Journal of Neuro-Oncology 27:127-132,1996. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

Laboratory Investigation

Reversion of phenotype of endothelial cells in brain tissue around glioblastomas

Rui Vaz ~, Nuno Borges, Ant6nio Sarmento and Isabel Azevedo Institute of Pharmacology and Therapeutics and t Department of Neurosurgery, Medical Faculty, 4200 Porto, Portugal

Key words: brain edema, peritumoral vessels, glioblastoma

Summary

With the aim of studying the putative involvement of peritumoral microvessels in the formation of brain edema, small pieces of peritumoral brain tissue were removed from six patients with glioblastoma multi- forme submitted to surgery. All patients had cerebral edema, as shown by preoperative C.T. and N.M.R. Control specimens were obtained from four patients undergoing ventriculo-peritoneal shunt. The tissue frag- ments were fixed in glutaraldehyde-osmium and embedded in Epon. In semi-thin sections observed under light microscopy peritumoral endothelial cells exhibited voluminous cytoplasm and nucleus.

Under the electron microscope, capillary cells from glioblastoma patients differed from controls mainly by showing nuclei rich in euchromatin, cytoplasm rich in pinocytotic vesicles and with occasional fenestrations. All these morphological characteristics are compatible with a process of reversion of phenotype of capillaries around glioblastomas to that of periphery as well as an increase in permeability. Both events may be due to diffusion of a tumoral vascular permeability/endothelial growth factor. This peripheral vessel phenotype of peritumoral microvessels supports their participation in the formation of brain edema and may provide a new clue for therapeutic intervention: for example it fits quite well to the known increase in permeability by leukotrienes and decrease in permeability by corticosteroids in tumoral edema.

Introduction

In 1913 Goldman created the term of blood-brain barrier, derived from observations that various com- pounds rapidly entering and distributing within most organs and tissues are excluded from the brain [1].

This barrier is created by the continuous layer of endothelial cells surrounding the vascular lumen, with tight intercellular junctions and very few pino- cytotic vesicles [2, 3]. These anatomical character- istics [4], quite distinct from those of peripheral mi- crovessels with their gaps, fenestrations and rich- ness in pinocytotic vesicles, correspond to equally distinct permeability characteristics between the two types of microvessels.

Everytime the blood-brain barrier is opened the volume regulation of neural tissue served by the damaged microvessels is disrupted, the hydrostatic pressure becomes the dominant force moving fluid from the blood to the brain and vasogenic edema is formed [5]. There is abundant evidence that edema occurs around cerebral gliomas [6]. The source of this abnormality seems to be an alteration in the function of the blood-brain barrier [7-13].

There is some controversy about the role of per- itumoral brain tissue in the origin of brain edema that is associated with some primary brain tumors. A few authors [14-17] conclude that the peritum- oral edema is due to alterations present within the tumor, the structure and function of capillaries in

128

the tumor periphery being relatively normal. In other studies [13, 18-21] it is suggested that the ab- normalities in the blood-brain barrier are not re- stricted to the tumor and that peritumoral brain tis- sue may contribute to the formation of edema. This is an important issue as alterations of peritumoral microvessels may provide a clue for new therapeut- ic interventions against the formation of edema as well as against the tumor spread.

Here we report a morphometric light microscopy and ultrastructural study of the capillary endothe- lial cells in peritumoral brain tissue from patients operated for glioblastoma multiforme.

Clinical material and methods

Patient population, imaging and surgery

Peritumoral brain tissue was obtained from six pa- tients with glioblastoma multiforme (patient data summarized in Table 1). Four adult patients under- going ventriculo/peritoneal shunt for 'communicat- ing' hydrocephalus were considered controls, as computed tomography (C.T.) and nuclear magnetic resonance images (N.M.R.) showed no abnormality

in the area (right frontal) where the ventricular catheter was introduced.

All operations were performed under general an- esthesia. Total intravenous anesthesia with propo- fol + sufentanyl and controlled ventilation were used. There were no peri-operative abnormalities.

The area of brain edema was determined from pre-operative C.T. and N.M.R. by a neuroradiolo- gist and the site of white matter biopsy was always annotated by the same neurosurgeon. All the pa- tients were in dexamethasone treatment (4 mg four times daily) and received no other medication.

All tumors were removed by microsurgical tech- nique. A sample of edematous white matter was taken near the tumor-brain junction, as atraumat- ically as possible, by means of a pituitary rongeur.

The protocol had been previously approved by the Ethics Committee of S. Jofio Hospital.

Morphological study

The methods used for this study were those previ- ously described for light and ultrastructural micros- copy [22, 23]. Briefly, the tissue samples were fixed by immersion in 3% glutaraldehyde in 0.1 mol/1 ca-

Table 1. Clinical data of evaluated patients

Case Age Symptoms and signs C.T. and N.M.R. findings

localization edema mass effect

1 59 headace

drowsiness L hemiparesis

2 55 dysphasia R hemiparesis

3 51 headache drowsiness L hemiparesis

4 74 headache R hemianopsis R hemiparesis

5 58 drowsiness L hemiparesis

6 63 headache drowsiness

R frontal ++ ++

L fronto-temporal ++ ++

R temporal ++ ++

L occipital + +

R frontal ++ +

R temporal ++ ++

+ indicates modera te and ++ marked edema or mass effect.

129

Fig. 1. Voluminous nuclei, rich in euchromat in and nucleoli (arrows) in peri tumoral capillary endothelial cells from a glioblastoma

multiforme. Clear spaces around blood vessels indicate the presence of edema fluid (-k). Bar = 10 gm.

codylate buffer, pH 7.3, at 4 ° C for 90 min. The tis- sue was then washed overnight in cacodylate buffer with 0.6 mol/1 sucrose at 4 ° C, post-fixed in a 2% so- lution of osmium tetroxide in 0.1 tool/1 cacodylate buffer, pH 7.3, at 4 ° C, for 1 h, dehydrated with eth- anol and propylene oxide and embedded in Epon 812 (Taab, Aldermaston, Berkshire, UK). Semi-thin

sections of 1 gm were obtained from five blocks of each case, taken at random, with a LKB Ultrotome, stained with 1% toluidine blue in 1% borax and ex- amined by light microscopy. Camera lucida draw- ings of all capillary endothelial cell nuclei in each section were made under a final amplification of x860. The largest diameter of each nucleus and the

Fig. 2. Capillary endothelial cell very rich in pinocytotic vesicles (arrowheads) in the peri tumoral tissue of a glioblastoma patient. Bar = 1 gm.

Fig. 3. Fenestrat ion (arrow) in a brain capillary from peri tumoral tissue of a glioblastoma patient. Bar = 1 gm.

130

Fig. 4. Endothelial cell (E) in a brain microvessel from a control patient. Note the smoothness of endothelial cell contours and the scarcity of pinocytotic vesicles. Bar = 1 gm.

axis perpendicular to it were measured on the paper drawings. These two values were then used to calcu- late the mean diameter (I)) for each nucleus [24].

All the morphometrical determinations were made by the same person in preparations identified by a code number, the meaning of which was un- known to the observer until all measurements were concluded.

Thin sections were cut in a LKB Ultrotome, stained with uranyl acetate and lead citrate and ob- served in a Siemens Elmiskope at 80 kV.

S ta t i s t i c s

The results are expressed as arithmetic means + SEM. Student's t test was applied to the difference between means (unpaired observations) and the difference considered significant if p < 0.05.

Results

In semi-thin sections observed under light micros- copy peritumoral capillary endothelial cells exhib- ited voluminous cytoplasm and nucleus (Fig. 1). Clear spaces indicating the presence of edema fluid were also observed. Morphometrical evaluation confirmed that capillary endothelial cell nuclei from glioblastoma patients were significantly larger

Fig. 5. Endothelial cell (E) in a heart capillary vessel. Note the roughness of its contours and the abundance of pinocytotic ves- icles (arrowheads). Bar = 1 gin.

than the controls (mean diameter 6.5 + 0.1 gm vs. 4.7 + 0.1 gm in controls, n = 160 each; p < 0.001).

Under the electron microscope, capillary endo- thelial cells from glioblastoma patients differed from controls mainly by showing nuclei rich in eu- chromatin, cytoplasm rich in pinocytotic vesicles (Fig. 2) and occasional fenestrations (Fig. 3). Elec- tron micrographs of a normal brain microvessel (Fig. 4) and of a cardiac (peripheral type) capillary (Fig. 5) are also shown.

Discussion

The purpose of the present study was to investigate the morphology of capillary endothelial cells in ar- eas surrounding human brain gliomas. The results showed striking morphological changes in these cells: a marked increase in the number of pinocytot- ic vesicle profiles and the appearance of occasional fenestrations directly indicate an increase in vessel permeability. Thus peritumoral microvessels in all probability contribute to formation of cerebral ede- ma in glioblastoma multiforme. Size of the endo- thelial cell nuclei, the abundance of nucleoli, the richness in euchromatin all indicate an increase in the synthetic activity of the cell [25, 26]. Taken to- gether, all changes exhibited by peritumoral capil- lary cells are compatible with a process of dediffe- rentiation of those ceils. As a matter of fact, cell dif-

ferent iat ion implicates, in general, a repression of

the cell nucleus activity with a decrease of global

prote in synthesis [25, 26], and, vice-versa, the acqui- sition of a synthetic pheno type denotes a process of dedifferent iat ion [27]. As to the morphologica l fea-

tures related with an increase in permeabi l i ty (i.e.,

pinocytot ic vesicles, fenestrat ions), they are remi-

niscent of per ipheral ra ther than brain microves-

sels, indicating, again, a dedifferent ia t ion f rom a b lood-bra in barr ier to a per ipheral vessel pheno-

type. Gl ioblas toma mult iforme, as well as o ther pri-

mary and metastat ic brain tumors, p roduces and re-

leases proteins which have been shown to increase

vascular permeabi l i ty [7, 8, 11-13, 28, 29] and to st imulate angiogenesis and endothel ial cell growth

[28, 30-34]. The endothel ial cell growth and dedif- ferent iat ion that we have now observed in per i tum-

oral brain microvessels may be due to the action of those vascular permeabi l i ty /endothel ia l growth fac-

tors diffusing f rom the tumor. This dedifferentiat-

ed, per ipheral vessel pheno type fits very well to the known phys iopharmacologica l characteristics of

brain tumora l /per i tumora l microvessels. For exam- ple, leukotr ienes are p roduced in brain tumors and

suspected of involvement in the fo rmat ion of ede-

ma [35, 36]. However , studies conduc ted to test that

hypothesis failed to demons t ra te any effect of leu- kotr ienes on b lood-bra in barr ier permeabi l i ty [37,

38]. As a mat te r of fact, leukotr ienes do not increase

the permeabi l i ty of brain microvessels but that of per ipheral microvessels [see 39, 40]. So, the failure

of leukotr ienes to increase the permeabi l i ty of the b lood-bra in barr ier [37, 38] does not invalidate the possibility of their involvement in cerebral edema

due to tumors , where capillaries dedifferent iate to

the per ipheral phenotype . Likewise, cort icosteroids

reduce cerebral edema in tumora l diseases proba- bly by decreasing microvascular permeabi l i ty

whereas they are ineffective in cerebral edemas due

to o ther causes [41]. This behavior, again, is compat - ible with the per ipheral vessel pheno type of tum- ora l /per i tumoral microvessels, as cort icosteroids decrease per ipheral microvascular permeabi l i ty [42].

In conclusion, the results of our study show that per i tumora l microvessel endothel ial cells have al-

131

tered anatomical and most p robably functional

characteristics. This modif ied pheno type may pro-

vide a new clue for the clarification of t umor growth or t umor spread pathogenesis and, hopefully, for di-

agnostic and therapeut ic interventions.

Acknowledgements

The authors would like to gratefully acknowledge

the technical assistance of M. Luisa Vasques. Sup-

por ted by JNICT.

References

1. Goldman E: Vitalfarbung am Zentralnervensystem. Beitrag zur Physiopathologie des Plexus Choroideus und der Hirn- haute. Abh Preuss Akad Wiss Phys-Math 1: 1-60, 1913

2. Geig NH: Brain tumors and the blood-brain barrier. In: Neuwelt EA (ed) Implications of the Blood-Brain Barrier and Its Manipulation, Vo112, Plenum Medical Book Com- pany, New York and London 77-106, 1989

3. Reese T, Karnovsky M: Fine ultrastructural localization of a blood-brain barrier to exogenous peroxidase. J Cell Bio134: 207-217, 1967

4. Wahl M, Unterberg A, Baethmann A, Schilling L: Media- tors of blood-brain barrier dysfunction and formation of vasogenic brain edema. J Cereb Blood Flow Metab 8: 621- 634, 1988

5. Klatzo I: Neuropathological aspects of brain edema. J Neu- ropathol Exp Neurol 26: 1-14, 1967

6. Cerv6s-Navarro J, Kannuki S, Nakagawa Y: Blood-brain barrier. Review from morphological aspect. Histol Histo- pathol 3: 203-213, 1988

7. Bruce J, Criscuolo GR, Merrill M J: Vascular permeability induced by protein product of malignant brain tumors: in- hibition by dexamethasone. J Neurosurg 67: 880-884, 1987

8. Criscuolo GR, Merrill M J, Oldfield EH: Further character- ization of malignant glioma-derived vascular permeability factor. J Neurosurg 69: 254-262, 1988

9. Quindlen EA, Bucher AP: Correlation of tumor plasmino- gen activator with peritumoral cerebral edema. J Neurosurg 66: 729-733, 1987

10. Robert AM, Godeau G, Miskulin M: Mechanism of action of collagenase on the permeability of the blood-brain bar- rier. Neurochem Res 2: 449-455, 1977

11. Senger DR, Galli SJ, Dvorak HF: Tumor cells secrete a vas- cular permeability factor that promotes accumulation of as- cites fluid. Science 219: 983-985,1983

12. Senger DR, Perruzzi CA, Feder J: A highly conserved vas- cular permeability factor secreted by a variety of human and rodent tumor cell lines. Cancer Res 46: 5629-5632, 1986

132

13. Ohnishi T, Sher PB, Posner JB, Shapiro WR: Capillary per- meabjlity factor secreted by malignant brain tumor. J Neu- rosurg 72: 245-251, 1990

14. Levin V, Freeman M, Landahl H: The permeability charac- teristics of brain adjacent to intracerebral rat tumors. Arch Neurol 32: 785-791, 1975

15. Long DM: Capillary ultrastructure and the blood-brain bar- rier in human malignant brain tumors. J Neurosurg 32:127- 144, 1970

16. Long DM: Capillary ultrastructure in human metastatic brain tumors. J Neurosurg 51: 53-58, 1979

17. Torack R: Ultrastructure of capillary reaction to brain tu- mors. Arch Neurol 5: 416-428, 1961

18. Gaetani P, Baena RR, Marzatico E Lombardi D, Knerich R, Paoletti P: 'Ex vivo' release of eicosanoid from human brain tissue: its relevance in the development of brain edema. Neurosurg 28: 853-858, 1981

19. Hiesiger EM, Voorhies RM, Basler GA: Opening the blood- brain and blood-tumor barriers in experimental rat brain tu- mors: the effect of intracarotid hyperosmolar mannitol on capillary permeability and blood flow. Ann Neurol 19: 50- 59, 1986

20. Pappius HM: Dexamethasone and local cerebral glucose utilization in freeze-traumatized rat brain. Ann Neurol 12: 157-162, 1982

21. Yamada K, Ushio Y, Hayakawa T, Arita N, Yamada N, Mo- gami H: Effects of methylprednisolone on peritumoral brain edema. J Neurosurg 59: 612-619, 1983

22. Azevedo I, Castro-Tavares J, Garrett J: Ultrastructural changes in blood vessels of perinephritic hypertensive dogs. Blood Vessels 18: 110-119, 1981

23. Sarmento A, Soares-da-Silva R Albino-Teixeira A, Azeve- do I: Effects of denervation induced by 6-hydroxydopamine on cell nucleus activity of arterial and cardiac cells of the dog. J Auton Pharmacol 7: 119-126, 1987

24. Weibel ER: Stereological techniques for electron micro- scopic morphometry. In: Hayat MA (ed) Principles and Techniques of Electron Microscopy, Vol III, Van Nostrand Reinhold Company, New York, 1973

25. Constantinides P: Functional electronic histology. Elsevier Scientific Publishing Company, Amsterdam, 1974

26. Ghadially FN: Ultrastructural pathology of the cell. Butter- worths, London, 1975

27. Branco D, Albino-Teixeira A, Azevedo I, Osswald W: Struc- tural and functional alterations caused at the extraneuronal level by sympathetic denervation of blood vessels. Naunyn Schmiedeberg's Arch Pharmacol 326: 302-312,1984

28. Connoly DT, Olander JV, Heuvelman D, Nelson R: Human vascular permeability factor. Isolation from U937 cells. J Biol Chem 264: 20017-20024, 1989

29. Whittle IR, Ironside JW, Piper IR, Miller JD: Neuropath-

ological and neurophysiological effects of interstitial white matter autologous and non-autologous protein containing solutions: further evidence for a glioma derived permeabil- ity factor. Acta Neurochir (Wien) 120: 164-174, 1993

30. Keck PJ, Hauser SD, Krivi G, Sanzo K: Vascular permeabil- ity factor, an endothelial cell mitogen related to PDGE Sci- ence 246: 1309-1312, 1988

31. Kim KJ, Li B, Winder J, Armanini M, Gillett N, Phillips HS, Ferrara N: Inhibition of vascular endothelial growth factor- induced angiogenesis supresses tumor growth in vivo. Na- ture 362: 841-844, 1993

32. Brem S, Tsanaclis AMC, Gately S, Gross JL, Herblin WF: Immunolocalization of basic fibroblast growth factor to the microvasculature of human brain tumors. Cancer 70: 2673- 2680, 1992

33. Li VW, Folkerth RD, Watanabe H, Yu C, Rupnick M, Barnes P, Scott RM, Black PMcL, Sallan SE, Folkman J: Mi- crovessel count and cerebrospinal fluid basic fibroblast growth factor in children with brain tumors. Lancet 344: 82- 86, 1994

34. Plate KH, Breier G, Weich HA, Risau W: Vascular endo- thelial growth factor is a potential tumor angiogenesis factor in human gliomas in vivo. Nature 359: 845-848, 1992

35. Baba T, Chio C-C, Black KL: The effect of 5-1ipoxygenase inhibition on blood-brain barrier permeability in experi- mental brain tumors. J Neurosurg 77: 403-406, 1992

36. Black KL, Hoff JT, McGillicuddy JE, Gebarski SS: In- creased leukotriene C4 and vasogenic edema surrounding brain tumors in humans. Ann Neuro119: 592-595, 1986

37. Unterberg A, Schmidt W, Wahl M, Baethmann A: Role of leukotrienes as mediator compounds in brain edema. Adv Neuro152: 211-214, 1990

38. Unterberg A, Schmidt W, Wahl M, Ellis EE Marmarou A, Baethmann A: Evidence against leukotrienes as mediators of brain edema. J Neurosurg 74: 773-780, 1991

39. Feurstein G: Leukotrienes and the cardiovascular system. Prostaglandins 27: 781-802, 1984

40. Piper PJ: Formation and actions of leukotrienes. Physiol Rev 64: 744-761, 1984

41. Nelson SR, Dick AR: Steroids in the treatment of brain ede- ma. In: Azarnoff DL (ed) Steroid Therapy, W.B. Saunders Co., Philadelphia 313-324, 1975

42. Haynes RC: Adrenocorticotropic hormone: adrenocortical steroids and their synthesis and actions of adrenocortical hormones. In: Goodman Gilman A, Rall TW, Nies AS, Tay- lor P (eds) Goodman and Gilman's The Pharmacological Basis of Therapeutics, Pergamon Press, New York 1431- 1462,1990

Address for offprints: 1. Azevedo, Institute of Pharmacology and Therapeutics, Medical Faculty, 4200 Porto, Portugal