~
~
~
~~~~
Short-term response of brain tissue to cerebrospinal fluid shunts in vivo and in vitro Marc R. Del Bigio* and Sergey Fedoroff Department of Anatomy, University of Manitoba, Winnipeg, and Department of Anatomy, University of Saskatchewan, Saskatoon, Canada The purpose of the studies was to determine how gross physical characteristics of cerebrospinal fluid (CSF) shunts and the cellular proliferative response to shunts contribute to shunt obstruction. Ventricular catheters with round holes, slots, and flanges were implanted into the lateral ventricles of rabbits for 4 weeks. All shunt designs were subject to ingrowth of tissue from the ventricle wall or choroid plexus. There were no qualitative or quantitative
differences between normal and hydrocephalic rabbits. Astroglial cells from newborn mice were cultured on shunt catheters for 2 or 4 weeks, The growth of these cells was poor, probably because the cells cannot attach well to the silicone rubber substrate. Contact between the shunt catheter and vascularized brain tissue is the most important factor in the genesis of shunt obstruction. 0 1992 John Wiley & Sons, Inc.
INTRODUCTION
Although effective in the treatment of hydrocephalus, cerebrospinal f h i d (CSF) shunting is of ten associated with complications that necessitate an average of two shunt revisions over the lifetime of a patient.' The most common complication is obstruction of the proximal catheter located in the lateral ventricle.2Ventricular catheters obtained at the time of revision have been found to be occluded by ependyma, glia, brain tissue, and choroid p l e x ~ s . ~ , ~ Furthermore, the tissues that form the walls of the rabbit lateral ventricle and rat fourth ventricle readily invade shunt catheters as a result of astroglial proliferation and mechanical forces.sb A variety of ventricular catheters have been designed to avoid ingrowth of tissue but their selection in clinical practice remains largely s~bjective.~ It is unclear whether the silicone rubber of which shunt catheters are manufactured can act as a stimulus to growth of astroglia. This study was intended to examine whether any design of shunt catheter is inherently resistant to invasion by brain tissue and whether astroglial cells can grow on shunts in ideal culture conditions. *To whom correspondence should be addressed at Division of Neuropathology, University of Toronto, c/o 1507-30 Hillsboro Ave., Toronto, Ontario Canada M5R 1S7. Journal of Biomedical Materials Research, Vol. 26, 979-987 (1992) ccc 0021-9304/92/080979-09$4.00 0 1992 John Wiley & Sons, Inc.
DEL BIG10 AND FEDOROFF
980 METHODS
In vivo studies Adult male New Zealand White rabbits (2.5-3.0 kg) were cared for in accordance with the guidelines set forth by the Canadian Council on Animal Care. Anesthesia was induced by intramuscular injections of ketamine (35 mg/kg) and xylazine (5 mg/kg). Five normal rabbits underwent unilateral implantation and five normal rabbits underwent bilateral implantation of 5-mm-long segments of cerebrospinal fluid shunt catheters into the frontal horns of the lateral ventricles as previously described: The implants included standard design shunt tubing (3 mm diameter with six round holes 0.5 mm in diameter) made of barium-impregnated ( n = 4), silver-impregnated (n = 2), or unimpregnated silicone rubber (n = 3), multiperforated flanged shunts (HeyerSchulte Corp.) ( n = 3), and slotted shunts (0.25-mm-wide slots; Holter-Hauser Int., Inc.) (n = 3) both made of barium-impregnated silicone rubber. There was no CSF flow through the shunts because the implants were entirely intraventricular. Postimplantation survival time was 4 weeks. Sixteen rabbits were made hydrocephalic by injection of silicone oil into the cisterna magna as previously described.' At 1 (n = 5), 4 (n = 2), or 8 (n = 9) weeks following injection a sterile barium-impregnated shunt catheter of standard design with round holes was inserted into the frontal horn of the lateral ventricle and the distal end was placed in the subcutaneous tissue. Two of the rabbits underwent ventriculoperitoneal shunting. Hydrocephalic rabbits were killed 1 or 4 weeks after shunting. Anesthetized rabbits were perfused by the transcardiac route with 2.0% glutaraldehyde-2.5% paraformaldehyde in 0.12M phosphate buffer with 0.02 mM calcium chloride added. The brains were removed from the skull intact and stored overnight in the same fixative at 4°C. Tissue blocks surrounding the lateral ventricles were excised and processed with the shunt tubing in situ. The Specimens were post-fixed in phosphate-buffered 2% aqueous osmium tetroxide for 2 h and dehydrated in ethanol. For light microscopy the tissue was embedded in Epon 812 and semithin sections were stained with methylene blue/azure XI. For scanning electron microscopy the specimens were critical-point dried in carbon dioxide and sputter-coated with gold : palladium (60.40). Among the normal rabbits, five were studied by light microscopy and five by scanning microscopy. Among the hydrocephalic rabbits, four were studied by light microscopy and ten by scanning microscopy.
In vitro studies The cerebral hemispheres of newborn CD-1 mice were isolated in an aseptic manner, the meninges were removed, and the neopallia were mechanically dissociated through sterile 75-pm nylon mesh. Cells were suspended in modified Eagle's minimum essential medium (mMEM) containing 5% adult horse serum. Unimpregnated and barium-impregnated silicone rubber ven-
BRAIN TISSUE RESPONSE TO SHUNTS
981
tricular catheters were split lengthwise and 4-cm lengths were sterilized. A preliminary study revealed no differences in cell growth on the concave or convex surfaces. An attempt to simulate the passage of the shunt through brain by smearing with brain tissue uniformly resulted in infected cultures. In five round culture dishes (Falcon, Becton-Dickinson Labware, Oxnard, CA), four or five shunt segments were anchored under a sterile glass rod. For culturing, nigrosine-excluding cells at a concentration of 2.5 X lo5 cells/mL in a total volume of 8 mL were added to the dishes. Cultures were incubated in a 5% C02/95% air atmosphere at 37°C. After 2 days the cell debris was washed away and the growth medium replaced. Growth medium was changed every 3 days until the time of fixation at 15 or 25-27 days. For fixation the specimens were rinsed twice in room-temperature Dulbecco's phosphate-buffered saline (PBS) then flooded with 2% glutaraldehyde in 0.125M phosphate buffer at pH 7.2 with 2mM magnesium chloride added. Immediately thereafter, an equal volume of 1%osmium tetroxide in the same buffer was added. The specimens were fixed for 30 min then rinsed in distilled water and dehydrated in graded ethanol solutions. Following critical-point drying in carbon dioxide, 1-cm segments of shunt tubing were sputter-coated with gold: palladium (60: 40). Six samples from the 15-day culture and seven samples from the 25- to 27-day culture were examined by scanning electron microscopy. The number of cells identified on the viewing screen at a magnification of X1500 were counted in ten randomly selected rectangular areas (130 X 100 pm) on the surface of each shunt segment. StatisticaI comparison between ceIl counts at the two growth periods were made using Student's t test. For immunocytochemical identification of the cells shunt specimens were washed in PBS, fixed in methanol for 8 min at -2O"C, washed in PBS, extracted for 10 min in 0.1% Triton X-100 in PBS, washed in PBS, incubated for 1 h with mouse monoclonal antibody to glial fibrillary acidic protein (GFAP) (Boehringer Mannheim Gmbh, FRG) diluted 1:25 in PBS, washed in PBS, and finally incubated for 30 min with rhodamine-labeled goat anti-mouse antibody (Miles-Yeda Ltd.; Rehovat, Israel) diluted 1:100 in PBS. Coverslips were affixed to the specimens with a PBS/glycerin mixture (50/50) and the specimens were examined by epif luorescence microscopy.
RESULTS
In vivo observations The nonhydrocephalic rabbits tolerated implantation of the shunt segments well and at necropsy the frontal horns of the lateral ventricles were of normal size. In control rabbits the ventricles were lined by cuboidal ependymal cells whose apical surfaces had microvilli and clusters of cilia. Adjacent to all implants, the ependymal cells were flattened and cilia were shortened or completely lost. Supraependymal macrophages identified by the presence of vacuoles, lysosomes, and branching processes were present on the surface of
DEL BIG10 AND FEDOROFF
982
the ventricle. Subependymal astrocytes with prominent bundles of intermediate filaments were more abundant and the extracellular spaces were enlarged in the region of the implants. Six of 15 implants assessed by either light or scanning electron microscopy were invaded by evaginations from the wall of the ventricle (Table I). These evaginations consisted of a vascularized core of astroglial tissue partially covered by flattened ependymal cells generally lacking cilia (Fig. 1).Seven of fifteen implants were invaded by choroid plexus (Table I and Fig. 2). Two shunts had both ventricle wall and choroid plexus ingrowths. Adherent to the surfaces of all shunts were numbers of macrophage-like cells. At the TABLE I Brain Tissue Growth into Shunts Implanted in Normal Rabbit Lateral Ventricles for 4 Weeks Shunt Type Sham control (no implant) Standard Slotted Flanged
Number Implanted 5
9 3 3
Evagination from Ventricle Wall 0 5” 2 0
Choroid Plexus -
5” 1 2
”One catheter in each group was invaded in two holes.
Figure 1. Light micrograph showing in longitudinal section a tissue evagination from the wall of the lateral ventricle overlying the caudate nucleus of a nonhydrocephalic rabbit. The ependymal cells (arrow) are flattened along the sides and absent at the apex. The core of the evagination is composed of glial cells and many blood vessels (arrowheads). Methylene blue/Azure I1 stain. Bar = 200 p m .
BRAIN TISSUE RESPONSE TO S H U N T S
983
Figure 2. Scanning electron micrograph showing choroid plexus entering the hole of a shunt catheter (large arrowhead). The pseudopodia-bearing cells (small arrows) on the choroid plexus surface and globular cells on the shunt surface are macrophages. Bar = 100 pm.
microscopic level, the unimpregnated silicone rubber was smooth and that impregnated with barium or silver was rough. However, there were no apparent differences in cell adhesion or gross tissue invasion. There were two shunt-related deaths among the hydrocephalic rabbits: one due to intracranial hemorrhage and one due to meningitis. Shunting significantly reduced the size of the ventricles in seven of eight surviving hydrocephalic rabbits that had been shunted for 1 week. The one shunt failure was caused by a tissue evagination from the wall of the ventricle. Only one of six hydrocephalic rabbits shunted for 4 weeks was successfully treated. In two rabbits failure was due to migration of the complete shunt apparatus into the peritoneal cavity. Three rabbits suffered complete proximal shunt obstruction, one due to an evagination from the wall of the ventricle, one due to invasion by choroid plexus, and one due to retrograde flow of silicone oil into the ventricular system. Hydrocephalus caused attentuation of the ependymal lining of the ventricle and subependymal astrogliosis. The area contacting the shunt suffered further erosion of the ependymal cells and more pronounced astrogliosis than that in the opposite lateral ventricle. Numerous supraependymal macrophages were situated on the surface of the ventricles. Four of the fourteen surviving shunted hydrocephalic rabbits exhibited evaginations from the wall of the ventricle that extended into the lumen of the shunt (Table 11). Among rabbits shunted for 1week one complete and two partial obstructions of the shunt resulted and among rabbits shunted for 4 weeks one complete obstruction resulted. In all of these the lateral ventricle has collapsed onto the shunt. The cellular architecture of the evaginations was similar to that in nonhydrocephalic rabbits although the ependymal coverage was less com-
DEL BIG10 AND FEDOROFF
984
TABLE I1 Brain Tissue Growth into Shunts Implanted i n Lateral Ventricles of Surviving Hydrocephalic Rabbits Shunt Duration (weeks)
Number of Animals
1
8
4
6
Evagination from Ventricle Wall
Choroid Plexus
3 1
0 1
plete. In one rabbit shunted for 1week there were prominent blood vessels on the surface of the evagination (Fig. 3).
In vitro observations As has been observed in culture systems without shunts: after 15 days of culturing the culture dish surface adjacent to the shunt segment was almost totally covered by large colonies of polygonal and stellate cells. By 25 days the cultured cells were confluent. More than 85% of these cells were immunoreactive for GFAP. On the surface of the shunt segments no cells with morphology similar to that seen on the adjacent dish surface were found. Rather, there were irregular globoid cells with only occasional cell processes spreading over the substrate surface (Fig. 4). Even over irregularities in the
Figure 3. Scanning electron micrograph showing a tissue evagination from the wall of the lateral ventricle of a hydrocephalic rabbit that had been shunted for 1 week. The evagination and ventricle surface (V) which is seen in the background are denuded of recognizable ependyrnal cells. Sizable blood vessels (arrowheads) accompany the glial tissue outgrowth whose apex (asterisk) had grown into the shunt lumen. Bar = 100 Km.
BRAIN TISSUE RESPONSE TO SHUNTS
985
Figure 4. Scanning electron micrograph showing a single astroglial cell on the surface of shunt tubing after 15 days in culture. The cell body is globoid and only a single flat process (arrow) extends along the substrate surface. Bar = 10 p m .
silicone rubber substrate, cellular adhesion appeared to be tenuous. The cells usually lay alone or in small groups and only in the 15day cultures were rare aggregates of more than 10 cells seen. Most of these cells were immunoreactive for GFAP but details of their cytoskeletons were vague as a result of considerable background fluorescence from the substrate. After 15 days in culture (n = 6) there were 27.17 f 6.5 (mean 2 standard error of mean) cells per 0.013 mm2 of shunt surface area. After 25-27 days (n = 7) the density of cells had significantly decreased to 3.86 ? 1.08 ( p < 0.01).
DISCUSSION
The in vivo experimental model simulates the situation in which the ventricle collapses onto the shunt, a factor believed to be important in the genesis of shunt obstruction.'" Contact of shunt catheters with the wall of the rabbit lateral ventricle results in astroglial proliferation and attenuation of ependyma that progress over a period of 16 weeks. Evaginations grow from the wall of the ventricle into the shunt lumen within 4 weeks.' Choroid plexus is reported to be the tissue that most commonly obstructs ventricular catheters in human hydrocephalic patients! Although some authors have advocated the use of flanged shunts to avoid this complication," others have failed to demonstrate any advantage.*' In this experimental study, all shunt designs were subject to invasion by either choroid plexus or evaginations from the wall of the ventricle. The response of hydrocephalic rabbit brains to shunts was qualitatively similar to that in normal animals. Evaginations from the wall of the ventricle were more commonly seen in rabbits with small ventricles than those with
DEL BIGIO AND FEDOROFF
986
persistently large ventricles showing again the importance of contact with the shunt. The relative paucity of choroid plexus ingrowth among hydrocephalic rabbits is possible because the larger initial size of the lateral ventricle allowed implantation at a greater distance from the choroid plexus. Whether the "flow-traction" effect caused by CSF flow from the ventricle into the shunt lumen'3 contributed to the growth of tissue is not clear. Texture and chemical characteristics of polymer implants have been shown to affect nonneural tissue rea~tivity.'~ Although impregnation of silicone rubber with radiopaque substances increases the surface roughness of shunt material, macrophage accumulation in viuo was not significantly enhanced. Adhesion of astroglial cells to the surface of shunt material was generally poor. This likely explains the decline between 2 and 4 weeks in the density of glial cells on the surface of the shunts in vitro. Adhesion to substrate is known to be necessary for spreading and proliferation of glial and other cell type^.'^-" In another in uitro study, moderate adhesion of glial and neural cells to Teflon has been observed.'' These findings suggest that the growth of evaginations from the ventricle wall or choroid plexus into holes of shunts requires physical contact with the shunts. The shunt surface itself does not appear to provide a substrate that promotes growth of astroglial cells. Ongoing growth of tissue that fills the shunt lumen is probably dependent on a vascularized pedicle which is in continuity with the brain. These observations support the clinical recommendation that placement of the shunt catheter tip well away from choroid plexus and avoidance of ventricle collapse onto the shunt are more important than the choice of shunt type.'uJ19 However, we cannot exclude the possibility that after longer periods of implantation growth of brain tissue into shunts also involves chemical'' and immunologicz1factors. We wish to thank Dr. J. E. Bruni for his criticism of this paper. This investigation was supported by grants from the Health Sciences Centre Research Foundation (Winnipeg) and the Medical Research Council of Canada (Grant #MT 4235 to S.F.).
References 1. H. J. Hoffman and M.S.M. Smith, "The use of shunting devices for cerebrospinal fluid in Canada," Can. 1. Neurol. Sci., 13, 81-87 (1986). 2. W. Serlo, E. Fernell, E. Heikkinen, H. Anderson, and L. von Wendt, "Functions and complications of shunts in different etiologies of childhood hydrocephalus," Childs Nerv. Syst., 6, 92-94 (1990). 3. S. H. Bigner, P. D. Elmore, A. L. Dee, and W. W. Johnston, "The cytopathology of reactions to ventricular shunts," Actu Cytol., 29, 391-396 (1985). 4. P. Collins, A. D. Hocklev, and D. H. M. Woolam. "Surface ultrastructure of tissues occluding ventricular catheters," 1. Neurosurg., 48, 609613 (1978). 5. J. E. Bruni and M. R. Del Bigio, "Reaction of periventricular tissue in 1
,
the rat fourth ventricle to chronically placed shunt tubing implants," Neurosurgery, 19, 337-345 (1986). 6. M. R. Del Bigio and J. E. Bruni, "Reaction of rabbit lateral periventricular tissue to shunt tubing implants," I. Neurosurg., 64, 932-940 (1986).
BRAIN TISSUE RESPONSE TO SHUNTS 7. 8. 9.
10. 11. 12. 13.
14. 15. 16. 17. 18. 19. 20.
21.
987
F. Epstein, ”How to keep shunts functioning, or ‘The Impossible Dream: ” Clin. Neurosurg., 32, 608-631 (1985). M. R. Del Bigio and J. E. Bruni, ”Periventricular pathology in hydrocephalic rabbits before and after shunting,” Acta Neuroptrfkol., 77, 186195 (1988). S. Fedoroff, J. Neal, M. Opas, and V. I. Kalnins, ’Astrocyte cell lineage. 111. The morphology of differentiating astrocytes in colony culture,” J. Neurocytol., 13, 1-20 (1984). K.G. Go, E. J. Ebels, and 0.H. van Woerden, “Experiences with recurring ventricular catheter obstructions,” Clin. Neurol. Neurosurg., 83, 4756 (1981). H. E. Portnoy, “New ventricular catheter for hydrocephalic shunts,” J. Neurosurg., 34, 702-703 (1974). J. Haase and R. Weeth, ”Multiflanged ventricular Portnoy catheter for hydrocephalus shunts,” Acta Neurockir., 33, 213-218 (1976). S. Hakim, ”Observations on the physiopathology of the CSF pulse and prevention of ventricular catheter obstruction in valve shunts,” Dev. Med. Child. Neurol. Suppl., 20, 42-48 (1969). B. D. Ratner, A. B. Johnston, and T. J. Lenk, ”Biomaterials surfaces,” 1. Biomed. Muter. Res., 21 (Suppl. Al), 59-90 (1987). J. FoIkman and A. Moscona, “Role of cell shape in growth control, ” Nature, 273, 345-349 (1978). F. Grinell, ”Cellular adhesiveness and extracellular substrata,” Int. Rev. Cytol., 53, 65-144 (1978). B. Westermark, ”Growth control in miniclones of human glial cells,” E x p . Cell Res., 111, 295-299 (1978). A. Ammar, C. Lagenaur, and P. Jannetta, ”Neural tissue compatibility of Tef Ion as an implant material for microvascular decompression,” Neurosurg. Rev., 13, 299-303 (1990). A. L. Albright, S. J. Haines, and F. H. Taylor, “Function of parietal and frontal shunts in childhood hydrocephalus,” 1. Neurosurg., 69, 883-886 (1988). D. E. Taylor and J. E. Penhallow, “Comparative biotolerance of polyacrylamide-agarose gel, silicone rubber, and microporous PTFE as soft tissue implants,” Biomateriuls, 7, 277-282 (1986). N. Kossovsky and R. B. Snow, “Clinical-pathological analysis of failed central nervous system shunts,” J. Biomed. Muter. Res., 23 (Suppl Al), 7386 (1989).
Received September 17, 1990 Accepted March 4, 1992
~
~
~~~~
Short-term response of brain tissue to cerebrospinal fluid shunts in vivo and in vitro Marc R. Del Bigio* and Sergey Fedoroff Department of Anatomy, University of Manitoba, Winnipeg, and Department of Anatomy, University of Saskatchewan, Saskatoon, Canada The purpose of the studies was to determine how gross physical characteristics of cerebrospinal fluid (CSF) shunts and the cellular proliferative response to shunts contribute to shunt obstruction. Ventricular catheters with round holes, slots, and flanges were implanted into the lateral ventricles of rabbits for 4 weeks. All shunt designs were subject to ingrowth of tissue from the ventricle wall or choroid plexus. There were no qualitative or quantitative
differences between normal and hydrocephalic rabbits. Astroglial cells from newborn mice were cultured on shunt catheters for 2 or 4 weeks, The growth of these cells was poor, probably because the cells cannot attach well to the silicone rubber substrate. Contact between the shunt catheter and vascularized brain tissue is the most important factor in the genesis of shunt obstruction. 0 1992 John Wiley & Sons, Inc.
INTRODUCTION
Although effective in the treatment of hydrocephalus, cerebrospinal f h i d (CSF) shunting is of ten associated with complications that necessitate an average of two shunt revisions over the lifetime of a patient.' The most common complication is obstruction of the proximal catheter located in the lateral ventricle.2Ventricular catheters obtained at the time of revision have been found to be occluded by ependyma, glia, brain tissue, and choroid p l e x ~ s . ~ , ~ Furthermore, the tissues that form the walls of the rabbit lateral ventricle and rat fourth ventricle readily invade shunt catheters as a result of astroglial proliferation and mechanical forces.sb A variety of ventricular catheters have been designed to avoid ingrowth of tissue but their selection in clinical practice remains largely s~bjective.~ It is unclear whether the silicone rubber of which shunt catheters are manufactured can act as a stimulus to growth of astroglia. This study was intended to examine whether any design of shunt catheter is inherently resistant to invasion by brain tissue and whether astroglial cells can grow on shunts in ideal culture conditions. *To whom correspondence should be addressed at Division of Neuropathology, University of Toronto, c/o 1507-30 Hillsboro Ave., Toronto, Ontario Canada M5R 1S7. Journal of Biomedical Materials Research, Vol. 26, 979-987 (1992) ccc 0021-9304/92/080979-09$4.00 0 1992 John Wiley & Sons, Inc.
DEL BIG10 AND FEDOROFF
980 METHODS
In vivo studies Adult male New Zealand White rabbits (2.5-3.0 kg) were cared for in accordance with the guidelines set forth by the Canadian Council on Animal Care. Anesthesia was induced by intramuscular injections of ketamine (35 mg/kg) and xylazine (5 mg/kg). Five normal rabbits underwent unilateral implantation and five normal rabbits underwent bilateral implantation of 5-mm-long segments of cerebrospinal fluid shunt catheters into the frontal horns of the lateral ventricles as previously described: The implants included standard design shunt tubing (3 mm diameter with six round holes 0.5 mm in diameter) made of barium-impregnated ( n = 4), silver-impregnated (n = 2), or unimpregnated silicone rubber (n = 3), multiperforated flanged shunts (HeyerSchulte Corp.) ( n = 3), and slotted shunts (0.25-mm-wide slots; Holter-Hauser Int., Inc.) (n = 3) both made of barium-impregnated silicone rubber. There was no CSF flow through the shunts because the implants were entirely intraventricular. Postimplantation survival time was 4 weeks. Sixteen rabbits were made hydrocephalic by injection of silicone oil into the cisterna magna as previously described.' At 1 (n = 5), 4 (n = 2), or 8 (n = 9) weeks following injection a sterile barium-impregnated shunt catheter of standard design with round holes was inserted into the frontal horn of the lateral ventricle and the distal end was placed in the subcutaneous tissue. Two of the rabbits underwent ventriculoperitoneal shunting. Hydrocephalic rabbits were killed 1 or 4 weeks after shunting. Anesthetized rabbits were perfused by the transcardiac route with 2.0% glutaraldehyde-2.5% paraformaldehyde in 0.12M phosphate buffer with 0.02 mM calcium chloride added. The brains were removed from the skull intact and stored overnight in the same fixative at 4°C. Tissue blocks surrounding the lateral ventricles were excised and processed with the shunt tubing in situ. The Specimens were post-fixed in phosphate-buffered 2% aqueous osmium tetroxide for 2 h and dehydrated in ethanol. For light microscopy the tissue was embedded in Epon 812 and semithin sections were stained with methylene blue/azure XI. For scanning electron microscopy the specimens were critical-point dried in carbon dioxide and sputter-coated with gold : palladium (60.40). Among the normal rabbits, five were studied by light microscopy and five by scanning microscopy. Among the hydrocephalic rabbits, four were studied by light microscopy and ten by scanning microscopy.
In vitro studies The cerebral hemispheres of newborn CD-1 mice were isolated in an aseptic manner, the meninges were removed, and the neopallia were mechanically dissociated through sterile 75-pm nylon mesh. Cells were suspended in modified Eagle's minimum essential medium (mMEM) containing 5% adult horse serum. Unimpregnated and barium-impregnated silicone rubber ven-
BRAIN TISSUE RESPONSE TO SHUNTS
981
tricular catheters were split lengthwise and 4-cm lengths were sterilized. A preliminary study revealed no differences in cell growth on the concave or convex surfaces. An attempt to simulate the passage of the shunt through brain by smearing with brain tissue uniformly resulted in infected cultures. In five round culture dishes (Falcon, Becton-Dickinson Labware, Oxnard, CA), four or five shunt segments were anchored under a sterile glass rod. For culturing, nigrosine-excluding cells at a concentration of 2.5 X lo5 cells/mL in a total volume of 8 mL were added to the dishes. Cultures were incubated in a 5% C02/95% air atmosphere at 37°C. After 2 days the cell debris was washed away and the growth medium replaced. Growth medium was changed every 3 days until the time of fixation at 15 or 25-27 days. For fixation the specimens were rinsed twice in room-temperature Dulbecco's phosphate-buffered saline (PBS) then flooded with 2% glutaraldehyde in 0.125M phosphate buffer at pH 7.2 with 2mM magnesium chloride added. Immediately thereafter, an equal volume of 1%osmium tetroxide in the same buffer was added. The specimens were fixed for 30 min then rinsed in distilled water and dehydrated in graded ethanol solutions. Following critical-point drying in carbon dioxide, 1-cm segments of shunt tubing were sputter-coated with gold: palladium (60: 40). Six samples from the 15-day culture and seven samples from the 25- to 27-day culture were examined by scanning electron microscopy. The number of cells identified on the viewing screen at a magnification of X1500 were counted in ten randomly selected rectangular areas (130 X 100 pm) on the surface of each shunt segment. StatisticaI comparison between ceIl counts at the two growth periods were made using Student's t test. For immunocytochemical identification of the cells shunt specimens were washed in PBS, fixed in methanol for 8 min at -2O"C, washed in PBS, extracted for 10 min in 0.1% Triton X-100 in PBS, washed in PBS, incubated for 1 h with mouse monoclonal antibody to glial fibrillary acidic protein (GFAP) (Boehringer Mannheim Gmbh, FRG) diluted 1:25 in PBS, washed in PBS, and finally incubated for 30 min with rhodamine-labeled goat anti-mouse antibody (Miles-Yeda Ltd.; Rehovat, Israel) diluted 1:100 in PBS. Coverslips were affixed to the specimens with a PBS/glycerin mixture (50/50) and the specimens were examined by epif luorescence microscopy.
RESULTS
In vivo observations The nonhydrocephalic rabbits tolerated implantation of the shunt segments well and at necropsy the frontal horns of the lateral ventricles were of normal size. In control rabbits the ventricles were lined by cuboidal ependymal cells whose apical surfaces had microvilli and clusters of cilia. Adjacent to all implants, the ependymal cells were flattened and cilia were shortened or completely lost. Supraependymal macrophages identified by the presence of vacuoles, lysosomes, and branching processes were present on the surface of
DEL BIG10 AND FEDOROFF
982
the ventricle. Subependymal astrocytes with prominent bundles of intermediate filaments were more abundant and the extracellular spaces were enlarged in the region of the implants. Six of 15 implants assessed by either light or scanning electron microscopy were invaded by evaginations from the wall of the ventricle (Table I). These evaginations consisted of a vascularized core of astroglial tissue partially covered by flattened ependymal cells generally lacking cilia (Fig. 1).Seven of fifteen implants were invaded by choroid plexus (Table I and Fig. 2). Two shunts had both ventricle wall and choroid plexus ingrowths. Adherent to the surfaces of all shunts were numbers of macrophage-like cells. At the TABLE I Brain Tissue Growth into Shunts Implanted in Normal Rabbit Lateral Ventricles for 4 Weeks Shunt Type Sham control (no implant) Standard Slotted Flanged
Number Implanted 5
9 3 3
Evagination from Ventricle Wall 0 5” 2 0
Choroid Plexus -
5” 1 2
”One catheter in each group was invaded in two holes.
Figure 1. Light micrograph showing in longitudinal section a tissue evagination from the wall of the lateral ventricle overlying the caudate nucleus of a nonhydrocephalic rabbit. The ependymal cells (arrow) are flattened along the sides and absent at the apex. The core of the evagination is composed of glial cells and many blood vessels (arrowheads). Methylene blue/Azure I1 stain. Bar = 200 p m .
BRAIN TISSUE RESPONSE TO S H U N T S
983
Figure 2. Scanning electron micrograph showing choroid plexus entering the hole of a shunt catheter (large arrowhead). The pseudopodia-bearing cells (small arrows) on the choroid plexus surface and globular cells on the shunt surface are macrophages. Bar = 100 pm.
microscopic level, the unimpregnated silicone rubber was smooth and that impregnated with barium or silver was rough. However, there were no apparent differences in cell adhesion or gross tissue invasion. There were two shunt-related deaths among the hydrocephalic rabbits: one due to intracranial hemorrhage and one due to meningitis. Shunting significantly reduced the size of the ventricles in seven of eight surviving hydrocephalic rabbits that had been shunted for 1 week. The one shunt failure was caused by a tissue evagination from the wall of the ventricle. Only one of six hydrocephalic rabbits shunted for 4 weeks was successfully treated. In two rabbits failure was due to migration of the complete shunt apparatus into the peritoneal cavity. Three rabbits suffered complete proximal shunt obstruction, one due to an evagination from the wall of the ventricle, one due to invasion by choroid plexus, and one due to retrograde flow of silicone oil into the ventricular system. Hydrocephalus caused attentuation of the ependymal lining of the ventricle and subependymal astrogliosis. The area contacting the shunt suffered further erosion of the ependymal cells and more pronounced astrogliosis than that in the opposite lateral ventricle. Numerous supraependymal macrophages were situated on the surface of the ventricles. Four of the fourteen surviving shunted hydrocephalic rabbits exhibited evaginations from the wall of the ventricle that extended into the lumen of the shunt (Table 11). Among rabbits shunted for 1week one complete and two partial obstructions of the shunt resulted and among rabbits shunted for 4 weeks one complete obstruction resulted. In all of these the lateral ventricle has collapsed onto the shunt. The cellular architecture of the evaginations was similar to that in nonhydrocephalic rabbits although the ependymal coverage was less com-
DEL BIG10 AND FEDOROFF
984
TABLE I1 Brain Tissue Growth into Shunts Implanted i n Lateral Ventricles of Surviving Hydrocephalic Rabbits Shunt Duration (weeks)
Number of Animals
1
8
4
6
Evagination from Ventricle Wall
Choroid Plexus
3 1
0 1
plete. In one rabbit shunted for 1week there were prominent blood vessels on the surface of the evagination (Fig. 3).
In vitro observations As has been observed in culture systems without shunts: after 15 days of culturing the culture dish surface adjacent to the shunt segment was almost totally covered by large colonies of polygonal and stellate cells. By 25 days the cultured cells were confluent. More than 85% of these cells were immunoreactive for GFAP. On the surface of the shunt segments no cells with morphology similar to that seen on the adjacent dish surface were found. Rather, there were irregular globoid cells with only occasional cell processes spreading over the substrate surface (Fig. 4). Even over irregularities in the
Figure 3. Scanning electron micrograph showing a tissue evagination from the wall of the lateral ventricle of a hydrocephalic rabbit that had been shunted for 1 week. The evagination and ventricle surface (V) which is seen in the background are denuded of recognizable ependyrnal cells. Sizable blood vessels (arrowheads) accompany the glial tissue outgrowth whose apex (asterisk) had grown into the shunt lumen. Bar = 100 Km.
BRAIN TISSUE RESPONSE TO SHUNTS
985
Figure 4. Scanning electron micrograph showing a single astroglial cell on the surface of shunt tubing after 15 days in culture. The cell body is globoid and only a single flat process (arrow) extends along the substrate surface. Bar = 10 p m .
silicone rubber substrate, cellular adhesion appeared to be tenuous. The cells usually lay alone or in small groups and only in the 15day cultures were rare aggregates of more than 10 cells seen. Most of these cells were immunoreactive for GFAP but details of their cytoskeletons were vague as a result of considerable background fluorescence from the substrate. After 15 days in culture (n = 6) there were 27.17 f 6.5 (mean 2 standard error of mean) cells per 0.013 mm2 of shunt surface area. After 25-27 days (n = 7) the density of cells had significantly decreased to 3.86 ? 1.08 ( p < 0.01).
DISCUSSION
The in vivo experimental model simulates the situation in which the ventricle collapses onto the shunt, a factor believed to be important in the genesis of shunt obstruction.'" Contact of shunt catheters with the wall of the rabbit lateral ventricle results in astroglial proliferation and attenuation of ependyma that progress over a period of 16 weeks. Evaginations grow from the wall of the ventricle into the shunt lumen within 4 weeks.' Choroid plexus is reported to be the tissue that most commonly obstructs ventricular catheters in human hydrocephalic patients! Although some authors have advocated the use of flanged shunts to avoid this complication," others have failed to demonstrate any advantage.*' In this experimental study, all shunt designs were subject to invasion by either choroid plexus or evaginations from the wall of the ventricle. The response of hydrocephalic rabbit brains to shunts was qualitatively similar to that in normal animals. Evaginations from the wall of the ventricle were more commonly seen in rabbits with small ventricles than those with
DEL BIGIO AND FEDOROFF
986
persistently large ventricles showing again the importance of contact with the shunt. The relative paucity of choroid plexus ingrowth among hydrocephalic rabbits is possible because the larger initial size of the lateral ventricle allowed implantation at a greater distance from the choroid plexus. Whether the "flow-traction" effect caused by CSF flow from the ventricle into the shunt lumen'3 contributed to the growth of tissue is not clear. Texture and chemical characteristics of polymer implants have been shown to affect nonneural tissue rea~tivity.'~ Although impregnation of silicone rubber with radiopaque substances increases the surface roughness of shunt material, macrophage accumulation in viuo was not significantly enhanced. Adhesion of astroglial cells to the surface of shunt material was generally poor. This likely explains the decline between 2 and 4 weeks in the density of glial cells on the surface of the shunts in vitro. Adhesion to substrate is known to be necessary for spreading and proliferation of glial and other cell type^.'^-" In another in uitro study, moderate adhesion of glial and neural cells to Teflon has been observed.'' These findings suggest that the growth of evaginations from the ventricle wall or choroid plexus into holes of shunts requires physical contact with the shunts. The shunt surface itself does not appear to provide a substrate that promotes growth of astroglial cells. Ongoing growth of tissue that fills the shunt lumen is probably dependent on a vascularized pedicle which is in continuity with the brain. These observations support the clinical recommendation that placement of the shunt catheter tip well away from choroid plexus and avoidance of ventricle collapse onto the shunt are more important than the choice of shunt type.'uJ19 However, we cannot exclude the possibility that after longer periods of implantation growth of brain tissue into shunts also involves chemical'' and immunologicz1factors. We wish to thank Dr. J. E. Bruni for his criticism of this paper. This investigation was supported by grants from the Health Sciences Centre Research Foundation (Winnipeg) and the Medical Research Council of Canada (Grant #MT 4235 to S.F.).
References 1. H. J. Hoffman and M.S.M. Smith, "The use of shunting devices for cerebrospinal fluid in Canada," Can. 1. Neurol. Sci., 13, 81-87 (1986). 2. W. Serlo, E. Fernell, E. Heikkinen, H. Anderson, and L. von Wendt, "Functions and complications of shunts in different etiologies of childhood hydrocephalus," Childs Nerv. Syst., 6, 92-94 (1990). 3. S. H. Bigner, P. D. Elmore, A. L. Dee, and W. W. Johnston, "The cytopathology of reactions to ventricular shunts," Actu Cytol., 29, 391-396 (1985). 4. P. Collins, A. D. Hocklev, and D. H. M. Woolam. "Surface ultrastructure of tissues occluding ventricular catheters," 1. Neurosurg., 48, 609613 (1978). 5. J. E. Bruni and M. R. Del Bigio, "Reaction of periventricular tissue in 1
,
the rat fourth ventricle to chronically placed shunt tubing implants," Neurosurgery, 19, 337-345 (1986). 6. M. R. Del Bigio and J. E. Bruni, "Reaction of rabbit lateral periventricular tissue to shunt tubing implants," I. Neurosurg., 64, 932-940 (1986).
BRAIN TISSUE RESPONSE TO SHUNTS 7. 8. 9.
10. 11. 12. 13.
14. 15. 16. 17. 18. 19. 20.
21.
987
F. Epstein, ”How to keep shunts functioning, or ‘The Impossible Dream: ” Clin. Neurosurg., 32, 608-631 (1985). M. R. Del Bigio and J. E. Bruni, ”Periventricular pathology in hydrocephalic rabbits before and after shunting,” Acta Neuroptrfkol., 77, 186195 (1988). S. Fedoroff, J. Neal, M. Opas, and V. I. Kalnins, ’Astrocyte cell lineage. 111. The morphology of differentiating astrocytes in colony culture,” J. Neurocytol., 13, 1-20 (1984). K.G. Go, E. J. Ebels, and 0.H. van Woerden, “Experiences with recurring ventricular catheter obstructions,” Clin. Neurol. Neurosurg., 83, 4756 (1981). H. E. Portnoy, “New ventricular catheter for hydrocephalic shunts,” J. Neurosurg., 34, 702-703 (1974). J. Haase and R. Weeth, ”Multiflanged ventricular Portnoy catheter for hydrocephalus shunts,” Acta Neurockir., 33, 213-218 (1976). S. Hakim, ”Observations on the physiopathology of the CSF pulse and prevention of ventricular catheter obstruction in valve shunts,” Dev. Med. Child. Neurol. Suppl., 20, 42-48 (1969). B. D. Ratner, A. B. Johnston, and T. J. Lenk, ”Biomaterials surfaces,” 1. Biomed. Muter. Res., 21 (Suppl. Al), 59-90 (1987). J. FoIkman and A. Moscona, “Role of cell shape in growth control, ” Nature, 273, 345-349 (1978). F. Grinell, ”Cellular adhesiveness and extracellular substrata,” Int. Rev. Cytol., 53, 65-144 (1978). B. Westermark, ”Growth control in miniclones of human glial cells,” E x p . Cell Res., 111, 295-299 (1978). A. Ammar, C. Lagenaur, and P. Jannetta, ”Neural tissue compatibility of Tef Ion as an implant material for microvascular decompression,” Neurosurg. Rev., 13, 299-303 (1990). A. L. Albright, S. J. Haines, and F. H. Taylor, “Function of parietal and frontal shunts in childhood hydrocephalus,” 1. Neurosurg., 69, 883-886 (1988). D. E. Taylor and J. E. Penhallow, “Comparative biotolerance of polyacrylamide-agarose gel, silicone rubber, and microporous PTFE as soft tissue implants,” Biomateriuls, 7, 277-282 (1986). N. Kossovsky and R. B. Snow, “Clinical-pathological analysis of failed central nervous system shunts,” J. Biomed. Muter. Res., 23 (Suppl Al), 7386 (1989).
Received September 17, 1990 Accepted March 4, 1992
