Expression of tenascin-C in various human brain tumors and its relevance for survival in patients with astrocytoma

Authors

  • Alexander Leins M.D.,

    1. Institute of Neuropathology, Ludwig-Maximilians-University, Munich, Germany
    2. Department of Microscopical Anatomy, University of Hamburg, Hamburg, Germany
    Current affiliation:
    1. Department of General and Thoracic Surgery, University of Kiel, Germany
    Search for more papers by this author
  • Pietro Riva M.D.,

    1. Department of Nuclear Medicine, Municipal Hospital “M. Bufalini,” Cesena, Italy
    Search for more papers by this author
  • Ragnar Lindstedt Ph.D.,

    1. Immunology Department, Sigma Tau S.p.A., Pomezia, Rome, Italy
    Search for more papers by this author
  • Michail S. Davidoff M.D.,

    1. Department of Microscopical Anatomy, University of Hamburg, Hamburg, Germany
    Search for more papers by this author
  • Parviz Mehraein M.D.,

    1. Institute of Neuropathology, Ludwig-Maximilians-University, Munich, Germany
    Search for more papers by this author
  • Serge Weis M.D.

    Corresponding author
    1. Institute of Neuropathology, Ludwig-Maximilians-University, Munich, Germany
    2. Laboratory for Brain Research and Neuropathology, Stanley Medical Research Institute, Bethesda, Maryland
    • Laboratory for Brain Research and Neuropathology, Stanley Medical Research Institute, 5430 Grosvenor Lane, Suite 200, Bethesda, MD 20814
    Search for more papers by this author
    • Fax: (301) 295-6276


Abstract

BACKGROUND

Tenascin-C (TN-C), a large extracellular matrix (ECM) glycoprotein with a molecular weight of 180–250 kilodaltons, is present in several normal adult tissues. TN-C is up-regulated during embryogenesis, in wound healing, and in tumor tissues. Glioblastoma multiforme (GBM) is the most frequent and malignant astrocytic tumor comprised of poorly differentiated, neoplastic astrocytes. Recently, TN-C-based radioimmunotherapy was administered to patients with GBM.

METHODS

In the current study, the authors used immunohistochemistry to conduct a systematic investigation of TN-C distribution patterns in normal human brain tissue and in a large variety of brain tumors (n = 485 tumors). Immunoreactivity for TN-C was assessed with regard to its localization within tumor cells, blood vessels, and ECM using three different monoclonal antibodies (clones BC2, BC4, and TN2).

RESULTS

In control human brains, a significant difference was noted in the expression of TN-C when comparing gray with white matter using either Western blot analysis or immunohistochemistry. TN-C was found in the white matter of the frontal, temporal, parietal, and occipital lobes and in the hippocampus, where the immunoreaction was especially strong in the hippocampal formation. In 181 astrocytomas of different grades (World Health Organization [WHO] Grade 2–4), TN-C immunopositivity was seen to varying degrees in the cellular and stromal components of the tumor and in tumor-associated vessels. Glioblastomas (n = 113 tumors) showed strong immunopositivity in the vessels and moderate immunopositivity of the ECM. A statistically significant reduction of TN-C immunopositivity in tumor-associated vessels or ECM was observed in anaplastic astrocytomas (WHO Grade 3) compared with GBM (WHO Grade 4). A Kaplan–Meier analysis showed that patients who had GBM lesions that lacked TN-C immunopositivity in the ECM had a significantly longer survival (median, 28 months; standard error, 7.8 months) (n = 12 patients) compared with patients who had GBM lesions with TN-C immunopositivity (median, 12 months; standard error, 1.6 months) (n = 87 patients). In meningiomas (n = 24 tumors), the neoplastic cells, the ECM of the tumor, and the vessels were TN-C negative. In schwannomas (n = 31 tumors), the tumor cells were TN-C negative; whereas, in > 50% of tumors, the vessels and the ECM of regressively altered tumor areas were positive. In metastatic carcinomas (n = 53 tumors), the tumor cells were negative; seldom were vessels stained positive for TN-C. Focal areas of the ECM, often accompanied with fibrotic changes, were immunopositive for TN-C.

CONCLUSIONS

The most constant TN-C immunopositivity was noted in the ECM of the fibrotic stroma in highly malignant brain tumors and along the tumor border, especially in high-grade astrocytomas. The current results suggest that TN-C expression may be correlated with the grade of malignancy in astrocytic tumors and that the presence or absence of TN-C expression in the stroma of astrocytic tumors may play a not yet clearly understood role in shortening or prolonging, respectively, the survival of patients. Cancer 2003. © 2003 American Cancer Society.

Tenascin C (TN-C), a large extracellular matrix (ECM) glycoprotein, was described first by Bourdon et al.1 as a tumor-specific antigen (glioma-mesenchymal ECM antigen); it is also known as myotendinous antigen, cytotactin, JI 220/200 glycoprotein, and hexabrachion.2, 3 TN-C is characterized by a six-armed quaternary structure and a modular construction.3 The molecular structure of TN-C is determined by a linear array of the following four subunits: 1) a cysteine-rich, amino terminal domain; 2) a sequence of epidermal growth factor (EGF)-like repeats; 3) a number of fibronectin type III (FN-III) repeats; and 4) a carboxy-terminal domain homologous to fibrinogen. There are several polymorphic isoforms of TN-C that result from alternative splicing of nine of the FN-III repeats.4 Large TN-C isoforms show a time dependent expression during embryogenesis at sites of tissue interactions and organogenesis, a restricted distribution in normal adult tissues, and a reexpression or neoexpression in pathologically altered tissues, including neoplasms. Epithelial-mesenchymal and neuronal-glial cell interactions are essential in the process of growth and differentiation during embryogenesis and fetogenesis.5, 6 Furthermore, ECM proteins act as substrate for tumor cell attachment and motility.7

Glioblastoma multiforme (GBM) is the most malignant astrocytic tumor composed of poorly differentiated, neoplastic astrocytes. GBM is the most frequent brain tumor, accounting for approximately 12–15% of all intracranial neoplasms. Bourdon et al.1 showed high levels of TN-C in GBM using the monoclonal antibody (MoAb) 81C6. Other researchers subsequently confirmed this finding.8–14 Because the clinical outcome of patients with GBM still is very poor, several groups started treating patients with GBM using radiolabeled monoclonal anti-TN-C antibodies, which were placed into the postoperative cavity using an indwelling (Rickam or Omaya) catheter.15–24 Riva et al.15 reported that the average overall survival of 9–10 months in patients with GBM could be prolonged to 19–25 months using radioimmunotherapy. No systematic data exist in the literature about the frequency and cellular localization or the relation of TN-C to overall survival in patients with other brain tumors, such as low-grade astrocytomas, meningiomas, and schwannomas. Therefore, in the current study, we conducted a comparative investigation on the immunoreactivity of TN-C in normal brain tissue and in a large number of various brain tumor entities to obtain data about the localization of TN-C in tumor cells, blood vessels, and ECM. Furthermore, we analyzed the possible role of TN-C expression in the overall survival of patients with GBM.

MATERIALS AND METHODS

Tissue Samples

Tissues from 21 normal human control brains (age range, 35–93 years; death by accident or by cardiovascular failure and absence of neuropathologic changes in the brain tissue), including the frontal, parietal, temporal, and occipital lobes as well as the hippocampal formation, thalamus, hypothalamus, basal ganglia, brain stem, and cerebellum, and from surgical samples of 483 intracranial tumors were investigated. A complete list of the examined tumors is provided in Table 1. Briefly, the following tumor entities were investigated: astrocytomas (World Health Organization [WHO] Grade 1–4), gliosarcomas, oligodendrogliomas, oligoastrocytomas, ependymal tumors, choroid plexus tumors, central neurocytomas, dysembryoblastic neuroepithelial tumors, primitive neuroectodermal tumors, medulloblastomas, pineocytomas, pineoblastomas, hemangioblastomas, germ cell tumors, meningiomas, hemangiopericytomas, lipomas, chondrosarcomas, rhabdomyosarcomas, sarcomas, paragangliomas, lymphomas, schwannomas, neurofibromas, malignant peripheral nerve sheath tumors, craniopharyngiomas, and metastatic tumors to the brain. Each sample underwent a conventional histopathologic staining procedure for diagnostic purposes. The applied criteria for tumor diagnosis followed the outlines of the WHO classification system25 and its recent, updated version26 and by the Atlas of Tumor Pathology of the Armed Forces Institute of Pathology.27

Table 1. Immunopositive Results for Tenascin-C in Different Tumor Entities
Tumor typePositive results
In and around vesselsIn the ECM
%Absolute no.%Absolute no.
  • ECM: extracellular matrix; WHO 1–4: World Health Organization Grade 1–4.

  • a

    Low immunopositivity.

  • b

    Strong immunopositivity.

  • c

    Tenasein-C-positivity only in fibrotic degenerations.

  • d

    Small (endothelial) vessels.

  • e Vascular channels.

  • f

    Only reticulin structure.

Astrocytoma (WHO 2)4216/389235a/38
Anaplastic astrocytoma (WHO 3)267/279626b/27
Glioblastoma multiforme (WHO 4)8596/11389101b/113
Giant cell glioblastoma (WHO 4)1002/21002b/2
Gliosarcoma (WHO 4)1001/11001c/1
Pilocytic astrocytoma (WHO 1)804/5603/5
Oligodendroglioma (WHO 2)5011/2210022/22
Oligoastrocytoma (WHO 2)1001/11001/1
Anaplastic oligodendroglioma (WHO 3)835/61006/6
Ependymoma9416/17295c/17
Choroid plexus papilloma00/4753c/4
Central neurocytoma00/100/1
Dysembryoblastic neuroepithelial tumor00/11001/1
Primitive neuroectodermal tumor1004/41004/4
Ependymoblastoma1001/11001/1
Medulloblastoma8113/16315/16
Meningioma41/24133c/24
Schwannoma (WHO 1)7122/313912/31
Neurofibroma (WHO 1)1006/6835/6
Malignant peripheral nerve sheath tumor00/11001a/1
Pineocytoma00/11001a/1
Pineoblastoma1002/21002/2
Hemangioblastoma4920d/415714/41
 886c41
Germinoma00/5603a/5
Hemangiopericytoma00/4753a/4
Paraganglioma00/100/1
Lymphoma (non-Hodgkin)8320/246716a/24
Craniopharyngioma00/228619c/22
Lipoma00/200/2
Chondrosarcoma00/4502a/4
Rhabdomyosarcoma00/21002/2
Sarcoma00/31003/3
Adenocarcinoma (metastasis)6018/306720a/30
Nephroblastoma (metastasis)674/6674a/6
Malignant melanoma (metastasis)8411/139212f/13
Undifferentiated metastasis to the brain00/4753c/4

Monoclonal Antibodies

The following MoAbs were used: 1) BC2 directed against the splicing domains A1 and A4 of human TN-C28 (provided by Ragnar Lindstedt, Ph. D., Immunology Department, Sigma Tau S.p.A., Pomezia, Rome, Italy), 2) BC4 directed against the EGF-like repeats of human TN-C28 (provided by Pietro Riva, M.D., Department of Nuclear Medicine, “M. Bufalini” Hospital, Cesena, Italy), and 3) TN2 directed against intact human TN-C (Dako Diagnostics GmbH, Hamburg, Germany).

Western Blot Analysis

For preparation of protein extracts, frozen brain tumor tissue and frozen normal human brain tissue were disrupted mechanically with Mikro Dismembrator (Braun Biotech International, Germany) and lysed in homogenization buffer (25 mM Na2HPO4/NaH2PO4, 5 mM ethylenediamine tetraacetic acid (EDTA), 0.1 mM dithiothreitol, and 1.0 mM phenyl methyl sulfonyl fluoride in distilled H2O). The homogenates were prepared by centrifugation at × 3000 g for 8 minutes at4 °C), and protein concentration was determined with a BCA Kit (Pierce Chemical Company, Rockford, IL). One hundred micrograms of protein from normal brain homogenates and 50 μg of brain tumor homogenates were loaded separately on a standard 5% sodium dodecyl sulfate polyacrylamide gel under reducing conditions. Fractionated proteins were blotted onto a Nitrocellulose membrane (Amersham Pharmacia Biotech, Buckinghamshire, U.K.). After blocking with low-fat milk for 2 hours, TN-C was detected using the 3 different monoclonal mouse anti-TN-C antibodies (BC2, BC4, and TN2) in a concentration of 0.436 ng/μL for each antibody. The secondary antibody was horseradish-peroxidase conjugated goat antimouse immunoglobulin G (Pierce Chemical Company). The peroxidase was visualized by reacting with the enhanced chemiluminescence detection reagent (Amersham Pharmacia Biotech) and Super RX film (Fuji Photo Film [Europe] GmbH, Germany). For the positive control, we loaded 1.25 μg purified human TN-C protein (Chemicon International Inc., Temecula, CA); for the negative control, we replaced the primary TN-C antibodies with nonimmune serum. Both probes were processed as described above.

Immunohistochemistry

For each case, 4-μm-thick sections were cut from paraffin blocks, 1 for neuropathologic evaluation and 1 for each MoAb for the immunohistochemical staining procedure. The sections were deparaffinized, rehydrated, and stained using a standard avidin-biotin-peroxidase method with diaminobenzidine as the final chromogen. For all primary antibodies, we used a concentration of 5.5 ng/μL. Predigestion with 0.02% protease (Protease XXIV; Sigma, Germany) in 0.05 M Tris-HCL buffer for 7 minutes was used for clones BC4 and TN2. For clone BC2, we used heating for antigen retrieval with a Tris-EDTA-citrate buffer in an autoclave (20 minutes, 1.5 bar, 135 °C). All immunohistochemical sections, except for the sections of melanoma brain metastases, were counterstained slightly with hematoxylin. For the samples of melanoma metastases, we used the Vector SG substrate staining kit (Vector Laboratories, Burlingame, CA) and counterstained with kernechtrot. For negative controls, the primary antibodies were replaced by nonimmune serum. To control the specificity of the MoAbs, sections from human liver were used, as described previously (Natali et al.13). To verify the specificity of the MoAbs, an affinity absorption test was performed using positive controls and preincubation with human TN-C purified protein (Chemicon International Inc.). Reduced immunoreactivity was observed at 0.025 mM of the human TN-C purified protein (Chemicon International Inc.). The sections were reviewed independently by two neuropathologists. The following localizations of the immunoreaction were considered: 1) intracellular, 2) perivascular/vascular, and 3) ECM. The immunostaining intensity was assessed separately using the following rating scale: no TN-C detected (−), low TN-C immunoreactivity (+), moderate TN-C immunoreactivity (++), and strong TN-C immunoreactivity (+++).

Evaluation of the Immunohistochemically Stained Tissues and Statistical Methods

For statistical purposes, the collected data were sorted into two groups, one with TN-C expression (low, moderate, and strong) and the other without TN-C expression. For each entity, the percentage of TN-C negativity and TN-C positivity in the three different localizations was determined. Differences between groups were assessed using chi-square statistics from the SPSS software package (SPSS, Inc., Chicago, IL). To analyze the differences in survival of patients with GMB, we used the survival analysis method of Kaplan and Meier.

RESULTS

Normal Brain Tissue

A statistically significant difference in TN-C immunoreactivity was noted between gray matter and white matter, as shown in Figure 1A and in Figure 2 (lanes 3 and 4). The ECM of the white matter of the telencephalic lobes in 70% of all 21 specimens displayed a low-to-moderate TN-C immunopositivity. In contrast, in only a few specimens (3 of 21 specimens) were the vessels stained positively. Weak TN-C immunopositivity of the ECM was found only in the first layer of the gray matter (6.87% of 21 specimens), but there was no TN-C staining in the vessels observed. Moderate-to-strong immunopositivity was detected in the temporal lobe that was pronounced markedly in the ECM of the CA-1 to CA-2 sectors and the perforant path. No TN-C positivity was detected in the vessels. Furthermore, we examined the hippocampus and temporal lobe in resected specimens from 12 patients with mesial temporal lobe epilepsy and noted that, in addition to the above findings, there was moderate immunopositivity in the ECM in areas with sclerotic degeneration. All other areas in the normal brain showed no TN-C expression.

Figure 1.

(A) Typical difference in staining for tenascin-C (TN-C) in the extracellular matrix (ECM) of the gray and white matter in normal human temporal cortex. (B) Typical staining for TN-C in the ECM of an anaplastic astrocytoma (World Health Organization [WHO] Grade 3). (C) Intracytoplasmic TN-C staining of astrocytes embedded in the TN-C negative ECM of a untreated patient with glioblastoma multiforme (GBM). (D) ECM around infiltrating tumor cells in the surrounding normal brain of a GBM lesion. (E) TN-C positivity in fibrotic changes in schwannoma. (F) Typical view of a primary central nervous system lymphoma with TN-C positivity along the onion ring-like and reticulin-like structures around cerebral vessels Original magnification × 20 (D–F); × 40 (A–C).

Figure 2.

Western blot showing negative control (Lane 1), tenascin-C (TN-C) protein (Lane 2), white matter (Lane 3), gray matter (Lane 4), glioblastoma (Lane 5), meningioma (Lane 6), and metastasis of an adenocarcinoma (Lane 7). An immunopositive band at 230 kilodaltons (arrow) shows TN-C in the corresponding homogenate.

Astrocytomas

Astrocytomas (WHO Grade 2) showed moderate TN-C immunopositivity in the ECM, whereas only focal TN-C staining was observed in and around vessels. In contrast, a significantly stronger staining intensity of TN-C in and around endothelial proliferation as well as in the ECM was noted in high-grade astrocytomas (WHO Grade 3 and 4). The stroma of anaplastic astrocytomas (WHO Grade 3) and GBM (WHO Grade 4) was moderately TN-C positive (Fig. 1B). Tumor cells did not express TN-C. In anaplastic astrocytomas and GBM, we observed immunopositivity for TN-C in the perivascular spaces. Furthermore, we found elevated intracytoplasmic TN-C expression in astrocytes in the peritumoral edematous areas of an untreated GBM patient (Fig. 1C). In addition, in GBM specimens, at the very edge of the tumor in normal appearing brain tissue, we observed moderate-to-strong TN-C positivity in the ECM around groups of tumor cells (Fig. 1D). Within the GBM group, there was a significant difference in survival comparing patients with and without TN-C expression. Patients without TN-C immunopositivity, on average, lived 13 months longer than patients with TN-C expression in the ECM, as shown in Figure 3. In fact, patients with GBM who had TN-C immunopositivity had a median survival of 28 months (standard error, 7.8 months), whereas patients without TN-C immunopositivity, on average, survived for 12 months (standard error, 1.6 months).

Figure 3.

Kaplan–Meier survival curves showing a 16-month survival benefit for patients without tenascin-C (TN-C) expression compared with patients with TN-C expression in the extracellular matrix (ECM) of their lesions. WHO 4: World Health Organization Grade 4.

Oligodendrogliomas

In oligodendrogliomas (WHO Grade 2), there was no TN-C staining in the tumor cells or the tumor ECM. There was low TN-C positivity in the network of delicate blood vessels (11 of 22 specimens). In contrast, anaplastic oligodendrogliomas (WHO Grade 3) showed moderate-to-strong immunostaining along the endothelial proliferations in > 50% of specimens. The ECM of the lesion was weakly reactive.

Ependymomas

Blood vessels in the center of the typical perivascular pseudorosettes expressed TN-C in nearly all specimens of ependymomas (16 of 17 specimens). We observed moderate-to-strong TN-C immunopositivity in the ECM of anaplastic ependymomas (WHO Grade 3) compared with TN-C negativity in the tumor stroma and only focal positivity along fibrotic changes.

Medulloblastomas

In contrast to the densely cellular, undifferentiated tumor nodules that were TN-C negative, nearly all the vessels in medullablastoma specimens displayed TN-C positivity. Only the desmoplastic areas (four specimens) showed strong staining for TN-C.

Meningiomas

No TN-C expression was seen by either Western blot analysis or with immunohistochemistry in the vessels and the ECM of either meningothelial or psammomatous meningiomas. In only three specimens, the ECM of regressive fibrotic areas showed low TN-C staining. The neoplastic cells did not show any TN-C positivity.

Schwannomas

The cytoplasm and nucleus of spindle-shaped cells of 31 schwannomas showed no TN-C expression. Nearly all blood vessels showed low-to-moderate TN-C staining. The ECM in 3 of 31 specimens, which showed regressive changes, had remarkable immunohistochemical staining for TN-C in these regressive areas (Fig. 1E). There was a statistically significant difference when comparing the TN-C expression in cerebral blood vessels from schwannomas (nearly every vessel was stained) with the TN-C expression in blood vessels from meningiomas (no staining of the vessels).

Hemangioblastomas

Most of the small endothelial vessels and stromal cells were negative for TN-C in hemangioblastoma specimens. Only the perivascular spaces of the ECM showed low TN-C positivity (14 of 41 specimens), notably only in areas of fibrotic changes.

Lymphomas

In the perivascular spaces, there was moderate-to-strong TN-C positivity along the onion-like and reticulin positive structures in nearly every specimen of lymphoma (20 of 24 specimens). In contrast, no TN-C staining was seen in the neoplastic cells (Fig. 1F). In addition, small reactive cells as well as larger neoplastic cells around blood vessels were TN-C negative.

Craniopharyngioma

In craniopharyngioma specimens, along the vessels, no TN-C could be detected. In many specimens, we found calcifications in association with epithelial cells. In the intermediate layer of epithelial cells, in nearly every specimen, there was low-to-moderate TN-C staining. The most persistent finding was TN-C positivity along the fibrotic areas between the cellular nodules of craniopharyngioma.

Secondary Tumors of the Central Nervous System

In contrast to astrocytomas, metastatic tumors did not show any TN-C expression in Western blot analyses. Also with immunohistochemistry, the tumor cells were negative, irrespective of the type of the metastatic lesion (i.e., primary tumor). In the ECM of the surrounding brain areas and in areas with reactive astrocytosis around metastatic tumor nodules, low-to-moderate TN-C positivity could be observed. In metastatic melanoma, in nearly all specimens, we found moderate perivascular TN-C immunopositivity in areas with induced neovascularization. In contrast to adenocarcinoma and renal cell carcinoma, in all specimens of malignant melanoma, a distinct reticulin-like, TN-C positive stroma was present. The reduced nonvascular network of adenocarcinoma and of renal cell carcinoma showed low TN-C positivity in > 50% of the specimens.

DISCUSSION

The results of the current study show for what to our knowledge is the first time the exact cellular and tissue localization of TN-C immunoreactivity in human control brain tissue and in a large number of different brain tumors. Immunoreactivity for TN-C was found predominantly in the following structures: In control brains, TN-C was found predominantly in the white matter of the frontal, temporal, parietal, and occipital lobes and in the hippocampus, with a significant difference in the expression of TN-C noted when comparing gray matter with white matter using either Western blot analysis or immunohistochemistry. In tumor tissue, the most constant TN-C immunopositivity was seen in the extracellular matrix of the fibrotic stroma of highly malignant tumors and along the tumor border, especially in high-grade astrocytomas. Tumor cells were usually TN-C negative.

The Normal Human Brain

With regard to the control brains, in general, the results of the current study confirm the previously published results. Authors described strong expression of TN-C restricted to the ECM and no TN-C expression in the blood vessels of the normal adult brain compared with the developing central nervous system.1, 4, 8, 13 Ventimiglia et al. 29 reported a detectable amount of TN-C (0.04 μg/mg brain tissue) in the temporal lobe and other cortical areas of adult brains. Perides et al. 30 described TN-C in the white matter of the central nervous system, comparing human and bovine brain tissues. We were able to confirm these observations by analyzing separately the gray matter and white matter of various regions from 21 control brains using Western blot analysis and immunohistochemistry with three different MoAbs against TN-C. The most consistent findings in control brains were a low-to-moderate TN-C positivity of the ECM in the white matter of nearly all specimens and the lack of TN-C immunoreactivity in intracerebral vessels. Furthermore, TN-C was expressed focally only in the first layer of the gray matter; this finding was confirmed by Western blot analyses, in which a weak signal for the gray matter homogenate could be seen at 230 kilodaltons. The expression of TN-C in the ECM of the normal brain tissue was interpreted as peritumoral in origin by some authors who used control brain specimens from surgical interventions for GBM.4, 13, 14 However, when using archival frozen tissues (n = 5 specimens) and formalin-fixed, paraffin-embedded blocks (n = 21 specimens) from control brains, we were able to detect TN-C expression constantly in the ECM of the white matter, a finding that is consistent with recently published data by Tews and Nissen.31 Based on these results, we conclude that, in the adult human brain, there still is a critical amount of total TN-C in the ECM of the white matter. This finding is highly important in the clinical setting, because it restricts systemic therapeutic approaches and forces us to develop highly specific MoAbs for radioimmunotherapy as applied in a locoregional therapeutic approach.

TN-C in Brain Tumors

Invasion and tumor cell-migration play crucial roles in disease progression and are biologic hallmarks of malignant tumors. These are the major causes of disease-associated morbidity and mortality. Cell biology of migrating fetal cells and invading neoplastic cells seems to possess similarities,32 because both cell types use the ECM and its proteins (e.g., TN-C) for attachment and motility.2, 7 It remains unknown whether the dissemination of glioma cells occurs along white fiber tracts because these tracts are the pathway of minor resistance or because of specific ligands located along these tracts.

It is well known that TN-C expression is up-regulated in tumor tissues.1, 4, 8, 13 In the current study, a comparison of the TN-C localization was made between tumors of astrocytic origin and tumors of other origin (meningiomas, metastases to the brain) by Western blot analysis. Furthermore, the TN-C distribution pattern in different regions of control brains and in a large variety of brain tumors using immunohistochemistry was used to establish whether there was an entity-specific localization of TN-C in tumor cells, blood vessels, or the ECM.

We found heterogeneous TN-C expression in the ECM or the perivascular area along fibrotic gliomesenchymal areas of nearly every examined nonastrocytic tumor entity (see Table 1). These expression patterns lead to the conclusion that TN-C is not important for malignant transformation of nonastrocytic tumors but may be involved in the ECM reaction around the tumor to prevent further tumor cell spread. It is well known that expression of TN-C is up-regulated actively by stimulated astrocytes in traumatic brain lesions.33, 34

Zagzag et al.35 observed significantly higher levels of TN-C in homogenates of GBM than in normal brain and observed enhanced TN-C expression in vascular hyperplasia in 59 astrocytomas. Those authors found a significant association between elevated TN-C expression and higher tumor grade. We also detected high levels of TN-C in GBM lesions and negative to low levels in tumors of nonastrocytic origin using Western blot analysis. Furthermore, we detected heterogeneous TN-C distribution in the ECM and in the perivascular compartment of different brain tumor entities, but the most consistent finding was observed in astrocytomas: Elevated TN-C expression was seen around blood vessels of the lesion of all tumor grades but was more pronounced around hyperplastic vessels in high-grade astrocytomas. Furthermore, there was a significant difference in the TN-C expression around vessels among the different astroglioma grades. Zagzag et al. observed elevated TN-C expression in the ECM at the infiltration zone of malignant glial tumors (astrocytoma, WHO Grade 2–4). We were able to confirm this observation in several specimens (Fig. 1D); in addition, we found TN-C immunopositivity in the cytoplasm of activated astrocytes in the peritumoral, edematous areas embedded in a TN-C negative ECM (Fig. 1C). Therefore, we conclude that TN-C generally may be up-regulated at the border of the tumor. Based on the adhesive activity of TN-C,36 the elevated TN-C expression in this zone leads to the hypothesis that the elevated expression of TN-C in areas of reactively transformed ECM may have an antiinfiltrative effect and may prevent further infiltrative tumor cell migration.

In patients with astrocytoma (WHO Grade 2; n = 19 patients) without TN-C expression around the vessels of the lesion, compared with patients who had TN-C positive astrocytoma, had a significant median survival benefit of 16 months (P = 0.04) according to Kaplan–Meier survival curves and log-rank test statistics. Among patients with GBM (WHO Grade 4; n = 99 patients), those with TN-C negative lesions (n = 12 patients) survived significantly longer (median: 16 months; P = 0.008) compared with patients who had elevated TN-C expression in the ECM of the lesion. Recently, Herold-Mende et al. 37 showed a significant correlation between a shorter disease free interval and perivascular staining for TN-C in patients with WHO Grade 2 and 3 gliomas

Because we observed a constant immunohistochemical finding of TN-C in the ECM of fibrotic changes in 33 different tumor entities, and especially elevated TN-C expression in the infiltrative tumor borders of astrocytic tumors, this antiinfiltrative TN-C up-regulation appears to be fatal in patients with astrocytic tumors, because these tumor cells spread easily—as in the developing brain—along TN-C boundaries into the nonlesioned brain. The postulated antiinfiltrative effect of elevated TN-C expression in nonastrocytic tumors protects the host brain from further nonastrocytic tumor cell invasion, whereas this antiinfiltrative effect seems to facilitate invasion of glioma cells at the tumor border in astrocytomas. This hypothesis is underlined by 1) the constant immunohistochemical finding of TN-C in the ECM of astrocytic tumor borders, 2) the intracytoplasmic TN-C expression of astrocytes in the peritumoral edematous areas of an untreated GBM (Fig. 1D), and 3) the obvious survival benefit of patients without TN-C expression in GBM.

Tumor cell invasion is defined as an active translocation of tumor cells through cell layers and ECM barriers of the host.38 In contrast to most malignant neoplasms, astrogliomas aggressively infiltrate the surrounding brain tissue as single cells but metastasize only very rarely.7 In astrogliomas, tumor cells most likely infiltrate and migrate to the neighboring brain parenchyma along the basement membrane of new endothelial proliferations as one possible pathway.7 Thus, the initiation of neovascularization is one critical step toward malignant progression.

Giese et al. 39 observed a five times higher migration rate of glioma cells on TN-C-monolayers than on collagen, fibronectin, or vitronectin layers using seven human glioma cell lines in a microliter scale assay. In vitro, endothelial cells attach to TN-C substrates and are able to elongate and extend as necessary for endothelial migration; in contrast, there was no spread on substrates, including fibronectin, collagen, vitronectin, or laminin.40 This suggests that, in gliomas, TN-C plays a critical role in modulating cell adhesion and motility in endothelial proliferation and tumor cell migration.

The observed histologic appearance of the TN-C staining in astrocytic tumors, with its strong immunopositivity along neovascularization and in the surrounding areas of highest cellular pleomorphism, the reduced TN-C expression in histologically lesser malignant areas, and the lack of TN-C immunostaining in vessels of control brains, lead to the conclusion that the adhesive/antiadhesive function of TN-C promotes endothelial cell motility and, thus, again plays a crucial role in glioma-induced neoangiogenesis. Current data show that endothelial cells of hyperplastic vessels can synthesize perivascular TN-C.35, 41 In vitro and in vivo experiments have shown the up-regulation of TN-C in the presence of basic fibroblast growth factor, transforming growth factor β, and platelet-derived growth factor.42, 43 Recently, Zagzag et al. 44 suggested that the expression of TN-C by migrating endothelial cells and the promotion of endothelial cell adhesion and migration by TN-C makes TN-C play a potential role in pathologic angiogenesis. Even if these factors stimulate angiogenesis and are associated with neovascularization in brain tumors, to date, it has not been proven that angiogenetic factors can stimulate TN-C up-regulation in brain tumors. Moreover, the elevation of TN-C expression seems to be a general reaction of the altered ECM of the brain induced by microglia and/or astrocytes, like what was seen with the elevated TN-C expression along fibrotic changes in the surrounding, nonaltered brain areas in four patients with bacterial meningoencephalitis.

The frequently observed infiltrative glioma growth in the cortex and especially the rapid glioma spread along white matter tracts, like the corpus callosum, anterior commissure, or optic radiation, could be mediated by TN-C, which was detected in the ECM of the white matter of nearly every examined region of the cortex in 21 control brains. Goldbrunner et al.7 postulated that gliomas obviously can produce their own ECM whenever necessary but also opportunistically use the host ECM whenever it is present and useful.

Varying degrees of TN-C expression are found in normal adult control brains and in astrocytomas with various degrees of malignancy. It appears the presence or absence of TN-C expression in the stroma of astrocytic tumors may play a not yet clearly understood role in patient survival. It is noteworthy that, in patients who had astrocytoma (WHO Grade 2; n = 19 patients) without TN-C expression around the vessels of the lesion, compared with patients who had TN-C positive astrocytoma, had a significant median survival benefit of 16 months (P = 0.04) according to Kaplan–Meier survival curves and log-rank test statistics. Among patients with GBM (WHO Grade 4; n = 99 patients), those with TN-C negative GMB (n = 12 patients) survived significantly longer (16 months; P = 0.008) compared with patients who had elevated TN-C expression in the ECM of the lesion.

Acknowledgements

The authors thank Ms. Angela Henn (Munich, Germany) and Ms. Christine Knies (Hamburg, Germany) for their skillful technical assistance. The help of Ida C. Llenos, M.D., in correcting the article is greatly appreciated. Furthermore, the authors thank H.A. Kretzschmar, M.D. (Department of Neuropathology, University of Munich, Germany) for supporting them with archival specimens of patients with astrocytoma (World Health Organization Grades 2–4).

Ancillary