Author pre-print version, final version Brain Pathol. 2008 Jul;18(3):344-53.

doi: 10.1111/j.1750-3639.2008.00127.x.

WT1 expression distinguishes astrocytic tumor cells from normal and reactive

astrocytes

Jens Schittenhelm1*, Michel Mittelbronn1*, Thai-Dung Nguyen*, Richard Meyermann*,

Rudi Beschorner*

*Institute of Brain Research, University Hospital of Tuebingen, Tuebingen, Germany

1 These two authors contributed equally to this study.

Running title: WT1 in reactive and neoplastic astrocytes

Corresponding author:

Jens Schittenhelm, MD

Institute of Brain Research

University of Tuebingen

Calwerstr. 3

D-72076 Tuebingen

Tel.: +49-7071-2982293

Fax: +49-7071-294846

Email: jens.schittenhelm@med.uni-tuebingen.de

Key words: WT1, astrocytoma, astrogliosis, neoplastic astrocytes, normal brain

This manuscript contains 4 tables and 6 figures.

Abstract

Particularly in small brain biopsies it might be difficult to distinguish reactive

astrogliosis from low-grade or infiltration zones of high-grade astrocytomas. So far no

immunohistochemical marker allows a reliable distinction. Recently the over-

expression of Wilms’ tumor gene product WT1 was reported in astrocytic tumor cells.

However, no sufficient data on WT1 expression in normal or reactive astrocytes are

available. Therefore, we investigated WT1 expression in paraffin-embedded brain

sections from 28 controls, 48 cases with astrogliosis of various etiology and 219

astrocytomas (WHO grade I to IV) by immunohistochemistry. In normal brains and in

astrogliosis expression of WT1 was restricted to endothelial cells. In astrocytomas,

WT1 positive tumor cells were found in pilocytic astrocytomas (66.7% of cases),

diffuse astrocytomas WHO grade II (52.7%), anaplastic astrocytomas (83.4%) and

glioblastomas (98.1%). Overall, the majority of all astrocytic neoplasms (84.5%)

expressed WT1. Establishing a cut off value of 0% immunoreactive tumor cells

served to recognize neoplastic astrocytes with 100% specificity and 68% sensitivity

and was associated with positive and negative predictive values of 1 and 0.68

respectively. Therefore, WT1 expression in astrocytes indicates a neoplastic origin

and might represent an important diagnostic tool to differentiate reactive from

neoplastic astrocytes by immunohistochemistry.

Introduction

In the central nervous system, morphological features of astrocytic neoplasms

overlap with reactive astrogliosis occurring within or around brain lesions of various

etiologies. Especially the distinction between astrogliosis and low-grade

astrocytomas or infiltration zones of high-grade astrocytomas may be difficult (9). In

the normal human brain astrocytes are associated with structural support, metabolic

homoeostasis and energy supply (13,45,49). They produce adhesion molecules,

neurotrophic factors and extracellular matrix substances that are essential for

structural integrity and repair (10). As a reaction to pathological changes they

increase in number and size and secrete additional extracellular matrix components

(1,38) through inhibition of membrane-bound Na+/K+-ATPase (5) or activation of

Epidermal Growth Factor Receptor (EGFR), a transmembrane receptor with tyrosine

kinase activity (30). Reactive (non-neoplastic) astrocytes may become very large and

may even display pleomorphic nuclei resembling neoplastic tumor cells (9). These

morphologic features are especially observed in progressive multifocal

leukoencephalopathy (PML) (12). Therefore, in small biopsy specimens this may

result in a diagnostic dilemma leading to inconclusive diagnosis (14) and might be

the main reason for interobserver variability (42). Especially if both atypical astrocytes

and mitotic figures are present in reactive lesions, an erroneous diagnosis of an

astrocytic neoplasm might occur (56). So far, no tool is available to reliably proof

neoplastic nature of astrocytes in pathology routine diagnosis. Expression

upregulation of glial fibrillary acidic protein (GFAP) (10), vimentin (53) and N-cadherin

(44) are the hallmarks of astrocytic activation and may also be observed in distant

areas of the lesion due to gap junction signalling (15). Under physiological conditions

the level of GFAP expression in CNS is low. The increase of GFAP in reactive

astrocytes makes GFAP an excellent immunohistochemical marker in histopathology

for labelling these cells (11). But its diagnostic usefulness in differentiating reactive

and neoplastic astrocytes is very limited since GFAP is also commonly observed in

astrocytomas (4,55). Previous studies have examined EGFR controlling cellular

functions in various gliomas (43). EGFR is not detected in normal astrocytes,

however reappearance occurs in reactive astrocytes under various

pathophysiological conditions (20,41). Other authors suggested that

immunohistochemical presence of p53 protein might detect neoplastic transformation

in glial tissue. However, it became evident that p53 is also expressed in a subset of

reactive astrocytes (26, 54). The cell-surface receptor for hyaluronic acid, CD44 was

considered to be significantly upregulated in astrocytic neoplasms compared to

normal brain (25). Later it became evident that normal brain, as well as neoplastic

tissues share the same splicing variant (35). In addition we were able to demonstrate

in a previous study that CD44 upregulation also occurs in reactive astrocytes at the

border of infiltrative neoplasms of non-glial origin (57). Therefore, a reliable method

to distinguish neoplastic from normal or reactive astrocytes suitable for routine

diagnosis has not yet been established.

The Wilms’ Tumor supressor gene encodes for a zinc-finger transcription factor

(WT1) and is associated with genito-urinary malformations due to reduced protein

levels in human development (8). Allelic losses or mutations of the gene result in

nephroblastoma, a common pediatric solid malignancy of the kidney (19). The WT1

protein is involved in proliferation, differentiation and apoptosis (21). Interaction with

other transcription factors like EGR-1 and p53 shows that WT1 has oncogenic

properties (31). Recent reports indicate that WT1 is expressed in glioblastoma cell

lines (32) and in astrocytic tumor tissue (39) among other types of solid cancer. In

addition it has been shown that WT1 might be useful to differentiate between

proliferating endothelial lesions and vascular malformations (28). In order to

characterize the astrocytic WT1 expression pattern and its diagnostic usefulness, we

investigated normal, reactive and neoplastic human CNS tissue specimens by

immunohistochemistry for differences.

Materials and Methods

Patient data

In total 295 cases from the tissue archives of the Institute of Brain Research

(Neuropathology), University Hospital of Tuebingen were enrolled in this study,

including normal brains (28 cases) and reactive CNS tissues (48 cases). The reactive

CNS tissue samples were derived from different pathological conditions such as

vascular malformations, brain metastasis, infiltrating meningiomas and ischemia

(details on cases are shown in table 1). Furthermore, astrocytic neoplasms (219

samples) consisting of WHO grade I pilocytic astrocytomas (n=24), WHO grade II

astrocytomas (n=36), WHO grade III anaplastic astrocytomas (n=54) as well as WHO

grade IV glioblastomas (n=105) were included in the present study. These patients

underwent surgical treatment between 1995 and 2005 (epidemiologic details are

shown in table 2). Histological diagnosis and grading were performed according to

the WHO classification system by at least two experienced neuropathologists (23).

Immunohistochemistry

After surgical removal all specimens were fixed in 4% formalin (pH 7.4) and

embedded in paraffin. Immunohistochemistry was performed using the established

monoclonal WT1-antibody, clone 6F-H2 (dilution 1:50, Dako, Glostrup, Denmark).

The antibody reacts with all isoforms of the full-length WT1 protein. In addition a

polyclonal GFAP antibody (Z0334, dilution 1:1000, Dako) was used. Sections were

deparaffinized, rehydrated and immunostained using the Benchmark

immunohistochemistry system (Ventana Medical Systems, Strasbourg, France). The

automated standard protocol is based on an indirect biotin-avidin system and

consists of a cell conditioning pretreatment, use of I-View inhibitor (Ventana) for 4

min, application of the WT1 antibody for 20 min or GFAP-antibody for 6 min followed

by an incubation of amplifier Systems A and B (Ventana), avidin-botin blocking for 4

min and one drop of I-View Biotin Ig (Ventana). For visualization the sections were

incubated with one drop of I-View SA-HRP for 8 min and then with DAB/H2O2 for

additional 8 min. The sections were finally incubated with a copper enhancer

(Ventana) for 4 min and counterstained with hematoxylin. Wilms tumor tissue

specimens were used as positive control displaying a distinct nuclear staining pattern

of tumor and endothelial cells. Negative control slides were processed in parallel with

each batch of staining. To exclude unspecific binding of the monoclonal antibody,

IgG isotype controls using the same IgG-concentrations as in the replaced primary

antibody solutions were performed (IgG1, Dako).

Evaluation and Statistics

Only samples showing WT1 expression in endothelial cells serving as internal control

were scored (28). Tumor cells were considered positive, when cytoplasmic staining of

cells matched at least the intensity observed in endothelial cells (47). The scores of

WT1+ tumour cells of the whole neoplastic specimen available was estimated

independently by three raters (JS, MM, RB) using a semiquantitative score including

0 (no staining), 1 (singular positive cells - <1%), 2 (2-19%), 3 (20-50%) and 4 (>50%)

allowing fast and easy scoring of the tissue. Only if two or more scores were identical

for the same case, this score was taken into account. Ordinal logistic regression was

performed for each comparison followed by a Likelihood Ratio Chi-square test. To

test interobserver variability of the three experienced neuropathologists in the WT1

assessment, a weighted kappa statistic was calculated. A summary kappa was

calculated both for each pairwise combination and for all raters as a group. Kappa

values of 0—0.20 have been defined as slight, 0.21—0.40 as fair, 0.41—0.60 as

moderate, 0.61—0.80 as substantial and 0.81—1.0 as almost perfect (27). JMP 7.0

was used for statistical analysis (* p≤0.05, ** ≤0.01, *** p≤0.0001).

Results

The applied WT1 scoring method is an easy and reproducible method.

Interobserver variability for all cases examined was at most one score level. The

agreement about the WT1 scores between the three raters taken as a group was

substantial (κ=0.683). Furthermore, all individual pairwise comparisons showed

substantial agreement ranging for κ=0.613 (SD=0.033) to κ=0.758 (SD=0.029).

Agreement across categories was almost perfect (κ=0.9560) for WT1 score 0 (no

WT1 expression) which is the most relevant factor for the diagnosis of positive and

negative cases.

WT1 expression in normal brain is restricted to endothelial cells

In all 28 normal human brains expression of WT1 was limited to endothelial cells of

small blood vessels, both in gray and in white matter (internal control). No astrocytes,

oligodendroglia, ependymal or neuronal cells showed WT1 immunoreactivity (score

0; figures 1A and 1B). Weak cytoplasmic staining of the epithelial choroid plexus

lining was considered nonspecific as it did not match endothelial staining of

capillaries and small vessels of CNS and leptomeninges (not shown). Regions

examined included samples from all four cerebral lobes, cerebellum, diencephalon

and brainstem. In Wilms’ tumor (positive control) a strong nuclear expression of WT1

was observed in tumor cells and in endothelial cells (figure 1C).

Reactive astrocytes lack WT1 expression

In all 48 brain specimens comprising of reactive astrogliosis due to various

pathological conditions (table 1) an increased number of astrocytes showed

gemistocytic morphology and marked expression of GFAP (figure 2). Reactive CNS

tissue around cavernomas (figure 2A) and arteriovenous malformations displayed

strong GFAP expression in reactive astrocytes close to the pathologic vessels (figure

2B). The same areas remained negative except for endothelial vessels on

consecutive sections stained with WT1 antibody (figure 2C). Gemistocytic astrocytes

around infarct lesions (figure 2D) showed prominent GFAP immunostaining (figure

2E) compared with the surrounding neuropil. These all showed a lack of WT1

expression (figure 2F). Non-primary brain tumors as invasive meningiomas,

melanomas and carcinoma metastases (figure 2G) showed reactive gliotic changes

at the tumor border with strong GFAP expression too (figure 2H) but remained

negative for WT1 (figure 2I). Expression of WT1 was always detectable in endothelial

cells of CNS and tumors (internal control) but was never observed in astrocytes

includig those with gemistocytic morphology (score 0). For the record, all 7

melanomas used in our study exhibited at least focally some cytoplasmic (7/7) or

nuclear (2/7) WT1 staining of tumor cells (not shown). Two cases of progressive

multifocal leucoencephalopathy (PML) with presence of JC-virus DNA confirmed by

molecular genetic methods did not express WT1 including their pleomorphic

astrocytes.

Expression of WT1 in astrocytomas differs significantly from normal brain and

reactive gliosis

Among all 219 glioma samples cytoplasmic WT1 imunoreactivity occurred in

endothelial cells and altogether 185 tumors (84.5%) showed WT1 expression in their

astrocytic cells. The distribution of WT1 positive tumor cells varied between different

WHO grades (table 3). Score variability in astrocytic tumors is exemplarily illustrated

in different glioblastomas (figure 3). Numbers of WT1 positive tumor samples were

higher in pilocytic astrocytomas (16/24) than in WHO grade II astrocytomas (19/36).

There was a significant increase of WT1 positive tumors in anaplastic astrocytomas

(45/54) with almost all glioblastomas showing WT1 positive tumor cells (103/105).

Likelihood Ratio Chi-square test showed that the distribution scores of astrocytic

tumors together and reactive CNS were significantly different (p<0.0001).

Comparison of WT1 expression in astrogliosis and astrocytomas together also

showed significant differences (table 4). Detailed p-values in comparison of normal

brain, gliosis and astrocytomas of different WHO grade are summarized in table 4.

Normal brain and astrocytomas (WHO grade I, II, III and IV, respectively) also

differed significantly (p<0.0001, each). Similarly, comparison of gliosis and

astrocytomas (WHO grade I, II, III and IV, respectively) revealed significant results

(p<=0.0001, each). The frequency distribution of different WT1 scores observed in

normal brain, gliosis and astrocytomas WHO grade I-IV is illustrated in figure 4. Using

score 0 as a cut off value, neoplastic astrocytes are recognized with 100% specificity

and 68% sensitivity and are associated with positive and negative predictive values

of 1 and 0.68 respectively. Using score 1 (<1% immunoreactive cells) instead for cut

off, neoplastic astrocytes are recognized with 100% specificity and 52% sensitivity

and are associated with positive and negative predictive values of 1 and 0.52,

respectively.

Expression patterns of WT1 in neoplastic astrocytes

In the majority of high-grade astrocytomas the tumor cells showed a strong

cytoplasmic staining including delicate cell processes. With few exceptions of

glioblastomas most astrocytomas were devoid of nuclear staining. A considerable

upregulation of WT1 immunoreactivity was observed in areas of pseudopalisading

around necrosis (figure 5A). Except for a thin internally located endothelial layer of

WT1 positive cells, proliferating vessels were immunonegative for WT1 (figure 5B).

Infrequently, astrocytes in close vicinity to microvessel proliferation also showed

increased WT1 staining compared to the more distant tumor areas (figure 5B). Cells

with exceptional anaplastic morphology (large, multinucleated tumor cells) in most

instances also displayed strong immunoreactive for WT1 (figure 5C). In anaplastic

astrocytomas there was usually a diffuse spread of WT1 positive tumor cells

intermingled with WT1 negative tumor cells (figure 5D), but in some cases clustering

of WT1-positive cells was present resulting in WT1-positive areas next to WT1-

negative areas. Similar WT1 immunoreactivity of tumor cells was present in

astrocytomas WHO grade II with exception of gemistocytic astrocytomas which

exhibited moderate to strong WT1 expression in most cases (figure 5E). Further,

most pilocytic astrocytomas frequently displayed WT1+ tumor cells including the

delicate piloid processes (figure 5F).

Discussion

Distinction between reactive astrogliosis and an astrocytic neoplasm is often difficult,

especially in cases when diffuse low-grade astrocytoma or the infiltration zone of a

high-grade astrocytoma is considered (9). So far no single immunohistochemical

marker exists to differentiate clearly in these cases. In recent years, attempts to

discriminate between reactive and neoplastic astrocytes have employed various

methods on molecular genetic level including in-situ hybridisations to detect

chromosomal aberrations absent in non-neoplastic tissue (50). However these

methods are elaborate, expensive and cannot replace the usefulness of

immunostaining practices.

There is a known period of transient WT1 upregulation in ependymal cells of brain

and spinal cord during embryogenesis suggesting involvement in differentiation of

glial cells (3,46). There is no evidence of WT1 reactivation in normal brain or reactive

astrogliosis so far (39). Expression of WT1 has previously been reported in

astrocytomas, although the number of tumors examined was rather low (32,39). In

the present study we were able to confirm that WT1 is indeed expressed in astrocytic

neoplasms. The consistent expression of WT1 in blood vessels of the brain even in

cases lacking WT1 expression in tumor cells serves as a good internal control for

immunohistochemistry and has been noted previously in mesotheliomas (24). WT1

expression in the vessels may come surprising at first glance, however recent studies

were able to demonstrate that WT1 transcriptionally regulates vascular endothelial

growth factor (VEGF) through interaction of its zinc finger DNA binding domain and

thus might be not only present in normal endothelium but also associated with

angiogenesis in neoplasia (16,17).

In our study we examined WT1 immunoreactivity in normal brain, reactive CNS tissue

with astrogliosis resulting from various underlying pathologies (e.g. vascular

malformations, brain metastasis, ischemia, PML) and astrocytic tumors of different

WHO grades. We were able to show that normal and reactive astrocytes do not

display any visible WT1 immunoreactivity. Our observations around ischemic CNS

lesions correlate with previous observations of Wagner and colleagues, describing

WT1 upregulation in kidney and heart but not in brain of rats kept under hypoxic

conditions (48).

In contrast to normal and reactive astrocytes, neoplastic astrocytes frequently

express WT1. In our series of 219 astrocytomas WT1+ tumor cells were found in

pilocytic astrocytomas (66.7% of cases examined), astrocytomas WHO grade II

(52.8%), astrocytoma WHO grade III (83.4%) and glioblastomas WHO grade IV

(98.1%). Overall, the majority of all astrocytic tumors (84.5%) expressed WT1 at least

focally. Thus, WT1 is significantly upregulated in astrocytomas of any WHO grade

when compared to normal brain or astrogliosis. Using score 0 (lack of WT1 positive

tumor cells) as a cut off astrocytic neoplasms can be diagnosed with 100% specificity

and 68% sensitivity. In cases where WT1 expression is present on astrocytic cells,

the diagnosis clearly points to an astrocytic neoplasm as there is an excellent

positive predictive value of 1. In cases lacking WT1 immunostaining differentiation

from reactive gliosis remains still difficult with a negative predictive value of 0.68. We

have employed WT1 in our routine diagnostic panel for glial neoplasms and reactive

gliosis for several months as well during the review process and have not

experienced a single false positive case so far. However we are very well aware that

there will be always an exception that proves the rule and expect someone to report

this. It is important to remember that WT1 is known to be expressed not only in

endothelial cells but also in fibroblasts (37). The presence of WT1 in brain tissue

therefore should not prompt a straightforward diagnosis of an astrocytoma. Rather to

rule out the presence of a glial scar on the one side and mesenchymal or

desmoplastic reaction on the other side, the pathologist should apply appropriate

additional stainings when in doubt. Moreover WT1 should not be regarded as an

astrocytoma-specific marker since its expression in other glial neoplasms

(ependymomas, oligodendrogliomas etc) has been demonstrated recently (18). Our

findings in astrocytic tumors are in accordance to a previous report by Oji et al. who

found expression of WT1 protein in 18/18 high-grade gliomas and in 5/6 low-grade

gliomas examined (39). We have limited the current immunohistochemical study to

astrocytic neoplasms since low-grade astrocytic neoplams are most likely prone to

misdiagnoses than other glial neoplasms. In the past months WT-1 staining helped to

clarify several challenging cases in our institute. Some of these cases are illustrated

in Figure 6. It is also noteworthy that the WT1 antibody can be applied on frozen

tissue as well.

It has previously been suggested by some authors that cytoplasmic staining

represents cross-reactivity with an epitope unrelated to WT1 (24). However recent

studies were able to demonstrate presence of WT1 in the cytoplasm via western blot

(36) and RT-PCR (32). In addition immunohistochemical staining of WT1 in the

cytoplasm of several types of cancer has been demonstrated (36). This suggests that

strong overexpression leads to inactive state of the WT1 transcription activity and

after phosphorylation it shuttles back into the nucleus, where it is able to bind DNA

(37). In addition recent identification of an immunogenic T-cell binding epitope within

WT1 suggests a specific tumor-antigen directed immunological approach in therapy

(6).

Tumor-restricted antigens that are not expressed in normal and/or reactive astrocytes

include the intracellular vimentin-binding 300-kilodalton intermediate filament

associated protein IFAP-300kDa (29). Its expression in radial glia of the embryonic

central nervous system is terminated postnatally and re-expression occurs only

during tumorigenesis as demonstrated in seven high-grade gliomas and cultured

cells (52). To date however no informations on expression in low-grade astrocytomas

or in reactive astrocytes have been provided. Integrin alpha3, a 140-kDa protein is

reported to be present in malignant gliomas but does not show antibody reaction

within the normal brain (22). As a cell membrane adhesion molecule its expression

might be a potential target for immunotherapy, yet its role in reactive astrocytes still

needs to be elucidated. One of the best evaluated proteins is the mutated isoform of

the Epidermal Growth Factor Receptor (EGFRvIII). Considered as a marker for

malignancy it is expressed in approximately half of all gliomas examined (2,51),

however immunohistochemical results on paraffin-embedded specimens remain poor

(7). Another interesting marker is the tumor-specific murine antibody 8H9 directed

against a 58 kDa surface antigen expressed among neuroectodermal, mesenchymal,

and epithelial tumors, which has been reported in 15/17 frozen glioblastoma and 6/8

frozen astrocytoma samples and does not react with normal human brain (33).

Recombinant immunotoxins raised against this epitope are currently tested against

various types of cancer (40). To date no information for 8H9 on paraffin-embedded

specimens exists. The common acute lymphoblastic antigen (cALLA, CD10), also

known as neutral endopeptidase was suggested not only as a marker of malignancy

progession but also considered lacking expression in reactive astrocytes in frozen

tissue sections (34). Although showing promising results, routine use of the

previously mentioned antibodies is limited because the staining of paraffin-embedded

specimens is the method of choice in diagnostic pathology.

In conclusion, WT1 immunohistochemistry is a valid and promising diagnostic tool to

distinguish neoplastic astrocytes from normal or reactive astrocytes in routinely

paraffin-embedded tissue. We have also shown that the used scoring method shows

substantial agreement and therefore is a valid and reproducible tool for quick

assessment of WT1 expression.

Acknowledgments

We wish to thank Gudrun Albrecht for the arrangement of the photomicrographs and

Andrea Bob-Schittenhelm for her help with the English.

References

1.Agapova OA, Ricard CS, Salvador-Silva M, Yang P, Hernandez MR (2001)

Expression of matrix metalloproteinases and tissue inhibitors of

metalloproteinases in human optic nerve head astrocytes. Glia 33:205-216.

2.Aldape KD, Ballman K, Furth A, Buckner JC, Giannini C, Burger PC,

Scheithauer BW, Jenkins RB, James CD (2004) Immunohistochemical detection

of EGFRvIII in high malignancy grade astrocytomas and evaluation of prognostic

significance. J Neuropathol Exp Neurol 63:700-707.

3.Armstrong JF, Pritchard-Jones K, Bickmore WA, Hastie ND, Bard JB (1993).

The expression of the Wilms' tumour gene, WT1, in the developing mammalian

embryo. Mech Dev 40:85-97.

4.Asano K, Kubo O, Tajika Y, Huang MC, Takakura K, Ebina K, Suzuki S (1997)

Expression and role of cadherins in astrocytic tumors. Brain Tumor Pathol 14;27-

33.

5.Aschner M, Sonnewald U, Tan KH (2002) Astrocyte modulation of neurotoxic

injury. Brain Pathol 12:475-481.

6.Asemissen AM, Keilholz U, Tenzer S, Müller M, Walter S, Stevanovic S, Schild

H, Letsch A, Thiel E, Rammensee HG, Scheibenbogen C (2006) Identification of

a highly immunogenic HLA-A*01-binding T cell epitope of WT1. Clin Cancer Res

12:7476-7482.

7.Biernat W, Huang H, Yokoo H, Kleihues P, Ohgaki H (2004) Predominant

expression of mutant EGFR (EGFRvIII) is rare in primary glioblastomas. Brain

Pathol 14:131-136.

8.Buckler AJ, Pelletier J, Haber DA, Glaser T, Housman DE (1991) Isolation,

characterization, and expression of the murine Wilms' tumor gene (WT1) during

kidney development. Mol Cell Biol 11:1707-1712.

9.Burger PC, Scheithauer BW, Vogel FS (2002) The Brain – Tumors. In: Surgical

Pathology of the nervous system and its coverings Burger PC, Vogel FC (eds), pp

160-378, 4th edn. Elsevier: Oxford.

10.Eddleston M, Mucke L (1993) Molecular profile of reactive astrocytes--

implications for their role in neurologic disease. Neuroscience 54:15-36.

11.Eng LF, Ghirnikar RS (1994) GFAP and astrogliosis. Brain Pathol 4:229-237.

12. Ellison D, Love S, Chimelli L, Harding BN, Lowe J, Vinters HV (2004)

Neuropathology – a reference text of CNS pathology, 2nd edn. Mosby: London.

13.Forsyth RJ (1996) Astrocytes and the delivery of glucose from plasma to

neurons. Neurochem Int 28:231-241.

14.Gaudin PB, Sherman ME, Brat DJ, Zahurak M, Erozan YS (1997) Accuracy of

grading gliomas on CT-guided stereotactic biopsies: a survival analysis. Diagn

Cytopathol 17:461-466.

15.Giaume C, Venance L (1998) Intercellular calcium signaling and gap junctional

communication in astrocytes. Glia 24:50-64.

16.Graham K, Li W, Williams BR, Fraizer G. Vascular endothelial growth factor

(VEGF) is suppressed in WT1-transfected LNCaP cells (2006) Gene Expr 13: 1-

14.

17.Hanson J, Gorman J, Reese J, Fraizer G (2007) Regulation of vascular

endothelial growth factor, VEGF, gene promoter by the tumor suppressor, WT1.

Front Biosci 12:2279-2290.

18. Hashiba T, Izumoto S, Kagawa N, Suzuki T, Hashimoto N, Maruno M,

Yoshimine T (2007) Expression of WT1 protein and correlation with cellular

proliferation in glial tumors. Neurol Med Chir 47:165-70.

19.Hastie ND (1994) The genetics of Wilms' tumor--a case of disrupted

development. Annu Rev Genet 28:523-558.

20.Junier MP (2000) What role(s) for TGFalpha in the central nervous system?

Prog Neurobiol 62:443-473.

21.Kerr JF, Winterford CM, Harmon BV (1994) Apoptosis. Its significance in

cancer and cancer therapy. Cancer 73:2013-2026.

22.Kishima H, Shimizu K, Tamura K, Miyao Y, Mabuchi E, Tominaga E,

Matsuzaki J, Hayakawa T (1999) Monoclonal antibody ONS-M21 recognizes

integrin alpha3 in gliomas and medulloblastomas. Br J Cancer 79:333-339.

23.von Deimling A, Burger PC, Nakazato Y, Ohgaki H, Kleihues P (2007) Diffuse

astrocytoma. In: WHO classification of Tumours: Pathology & Genetics: Tumours

of the Nervous System. DN Louis, H Ohgaki, OD Wiestler, WK Cavenee (eds) pp

25-29. International Agency for Research on Cancer: Lyon.

24.Kumar-Singh S, Segers K, Rodeck U, Backhovens H, Bogers J, Weyler J, Van

Broeckhoven C, Van Marck E (1997) WT1 mutation in malignant mesothelioma

and WT1 immunoreactivity in relation to p53 and growth factor receptor

expression, cell-type transition, and prognosis. J Pathol 181:67-74.

25.Kuppner MC, Van Meir E, Gauthier T, Hamou MF, de Tribolet N (1992)

Differential expression of the CD44 molecule in human brain tumours. Int J

Cancer 50:572-577.

26.Kurtkaya-Yapicier O, Scheithauer BW, Hebrink D, James CD (2002) p53 in

nonneoplastic central nervous system lesions: an immunohistochemical and

genetic sequencing study. Neurosurgery 51:1246-1254.

27.Landis JR, Koch GG (1977) The Measurement of Observer Agreement for

Categorical Data. Biometrics 33: 159-174.

28.Lawley LP, Cerimele F, Weiss SW, North P, Cohen C, Kozakewich HP,

Mulliken JB, Arbiser JL (2005) Expression of Wilms tumor 1 gene distinguishes

vascular malformations from proliferative endothelial lesions. Arch Dermatol

141:1297-1300.

29.Lieska N, Yang HY, Goldman RD (1985) Purification of the 300K intermediate

filament-associated protein and its in vitro recombination with intermediate

filaments. J Cell Biol 101:802-813.

30.Liu B, Chen H, Johns TG, Neufeld AH (2006) Epidermal growth factor receptor

activation: an upstream signal for transition of quiescent astrocytes into reactive

astrocytes after neural injury. J Neurosci 26:7532-7540.

31.Maheswaran S, Park S, Bernard A, Morris JF, Rauscher FJ 3rd, Hill DE, Haber

DA (1993) Physical and functional interaction between WT1 and p53 proteins.

Proc Natl Acad Sci U S A 90:5100-5104.

32.Menssen HD, Bertelmann E, Bartelt S, Schmidt RA, Pecher G, Schramm K,

Thiel E (2000) Wilms' tumor gene (WT1) expression in lung cancer, colon cancer

and glioblastoma cell lines compared to freshly isolated tumor specimens. J

Cancer Res Clin Oncol 126:226-232.

33.Modak S, Kramer K, Gultekin SH, Guo HF, Cheung NK (2001) Monoclonal

antibody 8H9 targets a novel cell surface antigen expressed by a wide spectrum

of human solid tumors. Cancer Res 61:4048-4054.

34.Monod L, Hamou MF, Ronco P, Verroust P, de Tribolet N (1992) Expression of

cALLa/NEP on gliomas: a possible marker of malignancy. Acta Neurochir 114:3-7.

35.Nagasaka S, Tanabe KK, Bruner JM, Saya H, Sawaya RE, Morrison RS

(1995) Alternative RNA splicing of the hyaluronic acid receptor CD44 in the

normal human brain and in brain tumors. J Neurosurg 82:858-863.

36.Nakatsuka S, Oji Y, Horiuchi T, Kanda T, Kitagawa M, Takeuchi T, Kawano K,

Kuwae Y, Yamauchi A, Okumura M, Kitamura Y, Oka Y, Kawase I, Sugiyama H,

Aozasa K (2006) Immunohistochemical detection of WT1 protein in a variety of

cancer cells. Mod Pathol 19:804-814.

37. Niksic M, Slight J, Sanford JR, Caceres JF, Hastie ND (2004) The Wilms'

tumour protein (WT1) shuttles between nucleus and cytoplasm and is present in

functional polysomes. Hum Mol Genet 13:463-471.

38.Norenberg MD (1994) Astrocyte responses to CNS injury. J Neuropathol Exp

Neurol 53:213-220.

39.Oji Y, Suzuki T, Nakano Y, Maruno M, Nakatsuka S, Jomgeow T, Abeno S,

Tatsumi N, Yokota A, Aoyagi S, Nakazawa T, Ito K, Kanato K, Shirakata T,

Nishida S, Hosen N, Kawakami M, Tsuboi A, Oka Y, Aozasa K, Yoshimine T,

Sugiyama H (2004) Overexpression of the Wilms' tumor gene WT1 in primary

astrocytic tumors. Cancer Sci 95:822-827.

40.Onda M, Wang QC, Guo HF, Cheung NK, Pastan I (2004) In vitro and in vivo

cytotoxic activities of recombinant immunotoxin 8H9(Fv)-PE38 against breast

cancer, osteosarcoma, and neuroblastoma. Cancer Res 64:1419-1424.

41.Planas AM, Justicia C, Soriano MA, Ferrer I (1998) Epidermal growth factor

receptor in proliferating reactive glia following transient focal ischemia in the rat

brain. Glia 23:120-129.

42.Prayson RA, Agamanolis DP, Cohen ML, Estes ML, Kleinschmidt-DeMasters

BK, Abdul-Karim F, McClure SP, Sebek BA, Vinay R (2000) Interobserver

reproducibility among neuropathologists and surgical pathologists in fibrillary

astrocytoma grading. J Neurol Sci 175:33-39.

43.Rasheed BK, Bigner SH (1991) Genetic alterations in glioma and

medulloblastoma. Cancer Metastasis Rev 10:289-299.

44.Redies C (2000) Cadherins in the central nervous system. Prog Neurobiol

61:611-648.

45.Rust RS (1994) Energy metabolism of developing brain. Curr Opin Neurol

7:160-165.

46.Scharnhorst V, Kranenburg O, van der Eb AJ, Jochemsen AG (1997)

Differential regulation of the Wilms' tumor gene, WT1, during differentiation of

embryonal carcinoma and embryonic stem cells. Cell Growth Differ 8:133-143.

47.Timár J, Mészáros L, Orosz Z, Albini A, Rásó E (2005) WT1 expression in

angiogenic tumours of the skin. Histopathology 47:67-73.

48.Wagner KD, Wagner N, Wellmann S, Schley G, Bondke A, Theres H, Scholz H

(2003) Oxygen-regulated expression of the Wilms' tumor suppressor Wt1 involves

hypoxia-inducible factor-1 (HIF-1). FASEB J 17:1364-1366.

49.Waxman SG, Sontheimer H, Black JA, Minturn JE, Ransom BR (1993)

Dynamic aspects of sodium channel expression in astrocytes. Adv Neurol 59:135-

155.

50.Wessels PH, Hopman AH, Ummelen MI, Krijne-Kubat B, Ramaekers FC,

Twijnstra A (2001) Differentiation between reactive gliosis and diffuse

astrocytoma by in situ hybridization. Neurology 56:1224-1227.

51.Wikstrand CJ, Hale LP, Batra SK, Hill ML, Humphrey PA, Kurpad SN,

McLendon RE, Moscatello D, Pegram CN, Reist CJ, Traweek ST, Wong AJ,

Zalutsky MR, Bigner DB (1995) Monoclonal antibodies against EGFRvIII are

tumor specific and react with breast and lung carcinomas and malignant gliomas.

Cancer Res 55:3140-3148.

52.Yang HY, Lieska N, Glick R, Shao D, Pappas GD (1993) Expression of 300-

kilodalton intermediate filament-associated protein distinguishes human glioma

cells from normal astrocytes. Proc Natl Acad Sci U S A 90:8534-8537.

53.Yamada T, Kawamata T, Walker DG, McGeer PL (1992) Vimentin

immunoreactivity in normal and pathological human brain tissue. Acta

Neuropathol 84:157-162.

54.Yaziji H, Massarani-Wafai R, Gujrati M, Kuhns JG, Martin AW, Parker JC

(1996) Role of p53 immunohistochemistry in differentiating reactive gliosis from

malignant astrocytic lesions. Am J Surg Pathol 20:1086-1090

55.Yung WK, Luna M, Borit A (1985) Vimentin and glial fibrillary acidic protein in

human brain tumors. J Neurooncol 3: 35-38.

56.Zagzag D, Miller DC, Kleinman GM, Abati A, Donnenfeld H, Budzilovich GN

(1993) Demyelinating disease versus tumor in surgical neuropathology. Clues to

a correct pathological diagnosis. Am J Surg Pathol 17:537-545.

57.Zeltner L, Schittenhelm J, Mittelbronn M, Roser F, Tatagiba M, Mawrin C, Kim

YJ, Bornemann A (2007) The astrocytic response towards invasive meningiomas.

Neuropathol Appl Neurobiol 33:163-168.

Figure legends:

Figure 1. WT1 immunoreactivity in normal brain and Wilms tumor.

A-C: In normal brain specimens expression of WT1 is restricted to endothelial cells,

both in gray (A) and in white matter (B). No WT1 expression is found in glial cells (A,

B). In Wilms tumor serving as positive control the vast majority of tumor cells and

endothelial cells showed strong nuclear expression of WT1 (C) (scale bar: 100µm).

Figure 2. WT1 immunoreactivity in reactive astrogliosis.

A-C: Thin pathological vessels of a cavernoma (A, H&E) are surrounded by reactive

astrocytes with eosinophilic cytoplasm (inset). Reactive astrocytes are strongly

positive for glial fibrillary acidic protein (GFAP, B), whereas WT1 expression remains

restricted to endothelial cells of capillaries and cavernomatous vessels (C). D-F:

Residual infarction aged 4 months shows glial scar formation with reactive astrocytes

(D, H&E), which express GFAP (E). In the same region WT1 immunoreactivity

remains also restricted to endothelial cells (F). G-I: Astrogliosis due to a brain

metastasis of an adenocarcinoma (G, HE). Reactive astrocytes express GFAP (H)

but not WT1 (I). Note WT1 expression in tumor vessels and in brain capillaries (scale

bars: A-I, 200µm; insets, 50µm).

Figure 3. Illustration of WT1 immunoreactivity scores.

Representative photomicrographs illustrating WT1 immunoreactivity scores in

glioblastomas. (A) WT1 expression is limited to endothelial cells of capillaries (score

0). (B) Beside few WT1 positive cell processes also single (<1%) WT1+ tumor cells

(arrows) are present (score 1). Glioblastomas showing WT1 expression in 1-19% (C,

score 2), in 20 to 50% (D, score 3) or in more than 50% of tumor cells (E, score 4).

Isotype control demonstrating specificity of the primary WT1 antibody (F) (scale bar:

A-F, 100µm).

Figure 4. Distribution of different WT1 scores in neoplastic glial cells.

Overall illustration of the frequency of different WT1 scores in astrocytomas grade I-

IV WHO. Note absence of WT1 expression in normal brain and astrogliosis (all cases

score 0, not shown). In astrocytomas of each WHO grade all WT1 scores are present

in variable frequency, respectively.

Figure 5. WT1 immunoreactivity in astrocytomas.

A-F: Immunostaining for WT1 in astrocytic neoplasms. A: Glioblastoma (grade IV

WHO) with areas of pseudopalisading necrosis (score 4), B: Glioblastoma with WT1

enhancement around pathological tumor vessels (score 3), C: Another glioblastoma

with WT1 expression predominating in pleomorphic tumor cells (score 3), D:

Anaplastic astrocytoma, grade III WHO (score 3), E: Gemistocytic astrocytoma grade

II WHO (score 2) and F: Pilocytic astrocytoma grade I WHO (score 3) show

cytoplasmic WT1 in varying number of positive cells (scale bars: A-F, 100µm).

Figure 6. Diagnostic value of WT1 in daily routine.

A-C: 69 year old patient with a left paraventricular mass lesion clinically foremost

considered being a glioblastoma. Stereotactic biopsy specimen showed weak to

moderate increase in cellularity and moderate nuclear pleomorphism but no mitotic

activity, necrosis or microvascular proliferation (A, HE stain). Therefore differential

diagnosis between reactive astrogliosis and infiltration zone of an astrocytoma was

difficult. Assumption of an infiltration zone being more likely was strengthened by

WT1 expression in astrocytic cells (B). The surgery specimen proved neoplastic

nature of the astrocytic lesion and was finally diagnosed as a glioblastoma (C, HE

stain).

D-F: 59-year old patient with a clinically unclear parahippocampal lesion foremost a

glioblastoma or a lymphoma. No steroids were applied prior to biopsy. Due to

complications during the stereotactic procedure only very small tissue specimen

could be achieved for intraoperative diagnosis. The frozen section revealed a

moderate increase in cell density with some gemistocytic astrocytes (D, HE stain).

WT1 expression on frozen-tissue was limited to endothelial cells (E) and there were

no Ki67-positive cells (not shown). Therefore a reactive astrogliosis was considered

to be more likely than an infiltration zone of a glioma. During the following weeks a

rapid spread of the lesion further favoured clinically a glioblastoma. Two months after

stereotaxis an open surgical biopsy was performed yielding the diagnosis of a highly

malignant primary cerebral B-cell lymphoma (F).

Tables

Table 1:

CNS tissue samples with reactive

astrogliosis

No. of cases

(n=76) male/female mean age

Cavernous hemangioma 12 5/7 37

Arteriovenous-Malformation 2 0/2 42

Carcinoma, metastatic 15 10/5 67.6

Meningioma, brain-invasive 6 4/2 60.0

Melanoma 7 3/4 62.1

Ischemia 4 1/3 65.7

Progressive multifocal leukoencephalopathy 2 2/0 55.0

Normal human brain 28 15/13 55.4

Table 2:

Tumor type WHO grade

No. of cases

(n=219) male/female

mean age

Pilocytic astrocytoma I 24 14/10 21.8

Diffuse astrocytoma II 36 24/12 43.9

Anaplastic astrocytoma III 54 36/18 56.1

Glioblastoma IV 105 63/42 58.3

Table 3:

Tissue type Score 0 Score 1 Score 2 Score 3 Score 4

Normal brain 28 0 0 0 0

Gliosis 48 0 0 0 0

Pilocytic astrocytoma 8 6 7 3 0

Diffuse astrocytoma 17 7 8 4 0

Anaplastic astrocytoma 9 13 15 12 5

Glioblastoma 2 8 29 37 29

Table 4:

Level A Level B p-value Gliosis Normal brain (n.s.) all values = 0 Gliosis WHO I <0.0001*** Gliosis WHO II <0.0001*** Gliosis WHO III <0.0001*** Gliosis WHO IV <0.0001*** Normal brain WHO I <0.0001*** Normal brain WHO II <0.0001*** Normal brain WHO III <0.0001*** Normal brain WHO IV <0.0001*** WHO I WHO II 0.3631 (n.s.) WHO I WHO III 0.04 (n.s.) WHO I WHO IV <0.0001*** WHO II WHO III 0.001** WHO II WHO IV <0.0001*** WHO III WHO IV <0.0001*** WHO I-IV Normal brain <0.0001*** WHO I-IV Gliosis <0.0001*** n.s. = not significant, values after Bonferroni-Holm correction