration of the second messenger pathway. Busija and Leffler(18) reported that pial arteriolar dilation to isoproterenol was not affected in piglets with SAH. Although isoproterenol is a cAMP-dependent vasodilator, this action is not mediated through the prostanoid pathway. In our piglets, a marked reduction of cerebrovascular reactivity was observed to PGE2 and histamine, both cAMP-dependent dilators(15). Prostanoids are involved in histaminergic cerebral vasodilation(27). The selective attenuation of prostanoid-dependent vasodilation in newborn piglets with SAH and IV/PV blood indicates that the mechanism may involve an effect on the prostanoid/cAMP signaling pathway(8, 18).

The CSF cAMP level was markedly reduced in our blood group when compared with the control group. We speculate that extravascular blood in the brain initiated a cascade of events generating by-products which have effects on cerebral adenylyl cyclase expression or activity. After erythrocyte breakdown, oxyhemoglobin is released, and elevated levels are maintained in the CSF for a few days(28). Oxyhemoglobin release is associated with oxygen free radical generation which may inhibit cAMP production by damaging mitochondrial function(24, 28–30). In addition, oxyhemoglobin stimulates synthesis of endothelin(31); CSF ET-1 levels are increased after intracranial hemorrhage(14, 32). ET-1 may inhibit adenylyl cyclase via a pertussis toxin-sensitive pathway resulting in decreased cAMP(33). Also, ET-1 may increase production of LPA which has an inhibitory effect on adenylyl cyclase activity. Cortical application of ET-1 in newborn pigs stimulates periarachnoid CSF LPA production(34), and LPA levels are found to be increased after experimental SAH(12, 34). Therefore, several different mechanisms, possibly interrelated, may ultimately contribute to inhibition of the adenylyl cyclase second messenger pathway.

Hypoxic/ischemic insult in premature infants can result in neurologic injury with or without IV/PV hemorrhage. Although the actual rupture of cerebral vessels has immediate effects on cerebral hemodynamics, these immediate effects may be compounded by altered vasogenic responses remote from the site of blood/hemorrhage. Long-term neurologic abnormalities from IV/PV hemorrhage may be augmented by prolonged ischemia secondary to delayed cerebral vasospasm or inability of cerebral vessels to respond appropriately to vasogenic stimuli.