Cellular proliferation and infiltration following interstitial irradiation of normal dog brain is altered by an inhibitor of polyamine synthesis

Int. J. Radiation

Oncology

Biol.

Pergamon

Phys.,

Vol. 32. No. 4, pp. 103% 1045, 1995 Copyright 0 1995 Elsevier Science Ltd Printed in the USA. All tights reserved 036@3016/95 $9.50 + .OO

0360-3016(95)00830-5

l

Biology Original Contribution CELLULAR

PROLIFERATION AND INFILTRATION FOLLOWING IRRADIATION OF NORMAL DOG BRAIN IS ALTERED INHIBITOR OF POLYAMINE SYNTHESIS

INTERSTITIAL BY AN

JOHN R. FIKE, PH.D.,**+)$ GLENN T. GOBBEL, D.V.M., PH.D.,**” DEAN CHOU, BAS P. L. WIJNHOVEN, B.S.,* MATTIA BELLINZONA, M.D.,* MINORU NAKAGAWA, AND

B.A.,* M.D.*

M. SEILHAN*

THERESA

*Brain Tumor ResearchCenter, Departmentof NeurologicalSurgery; +Departmentof RadiationOncology; *Departmentof Radiology; ‘Departmentof Neurology, School of Medicine, University of California, San Francisco,CA 94143-0520 Purpose: The objectives of this study were to quantitatively define proliferative and intiltrative cell responses after focal ‘%I irradiation of normal brain, and to determine the effects of an intravenous infusion of LYdifluoromethylomithine (DFMO) on those responses. Methods and Materials: Adult beagle dogs were irradiated using high activity ‘-1 sources. Saline (control) or DFMO (150 mg/kg/day) was infused for 18 days starting 2 days before irradiation. At varying times up to 8 weeks after irradiation, brain tissues were collected and the cell responses in and around the focal lesion were quantified. Immunohistochemical stains were used to label astrocytes (GFAP), vascular endothelial cells (Factor VIII), polymorphonuclear leukocytes (PMNs; MAC 387) and cells synthesizing deoxyribonucleic acid (DNA) (BrdU). Cellular responses were quantified using a bistomorphometric analysis. Results: After radiation alone, cellular events included a substantial acute inflammatory response followed byeased BrdU labeling and progressive increases in numbers of capillaries and astrocytes. cu-Difluoromethylomithine treatment significantly affected the measured cell responses. As in controls, an early inflammatory response was measured, but after 2 weeks there were more PMNs/unit area than in controls. The onset of measurable BrdU labeling was delayed in DFMO-treated animals, and the magnitude of labeling was significantly reduced. Increases in astrocyte and vessel numbers/mm’ were observed after a 2-week delay. At the site of implant, astrocytes from DFMO-treated dogs were significantly smaller than those from controls. Conclusions: There is substantial cell proliferation and infiltration in response to interstitial irradiation of normal brain, and these responses are significantly altered by DFMO treatment. Although the precise mechanisms by which DFMO exerts its effects in this model are not known, the results from this study suggest that modification of radiation injury may be possible by manipulating the response of normal cells to injury. Normal brain, Interstitial irradiation, Leukocytes, Immunohistochemistry,

Astrocytes, Morphometric

Polyamines, analysis.

INTRODUCTION

DFMO,

Endothelial

cells, Microvasculature,

tissue injury, focal irradiation is being used in addition to conventional fractionated teletherapy (23, 45). Although this approach has been shown to be effective, 40-50% of patients so treated require surgical and/or prolonged steroid therapy to manage treatment-induced injury to normal brain tissues(33,45). It is of importance, therefore, to develop new strategies to reduce injury to dose-limiting tissues of normal brain.

In the management of malignant brain tumors, radiation therapy is the single most effective treatment after surgical resection (54). However, the maximum radiation dose that can be used is limited by the tolerance of normal tissues surrounding the tumor (35, 48). In an effort to optimize the use of radiation while minimizing normal

Reprint requeststo: John R. Fike, Ph.D., Brain Tumor Research Center, Box 0520, University of California, San Francisco, San Francisco, CA 94143. E-mail address: [email protected]. Acknowledgemenrs-Supported by NIH Grant CA 13525 and by a gift from the Phi Beta Psi Sorority. We would like to thank

Dr. Keith Weaver for supplying the lz51 sources and Dr. Richard L. Davis for his helpful comments regarding the histopathology of radiation injury. Accepted for publication 20 January 1995.

1035

I. J. Radiation Oncology

1036

l

Biology 0 Physics

There are a number of biochemical changes observed after central nervous system (CNS) injury, some of which are associatedwith polyamine biosynthesis (9, 14,41,42, 46). The first step in the biosynthetic pathway of the polyamines, the conversion of omithine to putrescine (PU), is catalyzed by omithine decarboxylase (ODC), and increases in ODC activity and PU levels are commonly observed after CNS damage (9, 41, 42). a-Difluoromethylomithine (DFMO) inhibits ODC activity and has been shown to be effective in reducing the severity of certain types of CNS injury ( 12, 19, 39, 46). Recently we proposedthat polyamines might play a role in radiation-induced damageof the CNS and that manipulation of polyamine levels by inhibiting ODC activity with DFMO might provide a way to modulatethe extent of injury (12, 19). Those studiesshowed that focal ‘25Iirradiation of normal brain increasedtissueand cerebrospinalfluid (CSF) levels of PU. In contrast, an intravenous infusion of DFMO lowered PU levels in both brain tissuesand the CSF of irradiated dogs, and overall volume of brain injury was reduced ( 12, 19). In our studiesthree types of tissuedamage, vasogenic edema,necrosis, and blood-brain barrier (BBB) breakdown, were measuredand shown to be affected by DFMO. We suggestedthat the effect of DFMO on edema volume was mediatedthrough alterations in vascular permeability (12), which agreed with studies performed in other in viva systems(46, 52). How DFMO infusion affected the extent of necrosis and the extent of BBB breakdown was less clear, but it was suggestedthat alterations in cell responseafter focal injury might be involved (12). Considerabledata exist suggestingthat polyamines play a critical role in cell proliferation and that DFMO treatment is cytostatic (27, 47). Therefore, in our previous studies, DFMO treatment may have affected lesionsize by temporarily inhibiting mitosis and allowing cells to recover from radiation-induced injury. Alternatively, cell movement into and out of the lesion site, with concomittant releaseof cytotoxic or growth factors, may have inhibited the normal regenerative responses.Becausethe alteration of the normal cell responsecould have a significant impact in the overall development and magnitude of a focal radiation lesion, we were interested in quantifying such an effect. Using tissue samplesobtained during our previous radiographic studies (12, 19), we have now been able to address,quantitatively, specific cell responsesafter focal irradiation of dogs treated with or without DFMO. METHODS

AND MATERIALS

Animals Male adult purebred beagle dogs 1- 1.5 years old were used. Dogs were housed and cared for in accordance with the United States Department of Health and Human Services Guide for the Care and Use of Laboratory Animals

’ Cadd 1 Ambulatory St. Paul, MN.

Infusion Pump, Pharmacia Deltec Inc.,

Volume 32. Number 4, 1995

and institutional guidelines for care and handling of laboratory animals. Dogs were randomly assignedto receive radiation plus saline infusion (controls), or radiation combined with DFMO infusion. cr-Difluoromethylomithinetreated dogs and controls were part of previously reported studies (12, 19). Specifics regarding saline and DFMO infusion have been previously described (19). An indwelling vena caval line was surgically implanted, routed under the skin to a point between the scapulaewhere it was exteriorized and connected to a small, battery operated infusion pump.’ The pump was placed in a custom made vest worn by the animal. cY-Difluoromethylomithine’ was dissolved in physiological saline, sterilized, and loaded into the infusion pump reservoir. a-Difluoromethylomithine dose was 150 mg/kg/day. A total of 8 ml of DFMO or saline solution was infused during each 24-h period. Infusions were started 2 days before ‘251implantation, and were continued during the irradiation and for 14 days after removal of the radiation source (12, 19). The surgical procedures for implantation and removal of 12’1sources, along with calibration, dosimetry, and treatment parameters, have been described in detail (12, 19, 53). In all dogs, a dose of 20 Gy was delivered to a reference point 0.75 cm away from the source; dose rate at the reference point averaged 0.45 Gy/h. Radiation dose to a tissueregion in the contralateral hemispherecomparable to the site of implantation was calculated to be 2.03.0 Gy. Total irradiation time was 40-44 h. In the control group (12, 19), tissueswere collected at 1, 2,4,6, and 8 weeks after irradiation; samplesizes were 2,2,5,5, and 2, respectively. Tissueswere collected from DFMO-treated dogs (19) at 2, 4, and 6 weeks; sample sizes were 2,2, and 3, respectively. Ninety minutes before euthanasia, animals were infused intravenously with 5bromo-2’-deoxyuridine? (BrdU) to label cells synthesizing DNA. BrdU dose was 20 mg/kg body weight and total infusion time was 30 min. Animals were euthanized with an overdose of barbiturate and brains were removed using a Stryker autopsy saw. Transverse sections (5-7 mm thick) of whole brain were cut, and the sections containing the radiation lesion and corresponding tissue from the contralateral hemisphere were placed into separate plastic holders and immersed in a 10% buffered formalin solution for 1 week. The tissue sampleswere dehydrated and embedded in paraffin using routine histopathologic methods. Six pm thick sectionswere cut from the paraffin blocks and placed on glass microscopic slides previously treated with polylysine. To assess morphologic changes, one section from each block was stainedwith hematoxylin and eosin (H & E). Immunohistochemistry GlialjbrilEary acidic protein (GFAP; Fig. lc). Deparaffinized and rehydrated sections were treated with

* Marion

Merrell

Dow Research Institute,

Cincinnati,

OH.

Cellular

proliferation

and infiltration

0 J. R. FIKE

ef al.

03

(4

(4 Fig. 1. Photomicrographs of tissues from normal or irradiated brain showing the areas used for morphometric analysis and examples of specific cell staining. (a) Irradiated white matter 2 weeks after removal of the ‘25I source, showing the lesion (L) and perilesion (P) areas used in the morphometric analyses. Hematoxylin and eosin, 100~ magnification. (b) Irradiated white matter 4.weeks after removal of the ‘? source. The lesion area (L) is more organized, while the perilesion (P) shows some tissue disruption and signs of resolving vasogenic edema (arrowheads). Normal-appearing tissues (N) can be seen farther away from the lesion. Hematoxylin and eosin, 100X magnification. (c) Tissue from the left hemisphere of an irradiated dog stained with GFAP antibody to show astrocyte cell bodies and their processes. Gill’s hematoxylin counterstain, 320~ magnification. (d) Blood vessels. darkly stained with Factor VIII antibody, in the tissue of a dog irradiated 4 weeks earlier with 12’I. Gill’s hematoxylin counterstain, 320X magnification. (e) Polymorphonuclear leukocytes, surrounding a blood vessel in irradiated white matter, were darkly stained with MAC 387. Some glial-cell nuclei are lightly stained with the Gill’s hematoxylin counterstain, 320~ magnification. (f) Darkly stained BrdU-positive nuclei appear in the lesion area of a dog irradiated 4 weeks earlier with lz51. Gill’s hematoxylin counterstain, 320~ magnification.

1038

1. J. Radiation Oncology 0 Biology 0 Physics

0.1% protease (Type XXIV) in 0.05 M Tris-buffered saline (TBS; pH 7.4) for 2 min, and incubated for 60 min with rabbit anti-cow GFAP antibody4 diluted 1:200 with phosphate-buffered saline (PBS). Sections were rinsed in PBS and incubated with alkaline phosphatase-conjugated swine anti-rabbit IgG4 diluted 1:40 with TBS for 60 min; bound antibody was detected using the alkaline phophatase substrate Kit II5 that stained GFAP positive cells brown. Microvasculature (Fig. Id). Deparaffinized and rehydrated sections were first treated with 0.1% protease for 2 min. Then they were incubated overnight at 4°C with rabbit anti-human Factor VIII antibody,4 which specifically labels vascular endothelial cells. The antibody was diluted 1:25 with PBS containing 0.1% bovine serum albumin. Finally, the sections were treated with swine anti-rabbit IgG (alkaline phosphataseconjugated, 1:40)4; bound antibody was detected using the alkaline phosphatasesubstrate Kit IV that stained the endothelial cells blue. Polymorphonuclear leukocytes (PMNs; Fig. le). Deparaffinized and rehydrated sections were treated for 10 min with a mixture of 0.25% trypsin and 0.25% protease and incubated overnight at 4°C with 10% normal horse serum in PBS. Sections were treated for 2 h with mouse anti-human myeloid/histiocyte antigen antibody (MAC 387)4 (13) that was diluted 1:250 with PBS containing 10% normal horse serum. After washing, sections were incubated with biotinylated horse anti-mouse IgG’ diluted 1:200 in PBS containing 10% normal horse serum. The bound antibody was detected using an ABC Kit’ after treatment with 3% H202. Sections were developed using 3,3’-diaminobenzidine6 (DAB), which stained cells brown. Cells synthesizing DNA (Fig. lf). Tissue sections were denatured with 4N HCl for 10 min, digested with 0.1% protease for 30 min, and incubated for 60 min with mouse anti-BrdU antibody’ (IU-4) (21) diluted 1:200 in PBS containing 5% normal goat serum. Sections were then exposed to gold-labeled anti-mouse IgG8 diluted 1:40 in PBS containing 0.1% bovine serum albumin for 60 min. Bound primary antibody was detected using a silver enhancement kit.* As a control for nonspecific staining during each procedure, tissue sections were prepared as described above, and normal serum from the species of origin of the primary antibody was substituted for the primary antibody. Morphometric analysis Quantitative analysis was performed using a modified Chalkley technique for quantitative tissue analysis (6). A 5 x 5 mm indexing grid was mounted in an ocular lens and a stage micrometer used to calibrate the size of the

’ Sigma,Inc., St. Louis, MO. 4Dako Corporation,Carpinteria,CA. ’ Vector Laboratories,Inc., Burlingame,CA.

Volume 32, Number 4, 1995

grid at a magnification of 320x. Data were expressed as either number of labeled cells or vessels per unit area (mm*) or percent area occupied by vesselsor specific cell types. For each antibody, the total number of cells or vessels positively stained were determined by counting multiple grids randomly placed on the tissue section. The total number of grids needed for a sufficient sample was establishedfor each cell type by determining the number of labeled cells or vessels/mm*as a function of total grids counted up to 500. In this study 100 (GFAP), 150 (BrdU), or 250 (Factor VIII, MAC 387) grids were sufficient. The total number of positive staining cells/mm* did not change when more grids were counted. For percent area measurements, grids were projected onto tissue sections and the number of grid intersections that overlaid a specific cell type or blood vessel were counted. The area occupied was calculated as described by Chalkley (6). To determine vessel and astrocyte size, the percent area measurements were divided by the measurementsof vessel or cell numbers/mm*. The tissue regions analyzed in this study could be distinctly defined microscopically (Fig. la and lb). The regions included “lesion” encompassing all areas within the outer border of frank necrosis (1, 2, and 4 weeks) or previously necrotic tissue (6 and 8 weeks); ‘ ‘perilesion,’ ’ the tissues 1-2 mm beyond the lesion border; and tissuesfrom the contralateral hemisphere, which in size and location were similar to lesion + perilesion. Statistics For each posttreatment time, arithmetic means, standard deviations (SD) and standard errors of the mean (SEM) were calculated for each treatment group. Analysis of variance (ANOVA) was used to compare the effects of treatment (DFMO vs. saline) and time after irradiation on cell or vessel labeling. To determine whether any treatment effect was dependent upon time after irradiation, an interactive term including both treatment and time was initially included in the model and eliminated if it did not approach significance (p > 0.10). To account for multiple comparisons, Scheffe’s F-test was used for posthoc comparisons between means. Significance was indicated by p < 0.05; apparent trends (p < 0.10) were also noted.

RESULTS The general histopathologic character of the radiation lesions was not different between controls and dogs treated with DFMO (19). Briefly, l-2 weeks following irradiation there was a focus of coagulation necrosis at the site of the ‘*‘I source, surrounded by a narrow rim of reactive tissue. The latter region was characterized by a variety of vascular changes, ranging from fibrinoid necrosis of capillary walls to normal-appearing and apparently

’ Sigma,Inc., St. Louis, MO. ’ Caltag, So. San Francisco, CA. * Amersham,Inc., Arlington Heights,IL.

Cellular proliferation and infiltration 0 J. R. FKE et al.

vessels. Swollen axis cylinders were noted along with considerable tissue rarefaction, presumably indicative of edema. Inflammatory cells, primarily PMNs, were numerous in the necrotic area and were also evident in the reactive tissues immediately adjacent. Tissues distal to the reactive rim appeared only slightly abnormal with some tissue elements separated from one another, again representing interstitial edema. In some cases, minimal changes in some axis cylinders (e.g., swollen axons) were noted away from the reactive rim. Four weeks after irradiation there was some necrosis still visible, and the reactive rim was more extensive and contained monocytes, many of which appeared to be filled with lipid-like material. Endothelial cell hyperplasia was noted, and there still were signs of interstitial edema. At 6 and 8 weeks, there were no apparent signs of necrosis, and the dead tissues were entirely replaced with new vessels, astrocytes, and mononuclear cells that appeared to be macrophages and/ or reactive microglia. Interstitial edema was still visible but was reduced relative to earlier time points. In the contralateral hemisphere of controls the number of BrdU-labeled cells significantly increased from l-2 weeks (p < 0.05), but at the maximum value averaged only 2.8 -+ 0.9 cells/mm’ (mean -+ SD) (Fig. 2a). At 4 weeks, average number of BrdU labeled cells/mm* in the contralateral hemisphere tended to be lower than at 2 weeks @ < 0.10; Scheffe posthoc), and at 6 weeks the values were significantly lower (p < 0.05; Scheffe posthoc). In the right implanted hemisphere of controls, there were substantial increases in labeling within the lesion and perilesion regions at 2 weeks when compared to the contralateral hemisphere; maximum labeling was observed 4 weeks after irradiation (Fig. 2b). In tissues from DFMO-treated dogs there was signficantly less BrdU labeling in both the lesion 0, < 0.01; ANOVA) and perilesion regions (p < 0.05; ANOVA) compared to controls (Fig. 2b). There was no increase in the number of BrdU-labeled cells observed until 4 weeks after irradiation, and at that time, the absolute numbers of labeled cells/mm* in the lesion area were significantly less than controls @ < 0.05; Scheffe posthoc). There was also a significant effect of DFMO on BrdU labeling in the contralateral hemisphere that was dependent upon time after irradiation (p < 0.01; ANOVA). Two weeks after irradiation, but not at 4 or 6 weeks after irradiation, there was a significant difference in BrdU labeling within the contralateral hemisphere between controls and DFMO-treated animals @ < 0.05; Scheffe posthoc). In addition, while labeling decreased from 2-6 weeks in tissues from the contralateral hemisphere of controls, the labeling in tissues from DFMO-treated animals tended to increase from 2-4 weeks (p < 0.10; Scheffe posthoc) and then fell significantly from 4-6 weeks @ < 0.05; Scheffe posthoc). Factor VIII antibody was used to label vascular endothelial cells and outline the microvasculature within in-adiated and nonirradiated brain tissue. In the contralateral hemsiphere of controls, the average number of vessels/ patent

0

2

4

6

Weeks After Irradiation (a)

60-

0

6

8

Wgeks Afte; Irradiation (W

0

6

6

WGeks Atte”r Irradiation b-2 Fig. 2. (a) Number of BrdU-positive cells/mm2 vs. time, in tissues from the contralateral hemisphere of dogs treated with radiation alone (0) and radiation plus DFh40 (A). In most cases, each datum point represents a mean of the measurements from two or more dogs; error bars are SEM. (b and c) Number of BrdU-positive cells/mm* in the lesion (b) and perilesion tissues (c) vs. time in dogs treated with radiation alone (0) and radiation plus DFMO (A). Each datum point represents a mean of the measurements from two or more dogs; error bars are SEM.

I.

2

J.

RadiationOncology 0

4

6

8

0 Physics

Biology

10

Weeks After Irradiation

Volume

32, Number

4, 1995

density vs. time curves were not significantly different (p > 0.10; ANOVA). Similar to controls, in the lesion and perilesion areas of DFMO-treated dogs mean vessel size stayed realtively constant over the entire follow-up period, averaging 2.9 + 0.7 and 2.5 + 0.4 mm2 (X 10m4; mean t SEM, n = 3), respectively. The mean number of GFAP-positive cells (i.e., astrocytes) in the contralateral hemispheredid not change as a function of time in controls, averaging 156.2 2 7.0 cells/mm2 (mean ? SEM, n = 5) over the entire followup period. The number of GFAP-labeled cells in the lesion area tended to increase (p < 0.10; ANOVA) with time after irradiation in controls, but did not reach the levels observed in the contralateral hemisphere (Fig. 4). Astrocyte number in the perilesion area was higher than that seen in the lesion area and stayed relatively constant after 2 weeks (Fig. 4). Astrocyte size did not change significantly with time. However, when ANOVA was used to examine whether astrocyte size differed depending upon region, it was determined that cell size

40 5 .

30UJ

5 0

20 -

.-$ 0

2

4

6

8

10

-5 2

10 -

2b

o-

Weeks After Irradiation

Fig. 3. Numberof vessels/m&in the Iesion(top) andperilesion tissues(bottom) vs. time in dogstreated with radiation alone (0) andradiationplus DEMO (A). “Normal” capillary density (mean+ SEM) measuredin tissuesfrom the contralateralhernisphereof controls is shown as the gray bar at the top of the plots. Each datumpoint representsa meanof the measurements from two or moredogs; error barsare SEM. mm2 did not change with time, averaging 76.0 +_ 3.0 vessels/mm’ (mean 2 SEM, iz = 5). Similarly, mean vessel size in the contralateral hemisphere of controls was stable with time, averaging 1.7 2 0.2 mm2 (X 10m4;mean + SEM, n = 5). After 2 weeks there was a progressive and significant (p < 0.01; ANOVA) increase in vessel density in the lesion area of controls, and normal vessel density, determined by values obtained from the contralatera1hemisphere, was approached by 8 weeks (Fig. 3). A similar pattern was observed in the perilesion area, although at the early time points the absolute number of vesselsper unit area was higher (Fig. 3). In the lesion and perilesion areas of controls, vessel size stayed relatively constant over the entire follow-up period, averaging 2.5 + 0.1 and 2.4 ? 0.3 mm2 (X 10p4; mean 2 SEM, n = 5), respectively. After DFMO treatment the general response in vessel number as a function of time paralleled that seen in controls. Although there appeared to be a 2-week delay in the onset of the response (Fig. 3) the shapesof the cell

I

0

I

2

4

6

0

10

Weeks After Irradiation

0

2

4

6

8

10

Weeks After Irradiation

Fig. 4. Numberof GFAP-positive cells (astrocytes)/rnn?in the lesion (top) and perilesiontissues(bottom) vs. time in dogs treated with radiation alone(0) andradiation plusDFMO (A). The “normal” numberof astrocytes/nnn’(mean+ SEM), measuredin tissuesfrom the contralateralhemisphereof controls, is shown as the gray bar in the perilesiondata. Each datum point representsa meanof the measurements from two or more dogs; error bars are SEM.

Cellular

proliferation

and infiltration

in the lesion area was significantly greater than in the contralateral hemisphere (Fig. 5; p < 0.001; Scheffe posthoc) or the perilesion (data not shown; p < 0.01). The mean number of astrocytes in the contralateral hemisphere of dogs treated with DIM0 plus radiation stayed relatively constant with time, and was not different from controls, averaging 146.3 -C 11.7 cells/mm*. Astrocyte number in the lesion area increased with time after irradiation in tissues from DFMO-treated dogs, but like controls, did not reach the levels observed in the nonirradiated hemisphere (Fig. 4). Although at all timepoints the average number of astrocytes was lower than controls, at no point were those differences significant (p > 0.10; ANOVA). Astrocyte number in the perilesion area was higher than that seen in the lesion area and stayed constant after 2 weeks (Fig. 4). Unlike controls, there was no increase in size of GFAP-positive cells within the lesion of DFMO-treated animals (Fig. 5). The size of GFAP-positive cells in the perilesion area of tissues from DFMO-treated animals was also comparable to that seen in the contralateral hemisphere (data not shown). The number of PMNs/mm2 determined using MAC 387 staining was compared to PMNs counted on H & E stained sections. Polymorphonuclear leukocytes were identified on the latter sections based on nuclear morphology. There were no differences in number of PMNs/mm* between the two staining methods (data not shown). Because of limited numbers of tissue sections from some individual animals, PMN density was determined in those situations using H & E stained sections. In the contralatera1 hemisphere of controls, the number of PMNs averaged about 1 cell/mm*. The number of PMNs within the right implanted hemisphere was substantially increased 1 week following irradiation, and decreased significantly with time @ < 0.001; ANOVA) (Fig. 6). Polymorphonu-

COll1K.l

DFMO

DFMO Lesion

COlltrOl L&Oll

Fig. 5. Astrocyte cell size in the lesion area as a function of treatment. Each bar represents the mean of all evaluable dogs from a given treatment: n = 13, 16, 5, and 7 for lesion area of controls, corresponding area in the contralateral hemisphere of controls, lesion area of DEMO-treated dogs, and corresponding area in the contralateral hemisphere of DPMO-treated dogs, respectively. Error bars represent SEM.

1041

0 J. R. FIKE et al.

clear leukocyte number/mm* continued to decrease, reaching baseline by the sixth week. In the perilesion area, PMN density was significantly elevated relative to the contralateral hemisphere, but was less than that within the lesion area. Polymorphonuclear leukocyte density within the perilesion decreased with time in a similar manner to that seen in the lesion area, and baseline values were observed 8 weeks after irradiation (Fig. 6). In tissues from DFMO-treated dogs, the number of PMNs/mm2 in the contralateral hemisphere was not different from controls. Similar to controls, the number of PMNs within the lesion of DFMO-treated dogs was highest soon after irradiation and thereafter continuously and significantly declined @ < 0.001; ANOVA). However, DFMO treatment did significantly alter the PMN response to irradiation (p < 0.05; ANOVA); the average number of PMNs within the lesion of DFMO-treated animals was consistently higher than that of controls, and these differences were significant at 2 weeks after irradiation @ < 0.01; Scheffe posthoc), though not at 4 or 6 weeks.

DISCUSSION The histologic effects of radiation on normal brain tissues have been studied by numerous investigators, and the morphologic character described has generally been the same regardless of species or the type of radiation used to induce the damage (2-4, 28, 40, 44, 51, 57). However, most of those reports dealt with external beam radiation (2-4,44,5 1,57) rather than interstitial radiation (28, 40), and few reports have actually quantified the cellular response after irradiation. In light of recent studies showing that the extent of radiation-induced damage can be modified (12, 19), an understanding of specific cellular responses should help in the comprehension of such processes. In this study our primary goal was to quantify the cell response after irradiation with 1251.Tissue samples from dogs treated with radiation plus DFMO were more limited but were studied with the motive of gaining some insight as to how that compound modifies the radiation response of normal brain. The results of our study make it possible to outline some of the basic cellular events occurring after irradiation with high-activity ‘25I sources. Within 1 week of irradiation, there is coagulation necrosis and substantial acute inflammation in the region of the radioactive implant. Based on quantitative radiographic analyses of the animals used here (12, 19), the region of coagulation necrosis progressively enlarges, becoming maximum 4 weeks after irradiation. Thus, at least for the first 4 weeks following irradiation, the perilesion area represents a region that will ultimately become necrotic. Within this region, there is an initial decrease in vessel density, from approximately 75 vessels/mm* to about 5-lO/mm’, a loss of 85-90%. In comparison, the decrease in GFAP-positive cell (astrocyte) density is only from about 150 cells/mm2 to an average of about 75 cells/mm2, a loss of 50%. This finding suggests that

1042

I. J. Radiation

Oncology

0 Biology

0 Physics

Weeks After Irradiation

, loo%

:

E 80-

g

60-

P

: 40 -

Weeks After Irradiation

Fig. 6. Numberof polymorphonuclearleukocytes(PMNs)/mm* in the lesion (top) and perilesiontissues(bottom) vs. time in dogstreatedwith radiationalone(0) and radiationplus DFMO (A). In most cases,each datumpoint representsa meanof the measurements from two or more dogs;error barsare SEM. the vessels may be more sensitive to interstitial radiation than astrocytes, and it is consistent with preliminary in vitro data suggesting that cerebral endothelial cells are more sensitive to radiation damage than astrocytes (18). There may also be more rapid proliferation of endothelial cells in response to radiation injury, which could lead to a faster depletion of this cell type through mitotic death. Cell proliferation in response to our focal radiation lesion, as indicated by increased BrdU labeling in the lesion and surrounding tissue, is observed by 2 weeks (Fig. 2), increases from 2-4 weeks, and remains at a high level as long as 6 weeks after irradiation. The cellular proliferation is associated in time with the infiltration of new vessels and astrocytes into the previously necrotic region. By 6-8 weeks the previous region of coagulation necrosis is completely filled in with new vessels, astrocytes, macrophages, and microglial cells. The large numbers of PMNs observed l-2 weeks after irradiation with interstitial ‘*? demonstrate that there is an immediate and vigorous inflammatory infiltration into the necrotic lesion and the surrounding perilesional tissue

Volume

32, Number

4, 1995

(Fig. 6). The presenceof these cells suggeststhat the role of acute inflammation in the responseof normal brain to interstitial irradiation may have been previously underappreciated. Although Groothuis et al. (22) also reported the infiltration of polymorphonuclear and monocytic leukocytes 1 week after interstitial placement of low-activity ‘25Isources, the number of PMNs was described as small. The difference in PMN response between that and the current study could be related to the lower activity of the sourcesusedby Groothuis et al. (22) and the lesserdegree of focal injury 1 week after irradiation. The PMN response in our study most likely represents a response to acute radiation necrosis rather than placement of the interstitial implant into the brain. As a control, a nonradioactive source was placed into a dog for 48 h, and tissues were collected 2 weeks after seed removal. There were no detectable PMNs within or surrounding the region of seed placement (data not shown). The increase in BrdU labeling in both the lesion and perilesional regions of controls clearly indicates that DNA synthesis and, presumably, cell proliferation, are important in the evolution of these focal lesions. Other investigators have also shown increasedDNA synthesisin cells of normal brain after irradiation, but the irradiation parameters, dose, and time before DNA synthesis was initiated were different than that reported here (7). The fact that increased BrdU uptake was observed in the contralatera1hemisphere (Fig. 2), albeit at levels many times lower than that seen near the implant site, suggeststhat radiation-induced stimulation of cell proliferation can occur even after relatively low doses; for example, 2-3 Gy. Alternatively, it is possible that the overall morphologic and physiological disruption of the right hemisphere in some way stimulates a low level of proliferative activity. BecauseBrdU labeling increasesin parallel with increases in lesion size in the right implanted hemisphere, it is possible that acute cell death may stimulate the proliferation of surrounding cells. Although the biochemical stimuli responsible for cell proliferation in our study are unknown, the byproducts of cell death, as well as growth factors and cytokines, are possible candidates. Both blood vessel and astrocyte proliferation appeared to account for at least some of the BrdU labeling. Other studieshave reported blood vessel or endothelial cell proliferation following interstitial irradiation (10,22,28,53), but those studies did not quantify the extent of neovascularization as a function of time after irradiation. In our study, the increase in blood vessel density in the lesion area 2-8 weeks following irradiation is consistent with our earlier findings of increased vascularity measuredusing a quantitative imaging approach (20), and reflects new vessel growth rather than an increase in vessel size. Although the size of the vessels within the lesion was generally larger than vessels in the contralateral hemisphere, and although vasodilation and other changes in the morphology of blood vesselsare commonly observed after irradiation of the brain (3, 5, 11, 43, 44, 51), the increased size observed here was constant over the entire

Cellular

proliferation

and infiltration

follow-up period. Therefore, our data confirms and expands upon previous data showing an effective and rapid neovascularization after focal radiation injury in the brain. Astrocytic reactions that have been reported after radiation exposure include hypertrophy (3), increased GFAP expression (7), and alterations in the number of astrocytic cells (3, 7, 28, 40). Although the average size of the repopulating astrocytes in the lesion area of our samples was significantly larger than those in the contralateral hemisphere, the cell size remained constant, indicating that the increase in astrocyte density shown in Fig. 4 truly reflected change in number. In the perilesion area, 1 week after irradiation, the number of GFAP positive cells/mm2 was lower than in normal brain and tended to decrease further at 2 weeks (p < 0.10; Scheffe posthoc). Such loss of cell density is consistent with mitotic-linked death and/or migration of cells into the more highly damaged lesion area. Although mature astrocytes can migrate after being stimulated with a graft target (25), recent studies have shown that native astrocytes do not migrate when stimulated only by wounding (24). Whether a radiation insult provides an environment suitable for astrocytic migration is not clear. Regardless, during the follow-up period used here, astrocyte cell density never reached normal values. This is in contrast to vessel density, which neared normal values by 8 weeks. At least within the time frame studied here, the new, proliferating vessels were generally devoid of surrounding astrocytes. Contact of astrocytes with cerebral endothelial cells has been shown to play an important role in the induction and maintenance of the BBB (29). These differences may account for the reported delay in reestablishment of the BBB following interstitial irradiation (53). Because both vessel and astrocyte number increase at the same time that necrosis volume increases (12, 19), lesion size may represent a function of the difference between abortive cell division (i.e., mitotic-linked cell death) and productive cell division. It follows that a change in the rate of either of these processes could alter lesion size. Because DFMO is well recognized as a cytostatic agent and has also been reported to inhibit angiogensis (37, 47, 50), these actions might, in part, account for the apparent decrease in radiation-induced necrosis and BBB breakdown in DFMO-treated animals (12, 19). Despite the fact that the total number of tissue samples available from DFMO-treated dogs was small, our results show that DFMO acts as a cytostatic agent in our model of radiation-induced brain injury; DFMO treatment significantly diminished BrdU labeling, an index of cellular proliferation, following irradiation. In terms of vessel and astrocyte cell density, DFMO treatment resulted in an apparent delay in response of about 2 weeks, the total time of DFMO treatment. After 2 weeks the increase in vessel or cell density appeared to increase at a rate comparable to that seen in controls. Even in the nonirradiated hemisphere a 2-week delay was observed in terms of BrdU labeling.

0 J. R. FKE et al.

1043

In addition to its cytostatic effects, DFMO may also have other biological actions. In our study, DFMO treatment apparently affected astrocytic hypertrophy following irradiation (Fig. 5). Increased astrocyte size is a common neurocellular response to disease or injury and to exposure to radiation (3, 8). It has been suggested that in human brain astrocyte size is controlled by the growth factor transforming growth factor pl (TGF-,01) (8). This growth factor is present in low amounts in normal brain (56), but increases in TGF-@l mRNA are seen after brain wounding (34). Transforming growth factor p plays an important role in peripheral wound healing, affecting cell migration, cell proliferation and extracellular matrix (34, 38), and at least in some tissues, may mediate the remodeling of stromal extracellular matrix after irradiation (1). What, if any, role TGF-P plays in our model of radiation injury is unclear, but preliminary qualitative immunohistochemical analyses of frozen tissue from control animals demonstrate the presence of this growth factor after irradiation (data not shown). If DFMO counteracts the release or the effects of TGF-P and the resultant cellular reaction and proliferation, this could alter radiation damage as described above. Further studies are needed to determine the role of TGF-P and other growth factors in the evolution and regeneration of focal radiation injury. Because astrocytes are capable of phagocytosis under certain conditions (26), the difference in size between astrocytes from controls and dogs treated with DFMO may also represent impaired uptake of serum proteins or other compounds in cells from DFMO-treated animals. cu-Difluoromethylomithine has been shown to affect phagocytosis in macrophages (30), but similar studies in astrocytes have not been done. Alternatively, the smaller size of astrocytes in tissues from DFMO-treated dogs may simply be due to the fact that less material was availabe for phagocytosis because of the overall reduction in BBB disruption due to DFMO treatment. Finally, DFMO may affect the overall extent of radiation injury by altering the inflammatory reaction observed early after irradiation (Fig. 6). An inflammatory response is generally observed after focal brain injury, and PMNs and mononuclear cells of local and systemic origin have been shown to play a major role in the pathogenesis of tissue damage (16, 17, 55). In the brain, PMN infiltration has been associated with greater levels of tissue injury (32), and it is possible that such effects are due in part to the ability of PMNs to release eisosanoids, free radicals, and proteases, which can enhance edema formation and exacerbate injury (15, 16,31,49). In the present study there was a substantial inflammatory response early after irradiation in controls and an even greater response in DFMO-treated animals. The latter findings are somewhat surprising considering our previous observations that DFMO reduces radiation damage (12, 19). Studies by other investigators have shown that the function of inflammatory cells, notably macrophages and lymphocytes, are impaired by DFMO treatment (30, 36, 37); therefore it might be possible that the reduction in overall lesion

1044

I. J. Radiation

Oncology

0 Biology

0 Physics

volume observed after DFMO treatment is due, in part, to the impaired ability of inflammatory cells to produce or secrete substances that cause tissue damage. The increased PMN number in DFMO-treated animals might then represent a compensation for decreased cell function in those animals. More studies are required to determine whether that hypothesis is tenable with respect to focal brain lesions due to irradiation. Our study provides quantitative information about certain critical cellular phenomena involved in radiation response of normal brain. It remains to be seen whether similar changes occur in irradiated normal tissues surrounding a brain tumor. Regardless, hopefully the type of information found in this study can be

Volume

32, Number

4, 1995

used to formulate new hypotheses regarding radiation injury and its modification. After focal irradiation of normal brain, there is a sequence of cellular events involving both cell death and tissue regeneration. These events include inflammation, DNA synthesis and mitosis, neovascularization, and astrocytic infiltration. The inflammatory infiltrate is increased by inhibition of polyamine synthesis by DFMO while cellular proliferation is decreased by this same treatment. The precise mechanisms by which DFMO treatment affects these cell responses is not clear, but may be related to its recognized cytostatic effects and/or to unrecognized alterations of biologic mediators such as growth factors or cytokines.

REFERENCES 1. Barcellos-Hoff, M. H. Radiation-induced transforming growth factor p and subsequent extracellular matrix reorganization in murine mammary gland. Cancer Res. 53:38803886; 1993. 2. Burger, P. C.; Mahaley, M. S. J.; Dudka, L.; Vogel, F. S. The morphologic effects of radiation administered therapeutically for intracranial gliomas. A postmortem study of 2.5 cases. Cancer. 44:1256-1272; 1979. 3. Calvo, W.; Hopewell, J. W.; Reinhold, H. S.; Yeung, T. K. Time- and dose-related changes in the white matter of the rat brain after single doses of X rays. Br. J. Radiol. 61: 1043- 1052; 1988. 4. Caveness, W. F. Pathology of radiation damage to the normal brain of the monkey. NC1 Monogr. 46:57-76; 1977. 5. Caveness, W. F.; Carsten, A. L.; Roizin, L.; Schade, J. P. Pathogenesis of x-irradiation effects in the monkey cerebral cortex. Brain Res. 7:1-117; 1968. 6. Chalkley, H. W. Method for the quantitative morphologic analysis of tissues. J. Natl. Cancer Inst. 4:47-53; 1943. 7. Chiang, C. S.; McBride, W. H.; Withers, H. R. Radiationinduced astrocytic and microglial responses in mouse brain. Radiother. Oncol. 29:60-68; 1993. 8. da Cunha, A.; Jefferson, J. J.; Tyor, W. R.; Glass, J. D.; Jannotta, F. S.; Vitkovic, L. Control of astrocytosis by interleukin-1 and transforming growth factor-p1 in human brain. Brain Res. 631:39-45; 1993. 9. Dienel, G. A.; Cruz, N. E. Induction of brain omithine decarboxylase during recovery from metabolic, mechanical, thermal or chemical injury. J. Neurochem. 42:10531061; 1984. 10. Fike, J. R.; Cann, C. E.; Phillips, T. L.; Bernstein, M.; Gutin, P. H.; Turowski, K.; Weaver, K. A.; Davis, R. L.; Higgins, R. J.; DaSilva, V. Radiation brain damage induced by interstitial 125-I sources: A canine model evaluated by quantitative computed tomography. Neurosurgery 16:530537; 1985. K.; Higgins, R. J.; 11. Fike, J. R.; Cann, C. E.; Turowski, Chan, A. S. L.; Phillips, T. L.; Davis, R. L. Radiation dose response of normal brain. Int. J. Radiat. Oncol. Biol. Phys. 14:63-70; 1988. 12. Fike, J. R.; Gobbel, G. T.; Marton, L. J.; Seilhan, T. M. Radiation brain injury is reduced by the polyamine inhibitor o-difluoromethylomithine (DFMO). Radiat. Res. 138:99106; 1994. 13. Flavell, D. J.; Jones, D. B.; Wright, D. H. Identification of tissue histiocytes on paraffin sections by a new monoclonal antibody. J. Histochem. Cytochem. 35:1217-1226; 1987. 14. Gilad, G. M.; Gilad, V. H. Polyamines in neurotrauma.

15. 16.

17. 18. 19.

20. 21. 22.

23.

24. 25. 26. 27. 28.

Ubiquitous molecules in search of a function. Biochem. Pharmacol. 44:401-407; 1992. Giulian, D. Ameboid microglia as effecters of inflammation in the central nervous system. J. Neurosci. Res. 18:155171; 1987. Giulian, D.; Chen, J.; Ingeman, J. E.; George, J. K.; Noponen, M. The role of mononuclear phagocytes in wound healing after traumatic injury to adult mammalian brain. J. Neurosci. 94416-4429; 1989. Giulian, D.; Vaca, K. Inflammatory glia mediate delayed neuronal damage after ischemia in the central nervous system. Stroke. 24(suppl. 1):184-190; 1993. Gobbel, G. T.; Chan, P. H. Response of normal rat astrocytes and cerebral endothelial cells to radiation. (Abstr.) J. Neuro-Oncol. 15(suppl.):SlO; 1993. Gobbel, G. T.; Marton, L. J.; Lambom, K.; Seilhan, T. M.; Fike, J. R. Modification of radiation-induced brain injury by a-difluoromethylomithine (DFMO). Radiat. Res. 128:306315; 1991. Gobbel, G. T.; Seilhan, T. M.; Fike, J. R. Cerebrovascular response after interstitial irradiation. Radiat. Res. 130:236240; 1992. Gratzner, H. G. Monoclonal antibody to 5-bromo-and 5iododeoxyuridine: A new reagent for detection of DNA replication. Science 218:474-475; 1982. Groothuis, D. R.; Wright, D. C.; Ostertag, C. B. The effect of “‘1 interstitial radiotherapy on blood-brain barrier function in normal canine brain. J. Neurosurg. 67:895-902; 1987. Gutin, P. H.; Leibel, S. A.; Wara, W. M.; Choucair, A.; Levin, V. A.; Phillips, T. L.; Silver, P.; Da Silva, V.; Edwards, M. S. B.; Davis, R. L.; Weaver, K. A.; Lamb, S. Recurrent malignant gliomas: Survival following interstitial brachytherapy with high-activity iodine- 125 sources. J. Neurosurg. 67:864-873; 1987. Hatton, J. D.; Finkelstein, J. P.; U, H. S. Native astrocytes do not migrate de novo or after local trauma. Glia. 9: 1824; 1993. Hatton, J. D.; Garcia, R.; U, H. S. The migration of grafted rat astrocytes: Dependence on source/target organ. Glia. 5:251-258; 1992. Iacono, R. F.; Berrie, M. I.; Lascano, E. F. A triple staining procedure to evaluate phagocytic role of differentiated astrocytes. J. Neurosci. Meth. 39:225-230; 1991. Janne, J.; Alhonen, L.; Leinonen, P. Polyamines: From molecular biology to clinical applications. Ann. Med. 23:241259; 1991. Janzer, R. C.; Kleihues, P.; Ostertag, C. B. Early and late effects on the normal dog brain of permanent interstitial

Cellular

29. 30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41. 42.

proliferation and infiltration 0 J. R. FIKE et a/.

iridium- 192 irradiation. Acta Neuropathol. (Berlin) 70:9 l102; 1986. Janzer, R. C.; Raff, M. C. Astrocytes induce blood-brain barrier properties in endothelial cells. Nature 325:253-257; 1987. Kierszenbaum, F.; Wirth, J. J.; McCann, P. P.; Sjoerdsma, A. Impairment of macrophage function by inhibitors of ornithine decarboxylase activity. Infect. Immun. 55:24612464; 1987. Koch, A. E.; Cho, M.; Burrows, J. C.; Polverini, P. J.; Leibovich, S. J. Inhibition of production of monocyte/macrophage-derived angiogenic activity by oxygen free-radical scavengers. Cell Biol. Int. 16:415-425; 1992. Kochanek, P. M.; Hallenbeck, J. M. Polymorphonuclear leukocytes and monocyteslmacrophages in the pathogenesis of cerebral ischemia and stroke. Stroke 23:1367-1379; 1992. Leibel, S. A.; Gutin, P. H.; Wara, W. M.; Silver, P.; Larson, D. A.; Edwards, M. S. B.; Lamb, S. A.; Ham, B.; Weaver, K. A.; Barnett, C.; Phillips, T. L. Survival and quality of life after interstitial implantation of removable high-activity iodine- 125 sources for the treatment of patients with recurrent malignant gliomas. Int. J. Radiat. Oncol. Biol. Phys. 17:1129-1139; 1989. Lindholm, D.; Castren, E.; Kiefer, R.; Zafra, F.; Thoenen, H. Transforming growth factor-p1 in the rat brain: Increase after injury and inhibition of astrocyte proliferation. J. Cell Biol. 117:395-400; 1992. Marks, J. E.; Baglan, R. J.; Prasad, S. C.; Blank, W. F. Cerebral radionecrosis: Incidence and risk in relation to dose, time, fractionation and volume. Int. J. Radiat. Oncol. Biol. Phys. 7:243-252; 1981. Messina, L.; Spampinato, G.; Arcidiacono, A.; Malaguarnera, L.; Pagano, M.; Kaminska, B.; Kaczmarek, L.; Messina, A. Polyamine involvement in functional activation of human macrophages. J. Leuk. Biol. 52:585-587; 1992. Monte, M.; Klein, S.; Jasnis, M. A.; Davel, L.; Algranati, I. D.; de Lustig, E. S. Inhibition of lymphocyte and tumorinduced angiogenesis by the administration of difluoromethylomithine. Cancer J. 6:147- 150; 1993. Mustoe, T. A.; Pierce, G. F.; Thomason, A.; Gramates, P.; Spom, M. B.; Deuel, T. F. Accelerated healing of incisional wounds in rats induced by transforming growth factor-p. Science 237:1333-1335; 1987. Muszynski, C. A.; Robertson, C. S.; Goodman, J. C.; Henley, C. M. DFMO reduces cortical infarct volume after middle cerebral artery occlusion in the rat. J. Cereb. Blood Flow. Metab. 13:1033-1037; 1993. Ostertag, C. B.; Weigel, K.; Wamke, P.; Lombeck, G.; Kleihues, P. Sequential morphological changes in the dog brain after interstitial iodine- 125 irradiation. Neurosurgery 13:523-528; 1983. Paschen, W. Polyamine metabolism in reversible cerebral ischemia. Cerebrovasc. Brain Metab. Rev. 4:59-88; 1992. Paschen, W.; Widmann, R.; Weber, C. Changes in regional polyamine profiles in rat brains after transient cerebral ischemia (single versus repetitive ischemia): Evidence for release of polyamines from injured neurons. Neurosci. Lett. 135:121-124; 1992.

1045

43. Reinhold, H. S.; Calvo, W.; Hopewell, J. W.; van den Berg, A. P. Development of blood vessel-related radiation damage in the fimbria of the central nervous system. Int. J. Radiat. Oncol. Biol. Phys. 18:37-42; 1990. 44. Reinhold, H. S.; Hopewell, J. W. Late changes in the architecture of blood vessels of the rat brain after irradiation. Br. J. Radiol. 53:693-696; 1980. 45. Scharfen, C. 0.; Sneed, P. K.; Wara, W. M.; Larson, D. A.; Phillips, T. L.; Prados, M. D.; Weaver, K. A.; Malec, M.; Acord, P.; Lambom, K. R.; Lamb, S.; Ham, B.; Gutin, P. H. High activity iodine-125 interstitial implant for gliomas. Int. J. Radiat. Oncol. Biol. Phys. 24:583-591; 1992. 46. Schmitz, M. P.; Combs, D. J.; Dempsey, R. J. Difluoromethylomithine decreases postischemic brain edema and blood-brain barrier breakdown. Neurosurgery 33:882888; 1993. 47. Seiler, N. Polyamine metabolism. Digestion 46(suppl. 2):319-330; 1990. 48. Sheline, G. E.; Wara, W. M.; Smith, V. Therapeutic irradiation and brain injury. Int. J. Radiat. Oncol. Biol. Phys. 6:1215-1228; 1980. 49. Sunderkotter, C.; Goebeler, M.; Schulze-Osthoff, K.; Bhardwaj, R.; Sorg, C. Macrophage-derived angiogenesis factors. Pharmacol. Ther. 51:195-216; 1991. 50. Takigawa, M.; Enomoto, M.; Nishida, Y.; Pan, H.; Kinoshita, A.; Suzuki, F. Tumor angiogenesis and polyamines: cY-Difluoromethylomithine, an irreversible inhibitor of omithine decarboxylase, inhibits B 16 melanoma-induced angiogenesis in ovo and the proliferation of vascular endothelial cells in vitro. Cancer Res. 50:4131-4138; 1990. 51. Tiller-Borcich, J. K.; Fike, J. R.; Phillips, T. L.; Davis, R. L. Pathology of delayed radiation brain damage: An experimental canine model. Radiat. Res. 110: 161- 172; 1987. 52. Trout, J. J.; Koenig, H.; Goldstone, A. D.; Lu, C. Y. Bloodbrain barrier breakdown by cold injury-Polyamine signals mediate acute stimulation of endocytosis, vesicular trasport, and microvillus formation in rat cerebral capillaries. Lab. Invest. 55:622-63 1; 1986. 53. Turowski, K.; Fike, J. R.; Cann, C. E.; Higgins, R. J.; Davis, R. L.; Gutin, P. H.; Phillips, T. L.; Weaver, K. A. Normal brain iodine- 125 radiation damage: Effect of dose and irradiated volume in a canine model. Radiology 158:833-838; 1986. 54. Walker, M. D.; Strike, T. A.; Sheline, G. E. An analysis of dose-effect relationships in the radiotherapy of malignant gliomas. Int. J. Radiat. Oncol. Biol. Phys. 5:1725-1731; 1979. 55. Welboum, R.; Goldman, G.; Kobzik, L.; Paterson, I.; Shepro, D.; Hechtman, H. B. Interleukin-2 induces early multisystem organ edema mediated by neutrophils. Ann. Surg. 214:181-186; 1991. 56. Wilcox, J. N.; Derynck, R. Localization of cell synthesizing transforming growth factor-alpha mRNA in mouse brain. J. Neurosci. 8:1901- 1904; 1988. 57. Zook, B. C.; Bradley, E. W.; Casarett, G. W.; Rogers, C. C. Pathologic findings in canine brain irradiated with fractionated fast neutrons or photons. Radiat. Res. 84:562578; 1980.