Subdural compartment in pigA morphologic study with blood and horseradish peroxidase infused subdurally.код для вставкиСкачать
THE ANATOMICAL RECORD 230:22-37 (1991) Subdural Compartment in Pig: A Morphologic Study With Blood and Horseradish Peroxidase Infused Subdurally JAN R. ORLIN, KIRSTEN K. OSEN, AND TORSTEIN HOVIG Division of Experimental Neurosurgery, Institute for Surgical Research (J.R.O.) and Electron Microscope Laboratory, Institute of Pathology (T.H.), The National Hospital and Anatomical Institute, University of Oslo (K.K.O.), Oslo 1 , Norway ABSTRACT The dura-arachnoid junction is examined in normal animals and in animals subjected to subdural infusion of blood immediately prior to death, simulating acute subdural hemorrhages. The Norwegian landrace pig is used a s the experimental animal. Horseradish peroxidase (HRP) has been added to the injected blood to serve as a macromolecular tracer. The material is studied by light and electron microscopy. Special attention is given to the level of the induced subdural cleavage plane, the total distribution of the infused blood, and the natural sites of drainage. The dura-arachnoid junction, identified here as the subdural compartment (the dural border layer of others), consists of a n avascular tissue with flake-like, relatively electron-lucent cells stacked upon each other in several layers with narrow intercellular clefts. Under normal conditions there is no evidence of a so-called “subdural space.” When under the present experimental conditions bleeding takes place into this cellular tissue, it splits without any particular, predestined cleavage plane, although most often close to the fibrous matter of the dura. The bleeding extends throughout the cerebral and spinal parts of the compartment and also along the spinal nerve roots. Contamination of the subarachnoid space occurs only in some cases subjected to large infusions and apparently only a t spinal levels. The HRP diffuses into the dura, but does not traverse the arachnoid barrier layer. The structure of the meninges has been described in numerous studies of animals (Pease and Schultz, 1958; Andres, 1967; Waggener and Beggs, 1967; Klika, 1967; McCabe and Low, 1969; Himango and Low; 1971; Akashi, 1972; Allen and Low, 1975; Fiedler and Drommer, 1976; Allen and DiDio, 1977; Nabeshima et al., 1975; Oda and Nakanishi, 1984) and humans (Key and Retzius, 1876; Anderson, 1969; Lopes and Mair, 1974; Rascol and Izard, 1976; Schachenmayr and Friede, 1978; Alcolado et al., 1988). There is no general agreement, however, regarding definition, ultrastructure, and function of the dura-arachnoid junction, i.e., the meningeal layer referred to by Nabeshima e t al. (1975) a s the “dural border layer,” but tentatively identified here as the “subdural compartment” (SDC). In spite of increasing evidence against the existence of a preformed, so-called “subdural space” under normal conditions (Andres, 1967; Klika, 1967; Waggener and Beggs, 1967; Anderson, 1969; Allen and Low, 1975; Nabeshima et al., 1975; Rascol and Izard, 1976; Schachenmayr and Friede, 1978; reviews by Haines, 1991; Haines and Frederickson, 19911, it is a fact that a space easily occurs a t the level of the SDC, either as a dissection artifact or as a result of pathological processes such a s subdural bleeding. There are, however, no experimental data available concerning the fine structure of the meninges immediately following such bleedings, and conflicting opinions exist a s to the exact 0 1991 WILEY-LISS, INC. level of the cleavage plane. The potentials for a n initial flow of a subdural bleeding across the midline, a s indicated recently by magnetic resonance imaging (Makiyama et al., 1985; Polman et al., 19861, and to spinal levels also need further anatomical investigation. Experimental evidence for natural sites of drainage of subdurally deposited blood is sparse (Mairova, 1965). The cellular tissue forming what we call the SDC has been shown electron microscopically to be directly contiguous with the stratified perineural sheaths of the spinal nerve roots (Andres, 1967; McCabe and Low, 1969; Himango and Low, 1971; Fiedler and Drommer, 1976). Although a flow of cerebrospinal fluid from the subarachnoid space (SAS) along the spinal nerves has been documented (e.g., Klatzo et al., 1964), we have found no paper testing the possibility of a drainage from the SDC along this route. In the present study these issues are investigated by a novel combination of methods. These include light microscopy and transmission and scanning electron microscopy of the meninges under normal conditions and immediately following subdural blood infusions over one hemisphere. The infusions were performed so Received July 16, 1990; accepted September 21, 1990. Address reprint requests to Dr. J a n R. Orlin, Knapstadveien 12, N-1823 Knapstad, Norway. SUBDURAL COMPARTMENT I N PIG 23 a s to simulate acute venous and arterial subdural feeding tube No. 8 (inner diameter 1.6 mm) bent into bleedings without concomitant brain contusion. In the 90” angle 2.5 cm from its tip. Under the dura a narrow first case, horseradish peroxidase (HRP) was added to space appeared with a small amount of clear fluid the injected blood as a tracer. forming a meniscus. The tube was inserted into this Previous animal experiments on subdural bleedings space with the tip pointing rostrally, and glued to the have been performed in the cat (Cohn, 1948; Mairova, dura to prevent leakage. 1965), the dog (Gardner, 1932; Watanabe et al., 1972), and the monkey (Watanabe et al., 1972). We chose the Test injections Four pigs were then turned to the supine position Norwegian landrace pig, which has also been used in a joint study (Orlin and Zwetnow, 1989) concerning the and prepared for perfusion by thoracotomy. Blood was pathophysiological effects of subdural bleedings. The withdrawn from the femoral vein and mixed with hepig has become a common experimental animal in med- parine (130 i.u. per ml blood) and HRP (Sigma type 11, ical research. Although a literature search, covering final concentration 0.1%).Between 7 and 15 ml of the the last two decades, gave no examples of its use in mixture was injected through the subdural tube in 6-8 neurosurgical research, we have found the pig partic- min. In order to avoid reactive tissue changes and ularly useful for our purpose because of its size. It is transport of HRP by phagocytosis (Nabeshima and appropriate also from a technical, economical, and eth- Reese, 19721, the animals were perfusion-fixed, as ical point of view. An ultrastructural examination of above, immediately after the infusion. The brain and normal pigs, however, proved necessary because of the the spinal cord with surrounding meninges and about 8 shortage of reports on the structure of the meninges in mm of the spinal nerve roots, including the spinal ganthis species (Fiedler and Drommer, 1976; Klika and glia, were removed en bloc and kept in the same fixaRichter, 1989). Fifteen years after Nabeshima et al.’s tive for 2 hr, and overnight in 0.1 M phosphate buffer, (1975) comprehensive electron microscopic study of the pH 7.4. The meninges usually remained intact except meninges in smaller mammals, we take advantage of a t the base of the skull where the dura and the cranial the “historical gain” and use their summary diagram nerve roots always ruptured during dissection. a s a starting point in our results. Blocks of the meninges and subjacent nervous tissue were taken from the convexity of the hemispheres, cauMATERIALS AND METHODS dal to the injection tube, and from the spinal cord, cut Seven pigs with body weight 17-20 kg were anesthe- in 80 pm slices on a vibratome or in 20 pm sections on tized by intraperitoneal injection of pentobarbital (20- a cryomicrotome, and reacted with diaminobenzidine 30 mg per kg), tracheotomized, artificially ventilated, (DAB)/H2O2(Graham and Karnovsky, 1966). The cryosections were counterstained with thionine and covand subjected to the procedures described below. erslipped for light microscopy. The vibratome slices Normal Specimens were osmicated and prepared for transmission electron Two pigs with intact meninges were perfused tran- microscopy. Because of poor penetration of the DAB, a scardially with 1.0 liter phosphate-buffered saline, pH positive HRP reaction was found only in the most su7.4, followed by 4.5 liter of glutaraldehyde 2.5%, perficial ultrathin sections. During blocking, the meninges tended to separate in paraformaldehyde 1%, and calcium chloride 0.003% in 0.1 M phosphate buffer, pH 7.4. The skull was opened the blood-containing space. Segments of the deep and and blocks of the meninges over the lateral part of the outer lining of the space were sampled for scanning hemispheres with the subjacent cortex were removed electron microscopy. for electron microscopy before further dissection. Blocks for transmission electron microscopy were post- Arterial shunt bleedings In one pig regular arterial blood was shunted difixed for 2 hr, treated with 1% osmium tetroxide in 0.1 M phosphate buffer, pH 7.4, for 1 hr, dehydrated in rectly from the abdominal aorta and through the subethanol and propyleneoxide, and embedded in Araldite dural tube with the animal in prone position. About 50 (TAAB). Semithin sections were stained with 1% tolu- ml of blood passed through the shunt before the animal idine blue in 1% sodium borate. Ultrathin sections died after 15 min. The brain and the spinal cord with were contrasted with uranyl acetate and lead citrate. the meninges were removed en bloc, as described In blocks used for scanning electron microscopy the above, and immersion-fixed in 10% formalin. The dismeninges were split mechanically in the SDC imm.edi- tribution of the injected blood was examined by inspecately after the perfusion, whereby the leptomeninges tion and stereomicroscopy, a s well as by light microsbecame curled up like a carpet from the dura. The tis- copy of 20 pm cryostat sections of the spinal cord. sue was postfixed and osmicated a s above, dried in a Before thionin staining, the sections were treated with critical point dryer system, and sputtered with gold- DAB, as above, for visualisation of the erythrocytes which react positively because of endogenous peroxipalladium. dase. The meninges of 21 subdurally injected pigs from Subdural Infusions the pathophysiological experiments (Orlin and ZwetThese were performed either a s “test injections” of a now, 1989) were examined in the same way. limited volume of HRP-containing blood or a s “arterial shunt bleedings” leading to death. In both instances a Number of samples Optimal fixation proved to be a problem, particularly burr hole with a diameter of 15 mm was made over the frontoparietal region of the left hemisphere. The dura after subdural injections which tended to impair the was carefully lifted with a hook and a small hole was perfusion. The present figures were selected from 75 made, big enough to pass through it a Sherwood argyle electron micrographs from four samples of normal ma- 24 J.R. ORLIN ET AL. with finely dispersed chromatin. The cytoplasm is rich in organelles, particularly mitochondria, free ribosomes, and short profiles of rough endoplasmic reticulum (rER) (Figs. 2A, 3). Infoldings of the cell membrane, probably endo- or exocytotic pits, are seen adjacent to the fibril-containing spaces (Fig. 3). RESULTS The subdural compartment, or “dural border layer,” Normal Ultrastructure varies in thickness, but is generally about 5 pm thick. The structure of the meninges in normal pig con- It is composed of flattened, flake-like cells, referred to forms, to a large extent, with Nabeshima et al.’s (1975) by us as “subdural cells” (after Akashi, 1972), stacked electron microscopic study in rodents, cat, and monkey. upon each other in several layers parallel to the dura. As indicated in Figure 1, we have adopted their strat- The extranuclear portion of each cell is about 1 pm ification and terminology with a few modifications: We thick, and there are more cell layers (varying from a introduce the term “subdural compartment” for their few up to 10-15) than expected from the number of “dural border layer” a s a modification of the widely nuclei in semithin sections (compare Figs. 1B and 2A). used term “subdural space,” to denote a soft tissue, but In electron micrographs the individual cells may be not necessarily a cavity, enclosed between tougher traced over distances more than 50 pm (Fig. 2A). The membranes. We emphasize the SDC as a structure dis- nucleus is usually flattened with finely dispersed chrotinct from the dura in order to make the term rational matin as in the fibrocytes of the dura. The cytoplasm is and, hopefully, to end the fruitless discussion of more electron-lucent and contains fewer organelles whether a process in this portion of the meninges than any other meningeal cells: scattered microfilashould be named intradural or subdural (for reviews, ments, a few mitochondria, and short profiles of rER. see Leary, 1939, Loew and Kivelitz, 1976; Haines, The intercellular clefts are of about the same width 1991). Furthermore, we introduce the term “arachnoid (15-40 nm) as between the dural fibrocytes, but there reticular layer” for the collagen-containing portion of are more dilatations filled with flocculent material the arachnoid situated between the arachnoid barrier (Figs. 2-4). Scattered adhering (desmosome-like)junclayer and the SAS, to stress its existence and possible tions may be seen (Fig. 31, but there is no basement role a s a mechanical reinforcement of the arachnoid. membrane and no fibrillar intercellular material reThe third portion of the arachnoid, i.e., the arachnoid lated to these cells. The cell borders are almost rectitrabeculae, and the pia are not further commented on linear. The innermost cells, however, have a more irregular deep surface accommodating the arachnoid here. Four meningeal layers are according to our defini- barrier layer. The corresponding intercellular cleft aption situated outside the SAS: the arachnoid reticular pears largely similar to the clefts between adjacent layer, the arachnoid barrier layer, the SDC, and the subdural cells, but shorter stretches are narrower and dura (Figs. 1-4). Although the cells of the four layers possess more numerous adhering junctions (Fig. 3). are ultrastructurally distinct, the classification of in- Under these conditions a “subdural space” is definitely dividual elements in the border zones may be difficult. lacking. The three first-mentioned layers are composed of avasThe arachnoid barrier layer is about 5 pm thick. The cular tissues, only traversed by the bridging veins near cells are less flattened than the subdural cells and form the dural sinuses. The dura, on the contrary, contains both blood vessels and lymphatics. The three layers therefore must be nourished (or perfusion-fixed) either via the cerebrospinal fluid or from the dura. In the Abbreviations following, the normal ultrastructure of each of the four AB arachnoid barrier cell layers is described separately, starting with the dura. ABL arachnoid barrier layer arachnoid reticular cell Only cerebral levels are illustrated in this context, but AR arachnoid reticular layer from previous studies and our own observations only ARL asa anterior spinal artery minor modifications are to be expected at spinal levels. c o collagen dural fibrocyte Measurements are not corrected for shrinkage during D E elastin preparation. epineurium The dura of pig is about 100 km thick over the hemi- EP Er erythrocyte spheres, thicker (up to about 300 km) toward the mid- F intercellular flocculent material Golgi apparatus line. It consists of a dense connective tissue (Fig. 1B). Go horseradish peroxidase As already shown by Key and Retzius (1876) by micro- HRP lacuna dissection of human and animal material, the dural LN nucleus fibrocytes are flattened parallel to the surface of the P1 plasma perineurium dura and have a few thin processes. Ultrastructurally Pn rootlet they are about 1 km thick and alternate with fibrillar r rER rough endoplasmic reticulum intercellular material composed of relatively thick col- SAS subarachnoid space lagen fibrils (about 100 nm diameter), microfibrils, and SD subdural cell subdural compartment elastin. In the deep part of the dura where the cells SDC subdural hematoma occur stacked upon each other, the intercellular clefts SDH p.c spinal cord are electron-lucent and about 15-40 nm wide with few sSp.G spinal ganglion if any junctions, but with occasional dilatations con- Sp.NR spinal nerve root vessel taining a flocculent material. The nucleus is flattened V terial and 200 electron micrographs from 12 samples of test-injected cases. The light micrographs are representative for about 100 tissue sections, including semithin sections of numerous tissue blocks originally prepared for electron microscopy. SUBDURAL COMPARTMENT IN PIG 25 A DURA SAS A and P \ Fig. 1. A slightly modified version of Nabeshima et al.’s (1975, P1. 17, reproduced by courtesy of the authors) diagram of the mammalian cerebral meninges in A is compared with a light photomicrographic montage of a corresponding field of a toluidine blue-stained semithin section from a normal pig in B. Corresponding borderlines in A and B are connected by straight lines. The boldface letters between the lines indicate the presently used terminology (see list of abbreviations), slightly modified after that of Nabeshima et al. Our SDC is distinct from the dura and corresponds to their dural border layer (Db). Our ABL equals their arachnoid barrier layer (B). Our ARL corresponds to the collagen-reinforced, sub-barrier part of the arachnoid, not specially named by them. In A the curved arrow, intended by Nabeshima et al. to illustrate the artificial, dissection-induced “subdural space” (SDS), also indicates the intended site of blood infusion in the present study (which usually proved to be closer to the dura). Cell junctions: desmosomes (d), hemidesmosomes (h), tight junctions (t), and gap junctions (g). In B small arrows indicate dural fibrocytes, arrowheads the dark intercellular line of the ABL (also illustrated in A), medium arrows the deep lining of the ARL, open arrow perhaps a migrating cell. 26 J.R. ORLIN ET AL. Fig. 2. Electron micrographs of the cerebral meninges of a normal pig. A shows part of the dura with dural fibrocytes (D), the SDC (between large arrows) with subdural cells, the layer of arachnoid barrier cells, and part of the arachnoid reticular layer with arachnoid reticular cells and lacunae. The thicker collagen fibrils of the dura are indicated by small arrows, the thinner of the arachnoid by double arrows. Each SD-cell spans almost the entire field of the picture (SD, labeled by asterisks). The same is the case for the deepest D-cell (D,). Thus, both SD- and D-cells seem more sheet-like than AB-cells that are thicker with thin, interdigitating cell processes (isolated profiles indicated by dots). Of the cell types present, the SD-cells are most electron-lucent, the AR-cells most electron-dense. The narrow, electron-dense intercellular clefts between AB-cells are indicated by arrowheads, a possible migrating cell by open arrow. B shows part of an AB-cell connected by a triple-adhering junction (curved arrow) to adjacent AR-cells. SUBDURAL COMPARTMENT IN PIG Fig. 3. Electron micrograph with details from Figure 2, showing the innermost part of the dura (top) with D-cells, the SDC (between large arrows) with SD-cells, arachnoid barrier layer with AB-cells, and the outermost part of the arachnoid reticular layer with AR-cells (bottom). The dura contains collagen fibrils, microfibrils (small arrows), and elastin. The electron-lucent intercellular clefts between the Dcells and between the SD-cells show dilatations with a flocculent material. The large vesicles with flocculent material in SD1 may repre- 27 sent infoldings of the cell membrane. SD1 abuts upon the subjacent AB-cells and has a more irregular deep surface than the other SDcells. The narrow intercellular clefts of the interdigitating AB-cells show several adhering junctions (curved arrows) and a possible gap junction (double-crossed arrow). The AR-cells are darker and have larger mitochondria than the other cells. The lacunae (L) contain thin collagen fibrils and microfibrils (small arrows). Hemidesmosomes are attached to remnants of a basement membrane (arrowheads). 28 J.R. ORLIN ET AL. Fig. 4. Electron micrograph from the cerebral arachnoid barrier layer flanked by SD-cells to the left and AR-cells to the right. The AB-cell in the center of the field shows a Golgi apparatus, short profiles of rER, a few microtubules (small arrows), and microfilaments. The wide intercellular cleft (indicated by large arrows) toward the adjacent subdural cell (SD1) shows a dilatation with a flocculent ma- terial (F). The SD-cells show scattered microfilaments in their electron-lucent cytoplasm. The intercellular clefts (arrowheads) between AB-cells show adhering junctions (curved arrows) and tight junctions (crossed arrow). The arachnoid lacunae (L) contain collagen and microfibrils. 29 SUBDURAL COMPARTMENT IN PIG a n epithelium-like tissue with only one row of nuclei, but with 1-4 layers of highly interdigitating cell processes (Figs. 2-4). The intercellular clefts are narrower (about 10 n m wide) and filled with a n electron-dense material. The cells are interconnected by adhering junctions (Figs. 3,4,curved arrows), occluding (tight) junctions (Fig. 4, crossed arrow), and communication (gap) junctions (Fig. 3, double-crossed arrow), a s described by Nabeshima et al. (1975; see our Fig. 1A). The nucleus is oval with coarser chromatin than in the subdural cells. The cytoplasm is slightly darker with more organelles, notably rER, free ribosomes, scattered dark bodies, and groups of small vesicles, probably Golgi apparatuses. The cells also contain a few microtubules and many filaments, some evidently being tonofilaments related to the adhering junctions (Fig. 4).Vesicles and surface pits speak in favor of endo- or exocytotic activities. The arachnoid reticular layer, as defined here, is about 10-20 pm thick and composed of irregularly branched cells which form a loose network with intercellular “lacunae” (“cisterns” in Nabeshima et al.’s terminology) of irregular shape and size. These lacunae contain microfibrils and bundles of thin collagen fibrils (diameter about 50 nm) oriented in a crisscross pattern parallel to the brain surface (Figs. 2, 3). The lacunae appear partly shut off from the SAS by sheet-like cell processes (Fig. 1B). Between the lacunae, the interdigitating cell processes are separated by a regular, 20 nm wide electron-lucent space (Fig. 2B). The cell nucleus is irregularly shaped with relatively coarse chromatin. The cytoplasm is more electron-dense with numerous microfilaments (Fig. 4),and the mitochondria appear larger than in the barrier cells (Fig. 2A). Scattered rER, quite a few polyribosomes, and possible endo- or exocytotic vesicles are usually present (Fig. 3). Adhering junctions are scattered both between the arachnoid reticular cells and between these and the barrier cells (Fig. 2B). Hemidesmosomes occur toward the lacunae, and near the barrier layer short pieces of a basement membrane adjoin both kinds of cell (Fig. 3). When only parts of cells are present in the micrographs, a distinction between arachnoid barrier and reticular cells is not always possible (Fig. 4). infiltrated by erythrocytes (Fig. 6B), and a blood-containing space, probably continuous with that of the SDC, was found within the perineural sheaths (Fig. 6C). Occasionally the endoneurium was also infiltrated. Because the dura ruptured a t the base of the skull during the dissection, the basal part of the cerebral SAS and the cranial nerve roots could not be examined. Test Injections Arterial Shunt Bleedings After injections of a smaller amount of HRPcontaining, heparinized venous blood, the distribution of the blood was largely the same a s after the shunt bleedings although the diastasis of the dura and arachnoid was smaller (Fig. 7). There was no contamination of the SAS, not even at spinal levels (Fig. 7C,D). At the outer lining of the blood-containing space, blood cells appeared either in direct contact with the collagen fibrils and small vessels of the dura or separated from these structures by some flat cells that could be fibrocytes and/or subdural cells (Fig. 7B,E). The blood corpuscles could also be seen squeezed in between such cells (Fig. 7E). At the inner lining of the space, erythrocytes and platelet aggregates were sometimes found close to, and perhaps in contact with, the arachnoid barrier layer (not illustrated). More often the blood cells were separated from the arachnoid by several layers of subdural cells (Fig. 8). These appeared thinner and more irregularly shaped than normally and had increased electron density of the cytoplasm. The intercellular spaces were widened with a flocculent content, some of which undoubtedly represents plasma proteins (Fig. 8B). HRP reaction product occurred throughout the dura, but not in the arachnoid or subjacent structures (Fig. 7A,D). In superficial ultrathin sections from the DABreacted slices, the HRP appeared to have diffused with the plasma in between the subdural cells and a certain distance in between the arachnoid barrier cells, but not deep to the latter (Fig. 8A). As after the shunt bleedings, erythrocyte infiltration occurred in the epineurium and perineurium of the spinal nerve roots. In the latter location rows of erythrocytes appeared squeezed in between the perineural lamellae (Fig. 6D). Shunt bleedings from the abdominal aorta led to flow of blood bilaterally throughout the entire SDC, spinal as well as cerebral. On dissection of the fixed specimen a cap-like subdural blood clot was found over both cerebral hemispheres. The cisterna magna, itself free of blood, was compressed by a thick dorsal “collar” of blood (Fig. 5A). The spinal cord was surrounded by blood all the way down to the dural sac (Fig. 5B). Usually more blood was located dorsally than ventrally (Fig. 6B), which cannot be explained by gravitation because of the animal’s prone position during the infusion. Blood often filled the “pocket” of the SDC between the dorsal and ventral roots (Fig. 6B). The SAS was clearly compressed by the distended SDC. Blood was occasionally present in the lateral part of the spinal SAS, among the emerging rootlets, despite the absence of blood in the cisterna magna (Figs. 5B, 6A). On gross inspection the spinal nerve roots were discolored by blood (Fig. 5B). In sections the epineurium was The tendency for the SDC to split close to the dura is supported also by scanning electron microscopy of the cleaved SDC in the normal pig and in test-injected specimens. In normal animals the deep surface of the cleavage space, i.e., on the arachnoid side, was covered by a continuous, smooth cell layer that could represent subdural cells and/or arachnoid barrier cells (Fig. 9A). The fibrillar network of the arachnoid reticular layer could be seen through ruptures in the surface lining. The outer surface of the space, i.e., on the dural side, showed coarse fibers only partially covered by sheetlike cells that could be dural fibrocytes and/or subdural cells (Fig. 9B). In the injected specimens erythrocytes occurred on both surfaces of the blood-containing space. In areas not completely concealed by erythrocytes, the deep surface appeared covered by a continuous layer of flat cells, as in normal cases, but with the nuclear sites Scanning Electron Microscopy 30 J.R. ORLIN ET AL. Fig. 5. Color photographs from the central nervous system of a pig with arterial shunt bleeding. A shows the empty cisterna magna in a dorsolateral view, covered by a partly removed subdural hematoma. The originally infolded arachnoid is artificially stretched. B shows the cranial surface of a slice of the cervical spinal cord with the nerve roots and spinal ganglia. The pia is marked by black arrows, the arachnoid by white arrows. Blood is present in the SDC as well as in between the rootlets in the lateral part of the SAS, and in the roots. There is no blood around the anterior spinal artery. protruding slightly above the surroundings (Fig. 9C, compare with 9A). The latter feature may correspond to the shrinkage of the cytoplasm observed by transmission electron microscopy in such cases (Fig. $). The outer surface also had a n appearance corresponding to the normal pattern, with erythrocytes adhering either to the naked collagen fibrils of the dura or to a cover of flat cells (Fig. 9D). we call the SDC, conforms with the general mammalian pattern. By subdural infusions the blood immediately spreads throughout the SDC, including spinal levels. In contrast to serous cavities, the cleavage plane of the SDC is not predestined. The stacks of flattened cells split randomly, although most often close to the fibrous matter of the dura. The injected blood immediately infiltrates the epineural and perineural sheaths of the nerve roots. Leakage into the SAS appears limited to some cases with arterial shunt bleeding. The two latter findings may be restricted to spinal levels, although reservations have to be made for the base of DISCUSSION According to the present study, the normal ultrastructure of the meninges in pig, including that portion SUBDURAL COMPARTMENT IN PIG 31 Flg. 6. Light photomicrographs of the spinal meninges. A-C are from DAB-reacted, Nissl-stained cryosections of two cases with arterial shunt bleedings. The arachnoid is indicated by arrowheads. The empty spaces in the blood-containing SDC are due partly to sectioning artifacts, partly to contraction of the coagulum. A is from the same case a s Figure 5 with blood both in the distended SDC and the compressed SAS. B and C are from a case with only traces of blood in the SAS (at arrow), but with bleeding into the nerve roots. On the right hand side is discoloring of the epineurium. On the left hand side, enlarged in C, is bleeding into the perineural sheaths (arrows), in one fascicle (asterisk) also into the endoneurium. The lateral pocket of the blood-containing SDC is indicated by a white asterisk, the dentate ligament by an open star. D is from a toluidine blue-stained semithin section of a spinal nerve root fascicle of a test-injected case with erythrocytes (arrows) squeezed in between the membranes of the perineural sheath (some myelinated fibers marked by asterisks). Magn.: A,C x80;B x 6 ; D ~ 7 5 0 . the brain where the meninges and nerves ruptured during dissection. The arachnoid forms a n effective barrier against HRP, while the dura is leaky. as in the meninges of other animals (Pease and Schultz, 1958, rat; Andres, 1967, dog, cat; Waggener and Beggs, 1967, rat, guinea pig; Himango and Low, 1971, rat; Akashi, 1972, rabbit; Nabeshima et al., 1975, mouse, rat, chinchilla, rabbit, cat, monkey; Oda and Nakanishi, 1984, mouse) and humans (Anderson, 1969; Lopes and Mair, 1974; Rascol and Izard, 1976; Schachenmayr and Friede, 1978; Alcolado et al., 1988). It may correspond to the spongy-looking “intermediate leptomeningeal layer” shown by scanning electron microscopy a t spinal levels in humans (Nicholas and Weller, 1988). Normal Features, Comparative Aspects The arachnoid reticular layer, in our terminology, has been described by others (e.g., Alcolado et al., 1988) as the “inner layer of the arachnoid membrane.” In pig it is regularly distinguishable deep to the arachnoid barrier layer of the cerebral meninges. Judged by previously published illustrations i t is present also at spinal levels in pig (Fiedler and Drommer, 1976) a s well 32 J.R. ORLIN ET AL. Fig. 7. Light photomicrographs from the meninges following test injections of heparin- and HRP-containing blood. The arachnoid is marked by arrowheads. A shows a DAB-reacted cryosection of the cerebral meninges with the subjacent cortex. The HRP reaction is restricted to the blood-containing SDC and the dura. B is a toluidine blue-stained semithin section from the outer lining of the SDC showing erythrocytes in direct contact with the collagen fibers of the dura. C and D are from a DAB-reacted cryosection from the spinal meninges showing blood in the SDC and HRP reaction in the dura. The boxed area in C is shown at higher magnification in D. E is from a corresponding part of the meninges as shown in D, but from a toluidine blue-stained semithin section of another animal. Erythrocytes and platelets in the SDC are in part situated between layers of sheetlike cells (arrows) which may be subdural cells and/or dural fibrocytes. Magn.: A,D x 104; B x 1,500;C x 8; E x 750. Although we have done no systematic sampling from the brain surface, the arachnoid reticular layer in our opinion may not merely constitute a layer of compressed arachnoid trabeculae. The arachnoid lacunae, although probably traversed by cerebrospinal fluid, are, a s described also by others (e.g., Nabeshima et al., 1975), partly shut off from the main part of the SAS. The collagen fibrils in the lacunae are oriented parallel to the brain surface (Allen and DiDio, 1977; Alcolado et al., 1988),and obviously serve to strengthen the arachnoid mechanically. As shown by Key and Retzius (1876) by microdissection and Alcolado et al. (1988) by scanning electron microscopy, the fibrils converge toward the arachnoid trabeculae and apparently also serve to anchor the arachnoid to the pia. The arachnoid barrier layer in pig conforms with Nabeshima e t al.’s (1975) description based on transmission electron microscopy and freeze fracture in rodents, cat, and monkey. The typical interdigitation of the cells, the many junctions, and the narrow intercellular clefts filled with electron-dense material have been illustrated in most of the animal studies referred to above, as well a s in human biopsies obtained during surgery (Rascol and Izard, 1976; Alcolado et al., 1988). The dark intercellular clefts appear less conspicuous in Schachenmayr and Friede’s (1978) autopsy material of human. When appearing as a continuous line, many authors tend to misinterprete this a s the border between the barrier layer and the SDC. The many adhering junctions, the cellular microfil- SUBDURAL COMPARTMENT IN PIG Fig. 8. Electron micrographs from DAB-reacted slices of the cerebral meninges following test injections of heparin- and HRP-containing blood. The SDC-ABL border is indicated by large arrows. A is from a superficial section of a slice with strong HRP reaction in the injected plasma (P1+ HRP). Electron-lucent platelets (Tr)are seen between shrunken SD-cells (also marked by asterisks). A thin layer of reaction product (arrowheads) is seen between the AB-cells, but not deep to them. B is from a deep section without HRP reaction products. The 33 SD-cells are more electron-dense than normal, shrunken and irregularly shaped, and widely separated from each other by a flocculent material, some of which is probably plasma. Also in the outer part of the arachnoid barrier layer the intercellular clefts are slightly wider than normal and here the adhering junctions (curved arrows) and occluding junctions (crossed arrow) are clearly seen. A communication junction (double-crossed arrow) is seen between AR-cells. 34 J.R. ORLIN ET AL. Fig. 9. Scanning electron micrographs from the cleaved SDC. A, B From a normal animal. C, D From a subdural test injection. In the latter case the lining of the blood-containing space is represented only by sites incompletely covered by erythrocytes. A and C are from the arachnoid side, B and D from the dural side of the space. A shows a continuous, smooth cell lining with an artificial rupture (arrowheads) exposing the fibrillar network of the subjacent arachnoid reticular layer. B shows the coarse fiber bundles of the dura, partially covered by flattened cells (asterisks). C shows a cellular lining reminiscent of that in A, but with protruding nuclear sites (N) and scattered adherent erythrocytes. D is similar to B apart from the presence of erythrocytes. aments, and the basement membrane toward the subjacent lacunae may serve to strengthen the barrier layer mechanically. The numerous occluding junctions are evidently responsible for the barrier function, as demonstrated by injections of macromolecular tracers into the SAS (Key and Retzius, 1876; Weed, 1923; Nabeshima and Reese, 1972) and the SDC (Key and Retzius, 1876; present study). The function of the electron-dense material in the intercellular clefts remains to the explained. The subdural compartment, as defined here, corresponds to the “dural border layer” of Waggener and Beggs (1967) and Nabeshima et al. (1975). The more common term “subdural space” denotes a pathological condition and therefore is a misnomer in the context of normal anatomy (for reviews, see Haines, 1991; Haines and Frederickson, 1991). Its persistent use may be as- cribed to the topographically important keyword “subdural,” as opposed to “epidural” and “subarachnoid.” Our terminology retains this word and makes it rational by defining the SDC a s a unique structure that is histologically distinct from both the dura and the arachnoid. The term SDC simply replaces “space” with “compartment,” which goes with many languages and is used also elsewhere for soft tissues enveloped by membranes, e.g., “muscle compartments.” Although there seems to be general agreement regarding the fundamental cellular composition of this portion of the meninges, opinions differ with respect to the electron density of the cells and the size of the intercellular clefts. As discussed below, the discrepancies are probably due to difficulties with the fixation of this avascular, not easily accessible tissue, rather than to species characteristics. SUBDURAL COMPARTMENT IN PIG In well-preserved, perfusion-fixed material, obtainable only in experimental animals, the subdural cells appear densely packed and more electron-lucent than the arachnoid cells (Fiedler and Drommer, 1976; Andres, 1967; Himango and Low, 1971; Oda and Nakanishi, 1984; present study). They are darker in animals perfused with high concentrations of glutaraldehyde (Waggener and Beggs, 1967,6.2%) known to cause tissue shrinkage. In Nabeshima et al.’s (1975) material the “dural border cells” were darker than the barrier cells following glutaraldehyde fixation (concentration not given), while with osmium as first fixative no such difference was found. They invariably appear thin and electron-dense with wide intercellular clefts in necropsy material (Schachenmayr and Friede, 1978) and biopsies (Lopes and Mair, 1974; Rascol and Izard, 1976; Alcolado et al., 1988) of human, as well a s in sections from improperly fixed animals (Pease and Schultz, 1958). I n pig we found both varieties: electronlucent subdural cells with narrow clefts in normal animals, and shrunken, electron-dense cells with wide clefts in infused cases. As also emphasized by Andres (1967), the cells in question appear to vary in fluid content during anoxic, osmotic, and mechanical stress. The shrunken and electron-dense subdural cells observed under such circumstances are reminiscent of the “dark neurons” following manipulation of improperly fixed brain tissue (Cammermeyer, 1962). The classification of the subdural cells is unclear. The name “subdural cell” was introduced by Akashi (1972) who in contrast to us, but in agreement with general practice as mentioned above, included in this term also the arachnoid barrier cells outside the dark intercellular cleft. Key and Retzius (1876) and Mallory (1920) described the cells as a “subdural endothelium” lining the “subdural space.” More common terms are “subdural neurothelium” (Andres, 1967; Rascol and Izard, 19761, “subdural mesothelium” (Pease and Schultz, 1958; Tripathi, 1973; Alcolado et al., 19881, and “dural border cells” (Waggener and Beggs, 1967; Nabeshima et al., 1975; Schachenmayr and Friede, 1978; Yamashima and Yamamoto, 1985). The terms endo-, neuro-, and mesothelium point to certain epithelial features of the subdural cells, i.e., the lack of fibrillary intercellular substance and the presence of adhesion junctions. The cells are, however, less densely packed and show fewer junctions than usual for epithelia. The layer also lacks a basement membrane toward the connective tissue of the dura. According to Nabeshima et al. (19751, the subdural cells (their “dural border cells”) are not significantly different from the dural fibroblasts. Also in the present material these two cell types have many features in common, but in our opinion the differences in electron density and amount of organelles justify a distinction. In support of the fibroblast nature of the subdural cells, it has been claimed that these cells contribute to the formation of the neomembranes encapsulating chronic subdural hematoma (Friede and Schachenmayr, 1978; Yamashima and Yamamoto, 1985; for a review, see Haines, 1991). Since the SDC and the arachnoid barrier layer are both avascular, the numerous blood vessels of the neomembranes (Voris, 1946; Watanabe et al., 1972; Apfelbaum et al., 1974; Loew and Kivelitz, 1976) must originate from the dura. Fi- 35 brocytes from this source could grow into the neomembranes along with the vessels. In a situation with reactive changes, the subdural cells and dural fibrocytes may be even more difficult to distinguish than under normal conditions. The subdural cells have also been considered a subtype of arachnoid cells (Klika, 1967; Alcolado et al., 1988). Schachenmayr and Friede (1978) proposed to pool them with the arachnoid barrier cells into one common “interface layer.” In our opinion that would represent a step backward in the search for a specific role of the various cell types in meningeal function. Distribution of the Infused Blood The free flow of blood throughout the SDC in the pig, although unexpected, may not be unique for this species. The possibility of a flow across the midline in human is supported by the frequent occurrence of bilateral subdural hematomas, 15-50% in adults (Kaump and Love, 1938; Leary, 1939)and up to 78% in infants (Gutkelch, 1971), and the successful treatment of such cases by unilateral subdural-peritoneal shunting (Aoki et al., 1985; Aoki and Masuzawa, 1988). Recent investigations with magnetic resonance imaging have also demonstrated a bilateral dissipation of acute unilateral bleeding (Makiyama et al., 1985; Polman et al., 1986). A volume competition obviously exists between the SDC and the SAS within the rigid framework of the dura. Since the spinal compartment of the SAS may accommodate up to 70% of fluid injected into the cisterna magna (Lofgren and Zwetnow, 1973), i t is reasonable that most of the blood infused subdurally tends to accumulate outside the cisterna magna and at spinal levels. As far a s we know, involvement of the spinal meninges has not been reported following subdural infusions in animals (Cohn, 1948; Gardner, 1932, Watanabe e t al., 1972) or in connection with subdural hematomas in humans. A negative evidence in this regard, however, may not be conclusive because spinal levels are seldom examined. On the other hand, the larger brain surface, the deeper falx cerebri (about 5 cm against 0.5 cm in the pig), and the connective tissue strands between the cervical dura and arachnoid (Key and Retzius, 1876) in human could perhaps contribute to a more restricted distribution of subdural hematoma. The different modes of bleeding, as well as possible concomitant brain contusion, are other decisive factors. The Subdural Cleavage Plane Although a “subdural space” does not exist normally (Andres, 1967; Klika, 1967; Waggener and Beggs, 1967; Anderson, 1969; Allen and Low, 1975; Nabeshima et al., 1975; Rascol and Izard, 1976; Schachenmayr and Friede, 1978; Haines, 1991; Haines and Frederickson, 1991; present study), i t is a fact that a cleavage of the SDC easily occurs either a s a dissection artifact or a s a result of pathological processes. The tendency for the meninges to split in the SDC was also directly demonstrated in the present study during the insertion of the infusion cannula. The small amount of fluid encountered in the artificial “subdural space” under the lifted and perforated dura may have been generated by leakage from the subdural cells or by pressure-generated 36 J.R. ORLIN ET AL. a role in shearing (sliding) movements between the brain and the skull. Judged by shape and orientation, the subdural cells should be able to slide sideward, both with respect to each other and to the adjacent meninges, especially the dura. The need for such movements may become particularly urgent a t the hemispheres where the arachnoid is tightly anchored to the pia over the gyri (Key and Retzius, 1876; Allen and Low, 1975). That such movements really occur under extreme conditions has been demonstrated in monkeys whose calvarium had been replaced by a transparent material (Pudenz and Shelden, 1946). Subdural bleedings may originate from dural capillaries, bridging veins, and arteries of adjacent tissues (break through bleedings). Rupture of the bridging veins may be caused by shearing movements in the SDC during acceleration or deceleration traumas, even without direct impact to the head (Gennarelli and Thibault, 19821, for example, subdural hematoma in infants caused by violent shaking (Gutkelch, 1971; Aoki and Masuzawa, 1986). If the SDC is expanded due to cerebral atrophy and the veins are stretched in adDrainage of the SDC vance (Doherty, 19881, shearing movements may be The diffusion of erythrocytes into the sheaths of the assumed to cause subdural bleeding even after otherspinal nerve roots may be explained by the normal ul- wise negligible traumas. trastructure of the transitional zone between the nerve Questions might be raised about whether the locaroot sheaths and the meninges (Andres, 1967; McCabe tion of the cannula tip during infusions corresponds to and Low, 1969; Himango and Low, 1971; Fiedler and the sites at which bleeding occurs as a result of trauma. Drommer, 1976). The root epineurium coalesces with This uncertainty is inherent in the experimental dethe dura while the perineurium splits, the outer layers sign used. However, the correspondence between the merge with the SDC and the inner layer with the experimental results and the pathology of traumatic sheath of the rootlets. Evidently the blood dissects its subdural hemorrhages suggests that the experimental way along these routes. In live individuals perhaps this data have clear relevance to this common neurosurgical problem. could cause root symptoms. The explanation for the contamination of the SAS in ACKNOWLEDGMENTS some cases of shunt bleeding is unclear. Since both the SDC and the SAS (Klatzo et al., 1964) drain along the The authors wish to thank K.M. Gujord, J.J. Knutspinal nerve roots, erythrocytes originating from the sen, T.A. Slyngstad, and R. Odegbrd for technical asSDC could perhaps pass centrally into the SAS along sistance, L. Schjerven and his staff for animal care, E. the root sheaths. The absence of a typical arachnoid Buntz and her staff for library service, C. Ingebrigtsen, barrier layer in the subarachnoid angle, where the G.F. Lothe, B. Onstad, and E. Aarseth for phototechniarachnoid reflects from the deep aspect of the dura onto cal work, K. Skullerud and N.N. Zwetnow for advice on the rootlets (Andres, 1967; McCabe and Low, 1969; Hi- the project design, and O.P. Ottersen and D.E. Haines mango and Low, 1971; Fiedler and Drommer, 1976), for constructive criticism. Finally we want to thank R. could perhaps also allow a migration of erythrocytes Raaken for typing the manuscript. directly from the SDC into the SAS. Mairova (1965) LITERATURE CITED injected small volumes (0.2-0.8 ml) of nucleated hen erythrocytes subdurally in cat and found uptake in the Akashi, Y. 1972 Electron microscope studies on the fine structure of the arachnoid membrane on the base of the rabbit brain. Acta dural vessels within a few hours, but, as in our small Anat. Nippon, 47r285-297. test injections, the cells did not invade the SAS. In view Alcolado, R., R.O. Weller, E.P. Parrish, and D. Garrod 1988 The craof the virtual absence of contamination of the SAS also nial arachnoid and pia mater in man: anatomical and ultrastrucin some of our animals with shunt bleedings, occatural observations. Neuropathol. Appl. Neurobiol., 14tl-17. sional rupture of the arachnoid during dissection has to Allen, D.J., and L.J.A. DiDio 1977 Scanning and transmission electron microscopy of the encephalic meninges in dogs. J. Submibe considered. crosc. Cytol., 9r1-22. flux from nearby intercellular spaces. The normal intracranial pressure in the pig is 7.5 mm Hg above atmospheric pressure (unpublished observation) as in dog (Lofgren et al., 1973). Although the level of the cleavage plane seems random, most authors (Pease and Schultz, 1958; Rascol and Izard, 1976; Allen and DiDio, 1977; Klika and Richter, 1989) agree that it tends to be located adjacent to the fibrillar matrix of the dura, which may be explained by the scarcity, or virtual absence, of adhering junctions between the outermost subdural cells and the dural fibrocytes. Others (Waggener and Beggs, 1967; Nabeshima et al., 1975; Schachenmayr and Friede, 1978) localize the cleavage within the “dural border layer,” our SDC, which is also likely from the present findings. A cleavage between the innermost subdural cells and the arachnoid barrier layer, only seldom observed by us, may be counteracted by the undulating cellular borders and more numerous adhering junctions (see, however, Nabeshima et al., 1975) at this level. Functional Considerations The absorption of intracranial shock probably takes place mainly in the SAS (Weed, 1935). Judged by its thickness, a contribution to this effect by the normal SDC (9 p m in the dog, Andres, 1967; and human, Rascol and Izard, 1976; 5 pm in the pig) may be negligible. The absence of nutritive vessels and the tendency to cleave by collection of intercellular fluid make the SDC particularly well suited as a n intracranial space compensator by brain shrinkage. I t could perhaps also play Allen, D.J., and F.N. Low 1975 Scanning electron microscopy of the subarachnoid space in the dog. J. Comp. Neurol., 161r515-540. Anderson, D.R. 1969 Ultrastructure of meningeal sheaths. Arch. Ophtalmol., 82t659-674. Andres, K.H. 1967 Uber die Feinstruktur der Arachnoidea und Dura mater von Mammalia. Z. Zellforsch., 79r272-295. Aoki, N., and H. Masuzawa 1986 Subdural hematomas in abused children: report of six cases from Japan. Neurosurgery, 18t475477. Aoki, N., and H. Masuzawa 1988 Bilateral chronic subdural hematomas without communication between the hematoma cavities: treatment with unilateral subdural-peritoneal shunt. Neurosurgery, 22t911-913. SUBDURAL COMPARTMENT IN PIG Aoki, N., H. Mizutani, and H. Masuzawa 1985 Unilateral subduralperitoneal shunting for bilateral chronic subdural hematomas in infancy. J . Neurosurg., 63:134-137. Apfelbaum, R.I., M.C. Gutkelch, and K. Shulman 1974 Experimental production of subdural hematomas. J. Neurosurg., 4Or336-346. Cammermeyer, J . 1962 An evaluation of the significance of the “ d a r k neuron. Ergeb. Anat. Entw. Gesch., 36:l-61. Cohn, R. 1948 Subdural hematoma. An experimental study. Arch. Neurol. Psychiatr., 59r360-367. Doherty, D.L. 1988 Posttraumatic cerebral atrophy as a risk factor for delayed acute subdural hemorrhage. Arch. Phys. Med. Rehabil., 69542-544. Fiedler, H.-H., and W. Drommer 1976 Licht- und elektronen-optische Untersuchungen an der Membrana limitans gliae superficialis und der Leptomeninx im Ruckenmark des Schweines. Zentralbl. Veterinamed. C. Anat. Hist. Embryol., 5r206-223. Friede, R.L., and W. Schachenmayr 1978 The origin of subdural neomembranes. 2. Fine structure of neomembranes. Am. J. Pathol., 92:69-84. Gardner, W.J. 1932 Traumatic subdural hematoma. Arch. Neurol. Psychiatr., 272347-858. Gennarelli, T.A., and L.E. Thibault 1982 Biomechanics of acute subdural hematoma. J . Trauma, 22:680-686. Graham, R.C., and M.J. Karnovsky 1966 The early stages of absorption of injected horseradish peroxidase in the proximal tubules of mouse kidney: Ultrastructural cytochemistry by a new technique. J . Histochem. Cytochem., 14r291-302. Gutkelch, A.N. 1971 Infantile subdural haematoma and its relationship to whiplash injuries. Br. Med. J . [Clin. Res.1,2:430,431. Haines, D.E. 1991 On the question of a subdural space. Anat. Rec., 230.1-16. Haines, D.E., and R.G. Frederickson 1991 The meninges. In: Meningeomas. 0. Al-Mefty, ed. Raven Press, New York, pp. 9-2. Himango, W.A., and F.L. Low 1971 The fine structure of a lateral recess of the subarachnoid space in the rat. Anat. Rec., 171:l-20. Kaump, D.H., and J.G. Love 1938 “Subdural” hematoma. Surg. Gynecol. Obstet., 67:87-93. Key, A., and G. Retzius 1876 Studien in der Anatomie des Nervensystems und des Bindegewebes. P.A. Norstedt & Soner, Stockholm. Klatzo, I., J . Miquel, P.J. Ferris, J.D. Prokop, and D.E. Smith 1964 Observations on the passage of the fluorescein labeled serum proteins (FLSP) from the cerebrospinal fluid. J . Neuropathol. EXD.Neural.. 23.18-35. Klika, E. -1967 The ultrastructure of meninges in vertebrates. Acta Univ. Carol. [Med.] (Praha), 1353-71. Klika, E., and M. Richter 1989 SEM of subdural space in mammals. Prog. Clin. Biol. Res., 295:389-393. Leary, T.1939 Subdural or intradural hemorrhages? Arch. Pathol. Lab. Med., 282308-820. Loew, F., and R. Kivelitz 1976 Injuries of the brain and skull, part 2. In: Handbook ofclinical Neurology. P.J. Vinken and G.W. Bruyn, eds. North Holland Publishing Company, Amsterdam, pp. 297327. Lofgren, J., C. Van Essen, and N.N. Zwetnow 1973 The pressurevolume curve of the cerebrospinal fluid space in dogs. Acta Neural. Scand., 49:557-574. Lofgren, J., and N.N. Zwetnow 1973 Cranial and spinal components of 37 the cerebrospinal fluid pressure-volume curve. Acta Neural. Scand., 4 9 5 7 5 4 8 5 . Lopes, C.A.S., and W.G.P. Mair 1974 Ultrastructure of the arachnoid membrane in man. Acta Neuropathol. (Bed.), 28r167-173. Mairova, N.A. 1965 An experimental investigation into absorption of erythrocytes from the subdural space of the brain. Vopr. Neirokhir., 29:lO-14. Makiyama, Y., Y. Katayama, Y. Ueno, Y. Koike, and T. Tsubokawa 1985 Acute subdural hematomas spontaneously disappearing within 3 days following closed head injury: report of two cases. Nihon Univ. J. Med.. 27t123-127. Mallory, F.B. 1920 The type cell ofthe so-called dural endothelioma. J . Med. Res., 4Ir349-364. McCabe, J.S., and F.N. Low 1969 The subarachnoid angle: An area of transition in peripheral nerve. Anat. Rec., I64t15-34. Nabeshima, S., and T.S. Reese 1972 Barrier to proteins within the spinal meninges. J. Neuropathol. Exp. Neurol., 31:176,177. Nabeshima, S., T.S. Reese, D.M.D. Landis, and M.W. Brightman 1975 Junctions in the meninges and marginal glia. J.Comp. Neurol., 164:127-170. Nicholas, D.S., and R.O. Weller 1988 The fine anatomy of the human spinal meninges. A light and scanning electron microscopy study. J. Neurosurg., 69:276-282. Oda, Y., and I. Nakanishi 1984 Ultrastructure of the mouse leptomeninx. J. Comp. Neurol., 225t448-457. Orlin, J.R., and N.N. Zwetnow 1989 Experimental arterial subdural bleeding. In: Intracranial Pressure VII. J.T. Hoff and A.L. Betz, eds. Springer-Verlag, Berlin, pp. 675-680. Pease, D.C., and R.L. Schultz 1958 Electron microscopy of rat cranial meninges. Am. J . Anat., 102r301-321. Polman, C.H., C.J. Gijsbers, J.J. Heimans, H. Ponssen, and J. Valk 1986 Rapid spontaneous resolution of a n acute subdural hematoma. Neurosurgery, 19r446-448. Pudenz, R.H., and C.H. Shelden 1946 The lucite calvarium-a method for direct observation of the brain. 11. Cranial trauma and brain movement. J. Neurosurg., 3:487-505. Rascal, M.M., and J . Izard 1976 The subdural neurothelium of the cranial meninges in man. Anat. Rec., 186r429-436. Schachenmayr, W., and R.L. Friede 1978 The origin of subdural neomembranes. 1. Fine structure of the dura-arachnoid interface in man. Am. J. Pathol., 92~53-68. Tripathi, R.C. 1973 Ultrastructure of the arachnoid mater in relation to outflow of cerebrospinal fluid. Lancet, 819~8-11. Voris, H.V. 1946 Subdural hematoma. JAMA, 132:686-692. Waggener, J.D., and J . Beggs 1967 The membraneous coverings of neural tissues: an electron microscopy study. J. Neuropathol. Exp. Neural., 26t412-426. Watanabe, S., H. Shimada, and S. Ischii 1972 Production of clinical form of chronic subdural hematoma in experimental animals. J. Neurosurg., 3 7:552-561. Weed, L.H. 1923 The absorption of cerebrospinal fluid into the venous system. Am. J . Anat., 9253-68. Weed, L.H. 1935 Certain anatomical and physiological aspects of the meninges and cerebrospinal fluid. Brain, 58:383-397. Yamashima, T., and S. Yamamoto 1985 The origin of inner membranes in chronic subdural hematomas. Acta Neuropathol. (Berl.), 67r219-225.