Abstract

OBJECTIVE:

The ultrastructure of perinidal capillaries in cerebral arteriovenous malformations (AVMs) was examined to clarify their pathomorphological features.

METHODS:

Fifteen AVM specimens were dissected and divided into perinidal and intranidal groups and processed for ultrastructural study immediately after surgical removal. Eleven of the patients had presented with hemorrhage. Tissue from four normal controls was also studied. Electron microscopy was used to compare features of the blood-brain barrier and endothelial cells (ECs) of capillaries in perinidal, intranidal, and controls.

RESULTS:

Perinidal capillaries demonstrated abnormal ultrastructure of the blood-brain barrier with no basement membranes and astrocytic foot processes. ECs had fenestrated luminal surfaces. Large gaps were observed at endothelial intercellular junctions. ECs contained numerous filopodia, large numbers of cytoplasmic processes, numerous micropinocytotic vesicles, and the cytoplasm contained more filaments than those observed in controls. Pericytes were rich in pinocytotic vesicles, vacuoles, and filaments. Their processes were in close contact with ECs. Weibel-Palade bodies were present in perinidal ECs.

CONCLUSION:

The absence of blood-brain barrier components in perinidal capillaries may contribute to extravasation of red blood cells into the surrounding brain in the absence of major hemorrhage and explain the gliosis and hemosiderin occasionally seen around AVMs. Cellular differentiation and proliferation in perinidal capillaries should be included in a systematic study aimed at a better understanding of the mechanisms underlying the recurrence of surgically removed AVMs.

Arteriovenous malformations (AVMs) of the brain are the most common clinically reported central nervous system vascular malformations, affecting approximately 40 million people worldwide (18). They are characterized by direct shunts between arteries and veins, without an intervening capillary bed (19, 21). Therapeutic options include endovascular embolization, radiosurgery, or excision. The operative approach consists of total removal of the lesion after occlusion of the feeding arteries, dissection of the nidus from normal brain tissue, and occlusion of the draining veins. Dissection of the nidus from normal brain is often the most difficult part of the operation because abnormal fragile vessels frequently cause bleeding (11). Shortly after complete resection of a nidus, postoperative bleeding or edema in the surrounding brain tissue occasionally occurs (1). Recurrence of AVMs has been reported after angiographically confirmed total excision of the nidus (10). The mechanisms of these two clinically important issues are poorly understood. Morphological studies of the perinidal parenchyma of AVMs are likely to add to the understanding of pathological mechanisms underlying their clinical behavior. The vascular structures in the perinidal tissue have been evaluated by light microscopy recently (2, 26). However, the ultrastructure of perinidal microvasculature using electron microscopy has not been reported. Our goal was to examine systematically the perinidal microvascular structures of human brain tissue surrounding AVM nidi, focusing on blood-brain barrier components and endothelial cells (ECs). The ultrastructural integrity of the former might contribute to the understanding of intraoperative and postoperative bleeding. The latter might be a hint for the involvement of growth factors and hemodynamic changes acting on perinidal capillaries and possibly nidal recurrence.

PATIENTS AND METHODS

Patients and Lesions

Using standard microsurgical techniques, 15 AVM specimens were resected from 11 patients with a history of hemorrhage and four patients without a history of hemorrhage. The patients gave informed consent for the use of tissue, and ethical permission was given by the Human Ethics Committee of the South Eastern Sydney Area Health Service (Sydney, Australia). Normal human control tissues were obtained from patients undergoing epilepsy surgery and patients dying from non-neurological causes. The patient and lesion characteristics are summarized in Table 1.

TABLE 1.

Patient and lesion characteristics

Specimen Preparation

Fresh surgical specimens were processed for ultrastructural study using a standard protocol (3). Immediately after removal, AVM specimens were dissected, and parenchyma tissue was divided into perinidal and intranidal groups. Five dozen 1-mm3 fragments were evenly taken from each perinidal or intranidal group and fixed in 2% paraformaldehyde and 1% glutaraldehyde in 0.1 mol/L sodium cacodylate at 4°C for 2 hours. The specimens were rinsed three times sequentially in 0.1 mol/L sodium cacodylate for 10 minutes and then postfixed by immersion in 1% OsO4 prepared in 0.1 mol/L sodium cacodylate and rotated at 25°C for 1 hour. After rinsing in 0.1 mol/L sodium cacodylate, the specimens were initially stained en bloc with 2% uranyl acetate for 1 hour and sequentially dehydrated in increasing concentrations of ethanol (EtOH; 50%, 70%, 90%, and 100% EtOH 10 min, respectively; absolute dry EtOH 30 min; a 1:1 dry EtOH-propylene oxide mixture 15 min; and propylene oxide 15 min). Before embedding, tissue was sequentially infiltrated by serial immersion in increasing concentrations of epoxy (a mixture of 1 part propylene oxide:1 part epoxy and 1:3 propylene oxide-epoxy for 1 h, respectively, and pure epoxy resin overnight). The specimens were embedded in pure epoxy resin and polymerized at 70°C for 72 hours. Control specimens were processed identically.

Perinidal and intranidal vessels were identified by light microscopy on semithin sections stained with 1% toluidine blue. Multiple, serial ultrathin sections of hundreds of capillaries in perinidal, intranidal tissue, and controls were cut with an ultratome, and 50-nm sections were collected on grids coated with polyvinyl formal support film. The sections were stained with Reynold's lead citrate solution to enhance contrast (23).

Electron Microscopy and Analysis

Transmission electron microscopy was used to evaluate and analyze the microvasculature, EC morphology, and cytoplasmic organelles. Structural features were identified and compared on similar size vessels among perinidal, intranidal, and controls by four observers blinded to the nature of the specimen and photographed using 50 electron micrographs per epoxy block and 50 epoxy blocks per specimen. Ultrastructural findings on electron micrographs were viewed by nine observers blinded to the sample nature.

Histology

Paraffin sections of specimens were stained with hematoxylin-eosin using a standard protocol. Perinidal and intranidal parenchyma were compared using a Leica DMR (Leica Microsystems, Wetzlar, Germany). Digital photographs were acquired using a Zeiss digital camera (Axiocam, Zeiss, Oberkochen, Germany).

RESULTS

Blood-Brain Barrier

Normal cerebral capillaries consisted of thin-walled tubes lined by a single layer of extremely flattened polygonal ECs. In their size and ultrastructure, they corresponded to capillaries with an endothelial lining of varying thickness and no fenestrations. The ECs were connected by tight cellular junctions. At these sites, the intercellular clefts showed closely opposed membranes of neighboring ECs. On their exterior surface, the ECs were coated with a glycoprotein basal lamina. Pericyte processes were also enclosed within the basement membrane complex. On capillary cross section (Fig. 1A), the endothelial layer was thin except in areas occupied by an oval nucleus. The cytoplasm in the perinuclear region contained a sparse distribution of mitochondria, free ribosomes, and short segments of endoplasmic reticulum. The plasma membrane was a typical unit membrane. The ECs rested on a basal lamina of approximately 80 nm in thickness. Perivascular cells were surrounded by a basal lamina that was shared by the endothelium at points of close apposition of the two cells. The cellular components of the normal blood-brain barrier were identifiable, namely, ECs, pericytes, and astrocyte foot processes (Fig. 1B).

FIGURE 1.

Normal human cerebral capillaries. A, transverse section of normal capillary. It consists of flat endothelial cells (EC) bound to one another through intercellular junctions. At the exterior side, ECs are coated with relatively thick basal lamina (bl). Pericytic cells are also covered by a single basement membrane. L, lumen; j, EC junction; n, nucleus; r, cytoplasm containing ribosomes; m, mitochondria; G, Golgi complex; v, a vesicle; Nce, nonfenestrated, continuous endothelium; Cp, cytoplasm of pericyte; Epj, endothelial-pericytic junction; Pfa, perivascular feet of astrocytes. Scale bar = 1 μm. B, insert of A, blood-brain barrier consists of tight junction (fusion of outer leaflets of apposing cell membranes, Tj), pericytes (P), and astrocyte foot processes (Afp). Scale bar = 1 μm.

FIGURE 1.

Normal human cerebral capillaries. A, transverse section of normal capillary. It consists of flat endothelial cells (EC) bound to one another through intercellular junctions. At the exterior side, ECs are coated with relatively thick basal lamina (bl). Pericytic cells are also covered by a single basement membrane. L, lumen; j, EC junction; n, nucleus; r, cytoplasm containing ribosomes; m, mitochondria; G, Golgi complex; v, a vesicle; Nce, nonfenestrated, continuous endothelium; Cp, cytoplasm of pericyte; Epj, endothelial-pericytic junction; Pfa, perivascular feet of astrocytes. Scale bar = 1 μm. B, insert of A, blood-brain barrier consists of tight junction (fusion of outer leaflets of apposing cell membranes, Tj), pericytes (P), and astrocyte foot processes (Afp). Scale bar = 1 μm.

Under the light microscope, the AVM nidus was a mass of abnormal arteries and veins with no intervening brain tissue. No capillary bed was observed in the nidus. Figure 2A illustrates part of a compact nidus, perinidal, and the transition area where perinidal capillary bed turns into relatively normal brain tissue. Small vessels branching from the nidal wall and extending through to the surrounding brain tissue formed perinidal capillary networks. Perinidal capillaries were located in brain tissue 1 to 6 mm from the nidal border. The size of perinidal capillary bed was positively correlated with the diameter of the nidus. The perinidal capillary networks formed two-way connections, not only to the nidus but also to the normal capillaries. Dilation was a main feature of perinidal capillaries, with the diameters 8 to 16 times of those normal capillaries. The caliber of these capillaries was negatively correlated with the distance from the nidus and eventually appeared as a normal size. The transmission electron microscope revealed that intervening brain tissue occurred in a compact nidus in all cases. Perinidal capillaries demonstrated abnormal ultrastructural features of the blood-brain barrier (Fig. 2, B and C), with no basement membranes and astrocytic foot processes seen. The ECs showed fenestrated processes, numerous filopodia, lysosomes, cytoplasmic vesicles, and vacuoles. Large vacuoles were observed near EC junctions. There was approximately one pericyte for every three EC.

FIGURE 2.

A nonbleeding AVM specimen. A, small vessels (arrowheads) branch from nidal wall and extend through to surrounding brain tissue, forming perinidal capillary networks (hematoxylin-eosin; scale bar = 200 μm). B, perinidal capillary wall absence of a competent blood-brain barrier. E, erythrocyte; Mon, monocyte; M, macrophage; f, filopodia; cv, cytoplasmic vesicle; V, vacuole; L, lysosome; P, pericytes. Scale bar = 5 μm. C, insert of B, endothelial junction complex of perinidal capillary. Overlapping endothelial processes (Oep). Scale bar = 2 μm.

FIGURE 2.

A nonbleeding AVM specimen. A, small vessels (arrowheads) branch from nidal wall and extend through to surrounding brain tissue, forming perinidal capillary networks (hematoxylin-eosin; scale bar = 200 μm). B, perinidal capillary wall absence of a competent blood-brain barrier. E, erythrocyte; Mon, monocyte; M, macrophage; f, filopodia; cv, cytoplasmic vesicle; V, vacuole; L, lysosome; P, pericytes. Scale bar = 5 μm. C, insert of B, endothelial junction complex of perinidal capillary. Overlapping endothelial processes (Oep). Scale bar = 2 μm.

Capillary Bleeding

Perinidal capillaries exhibited evidence of previous microhemorrhage (Fig. 3). Hemosiderin stained the fringe of normal brain tissue surrounding the capillaries (Fig. 3A). There was separation of the endothelium and extravasation of blood cells from the vascular wall (Fig. 3C). Free erythrocytes at different stages of degradation were present in the collagenous stroma, distributed among macrophages and neutrophils (Fig. 3B), neurons and neuropil (Fig. 3C), fibroblasts, and connective tissue (Fig. 3D) surrounding the capillaries. There were more macrophages in hemorrhagic AVMs than in nonhemorrhagic AVMs.

FIGURE 3.

Microhemorrhage from a perinidal capillary. A, vessels showed hemosiderin ring (arrowheads) demonstrating previous hemorrhage history (hematoxylin-eosin; scale bar = 100 μm). B and C, erythrocytes (E) undergo degradation (pale and reduction in size) in molecular layer of cerebral cortex. L, lumen; Lym, lymphocyte; M, macrophage; F, fibroblast; N, neuron; Neu, neuropil of molecular layer of cerebral cortex; Nec, necrotic ECs. Endothelial gaps (arrowheads). B, scale bar = 5 μm; C, scale bar = 5 μm. D, erythrocytes (E) observed in molecular layer of cerebral cortex (M) (scale bar = 5 μm).

FIGURE 3.

Microhemorrhage from a perinidal capillary. A, vessels showed hemosiderin ring (arrowheads) demonstrating previous hemorrhage history (hematoxylin-eosin; scale bar = 100 μm). B and C, erythrocytes (E) undergo degradation (pale and reduction in size) in molecular layer of cerebral cortex. L, lumen; Lym, lymphocyte; M, macrophage; F, fibroblast; N, neuron; Neu, neuropil of molecular layer of cerebral cortex; Nec, necrotic ECs. Endothelial gaps (arrowheads). B, scale bar = 5 μm; C, scale bar = 5 μm. D, erythrocytes (E) observed in molecular layer of cerebral cortex (M) (scale bar = 5 μm).

ECs and Pericytes

ECs in the microvasculature of perinidal capillaries were heterogeneous (Fig. 4, A and B). In a transverse section of a microvessel shown in Figure 4A, ECs were hyperactive in Area I, rich in filopodia on the luminal surface, contained numerous cytoplasmic vesicles and vacuoles, and had EC junctions with multiple focal dilatations and large vacuoles. In the same cross-section in Area II, ECs were multilayered with widened junction spaces between layers. In Area III, ECs were hypoactive, appearing pale and swollen, and the microvessel wall was partially collapsed. In another cross-section of a capillary network (Fig. 4B), ECs were immature in Area I where EC junctions showed widening at intercellular junctions with intercellular clefts between neighboring ECs. ECs were hyperactive and differentiating in Area II, where ECs showed numerous filopodia, cytoplasmic processes, pinocytotic vesicles, lysosomes, and large cytoplasmic vacuoles.

FIGURE 4.

Perinidal capillary wall lined by heterogeneous ECs. A, capillary lined by a discontinuous layer of hyperactive ECs (I), multilayered ECs (II), partially collapsed necrotic endothelial processes (III). Intercellular gaps between multiple abnormal layers of EC processes (*); a junction gap (arrowhead) between adjacent ECs (scale bar = 5 μm). B, capillary network lined by fenestrated, discontinuous, immature endothelial processes (I), fenestrated, hyperactive and differentiating ECs (II). Gaps (arrowheads) between fragments of immature endothelial processes (scale bar = 3 μm). C and D, capillary lined by an extremely thin, discontinuous fenestrated endothelium (I), thick endothelial processes (II), pilopodia toward both exterior and interior surfaces (III). Pinocytotic vesicles (Pv) (scale bar = 3 μm). E, abnormal pericytes in a perinidal capillary. Cytoplasm of pericyte is large and rich in organelles, denoting cellular hyperactivity, and contains various inclusions (scale bar = 1 μm). F, endothelial processes are in close contact with pericyte (arrowhead), cytoplasmic process (cp), filaments (f) (scale bar = 1 μm).

FIGURE 4.

Perinidal capillary wall lined by heterogeneous ECs. A, capillary lined by a discontinuous layer of hyperactive ECs (I), multilayered ECs (II), partially collapsed necrotic endothelial processes (III). Intercellular gaps between multiple abnormal layers of EC processes (*); a junction gap (arrowhead) between adjacent ECs (scale bar = 5 μm). B, capillary network lined by fenestrated, discontinuous, immature endothelial processes (I), fenestrated, hyperactive and differentiating ECs (II). Gaps (arrowheads) between fragments of immature endothelial processes (scale bar = 3 μm). C and D, capillary lined by an extremely thin, discontinuous fenestrated endothelium (I), thick endothelial processes (II), pilopodia toward both exterior and interior surfaces (III). Pinocytotic vesicles (Pv) (scale bar = 3 μm). E, abnormal pericytes in a perinidal capillary. Cytoplasm of pericyte is large and rich in organelles, denoting cellular hyperactivity, and contains various inclusions (scale bar = 1 μm). F, endothelial processes are in close contact with pericyte (arrowhead), cytoplasmic process (cp), filaments (f) (scale bar = 1 μm).

Endothelial processes varied in thickness and structure (Fig. 4, C and D). Fenestrated endothelial processes were observed in Area I. Relatively thick endothelial processes were seen in Area II. Endothelial processes with filopodia toward both exterior and interior surfaces were noted in Area III.

Pericytes were less common in perinidal capillaries than in control vessels (Figs. 2, B and C, and 4, B and C). They were heterogeneous, not only rich in pinocytotic vesicles and vacuoles (Fig. 4E), but also abundant in filaments, and formed filament bundles (Fig. 4F). Their processes were sometimes in close contact with ECs.

Intracellular Organelles

Weibel-Palade bodies were identified in ECs near interendothelial junctions of perinidal capillaries in all the specimens (Fig. 5), but not in control vessels. Weibel-Palade body membranes often contacted small vesicles that were the continuation of long invaginations of the plasma membrane. They exhibited the typical pattern of a rod-shaped cytoplasmic component consisting of a bundle of fine tubules, enveloped by a tightly fitted membrane. They were approximately 144 nm thick, measured 450 nm in length, and contained several small tubules embedded in a dense matrix and were disposed parallel to the long axis of the rod.

FIGURE 5.

Weibel-Palade bodies in a perinidal capillary. Weibel-Palade body (arrowhead) was identified as a membrane-bound, rod-shaped structure containing parallel microtubules in junction area between endothelial cells (EC). L, lumen; F, fibroblast. Scale bar = 1 μm.

FIGURE 5.

Weibel-Palade bodies in a perinidal capillary. Weibel-Palade body (arrowhead) was identified as a membrane-bound, rod-shaped structure containing parallel microtubules in junction area between endothelial cells (EC). L, lumen; F, fibroblast. Scale bar = 1 μm.

Differences between Nonbleeding and Bleeding AVMs

Histological and electromicroscopic differences between the perinidal tissue in AVMs with hemorrhage and those without hemorrhage are summarized in Table 2. Histological specimens without hemorrhage history showed the typical characteristics of AVMs (Fig. 2A). Abnormal arteries and veins coexisted in the AVM nidus. Red blood cells were maintained within the vascular channels. Specimens with bleeding history unveiled hemosiderin staining of the fringe of normal brain parenchyma surrounding the capillaries (Fig. 3A). Free red blood cells were observed in the normal brain tissue adjacent to perinidal capillaries (Fig. 3, B–D). Infiltration of microphages appeared in both nonbleeders and bleeders, but a greater number of microphages were found in specimens with bleeding history.

TABLE 2.

Structural comparison of perinidal tissue in arteriovenous malformations with hemorrhage and those without hemorrhage

DISCUSSION

Previous ultrastructural studies of AVMs focused entirely on the nidus, and, according to Wong et al. (33), the blood vessels within the nidus have preserved vessel wall integrity. Recent light microscopy studies, however, have suggested the AVM nidus is accompanied by a perinidal dilated capillary network (2, 26). To the best of our knowledge, studies addressing the ultrastructure of the microvasculature in the perinidal parenchyma have not appeared in the literature. The ultrastructure of perinidal dilated capillary networks could be important in the understanding of mechanisms underlying intraoperative and postoperative bleeding, growth, and recurrence of surgically removed cerebral AVMs.

Incompetent Blood-Brain Barrier

The competency of the blood-brain barrier is controlled by the biochemical properties of the plasma membranes of the capillary ECs, pericytes, and astrocyte foot process. In blood-brain barrier capillaries, the ECs are joined by continuous tight junctions produced in response to paracrine signals from astrocytes (20). These tight junctions prevent molecules from diffusing through the gaps between the cells (i.e., the paracellular route). These ECs lack fenestrations and have a very low density of endothelial vesicles that keep the diffusion of molecules across the cell to a minimum (5).

Total resection of AVM nidus is occasionally followed by hemorrhage and edema of the surrounding brain tissue. This is possibly related to the abnormal perinidal capillaries and the deficient blood-brain barrier. We have demonstrated that astrocytic foot processes are absent in perinidal capillaries, a finding that has also been reported in an experimental model of chronic steal related to an arteriovenous fistula (27). In the perinidal capillary endothelium, a proportion of ECs have changes similar to those of necrosis and apoptosis leading, as noted, to the activation of death-associated protein 3, caspase 4, and tumor necrosis factor and up-regulation of cell adhesion molecules (9), leaving infrequent, but large gaps in the vessel wall (Figs. 3C and 4, A and B). These large gaps could be responsible for significant fluid extravasation and microhemorrhages from perinidal capillaries when there are changes in blood pressure. The abnormal antiendothelial junctions are “shortened,” with a restricted area of overlap between adjoining ECs, suggesting a shorter, more direct, paracellular diffusion path. Some interendothelial junctions are “open,” suggesting a widened paracellular channel. These findings support the hypothesis that paracellular movement of fluid is a significant cause of blood-brain barrier opening in postoperative brain edema.

Fenestrae were present in EC in the perinidal capillaries (Figs. 3B and 4, C and D). Fenestrations are “windows” in attenuated areas of the endothelial cytoplasm 60 to 70 nm in diameter that may or may not be closed by a membranous diaphragm. Fenestrae form a second permeability route for the movement of fluids into the surrounding brain tissue. We observed numerous vesicles and large vacuoles in the perinidal capillary ECs and pericytes (Figs. 2B, 3, B and C, and 4, A–E). They may form a third permeability route for the movement of fluids into the surrounding brain tissue.

The blood-brain barrier opening in the perinidal capillaries could be biphasic and different for small and large molecules. This may explain the usual lack of enhancement of the perinidal region on axial imaging with contrast agents. An alternative explanation is that the abnormal perinidal vessels are beyond the resolution of current imaging modalities.

The pathogenic mechanisms that lead to the formation of an incompetent blood-brain barrier in the perinidal capillaries are not well understood. However, at least two factors can be considered. In normal cerebral capillaries, it is possible that inductive signals from brain cells initiate and maintain the development of the blood-brain barrier phenotype (8). The perinidal capillaries might fail to elaborate normal inductive factors that are necessary for the development and maintenance of blood-brain barrier features in new vessels. Alternatively, cells in the actual nidus may secrete inductive signals that trigger the formation of abnormal capillaries without the blood-brain barrier phenotype. Many growth factors related to abnormal angiogenesis enhance vascular permeability either directly or indirectly. Vascular endothelial growth factor (VEGF) is a particularly interesting candidate because its receptors Flt-1, Flk-1, and Flt-4 are expressed in the cytoplasm of vascular endothelium of the AVM nidus (14). During embryogenesis, VEGF is secreted by ventricular neuroepithelial cells at the time when ECs proliferate rapidly but is reduced in adult brain when EC proliferation has ceased (4). It is possible that a developmental failure of capillaries in the early and middle phases of the embryonic period is related to the development of AVMs (17). When intracranial AVMs grow or recur, VEGF expression is again up-regulated in the cytoplasm of reactive astroglia, and its receptors are up-regulated in the cytoplasm of the vascular endothelium within the AVM nidus (14). VEGF was also reported as a vascular permeability factor isolated from ascites fluid that caused marked vascular extravasation after intradermal injection (28). It is likely that the VEGF-VEGF receptor system is related to both angiogenesis and the increased vascular permeability in the nidus and its perinidal capillaries.

The finding of extravasated erythrocytes could be an artifact resulting from the effects of surgery. We think this is unlikely because the erythrocytes observed were at different stages of degradation. Some of them were phagocytized by macrophages.

ECs and Pericytes

It is apparent that ECs are heterogeneous and can vary in phenotype, function, antigenic composition, metabolic properties, and in their response to growth factors. Species heterogeneity has been reported (24), and within species, ECs differ depending upon the size, function, and location of the vessel and can even vary within discrete segments of a single microcirculatory loop (31). Although various aspects of the endothelium have been published, none have reported on EC heterogeneity of the same ultrathin section. ECs exhibited differences in viability (Figs. 3C and 4A) layers/intercellular spaces (Fig. 4, A and B), maturity (Fig. 4B), and differentiation (Fig. 4, C and D), resulting in heterogeneity of thickness, strength, and permeability in the perinidal capillaries. On the basis, primarily, of the fine structure of the endothelium and its basal lamina, three types of capillaries (continuous, fenestrated, and discontinuous) were observed in the same capillary.

Because of the seemingly important role of pericytes in the blood-brain barrier, it will be worthwhile to discuss the heterogeneity in the subcellular morphology of this cell type. For example, pericytes may be "granular" or "agranular," that is, cytoplasmic lysosomes may be abundant or sparse (30). Furthermore, lysosomes of brain pericytes are very strongly reactive for acid phosphatase. It is likely, therefore, that the proposed phagocytic function of pericytes is especially developed in perinidal capillaries. Experimentally, the number of lysosomes in brain pericytes can be increased by disrupting the blood-brain barrier (e.g., with hyperosmotic mannitol together with the administration of adriamycin) (15). In the perinidal capillaries, the increased number of lysosomes (Fig. 4E) may be related to blood-brain barrier opening.

In perinidal capillaries, pericytes are rich in plasmalemmal and cytoplasmic vesicles often associated with ECs. The majority of vesicles in pericytes are not independent but interconnected and continuous with the cell surface. Most of the vesicles were connected with the surface facing the EC; only 20% of vesicles were connected with the adventitial surface (Fig. 4E). In normal capillaries, most of the vesicles are connected with the adventitial surface of the capillary, whereas approximately 10% contact the cell surface facing the capillary lumen (30). Pericytes also contain microfilament bundles, which have been assumed to be contractile (31), and are concentrated in the cytoplasm adjacent to the EC (Fig. 4F).

Pericytes contribute to the joint EC-pericyte basal lamina by synthesizing basal lamina collagen, laminin, and glycosaminoglycans, as demonstrated in retinal microvessel pericytes in vitro (29). Pericytes, therefore, seem to modify the basal lamina as a charged matrix with permeability barrier properties.

Differentiation in Perinidal Capillaries

After complete AVM resection, there is the possibility of nidal recurrence at the primary location (7, 10). Cellular differentiation and proliferation of perinidal capillaries should be studied as part of an investigation into the mechanisms underlying nidal recurrence. Ultrastructural evidence of cellular differentiation and proliferation in perinidal capillaries includes numerous filopodia (Figs. 2B and 4, A–C), large numbers of cytoplasmic processes (Figs. 2B; 3, B and C; and 4, A and C), numerous micropinocytotic vesicles (Figs. 2B; 3, B and C; and 4A), presence of Weibel-Palade bodies (Fig. 5), and the cytoplasm contains more filaments than those observed in the controls (Fig. 4, B, C, and E). These cytoplasmic filaments and microtubules in Weibel-Palade bodies are similar to those occurring in the ECs in cerebral arteries during hypertension, presumably related to the movement of Weibel-Palade bodies (16) or degenerative changes induced by increased blood flow. These ultrastructural changes could be caused by the hemodynamic effects of the AVM, including relative ischemia in the tissue adjacent to the AVM. An alternative explanation is the influence of over-expression of vascular growth factors (possibly responsible for the original AVM development). VEGF and basic fibroblast growth factor 2 are expressed in cerebrovascular malformations (13, 25). Basic fibroblast growth factor 2 is thought to involve initiation of vasculogenesis, and VEGF is an early positive regulator of vasculogenesis, acting as a mitogen with primary specificity for ECs. They are expressed at high concentrations during embryonic development and are normally undetectable in adult cerebral vasculature (6).

Perinidal capillaries show marked variations in size and shape. Most of their lining ECs are similar to these of growing and recently formed blood vessels. They are large, of irregular shape, with prominent nuclei (Figs. 2B and 3B) discontinuities and gaps of the vascular wall (Fig. 4A). They contain increased number of organelles (Fig. 5) and prominent rough endoplasmic reticulum indicative of active metabolism. Of special interest is the presence of pericytes adjacent to ECs (Fig. 4F). Some are large, with numerous organelles, indicating that these cells are metabolically active. The pericytes are thought to be multipotent and capable of transforming into other mesenchymal cells (29).

The present study provides the first direct evidence of Weibel-Palade bodies in cerebral perinidal capillaries (Fig. 5). These membrane-bound organelles contain longitudinally arranged tubular structures that represent highly organized polymers of the adhesive glycoprotein von Willebrand factor (32). Weibel-Palade bodies are present in ECs of systemic vessels but are not present or rarely seen in cerebral ECs. They are identified in the vasculature of cerebral tumors (12, 22).

CONCLUSIONS

The absence of blood-brain barrier components in perinidal capillaries may lead to leakage of red blood cells into the surrounding brain in the absence of major hemorrhage. The finding of abnormally differentiated and proliferative ECs in perinidal capillaries raises the possibility that these are involved in the occasional recurrence of surgically removed AVMs.

REFERENCES

1.
Al-Rodhan
NR
,
Sundt TM Jr, Piepgras
DG
,
Nichols
DA
,
Rufenacht
D
,
Stevens
LN
Occlusive hyperemia: A theory for the hemodynamic complications following resection of intracerebral arteriovenous malformations
.
J Neurosurg
 
78
:
167
175
,
1993
.
2.
Attia
W
,
Tada
T
,
Hongo
K
,
Nagashima
H
,
Takemae
T
,
Tanaka
Y
,
Kobayashi
S
Microvascular pathological features of immediate perinidal parenchyma in cerebral arteriovenous malformations: giant bed capillaries
.
J Neurosurg
 
98
:
823
827
,
2003
.
3.
Bozzola
JJ
,
Russell
LD
Specimen preparation for transmission electron microscopy
, in
Bozzola
JJ
,
Russell
LD
(eds)
Electron Microscopy
.
Boston
,
Jones and Bartlett
,
1992
, pp
14
38
.
4.
Breier
G
,
Albrecht
U
,
Sterrer
S
,
Risau
W
Expression of vascular endothelial growth factor during embryonic angiogenesis and endothelial cell differentiation
.
Development
 
114
:
521
532
,
1992
.
5.
Brightman
MW
,
Reese
TS
Junctions between intimately apposed cell membranes in the vertebrate brain
.
J Cell Biol
 
40
:
648
677
,
1969
.
6.
Gault
J
,
Sarin
H
,
Awadallah
NA
,
Shenkar
R
,
Awad
IA
Pathobiology of human cerebrovascular malformations: Basic mechanisms and clinical relevance
.
Neurosurgery
 
55
:
1
17
,
2004
.
7.
Hashimoto
T
,
Lawton
MT
,
Wen
G
,
Yang
GY
,
Chaly
T
,
Stewart
CL
,
Dressman
HK
,
Barbaro
NM
,
Marchuk
DA
,
Young
WL
Gene microarray analysis of human brain arteriovenous malformations
.
Neurosurgery
 
54
:
410
423
,
2004
.
8.
Hashimoto
N
,
Nozaki
K
Do cerebral arteriovenous malformations recur after angiographically confirmed total extirpation?
Crit Rev Neurosurg
 
9
:
141
146
,
1999
.
9.
Ikeda
E
,
Flamme
I
,
Risau
W
Developing brain cells produce factors capable of inducing the HT7 antigen, a blood-brain barrier-specific molecule, in chick endothelial cells
.
Neurosci Lett
 
209
:
1
4
,
1996
.
10.
Kader
A
,
Goodrich
JT
,
Sonstein
WJ
,
Stein
BM
,
Carmel
PW
,
Michelsen
WJ
Recurrent cerebral arteriovenous malformations after negative postoperative angiograms
.
J Neurosurg
 
85
:
14
18
,
1996
.
11.
Kato
Y
,
Sano
H
,
Nagahisa
S
,
Yoshimoto
J
,
Ninomiya
T
,
Kanaoka
N
,
Kanno
T
Control of hemorrhage during AVM surgery: With special reference to treatment of dilated capillaries and arteries around the nidus
.
Minim Invasive Neurosurg
 
41
:
62
65
,
1998
.
12.
Kawamura
J
,
Garcia
JH
,
Kamijyo
Y
Cerebellar hemangioblastoma: Histogenesis of stroma cells
.
Cancer
 
31
:
1528
1540
,
1973
.
13.
Kılıc
T
,
Pamir
MN
,
Kullu
S
,
Eren
F
,
Ozek
MM
,
Black
PMcL
Expression of structural proteins and angiogenic factors in cerebrovascular anomalies
.
Neurosurgery
 
46
:
1179
1192
,
2000
.
14.
Koizumi
T
,
Shiraishi
T
,
Hagihara
N
,
Tabuchi
K
,
Hayashi
T
,
Kawano
T
Expression of vascular endothelial growth factors and their receptors in and around intracranial arteriovenous malformations
.
Neurosurgery
 
50
:
117
126
,
2002
.
15.
Kondo
A
,
Inoue
T
,
Nagara
H
,
Tateishi
J
,
Fukui
M
Neurotoxicity of adriamycin passed through the transiently disrupted blood-brain barrier by mannitol in the rat brain
.
Brain Res
 
412
:
73
83
,
1987
.
16.
Manneville
JB
,
Etienne-Manneville
S
,
Skehel
P
,
Carter
T
,
Ogden
D
,
Ferenczi
M
Interaction of the actin cytoskeleton with microtubules regulates secretory organelle movement near the plasma membrane in human endothelial cells
.
J Cell Sci
 
116
:
3927
3938
,
2003
.
17.
Mayer
PL
,
Kier
FL
The ontogenetic and phylogenetic basis of cerebrovascular anomalies and variants
, in
Apuzzo
ML
(ed)
Brain Surgery: Complication Avoidance and Management
 .
New York
,
Churchill Livingstone
,
1993
, pp
691
792
.
18.
Michelson
WJ
Natural history and pathophysiology of arteriovenous malformations
.
Clin Neurosurg
 
26
:
307
313
,
1978
.
19.
Mullan
S
,
Mojtahedi
S
,
Johnson
DL
,
MacDonald
RL
Embryological basis of some aspects of cerebral vascular fistulas and malformations
.
J Neurosurg
 
85
:
1
8
,
1996
.
20.
Neuwelt
EA
Mechanisms of disease: The blood-brain barrier
.
Neurosurgery
 
54
:
131
140
,
2004
.
21.
Padget
DH
The cranial venous system in man in reference to development, adult configuration, and relation to the arteries
.
Am J Anat
 
98
:
307
340
,
1956
.
22.
Popoff
NA
,
Malinin
TI
,
Rosomoff
HL
Fine structure of intracranial hemangiopericytoma and angiomatous meningioma
.
Cancer
 
34
:
1187
1197
,
1974
.
23.
Reynold
ES
The use of lead citrate at high pH as an electron-opaque stain in electron microscopy
.
J Cell Biol
 
17
:
208
212
,
1963
.
24.
Rhodin
JA
Ultrastructure of mammalian venous capillaries, venules and small collecting veins
.
J Ultrastruct Res
 
25
:
452
500
,
1968
.
25.
Rothbart
D
,
Awad
IA
,
Lee
J
,
Kim
J
,
Harbaugh
R
,
Criscuolo
GR
Expression of angiogenic factors and structural proteins in central nervous system vascular malformations
.
Neurosurgery
 
38
:
915
925
,
1996
.
26.
Sato
S
,
Kodama
N
,
Sasaki
T
,
Matsumoto
M
,
Ishikawa
T
Perinidal dilated capillary networks in cerebral arteriovenous malformations
.
Neurosurgery
 
54
:
163
170
,
2004
.
27.
Sekhon
LH
,
Morgan
MK
,
Spence
I
Normal perfusion pressure breakthrough: The role of capillaries
.
J Neurosurg
 
86
:
519
524
,
1997
.
28.
Senger
DR
,
Galli
SJ
,
Dvorak
AM
,
Perruzzi
CA
,
Harvey
VS
,
Dvorak
HF
Tumor cells secrete a vascular permeability factor that promotes accumulation of ascites fluid
.
Science
 
219
:
983
985
,
1983
.
29.
Sim
DE
The pericyte: A review
.
Tissue Cell
 
18
:
153
174
,
1986
.
30.
Sim
DE
Recent advances in pericyte biology-implications for health and disease
.
Can J Cardiol
 
7
:
431
443
,
1991
.
31.
Stewart
PA
,
Wiley
MJ
Developing nervous tissue induces formation of blood-brain barrier characteristics in invading endothelial cells: A study using quail-chick transplantation chimeras
.
Dev Biol
 
84
:
183
192
,
1981
.
32.
Wagner
DD
,
Saffaripour
S
,
Bonfanti
R
,
Sadler
JE
,
Cramer
EM
,
Chapman
B
,
Mayadas
TN
Induction of specific storage organelles by von Willebrand factor propolypeptide
.
Cell
 
64
:
403
413
,
1991
.
33.
Wong
JH
,
Awad
IA
,
Kim
JH
Ultrastructural pathological features of cerebrovascular malformations: A preliminary report
.
Neurosurgery
 
46
:
1454
1459
,
2000
.

Acknowledgments

This study was supported in part by the National Stroke Foundation, Australia, Australian Brain Foundation, and Ramaciotti Foundations. We thank Mr. Collin Yeo and Mrs. Emma Kettle for their expert technical advice.

COMMENTS

The authors analyzed ultrastructural features of capillaries in perinidal brain parenchyma and found abnormal findings, including the absence of blood-brain barrier (BBB) components. They suggested that insufficient BBB in perinidal dilated capillaries might explain the presence of hemosiderin around the nidus, postoperative brain edema, and recurrence after apparent complete resection of the nidus. Arteriovenous malformation (AVM) surgery always encounters fragile vessels around the nidus that are difficult to coagulate, and it is often necessary to resect the nidus with surrounding brain tissues. The finding that perinidal capillaries exist between normal capillaries and the nidus is also interesting, and we must pay attention to the possibility of incomplete resection of “abnormal vessels” during AVM surgery. The biological response of perinidal capillaries to radiosurgery is another interesting issue to answer. However, we do not know that the absence of normal BBB has any clinical meaning, and further functional and physiological analysis should be performed to clarify clinical relevance of these authors' findings.

Nobuo Hashimoto

Kyoto, Japan

This study presents some novel ultrastructural data regarding perinidal capillaries associated with cerebral AVMs. Although previous ultrastructural studies have investigated the AVM nidus, and light microscopic studies of the perinidal AVM tissue have been reported, no previous studies have focused on the electron microscopic pathomorphology of the perinidal AVM region. The current study demonstrates abnormal perinidal capillary ultrastructure with absence of normal BBB components, including a lack of basement membranes and astrocytic foot processes; endothelial cells with numerous filopodia, cytoplasmic processes, vesicles, filaments, and Weibel-Palade bodies; and pericytes containing vesicles, vacuoles, filaments, and with processes in close contact with endothelial cells. Despite the accepted problems with small numbers of specimens and selection bias of tissue examined that is inherent in electron microscopic studies, these current observations are noteworthy.

The authors conclude that the lack of normal BBB components in the perinidal capillaries may contribute to extravasation of red blood cells into the surrounding brain, even in the absence of major hemorrhage. They further suggest that differentiation and proliferation of abnormal endothelial cells in perinidal capillaries may contribute to occasional recurrence of surgically removed AVMs. As suggested, but unproven by the data presented, these interesting hypotheses should be investigated further in future studies, perhaps elucidating mechanisms of minor AVM hemorrhage and recurrence.

Gary K. Steinberg

Stanford, California

Tu et al. have provided very interesting observations regarding the pathoanatomy of perinidal capillaries surrounding brain AVMs. These capillaries exhibit defective BBB structural components and manifest vacuoles, poor tight junctions, and other features of leaky immature vasculature. The images seem convincing, although it is hard to control for selection bias when examining relatively small sample areas from a huge surrounding field. There is a tendency to see the abnormalities we are looking for, unless a strict number of vessels is set out to be sampled, and the criteria for sampling and grading of anatomic features are articulated a priori. With the powerful magnification of electron microscopy, it is a bit like looking at leaves on trees in the Amazon forest from a satellite picture and describing the presence of “yellow leaves” in the overall sample. Blinding observers to vessel type was described, but this is not very much feasible because a trained observer can rapidly tell what they are looking at, and much of the bias occurs when selecting what vessels to examine.

Our group had reported ultrastructural features of cavernous malformation and AVM nidus vessels (4) in comparison with control vasculature, but we did not sample or describe perinidal capillaries of AVMs in that study. Within AVM nidus, there was a paucity of true capillaries, and the smallest nidal vessels did have features of intact BBB, including a mature subendothelial structure and normal interendothelial tight junctions. In contrast, cavernous malformation nidal vessels did exhibit immature and leaky capillary phenotype, similar to that described by Tu et al. in AVM perinidal capillaries.

We concur with the authors' suggestion that these features may predispose to edema and hemorrhage near AVM nidi and reflect another potential substrate for “perfusion breakthrough” in brain adjacent to AVMs. In other research from our laboratory, we demonstrated overexpression of angiogenesis factors (2) and receptors (3) in AVM perinidal white matter, consistent with activation of angiogenesis outside the nidus boundaries. We speculated that this may contribute to the recruitment of new vessels into the nidus (1). We can now also add that these perinidal capillaries exhibit features of immature proliferating capillaries, also consistent with active angiogenesis. Investigators should now be more attentive to the pathobiology of these perinidal vessels, as well as the nidus, in future research.

Issam A. Awad

Evanston, Illinois

1.
Gault J, Sarin H, Awadallah NA, Shenkar R, Awad IA: Pathobiology of human cerebral vascular malformations: Basic mechanisms and clinical relevance. Neurosurgery 55:1–17, 2004.
2.
Rothbart D, Awad IA, Lee J, Kim J, Harbaugh R, Criscuolo G: Expression of angiogenic growth factors and structural proteins in central nervous system vascular malformations. Neurosurgery 38:915–925, 1996.
3.
Uranishi R, Baev NI, Ng P-Y, Kim JH, Awad IA: Expression of endothelial cell angiogenesis receptors in human cerebral vascular malformations. Neurosurgery 48:359–368, 2001.
4.
Wong J, Kim J, Awad IA: Ultrastructural pathology of cerebral vascular malformations: A preliminary report. Neurosurgery 46:1454–1459, 2000.