Cellular characterization of the peritumoral edema zone in malignant brain tumors Blackwell Publishing Asia
Tobias Engelhorn,1,8 Nic E. Savaskan,2,3,8 Marc A. Schwarz,4 Jürgen Kreutzer,4,9 Eric P. Meyer,5 Eric Hahnen,6 Oliver Ganslandt,4 Arnd Dörfler,1 Christopher Nimsky,4,10 Michael Buchfelder4 and Ilker Y. Eyüpoglu4,7 1
Department of Neuroradiology, University of Erlangen-Nuremberg, Erlangen, Germany; 2Brain Research Institute, Department of Biology, Swiss Institute of Technology (ETH) and University of Zurich, Zurich, Switzerland; 3Institute of Cell Biology and Neurobiology, Center for Anatomy, Charité-Universitätsmedizin Berlin, Berlin, Germany; 4Department of Neurosurgery, University of Erlangen-Nuremberg, Erlangen, Germany; 5Department of Zoology, University of Zurich, Zurich, Switzerland; 6Institute of Human Genetics, Institute of Genetics and Center for Molecular Medicine Cologne (CMMC), University of Cologne, Cologne, Germany (Received January 16, 2009/Revised June 5, 2009/Accepted June 16, 2009/Online publication July 21, 2009)
Brain edema is a hallmark of human malignant brain tumors and contributes to the clinical course and outcome of brain tumor patients. The so-called perifocal edema or brain swelling imposes in T2-weighted MR scans as high intensity areas surrounding the bulk tumor mass. The mechanisms of this increased fluid attraction and the cellular composition of the microenvironment are only partially understood. In this study, we focus on imaging perifocal edema in orthotopically implanted gliomas in rodents and correlate perifocal edema with immunohistochemical markers. We identified that areas of perifocal edema not only include the tumor invasion zone, but also are associated with increased glial fibrillary acidic protein (GFAP) and aquaporin-4 expression surrounding the bulk tumor mass. Moreover, a high number of activated microglial cells expressing CD11b and macrophage migration inhibitory factor (MIF) accumulate at the tumor border. Thus, the area of perifocal edema is mainly dominated by reactive changes of vital brain tissue. These data corroborate that perifocal edema identified in T2-weighted MR scans are characterized with alterations in glial cell distribution and marker expression forming an inflammatory tumor microenvironment. (Cancer Sci 2009; 100: 1856–1862)
M
alignant gliomas are common tumors of the central nervous system (CNS).(1) Aside from uncontrolled cell proliferation and diffuse tissue invasion, brain edema represents a hallmark of these tumors and renders complete surgical resection almost impossible.(1–4) Additional therapeutic strategies are radiotherapy and chemotherapy regimens. Nevertheless, these non-surgical treatments have not achieved nearly the success rates seen in non-CNS tumors.(5) The polyclonal character of malignant gliomas has hampered biochemical, cellular, and molecular analysis because of the phenotypic diversity of the multiple cell characteristics within a single tumor.(3) Further, the feature to artificially increase their total volume by fluid accumulation as so-called perifocal or perilesional edema contributes to the clinical course and outcome of brain tumor patients.(6–9) Peritumoral edema in vivo is routinely determined by magnetic resonance imaging (MRI) as a standard method in clinical practice.(10) To visualize the proper expansion of brain lesions, T2-weighted scans are used to depict perifocal edema as high intensity areas, while T1-weighted images in combination with paramagnetic contrast administration identify the bulk tumor mass. The mechanisms of these malignant processes, i.e. increased proliferation, diffuse tissue invasion, and induction of perifocal edema, are only partially understood. However, it is thought that neurotoxic levels of glutamate play an important role in glioma-induced cell death(11–13) and induce subsequently peritumoral edema.(13) The determination between pure perifocal edema of vital brain tissue and the border of glioma invasion represent one major clinical problem. It is still a matter of debate Cancer Sci | October 2009 |
vol. 100 | no. 10 | 1856–1862
how far areas of perifocal edema represent tumor-invaded districts or whether this area represents solely reactively altered vital brain tissue. Moreover, the tumor surrounding tissue as so-called perifocal edema is only marginally characterized. To dissect peritumoral vital brain tissue from glioma-infiltrated areas, we identified perifocal edema using a standard clinical 1.5-Tesla MRI and subsequently analyzed these areas with different immunohistochemical markers. This study revealed that peritumoral areas visualized by MRI scans reflect an inflammatory tumor microenvironment with reactive astrocytes and microglia surrounding the bulk tumor mass. Materials and Methods F98 glioma cell line. The rat glioma cell line F98 (passage 50 – 70) was established more than 20 years ago by ethylnitrosoureainduced carcinogenesis in CDF Fischer rats. Growth characteristics, cytological, and immunohistochemical features of this cell line have been extensively reported.(14,15) Cells were cultivated in a humidified atmosphere at 37°C and 5% CO2 in DMEM ( Invitrogen, Karlsruhe, Germany) supplemented with 10% fetal calf serum (Biochrom, Berlin, Germany). Transfection of GFP and RFP into F98 glioma cells. F98 glioma cells were transfected either with peGFP-N1 (Takara BioCompany Clontech, Saint-Germain-en-Laye, France) or with pmRFP (monomeric version of Discosoma red fluorescent protein) using Roti-Fect transfection reagent (Roth, Karlsruhe, Germany) according to the manufacturer’s protocol. After 48 h of transfection, the cells were fed with fresh medium. Transfected cells were cultured in a selection medium (containing 700 μg G418/mL) for 1 week and subsequently sorted (MoFlo, DakoCytomation, Germany) twice to accelerate the selection process. To control unwanted side effects or any challenges due to transfection, wild-type and stable transfected F98 gliomas were implanted into organotypic brain slice cultures and invasion and growth was monitored. Neither morphological, proliferative, nor invasive characteristics of F98 malignant glioma cells were altered in the ex vivo organotypic glioma invasion model compared to wild-type F98 cells.(16) Glioma implantation in vivo. Animal experiments were done in congruence with the European Union guidelines for the use of laboratory animals. The protocol for animal experimentation was approved by the Government of Central Franconia (permission
7
To whom correspondence should be addressed. E-mail:
[email protected] or
[email protected] Both authors contributed equally to this work and are listed in alphabetical order. 9 Present address: Department of Neurosurgery, Technical University of Munich, Germany 10 Present address: Department of Neurosurgery, University of Marburg, Germany 8
doi: 10.1111/j.1349-7006.2009.01259.x © 2009 Japanese Cancer Association
no. 54.2531.31-8/06). Female Fischer rats weighing 150–200 g (Charles River, Sulzfeld, Germany) were anesthetized using intraperitoneal injection (mixture of 70 mg/kg BW ketamine [Pfizer, Karlsruhe, Germany], 15 mg/kg BW xylazin [Bayer, Leverkusen, Germany], and 0.05 mg/kg BW atropine [Braun, Kronberg / Taunus, Germany]) before being fixed in a stereotactic frame (David Kopf Instruments, Bilaney Consultants, Duesseldorf, Germany). F98 rat glioma cells in a volume of 5 μL (1.5 × 105) were stereotactically implanted with a Hamilton syringe (VWR, Darmstadt, Germany) into the right frontal lobe of the animals (2 mm lateral to bregma, depth 4 mm from dura). Tumor implantation was monitored 10 days after surgery using 1.5Tesla MRI. Animals were sacrificed after 1.5-Tesla MRI control (see below) and brains were collected for histological evaluation. Magnetic resonance imaging (MRI) in vivo and postmortem and volume determination. Imaging was performed on a 1.5-Tesla
MR scanner unit (Sonata; Siemens, Erlangen, Germany) with a 40-mm-diameter, small field-of-view orbita surface coil as receiver. Scout images and a 3D-CISS sequence (repetition time = 9 ms, echo time = 5 ms, reconstructions with a slice thickness of 0.4 mm) were obtained in coronal, axial, and transverse planes to position the slices accurately. Ten coronal T1- and T2-weighted slices, each with 1-mm thickness and 0.2-mm separation (inter-slice gap) were then positioned on the transverse scout images to cover the tumor (in most cases on three to four slices). The T1-weighted images were acquired with a 256 × 256 matrix, field-of-view = 40 mm, repetition time = 507 ms, echo time = 17 ms, and a total scan time of 3 min 42 s. For contrast-enhanced images, each animal received 1 mL/kg body weight of contrast media (Magnevist; Schering, Berlin, Germany) i.p. 10 min prior to the acquisition of T1weighted sequences. The T2-weighted images were acquired with a 512 × 512 matrix, field-of-view = 52 mm, repetition time = 4500 ms, echo time = 158 ms, and a total scan time of 6 min 12 s. Signal to noise was always 0.05). (e) T1-weigthed MR images from postmortem (ex vivo) brain samples (T1pm, left), corresponding fluorescence image (middle), and superimposed image (right, merge). Arrows mark the transition zone from tumor mass to vital brain tissue (f) T1-weigthed MR images from postmortem (ex vivo) brain samples (T1pm, left), and corresponding H&E-stained histopathological section (middle), and merged image (right). Arrows indicate the tumor expansion and tumor front. (g) Monitoring fluorescent signals of GFP-positive F98 glioma cells and comparison with H&E staining. Given are representative H&Estained and fluorescence microscopic images of a tumor-bearing brain section (10-μm thick). Sections of F98 glioma implanted rat brains were stained by H&E (left) and indicate that glioma cells migrate into the adjacent brain tissue along radial oriented trails. Respective fluorescent signals (right) confirm centrifugal migration streams as well as an infiltrative growth patterns. Arrow heads mark the putative outer invasion border. Scale bar represents 130 μm (left), 250 μm (right).
striatum of female Fischer rats and cranial MR imaging was performed 10 days thereafter. By generation 3D data sets using a CISS sequence we first analyzed the tumor expansion and defined the best sectional plane defining anatomical landmarks for further immunohistochemical analysis (Fig. 1a). Using this approach, the coronal plane revealed the best anatomical orientation for tumor and peritumoral visualization and thus we proceeded with this plane for immunohistological sectioning and superimposition analysis. We extended our analysis by further imaging T1-weighted postmortem (or ex vivo) brain samples from the same individuals and compared those with images from the in vivo approach, since T2-weighted images from postmortem samples did not give sufficient contrast (Fig. 1b). Intraperitoneal injection of contrast medium visualized the bulk tumor mass in the right frontal lobe in T1-weighted images of in vivo imaged animals as a high signal intensity area (Fig. 1b). Spreading of tumor cells to other brain areas was not detected. MR imaging of postmortem brain samples was executed afterwards (Fig. 1b). Contrast and tumor borders were preserved 1858
in postmortem imaged brains and these images visualized an even clearer contrast of tumor brain borders. In addition, contrast enhancement and tumor expansion appeared slightly expanded in postmortem MR images. Next we compared MRI volumetry of brain tumors determined on T1-weighted images after contrast injection with different histological procedures. Interestingly, MRI volumetry resulted in a mean tumor volume of 64 ± 29 mm3 (Fig. 1c,d), H&E staining revealed a mean tumor volume of 76 ± 13 mm3 in gliomaimplanted rats (Fig. 1c,d). Histological analysis of the tumor bulk using fluorescence signals generally resulted in a high degree of consistency between all four groups. Gliomas expressing GFP revealed a mean tumor bulk of 84 ± 6 mm3 (Fig. 1d). The tumor volume determined on T1-weighted images after contrast injection appeared slightly smaller compared with the tumor volume measured in histological sections (H&E, GFP), although differences were not statistically significant. This non-significant difference was consistent in all tested samples (t-test: P > 0.05). Furthermore, tumor borders were unambiguously identified doi: 10.1111/j.1349-7006.2009.01259.x © 2009 Japanese Cancer Association
Fig. 2. Aquaporin-4 expression in peritumoral zones of gliomas. T1-weighted MR sequences after contrast injection reveal tumor expansion in orthotopic tumor–bearing rats 10 days after tumor implantation (a). Perifocal edema in T2-weigthed MR scans impresses as hyperintensive area surrounding the bulk tumor mass and indicates that glioma cells are still expanding. T1- and T2-derived volumes were quantified by means of MR (b). Quantification of tumor mass (Tumor) and perifocal edema (Edema) (b). Data are given as means ± SDs from 12 tumor-implanted rats and statistical analysis was performed with Student’s t-test. (c) Sections of brain tissues stained for aquaporin-4 (red), while F98 glioma cells appear green due to their genetically introduced GFP expression. Scale bar represents 5 mm. (d) Higher magnifications identifies increased aquaporin-4 expression almost exclusively at the border of the bulk tumor mass. Nuclei are shown in blue (Hoechst staining). Right image is a superimposed merge of all three fluorescence channels. Scale bar represents 100 μm. (e) Statistical analysis of aquaporin-4 expression. Tumor implanted animals show significantly elevated aquaporin-4 expression compared to controls. Statistical analysis was performed with Student’s t-test, *P < 0.05, and data are given as means ± SDs from eight tumor-implanted rats.
in light and fluorescence microscopic sections, whereas in MR images these structures were less clear to distinguish easily (Fig. 1b,c) and thus may account for the differences in tumor volume (Fig. 1d). Interestingly, MR images from postmortem brain samples gave overall accurate superimpositions with fluorescence microscopy images and H&E-stained histological sections (Fig. 1e,f). However, although anatomical landmarks overlapped unambiguously in MR images with histopathological sections, tumor borders revealed slight differences in MR images compared to microscopic images which in addition may explain differences in measurements for tumor volume (Fig. 1d). Identification of invasive glioma growth in vivo. A general observation was that syngeneic tumor transplants in vivo predominantly grow by local expansion rather than diffuse infiltration, although invasive growth is a hallmark of human malignant gliomas.(1) Whether this is due to particular cell line characteristics or technical reasons (unheralded invading tumor cells) has yet to be determined. To further analyze glioma brain invasion in vivo, glioma-implanted CNS tissues of rat brain were processed for histochemistry (Fig. 1e–g). At the tumor border we found infiltrative growth patterns of gliomas into brain parenchyma (Fig. 1e–g). We thereafter compared the H&E staining with the respective fluorescent signals. Fluorescence microscopic analysis revealed single outgrowing glioma cells infiltrating normal brain parenchyma (Fig. 1g). However, dissecting vital brain tissue from F98 glioma growth was difficult when H&E staining was solely used as a histological approach, while analysis of fluorescent signals of transfected glioma cells simplified the identification of infiltration patterns and geographically escaped tumor cells. Engelhorn et al.
Perifocal edema is associated with glial alterations. To control tumor loading we monitored tumor growth in vivo by MRI 10 days after implantation. Tumor volume assessed by T1-weighted images at 1.5 Tesla after paramagnetic contrast administration revealed no differences in all analyzed animals (Fig. 2a,b). Measuring tumor volumetry resulted in a glioma mass of 64 ± 29 mm3 (Fig. 2b). To unmask the proper expansion of tumor lesions, T2-weighted scans were acquired depicting perifocal edema as high-intensity areas (Fig. 2a). Bulk tumor mass and perifocal edema engaged a volume of 82 ± 33 mm3 (Fig. 2b). The cerebrospinal fluid was seen as high intensity areas on T2-weighted images. A combination of both sequences offered the possibility to quantify the perifocal edema, exclusively (Fig. 2b). While tumor volumes identified by T1-weighted images correlate tightly with those determined by immunohistochemistry (Fig. 1), one can assume that the so-called perifocal edema might be a transitional zone including invasive tumor cells (Fig. 2). However, the area of perifocal edema was consistently greater than the identified tumor invasion zone. This points to the possibility that perifocal edema might also include alterations of vital brain parenchyma such as pathological vessel formation (tumor angiogenesis), glial alteration, and blood–brain barrier disruption. We therefore analyzed the perifocal tumor zone by means of immunohistochemistry. We found alterations in astrocytes surrounding the tumor border and elevated aquaporin-4 expression corroborating the in vivo findings of perifocal edema in tumor-implanted animals (Fig. 2c–e). Fusion of T1-weighted images and histology sections demonstrated that the tumor bulk revealed by MR images matches with the distribution of fluorescence signals (Fig. 3a). In addition, infiltrative growing tumor cells outside of the bulk tumor mass
Cancer Sci | October 2009 | vol. 100 | no. 10 | 1859 © 2009 Japanese Cancer Association
Fig. 3. Astroglial cells accumulate at the peritumoral edema zone surrounding the bulk tumor mass. (a) T1-weighted MR sequences after contrast agent injection (left) and fluorescent signal of GFP-expressing F98 glioma cells (green, middle). Left, superimposed MR and fluorescence image. (b) MR sequences were matched with immunohistochemical staining. Superimposed T2-weighted MR sequences overlaid with the corresponding aquaporin-4 immunostaining (red). (c) glial fibrillary acidic protein (GFAP)-stained brain section (red), while F98 glioma cells appear in green due to constitutively GFP expression. (d) Higher magnification reveals swollen astrocytes (red) surrounding the bulk tumor mass (green). Nuclear bodies are given in blue (Hoechst staining). (e) Statistical analysis of GFAP expression in tumor-bearing brains compared to control brain sections. Statistical analysis revealed a significant increase in GFAP expression in peritumoral regions compared to the contralateral tumor-free region and control animals. Data are given as means ± SDs from eight tumor-implanted rats. *P < 0.05; Student’s t-test.
have almost been found within the perifocal edema zone (Figs 2 and 3a). Moreover, using the same approach by fusion of T2-weigthed images with aquaporin-4-immunostained sections revealed that perifocal edema correlates with astrocytic aggregation and increased expression of aquaporin-4 within these zones (Fig. 3b,c). These data were further corroborated by immunohistochemistry: Immunostaining for GFAP as a marker for reactive astrocytes revealed increased GFAP expression after tumor implantation (Fig. 3c). Higher magnification of perifocal tumor zones further identified reactive and swollen astrocytes, respectively (Fig. 3d,e). Hence, we analyzed the distribution and activation status of microglia in glioma-implanted brains. A high number of microglial cells accumulated within the tumor core (Fig. 4a,b). However, most microglial cells were located at the tumor periphery and showed an altered morphology compared to microglia cells without contact with malignant gliomas (Fig. 4c). In addition, augmented levels of IB4 intensity (Fig. 4c), a marker for microglial cells, as well as an increase of the total amount of microglial cells (Fig. 4c), was found in microglial cells exposed to the tumor compared to controls. Microglia within the perifocal edema zone showed also an altered morphology from a ramified, inactive into an amoeboid, active morphology as quantified by the index of ramification (Fig. 4c),(18) indicating an inflamed peritumoral microenvironment. In addition, the inflammatory 1860
cytokine MIF was increased in microglia in tumor-implanted brains compared to control brain sections (Fig. 4d), indicating an inflammatory tumor microenvironment. Further, the microglial activation marker CD11b was significantly up-regulated in those microglial cells surrounding the bulk tumor mass in the zone of the perifocal edema (Fig. 4e,f ). Discussion
Understanding glioma biology and in particular glioma brain parenchyma interaction still represents a challenge in neurooncological research. (19,20) Nevertheless, tracking glioma progression and analysis of tumor-host interactions are difficult to study with respect to the complex organization of the CNS.(21) Thus, in vivo monitoring and cellular imaging represent a relevant step in neuro-oncological studies.(9,22) In this study we combined MRI approaches equivalent to those used in clinical settings with genetic tools (in vivo fluorescence reporter expression) and standard immunostaining and histology. These data were correlated with immunohistochemical procedures and in vitro tools were used, i.e. stable fluorescent transfection of glioma cells, to visualize all tumor cells. Interestingly, these tumor volume correlations acquired from MRI and histopathological specimens revealed slight differences in the way that MR images gave smaller values. These volumetry data analyses were based doi: 10.1111/j.1349-7006.2009.01259.x © 2009 Japanese Cancer Association
Fig. 4. Microglia activation and accumulation at peritumoral regions. Invasive malignant glioma growth induces microglia accumulation at the front of the bulk tumor. (a) Representative images of IB4-stained microglial cells (red, left) in tumorimplanted brain parenchyma. The tumor is revealed by GFP expression (green). Nuclei are stained in blue (Hoechst). (b) Microglia distribution (red) in normal brain parenchyma. Nuclear staining is given in blue. Scale bar represents 80 μm. (c) Quantification of microglial activation. Left, quantification of IB4 staining intensity, microglial cell density (middle), and quantification of the index of ramification. Data are given as means ± SDs from eight tumor-implanted rats. *P < 0.05; Student’s t-test. TI, tumor implanted. (d) MIF (macrophage migration inhibitory factor) expression in control brain sections (con, left), and tumor-bearing brain samples ( TI, right). Arrows mark the accumulation zone of MIF+ microglia with vessels. Scale bar represents 60 μm. (e) CD11b expression (red) in control brain sections (con, left), and tumor-bearing brain samples ( TI, right). Tumor is given in green. Scale bar represents 500 μm. (f ) Quantification of CD11b immuno-intensity in controls (con) and tumor-implanted brain samples (TI). Data are given as means ± SDs from six tumor-implanted rats. *P < 0.05; Student’s t-test.
on a standard clinical 1.5-Tesla MR scanner equipped with a standard orbital surface coil. Explanation for these differences may lie in the less-sharp identification of tumor borders in MR images, whereas light and fluorescence microscopic images visualized clear-cut tumor borders reliably identified by observers as well as by digital data analysis. In addition, standard immunohistochemistry gave essentially the same volumetric data as immunofluorescence analysis. However, dissecting vital brain tissue from glioma growth was difficult when standard immunohistochemistry was used, while analysis of fluorescent tumor cells simplified the identification of infiltration and quantification of invasive glioma growth. Thus, clear morphological distinction of fluorescently labeled tumor cells from normal brain tissue revealed an even more detailed glioma invasion pattern and also unmasked geographically dodged tumor cells. However, the major goal of this study was a characterization of the so-called perifocal edema, a microenvironment surrounding the bulk tumor mass including affected brain tissue as well as invading tumor cells. Therefore, T2-weighted images, which are ideally suitable for depicting areas of increased fluid accumulation, were performed and matched with corresponding immunohistochemistry, i.e. glial as well as water-channel-associated markers. Engelhorn et al.
We identified that MRI displayed perifocal edema tightly correlated with glial alteration. Accumulation of microglial cells surrounding the bulk tumor mass as well as reactively altered astrocytes appeared to be characteristic for the perifocal edema. Moreover, increased expression of aquaporin-4 was detected almost exclusively surrounding the bulk tumor mass, indicating that dynamic changes of liquid in extra- and intracellular spaces are relevant steps in the development of perifocal edema. Thus, perifocal edema not only includes invasive growing glioma, but also is associated with glial alterations, i.e. with the appearance of reactive astrocytes and activated microglia. Indeed, microglial cells exposed to brain tumors altered their morphology from a ramified inactive into an amoeboid active morphology as indicated by an augmented index of ramification. In addition, microglial activation was detectable by increased expression of the microglia markers IB4 and up-regulated CD11b. The latter one is well known as being expressed, especially in activated microglia. Moreover, enhanced expression of the inflammatory cytokine MIF was identified in microglia surrounding the bulk tumor mass. Although we found elevated activation of microglial cells in tumor implanted brain samples, phagocytozed tumor cells were not consistently found by means of GFP Cancer Sci | October 2009 | vol. 100 | no. 10 | 1861 © 2009 Japanese Cancer Association
fluorescent signals in activated microglia. Microglia in particular have been suspected to promote tumor growth and angiogenesis rather than defend the brain against gliomas.(23–28) In addition, activated microglia release cytokines which stimulate angiogenesis and vasculogenesis.(27–29) Whether tumor-induced microglial activation and increased MIF secretion affects the development of brain swelling remains unknown. However, microglial COX-2 inhibition with subsequent Prostaglandin E2 (PGE2) reduction reduced glioma-induced brain edema formation.(30) This is further supported by the findings that the routinely used dexamethasone reduces cerebral edema, and also suppresses microglial activation.(6) However, under which conditions activated microglia promotes perifocal brain swelling and whether microglial cells are actively involved in perifocal and tumorinduced angiogenesis has to be defined in the future. Another unexpected cell type occurred in peritumoral zones, reactive astrocytes. Astrocytic cells balance the extracellular fluids and therefore protect neuronal cells from heavy osmotic variations in normal brain tissue. However, malignant gliomas disrupt this fluid balance by secretion of neurotoxic compounds and subsequently induce perifocal edema.(13) Here, we identified that astrocytes alter their morphology showing astrocytic swelling and increased expression of water channels, i.e. aquaporin-4, References 1 DeAngelis LM. Brain tumors. N Engl J Med 2001; 344: 114–23. 2 Hanahan D, Weinberg RA. The hallmarks of cancer. Cell 2000; 100: 57–70. 3 Maher EA, Furnari FB, Bachoo RM et al. Malignant glioma: genetics and biology of a grave matter. Genes Dev 2001; 15: 1311–33. 4 Papadopoulos MC, Saadoun S, Binder DK, Manley GT, Krishna S, Verkman AS. Molecular mechanisms of brain tumor edema. Neuroscience 2004; 129: 1011–20. 5 Stewart LA. Chemotherapy in adult high-grade glioma: a systematic review and meta-analysis of individual patient data from 12 randomised trials. Lancet 2002; 359: 1011– 8. 6 Badie B, Schartner JM, Paul J, Bartley BA, Vorpahl J, Preston JK. Dexamethasone-induced abolition of the inflammatory response in an experimental glioma model: a flow cytometry study. J Neurosurg 2000; 93: 634– 9. 7 Badie B, Schartner J. Role of microglia in glioma biology. Microsc Res Tech 2001; 54: 106 –13. 8 Wick W, Kuker W. Brain edema in neurooncology: radiological assessment and management. Onkologie 2004; 27: 261–6. 9 Jain RK, di Tomaso E, Duda DG, Loeffler JS, Sorensen AG, Batchelor TT. Angiogenesis in brain tumours. Nat Rev Neurosci 2007; 8: 610–22. 10 Lin TN, He YY, Wu G, Khan M, Hsu CY. Effect of brain edema on infarct volume in a focal cerebral ischemia model in rats. Stroke 1993; 24: 117–21. 11 Takano T, Lin JH, Arcuino G, Gao Q, Yang J, Nedergaard M. Glutamate release promotes growth of malignant gliomas. Nat Med 2001; 7: 1010 –5. 12 Ye ZC, Sontheimer H. Glioma cells release excitotoxic concentrations of glutamate. Cancer Res 1999; 59: 4383– 91. 13 Savaskan NE, Heckel A, Hahnen E et al. Small interfering RNA-mediated xCT silencing in gliomas inhibits neurodegeneration and alleviates brain edema. Nat Med 2008; 14: 629 –32. 14 Ko L, Koestner A, Wechsler W. Morphological characterization of nitrosoureainduced glioma cell lines and clones. Acta Neuropathol (Berl) 1980; 51: 23–31. 15 Reifenberger G, Bilzer T, Seitz RJ, Wechsler W. Expression of vimentin and glial fibrillary acidic protein in ethylnitrosourea-induced rat gliomas and glioma cell lines. Acta Neuropathol (Berl) 1989; 78: 270 –82. 16 Eyupoglu IY, Hahnen E, Heckel A et al. Malignant glioma-induced neuronal cell death in an organotypic glioma invasion model. Technical note. J Neurosurg 2005; 102: 738– 44.
1862
which may reflect a desperate approach to restore the extracellular balance of fluids. The role of astrocytic cells in this context is still unknown. Therefore, a novel focus in future studies should be further centered on the role of reactive astrocytes surrounding the bulk tumor mass. In conclusion, we identified that the so-called perifocal edema not only includes invading tumor cells but also is associated with glial alterations in vital brain tissue, i.e. astrocytic swelling, microglial accumulation, and microglial activation. The role of these inflammatory changes remains to be uncovered and may represent a relevant line in neuro-oncological research to better understand the mechanisms of peritumoral brain swelling and distortion. Acknowledgments We are deeply grateful to Frank Bittner for the digital artwork and programming work. We thank Nadine Scheufler and Irene Emtmann for excellent technical assistance. N.E.S. was supported by the International Human Frontiers Science Program Organization (HFSPO). This study was funded by a grant from the Wilhelm Sander-Stiftung (no. WSS 2005.089.1 to I.Y.E.) and the Institut Danone Ernährung für Gesundheit e.V. to I.Y.E. & N.E.S.
17 Kim B, Chenevert TL, Ross BD. Growth kinetics and treatment response of the intracerebral rat 9L brain tumor model: a quantitative in vivo study using magnetic resonance imaging. Clin Cancer Res 1995; 1: 643– 50. 18 Eyupoglu IY, Bechmann I, Nitsch R. Modification of microglia function protects from lesion-induced neuronal alterations and promotes sprouting in the hippocampus. Faseb J 2003; 17: 1110–1. 19 Berens ME, Giese A. ‘... those left behind.’ Biology and oncology of invasive glioma cells. Neoplasia 1999; 1: 208–19. 20 von Deimling A, Louis DN, Wiestler OD. Molecular pathways in the formation of gliomas. Glia 1995; 15: 328–38. 21 Pilkington GJ, Bjerkvig R, De Ridder L, Kaaijk P. In vitro and in vivo models for the study of brain tumour invasion. Anticancer Res 1997; 17: 4107– 9. 22 Neeman M, Gilad AA, Dafni H, Cohen B. Molecular imaging of angiogenesis. J Magn Reson Imaging 2007; 25: 1–12. 23 Farber K, Synowitz M, Zahn G et al. An alpha5beta1 integrin inhibitor attenuates glioma growth. Mol Cell Neurosci 2008; 39(4): 579–85. 24 Nishie A, Ono M, Shono T et al. Macrophage infiltration and heme oxygenase-1 expression correlate with angiogenesis in human gliomas. Clin Cancer Res 1999; 5: 1107 –13. 25 Roggendorf W, Strupp S, Paulus W. Distribution and characterization of microglia / macrophages in human brain tumors. Acta Neuropathol (Berl) 1996; 92: 288 –93. 26 Leung SY, Wong MP, Chung LP, Chan AS, Yuen ST. Monocyte chemoattractant protein-1 expression and macrophage infiltration in gliomas. Acta Neuropathol 1997; 93: 518–27. 27 Bingle L, Brown NJ, Lewis CE. The role of tumour-associated macrophages in tumour progression: implications for new anticancer therapies. J Pathol 2002; 196: 254 –65. 28 Platten M, Kretz A, Naumann U et al. Monocyte chemoattractant protein-1 increases microglial infiltration and aggressiveness of gliomas. Ann Neurol 2003; 54: 388 –92. 29 Yao Y, Kubota T, Sato K, Kitai R. Macrophage infiltration-associated thymidine phosphorylase expression correlates with increased microvessel density and poor prognosis in astrocytic tumors. Clin Cancer Res 2001; 7: 4021– 6. 30 Badie B, Schartner JM, Hagar AR et al. Microglia cyclooxygenase-2 activity in experimental gliomas: possible role in cerebral edema formation. Clin Cancer Res 2003; 9: 872–7.
doi: 10.1111/j.1349-7006.2009.01259.x © 2009 Japanese Cancer Association
