Neuropathology and Applied Neurobiology (2008), 34, 316–329
doi: 10.1111/j.1365-2990.2007.00898.x
Tenascin-C expression relates to clinicopathological
features in pilocytic and diffuse astrocytomas
C. Maris*, S. Rorive*, F. Sandras*, N. D’Haene*, N. Sadeghi†, I. Bièche‡, M. Vidaud‡,
C. Decaestecker§¶ and I. Salmon*
Departments of *Pathology and †Radiology, Erasme University Hospital, and §Laboratory of Toxicology, Institute of
Pharmacy, Université Libre de Bruxelles, Brussels, Belgium, ‡Laboratoire de Génétique Moléculaire, Faculté des Sciences
Pharmaceutiques et Biologiques, Université Paris V, Paris, France, and ¶Fonds National de la Recherche Scientifique,
Brussels, Belgium.
C. Maris, S. Rorive, F. Sandras, N. D’Haene, N. Sadeghi, I. Bièche, M. Vidaud, C. Decaestecker and I. Salmon
(2008) Neuropathology and Applied Neurobiology 34, 316–329
Tenascin-C expression relates to clinicopathological features in pilocytic and diffuse astrocytomas
Aims: Tenascin-C (TN-C) is an extracellular matrix brain
glycoprotein for which conflicting in vitro and in vivo
results are reported in the literature dealing with its
involvement in astrocytoma aggressiveness, in particular
astrocytoma invasion. In view of these conflicting results
and the lack of data available on low-grade astrocytomas,
the present study focuses on pilocytic World Health Organization (WHO) grade I, and diffuse WHO grade II astrocytomas, that is, two histological entities associated with
very different invasive abilities. Methods: Using real-time
reverse transcription polymerase chain reaction and
immunohistochemistry, we analysed the TN-C expression
in normal brain tissue as well as in a series of 54 pilocytic
and 53 grade II astrocytomas. Conclusions: Our data on
normal brain showed that while TN-C is largely expressed
in supratentorial white matter, it was largely absent in
infratentorial white matter. Paralleling these observations, we showed that TN-C expression in low-grade astrocytomas similarly varies according to tumour site. Cox
regression analysis evidenced that TN-C provided an independent prognostic value which is enhanced in the case of
grade II astrocytomas for which positive TN-C expression
is associated with a higher risk of recurrence. We also
analysed TN-C expression specifically in vascular areas of
low-grade astrocytomas without demonstrating any prognostic value for this additional feature. Results: Similarly
to normal brain, low-grade astrocytomas exhibit variations in TN-C expression with site, and this expression is
associated with an independent prognostic value in terms
of recurrence.
Keywords: extracellular matrix, low-grade astrocytoma, multivariate analysis, prognosis, tenascin-C
Introduction
Astrocytomas can be grouped into histological entities
associated with very different clinical behaviour patterns
and pathological findings, and are classified into four
categories (grade I–IV) according to the World Health
Organization’s (WHO) grading system [1,2]. Grade IV
astrocytomas (glioblastomas) constitute the most aggresCorrespondence: Isabelle Salmon, Departments of Pathology, Erasme
University Hospital, Université Libre de Bruxelles, Brussels, Belgium.
Tel: +322 553115; Fax: +322 554790; E-mail: isalmon@ulb.ac.be
316
sive type, and their biological characteristics are widely
studied. While grade II diffuse astrocytomas constitute
the first step in the malignancy development continuum
within the so-called group of diffuse astrocytic tumours
(that is, WHO grade II–IV), grade I pilocytic astrocytomas
are generally well delineated and do not progress to higher
grades. However, aggressive progression may be observed
in a number of pilocytic astrocytomas with an infiltration
pattern often leading to tumour recurrence, even though
these tumours do not necessarily share the biological
features of diffuse high-grade astrocytomas [2–4].
© 2007 Blackwell Publishing Ltd
Tenascin-C expression in astrocytomas
Aggressiveness and poor prognosis in the case of patients
suffering from astrocytomas are mainly associated with
the ability of astrocytoma cells to invade the surrounding parenchyma; this ability is modulated by the tumour
environment, especially the extracellular matrix (ECM)
[5]. In addition, the ECM plays a number of central roles
in other biological processes, such as proliferation and
angiogenesis, which are involved in astrocytoma development [5–8]. While the composition and the roles of
the ECM in the central nervous system (CNS) are not well
defined, various authors agree that the ECM of the brain
is different from that of the other organs. While collagen,
fibronectin and laminin are the predominant ECM components in non-CNS tissue, hyaluronan, proteoglycans,
tenascin-C (TN-C) and thrombospondin are the principal
elements in the CNS ECM [8,9]. TN-C is a glycoprotein
characterized by a hexameric structure. Its aminoterminus includes a series of cysteines and heptad
repeats, both of which are involved in multimerization.
This terminal part is followed by 14.5 epidermal growthfactor-like and 17 fibronectin type III-like repeats. Finally,
the carboxy-terminus consists of a COOH-terminal knob
made up of a sequence homology with the globular
domain of the b and g chains of human fibrinogen
[10,11]. While TN-C is encoded by a single gene under
the control of a single promoter, a number of structurally and functionally different human TN-C isoforms are
generated by the alternative splicing of the TN-C transcript at the fibronectin type III repeat level [10,11]. By
regulating the adhesive and signalling properties of cells,
TN-C is able to play a number of morphoregulatory roles
during the processes of development and tissue remodelling as well as in disease [10–13]. However, conflicting in
vitro and in vivo results are reported in the literature concerning the role that TN-C plays in the different processes
such as cell adhesion, motility and invasion that lead to
astrocytoma cell migration – see the Discussion [14–21].
It should be emphasized that most TN-C-related studies
concern high-grade astrocytomas. In fact, very limited
data are available on low-grade astrocytic tumours, and
in particular on pilocytic (WHO grade I) astrocytomas,
which display very different behavioural patterns as compared with the diffuse (WHO grade II–IV) group. This
motivated us to analyse the TN-C expression in normal
brain tissue as well as in a series of 107 low-grade astrocytomas [by means of real-time reverse transcription
polymerase chain reaction (RT-PCR) and immunohistochemistry] and to focus on the potential involvement of
317
TN-C in the magnetic resonance imaging (MRI) aspects
of low-grade astrocytomas. To this end, we used multivariate data analyses that enabled us: (i) to study the
case distribution in a multivariate (clinical and anatomopathological) feature space in which the feature interactions were analysed (by means of Log-linear models);
and (ii) to carry out multivariate prognosis analyses (by
means of Cox regression).
Materials and methods
Clinical and histopathological data
The investigations using normal tissue were carried out
on samples from five normal human post mortem brains
(without neuropathological alterations) obtained within
24 h of death. Six samples were taken from six different
areas: grey and white matter from the cerebral hemispheres (frontal lobes) and grey and white matter from the
cerebellar hemispheres, the brainstem and the cervical
spinal cord. Three samples from each site were stored
at -80°C for RT-PCR analyses and three others were
embedded in paraffin.
A series of 107 low-grade astrocytomas was investigated in parallel. This series consisted of archival
formalin-fixed and paraffin-embedded samples obtained
from the Laboratory of Pathology of the Erasme University Hospital (Brussels, Belgium) and collected between
1984 and 2005. As detailed in Table 1, all the cases were
classified by two pathologists according to the WHO classification [1,2]. Frozen tumour samples from astrocytomas were also available for RT-PCR analyses (see below).
These samples were used subject to the approval of the
Université Libre de Bruxelles Hôpital Erasme Ethics
Committee.
The available clinical data included patients’ ages and
genders, their tumour sites, the extent of their surgical
resections and their pre- and post-surgical adjuvant treatments and follow-ups, as detailed in Table 1.
The preoperative neuroimaging data (including fluid
attenuated inversion recovery and enhanced T1-weighted
images when available) were retrospectively reviewed and
enabled 100 astrocytomas to be classified as either well- or
ill-circumscribed (see MRI status in Table 1). Because of
the absence of normal tissue in the available materials,
histopathological infiltration could not be assessed for a
large number of cases and was thus not considered in the
present study.
© 2007 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 34, 316–329
318
C. Maris et al.
Table 1. Clinicopathological data for 107 patients with low-grade
astrocytomas
Number of patients
Age: child/adult*
Gender: male/female
Sites
Cerebral hemisphere
Diencephalon
Cerebellum
Brain stem
Spinal cord
Others†
MRI status
Ill-circumscribed
Well-circumscribed
Not specified
Surgical resection
Total
Partial
Biopsy
Adjuvant therapy
Pre-surgery
Post-surgery
Follow-up (months)
Range
Median
Recurrence‡ (median delay)
Death
210 kDa
Pilocytic
Grade II
181.8 kDa
54
29/25
29/25
53
14/39
30/23
115.5 kDa
12
7
20
10
3
2
22
2
9
4
15
1
23
24
7
46
7
0
30
22
2
8
40
5
7
7
4
17
1–335
44
14 (34)
4
1–348
39
20 (17)
8
The data present numbers of cases in the different categories except
where other measurements are indicated (such as range).
*Cut-off value of 18 years.
†optic nerve, corpus callosum.
‡cases with total or partial surgery.
MRI, magnetic resonance imaging.
The recurrences concerned patients having benefited of
a total or subtotal surgery, and were defined as cases
presenting magnetic resonance imaging (MRI) evidence
of progression that required new surgery or adjuvant
treatments.
Immunohistochemistry
Five-micrometer-thick sections were submitted to standard immunohistochemistry as previously detailed
[22,23], with the immunohistochemical expression being
visualized by means of streptavidin—biotin–peroxidase
complex kit reagents (BioGenex, San Ramon, CA, USA)
with diaminobenzidine/H2O2 as the chromogenic substrate. Counterstaining with haematoxylin concluded the
processing. The TN-C expression was evidenced by means
of a murine monoclonal anti-TN-C antibody clone DB7
82.2 kDa
1
2
3
4
5
6
7
Figure 1. Western blot analysis showing negative control (lane 2);
TN-C protein (lane 3); normal supratentorial grey matter (lane 4);
normal supratentorial white matter (lane 5); and glioblastomas
(lanes 6 and 7). TN-C, tenascin-C.
(Chemicon Int, Temecula, CA, USA; dilution 1:200).
Figure 1 illustrates the antibody specificity (1:750 dilution) proven by means of Western blot analysis using
purified human TN-C protein (1.25 mg; Chemicon Int,
Temecula, CA, USA; Figure 1 lane 3), 200 mg of proteins
from the normal supratentorial grey (lane 4) and white
(lane 5) matter, and 50 mg of proteins from two glioblastomas (lanes 6 and 7). For immunohistochemistry purposes, a negative control was carried out by replacing the
primary TN-C antibody with non-immune serum (Dako,
Glostrup, Denmark).
Evaluation of the immunohistochemical
TN-C expression
The TN-C immunostaining intensity was assessed by two
independent observers (CM, SR) using standard lightmicroscopy. Where discrepancies were encountered, the
cases were settled by consensus with a third observer (IS).
In view of the staining patterns observed (see Results),
TN-C immunostaining in astrocytomas was systematically
evaluated in the ECM, distinguishing between a negative to
low expression level (labelled as ‘negative’) and a moderate
to high one (labelled as ‘positive’). Additional investigations were carried out in vascular areas. This enabled
us to identify in a number of cases presenting a particular
pattern of TN-C expression (illustrated in Results). This
particular pattern is defined by a major enhancement of
TN-C expression in the vessel walls and/or perivascular
ECM as compared with neighbouring tumour ECM, and
was mentioned as vascular enhancement (VE) in the following. As detailed in the results, this latter evaluation was
possible with certainty in 50 pilocytic and 44 grade II
astrocytomas only. The two TN-C-related features were
labelled ECM TN-C and VE TN-C respectively.
© 2007 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 34, 316–329
Tenascin-C expression in astrocytomas
Real-time RT-PCR
Real-time RT-PCR analyses were carried out on a series
of tissue samples in order to evaluate the TN-C mRNA
expression. RNA extraction, cDNA synthesis and PCR
reaction conditions have been described previously [23].
After samples that did not satisfy the quality controls had
been excluded, the remaining series consisted of 40
normal brain tissue samples (10 cerebral grey and 12
white matter, 10 cerebellar grey and eight white matter
samples), nine pilocytic astrocytomas (four supratentorial and five infratentorial including three cerebellar)
and nine grade II astrocytomas (five supratentorial and
four infratentorial including one cerebellar). Quantitative
values were obtained from the threshold cycle number
(Ct value), at which the increase in the fluorescent signal
associated with an exponential growth of PCR products
began to be detected by the laser detector of the ABI
Prism 7700 Sequence Detection System (Perkin-Elmer
Applied Biosystems, Foster City, CA, USA) using the PE
Biosystems analysis software according to the manufacturer’s manuals. The precise amount of total RNA added
to each reaction mix (based on optical density) and its
quality were both difficult to assess. Because of this, we
also quantified the transcripts of RPLP0, an endogenous
RNA control gene [24]. As previously detailed [23], the
results were termed ‘Ntarget’ and expressed as N-fold differences in target gene expression relative to the RPLP0
gene. Primers for the RPLP0 and TN-C were chosen
with the assistance of the Oligo 5.0 computer program
(National Biosciences, Plymouth, MN, USA). We carried
out searches in the dbEST and nr databases to confirm
the absence of single nucleotide polymorphisms and the
total gene specificity of the nucleotide sequences chosen
as primers. To avoid the amplification of contaminating
genomic DNA, one of the two primers was placed at the
junction between two exons. The nucleotide sequences of
the primers used were as follows: RPLP0-U (5′-GGC GAC
CTG GAA GTC CAA CT-3′) and RPLP0-L (5′-CCA TCA
GCA CCA CAG CCT TC-3′) for RPLP0 trancripts (PCR
product of 149 bp), TNCtot-U (5′-GAG GGT GAC CAC
CAC ACG CTT-3′) and TNCtot-L (5′-CAA GGC AGT GGT
GTC TGT GAC ATC-3′) for total TNC transcripts (PCR
product of 73 bp).
Gel electrophoresis was used to verify the specificity of
the PCR amplicons. For each primer pair, we performed
no-template control and no-reverse-transcriptase control
assays, which produced negligible signals (usually >40 in
319
Ct value), suggesting that the effects of the primer–dimer
formation and genomic DNA contamination were
negligible.
Data analysis
All the statistical analyses were carried out using Statistica (Statsoft, Tulsa, OK, USA).
The first step was to study the relationships between
pairs of qualitative variables by means of contingency
tables. The significance of the potential associations was
evaluated by means of either the c2-tests or Fisher’s exact
tests (in 2 ¥ 2 cases only).
The second step involved the log-linear analysis technique, which is a multivariate extension of the c2-test of
independence [25]. The major task was to establish the
best possible fit for the cell frequencies of a multiway
contingency table by means of a log-linear model. We
identified the simplest model that fitted the data (that is,
explained the multivariate data distribution) by using a
methodology similar to that detailed in [26] and based
on the ‘Automatic best model selection’ procedure
included in the log-linear analysis module of the Statistica software. Finally, the standard Cox regression analysis was also used to fit an explanatory model to the
relapse-free survival data. This method enabled the
possible simultaneous influence of several variables on
the survival period to be tested. In addition, the nonparametric Mann–Whitney test was used to compare
independent groups of quantitative values provided by
real-time RT-PCR.
Results
Expression of TN-C in the normal human brain
As shown in Figure 2, in the cerebral white matter area,
there is a highly heterogeneous level of TN-C expression
with a high level of immunopositivity, whereas the cortex
displayed a low level of immunopositivity limited to the
first layer and the external glia limitans (Figure 2A).
While the leptomeningeal vessels (smooth muscle cells)
showed moderate TN-C immunopositivity, the capillaries
of the white and grey matter did not offer any immunopositivity at all. In contrast, in the cerebellum, we observed a
completely different staining pattern consisting of the
absence of any expression in the white matter (Figure 2B),
whereas the ECM of the molecular layer sometimes
© 2007 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 34, 316–329
320
C. Maris et al.
Figure 2. Illustrations of immunohistochemical TN-C expression in normal human brains and astrocytomas. (A) Major difference in
staining for TN-C of the grey and white matter in normal frontal lobes. (B) Absence of TN-C expression in the normal human cerebellum
(molecular layer and white matter). (C) Diffuse and strong TN-C expression in the ECM of a supratentorial pilocytic astrocytoma in contrast
to the weak or wholly absent TN-C expression in the ECM of an infratentorial pilocytic astrocytoma (D). Same variation of staining observed
between a supratentorial (E) and an infratentorial grade II astrocytoma (F). Two patterns of vascular enhancement of TN-C expression
(VE TN-C) in astrocytomas showing strong TN-C expression in vessels walls with (G) or without (H) perivascular enhancement. Original
magnification ¥40 (A–B); ¥100 (G); ¥200 (C–F, H). ECM, extracellular matrix; VE TN-C, vascular enhancement tenascin-C.
© 2007 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 34, 316–329
Tenascin-C expression in astrocytomas
321
Figure 3. Real-time RT-PCR results of total TN-C mRNA expression in normal brain tissue (A) and low-grade astrocytoma samples (B). (A)
The analyses were carried out on 40 samples of cerebral and cerebellar grey and white matter (10/12 cerebral grey/white matter, 10/8
cerebellar grey/white matter). (B) The analyses were carried out on nine supratentorial (four pilocytic and five grade II) astrocytomas and
nine infratentorial (five pilocytic and four grade II) ones. The data are expressed as means ⫾ SEM. RT-PCR, reverse transcription polymerase
chain reaction; TN-C, tenascin-C; SEM, standard error mean.
displayed a very weak level of expression. No TN-C immunopositivity was detected in the cerebellar vessels. While
the ECM of the brainstem and spinal cord specimens was
negative, we did observe intracytoplasmic TN-C expression
in focal neurones. These results were globally confirmed
at the mRNA expression level by means of quantitative
RT-PCR. As shown in Figure 3A, the TN-C mRNA levels
increased in the cerebral white matter as compared with
the cerebellar (P = 0.047), with very low (if any) mRNA
levels in the grey matter regardless of its origin.
Expression of TN-C in astrocytomas
To ensure certainty in the detection of the presence or
absence of VE TN-C, the evaluation was carried out twice
by the same observer (IS), after excluding six cases
because of the limited size of the available material. Of the
101 remaining cases, 94 cases (50 pilocytic and 44 grade
II) were concordantly evaluated (that is, 93% of concordance), of which 33 cases exhibited VE TN-C as illustrated
in Figure 2G,H. In the following, we restricted any analysis related to VE TN-C to these 94 concordant cases. It
should also be noted that the distinction between the different types of VE TN-C (observed in perivascular ECM
only, or vascular cells only, or both) did not add additional
information (data not shown). In addition, in the case of
very strong VE TN-C, the identification of the vascular cell
types expressing TN-C was not possible.
A first analysis showed no significant dependence
between the two TN-C-related features (ECM and VE
TN-C), revealing that these two features produced
different information on TN-C expression in lowgrade astrocytomas. The following results detailed
the difference observed between these two features in
relation to the other clinicopathological features
analysed.
We also showed that there was no need to distinguish
between the tumours of the patients submitted to presurgery adjuvant therapies (n = 11) and those of the
patients who had not undergone any such therapies (see
Table 1), because no difference in TN-C expression (ECM
and VE TN-C) was evident across these two groups (data
not shown).
In relation to tumour type The data concerning the immunohistochemical TN-C expression in the low-grade astrocytomas are detailed in Tables 2 and 3 and illustrated in
Figure 2C–F, while the quantitative RT-PCR assessments
are shown in Figure 3B. As detailed in Table 2, the total
percentages of cases with positive TN-C expression in
tumour ECM slightly increased from 46% for the pilocytic
to 68% for the grade II (P = 0.03). At the mRNA level, no
difference was observed between pilocytic and grade II
cases when their locations were not taken into account
(data not shown). As explained below, this was not the
case when comparing infratentorial and supratentorial
tumours.
Although VE TN-C seemed to be more prevalent in pilocytic (44%) than in grade II (25%) astrocytomas, this
tendency was not significant (P = 0.08).
© 2007 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 34, 316–329
322
C. Maris et al.
Table 2. TN-C expression in low-grade astrocytomas
Pilocytic
Sites
Supratentorial
Hemisphere
Diencephalon
Subtotal
Infratentorial
Cerebellum
Brain stem
Spinal cord
Subtotal
Others
Total
Grade II
ECM
VE
ECM
83 (10/12)
57 (4/7)
74 (10/19)
64 (7/11)
50 (3/6)
59 (10/17)
91 (20/22)
100 (2/2)
92 (22/24)
20
40
67
30
33
50
33
39
(4/20)
(4/10)
(2/3)
(10/33)
(1/2)
46 (25/54)
(6/18)
(5/10)
(1/3)
(12/31)
(0/2)
44 (22/50)
VE
6 (1/18)
0 (0/2)
5 (1/20)
22
50
67
50
(2/9)
(2/4)
(10/15)
(14/28)
(0/1)
68 (36/53)
44
75
33
43
(4/9)
(3/4)
(3/10)
(10/23)
(0/1)
25 (11/44)
The data are expressed as the percentages of cases exhibiting positive TN-C expression in ECM or VE. The exact ratios are also mentioned.
TN-C, tenascin-C; extracellular matrix; VE, vascular enhancement.
Table 3. MRI status of low-grade astrocytomas in relation to their sites and TN-C expression
(A) Occurrence of ill-circumscribed astrocytomas in relation to tumour sites
Pilocytic
Supratentorial
Infratentorial
Cerebral hemisphere
Diencephalon
Cerebellum
Brain stem
Spinal cord
44
43
19
80
100
Others
Grade II
(4/9)
(3/7)
(3/16)
(8/10)
(3/3)
(2/2)
95
100
56
75
93
}44%
}85%
(21/22)
(2/2)
(5/9)
(3/4)
(14/15)
(1/1)
}96%
}89%
(B) Positive TN-C expression in relation to MRI status
Pilocytic
TN-C Expression
Well-circumscribed
Ill-circumscribed
Grade II
ECM
29 (7/24)
61 (14/23)
VE
27 (6/22)
62 (13/21)
ECM
57 (4/7)
70 (32/46)
VE
14 (1/7)
27 (10/37)
The data in (A) are expressed as the percentages of ill-circumscribed tumours in each category (the exact ratios are also mentioned), while the
data in (B) are expressed as the percentages of cases exhibiting positive TN-C expression in ECM or VE (the exact ratios are also mentioned).
TN-C, tenascin-C; MRI, magnetic resonance imaging; ECM, extracellular matrix; VE, vascular enhancement.
In relation to site Because in the normal brain TN-C was
expressed differently in terms of site, we analysed the
possible variations in astrocytoma TN-C expression in
relation to tumour site. The results shown in Table 2
indicate that the slight increase in ECM TN-C expression
from the pilocytic to the grade II astrocytomas was similarly observed in the supratentorial and infratentorial
astrocytomas. As also illustrated in Figure 2C–F, ECM
TN-C was expressed in a significantly larger number of
supratentorial pilocytic (74%) and grade II (92%) astrocytomas than their infratentorial counterparts (30% and
50% respectively; P = 0.004 for pilocytic, P = 0.002 for
grade II and P < 10-6 over the total group of low-grade
tumours). The ‘Others’ group containing only three
cases was not taken into account in the statistical
analyses.
The same tendency was observed at the mRNA level.
When grouping the low-grade cases per location, a
© 2007 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 34, 316–329
Tenascin-C expression in astrocytomas
decrease was observed in infratentorial as compared with
supratentorial cases (Figure 3B) without being significant. This decrease was probably due to the relative heterogeneity detailed in Table 2 and the small number of
cases analysed.
Regarding VE TN-C, the pilocytic astrocytomas did not
exhibit any significant variation in terms of tumour site
(Table 2). This contrasts with the significant variations
(P = 0.005) evidenced in grade II astrocytomas for which
VE was observed in infratentorial locations (including the
cerebellum) almost exclusively (see Figure 2F).
In relation to patients’ ages We also investigated whether
paediatric vs. adult astrocytomas were characterized by
different TN-C expression levels. We observed that grade II
astrocytomas in children generally had lower levels of
ECM TN-C expression than those in adults (P = 0.006). We
did not observe this relation in the case of the pilocytic
astrocytomas (P = 0.10). No variation in VE TN-C was
evident between paediatric and adult astrocytomas.
In relation to the MRI status Finally, we focused on
the MRI status (well- vs. ill-circumscribed). The results
detailed in Table 3A show the variations in the MRI status
of the low-grade astrocytomas in relation to their sites,
while bringing out the differences between the supratentorial and the infratentorial sites where the cerebellar sites
must be distinguished. The ill-circumscribed pilocytic
tumours were found particularly in the brainstem and the
spinal cord (85%), were less frequently observed in the
supratentorial locations (44%) and even less so in the cerebellum (19%) (P = 0.002). As expected, the diffuse grade
II tumours were generally ill-circumscribed regardless of
site (96% of supratentorial cases and 89% of brainstem
and spinal cord ones), except in the cerebellum where only
56% were ill-circumscribed (P = 0.009).
When the pilocytic and grade II astrocytomas were
grouped, a significantly higher number of ECM TN-Cpositive cases were shown in the group of the illcircumscribed tumours (66% vs. 35% for the
well-circumscribed astrocytomas, P = 0.005). As detailed
in Table 3B, this relation was essentially due to the pilocytic astrocytomas (61% vs. 29% P = 0.04), for which a
significant association was also evident between VE TN-C
and the MRI status (62% vs. 27%, P = 0.03). In contrast,
no significant relation was demonstrated between VE
TN-C and the MRI status in grade II astrocytomas (or after
grouping the pilocytic and grade II astrocytomas).
323
Analysis of the multivariate interactions
between TN-C expression, tumour grades, sites
and MRI status and patients’ ages
The results reported above led us to consider that interdependencies exist between the different variables, particularly in the case of low-grade astrocytomas. We decided to
perform a multivariate analysis in order to analyse these
interdependencies systematically. We therefore carried
out a first log-linear analysis on the five-dimensional contingency table, cross-classifying ECM TN-C expression,
patients’ ages (child vs. adult), tumour grades (I vs. II),
sites (distinguishing between supratentorial, cerebellar
and other infratentorial tumours) and MRI status (ill- vs.
well-circumscribed). For the set of 97 tumours so concerned (located in the four major sites and for which the
MRI status were available), the log-linear analysis was
able to test for statistical significance in the case of each
possible interaction between two or more features; Table 4
describes the results obtained. To begin with, we evaluated
the order of the significant feature interactions (2, 3 or 4)
that have to be taken into account in a log-linear model to
explain the data correctly (that is, without any significant
loss of information). The results shown in Table 4A
indicate that only the two-feature interactions were significant. This confirms both the presence of a degree
of interdependence between pairs of features and the
absence of more complex relationships. Second, the information provided by the different feature interactions was
analysed by evaluating the partial and marginal associations, as detailed in Table 4B. Briefly, the partial association between two features evaluates the loss of information due to the exclusion of the feature interaction of
interest from a complete log-linear model of order 2
(including all the effects due to the individual features and
the two-feature interactions). In addition, the marginal
association between two features evaluates the gain of
information by adding this particular interaction to a loglinear model including only the effects of order 1 (due to
the individual features). The results (detailed in Table 4B)
showed significant (partial and/or marginal) associations
for each pair of features except between tumour grade and
site (interaction 24). It should be noted that of all the
partial associations involving ECM TN-C expression (interactions 12–15), only one was significant, that is, interaction 14 between ECM TN-C expression and tumour site.
This means that failing to take account of astrocytoma
sites in the evaluation of ECM TN-C expression leads to a
© 2007 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 34, 316–329
324
C. Maris et al.
Table 4. Log-linear analysis of the multivariate interactions between features
(A) Fitting of all K-feature interactions
K-feature interaction
Degrees of freedom
Max.Lik. c2
P-value
1
2
3
4
5
6
14
16
9
2
19.74
75.05
11.76
5.51
0.30
0.003
<10-6
0.76
0.79
0.86
(B) Partial and marginal association – identification of the best model
Effect
Degrees of
freedom
Partial Ass.
c2
Partial Ass.
P-value
Mrg. Ass.
c2
Mrg. Ass.
P-value
Initial
model
Order 1
1
2
3
4
5
1
1
1
2
1
1.86
0.41
4.40
2.80
10.27
0.17
0.52
0.04
0.25
0.001
1.86
0.41
4.40
2.80
10.27
0.17
0.52
0.04
0.25
0.001
X
Order2
12
13
14
15
23
24
25
34
35
45
1
1
2
1
1
2
1
2
1
2
0.87
3.64
13.00
1.31
2.44
0.03
8.78
3.09
0.55
9.30
0.35
0.06
0.002
0.25
0.12
0.98
0.003
0.21
0.46
0.01
4.93
10.23
20.48
7.54
6.63
3.33
14.01
10.01
6.46
16.23
0.03
0.001
0.00004
0.006
0.01
0.19
0.0002
0.007
0.01
0.0003
Best
model
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
(A) Tests on the K-factor interactions (K = 1, 2, 3 or 4). A P-value <0.05 indicates that the corresponding interaction is significant. (B) Features:
1, ECM TN-C expression; 2, tumour grade; 3, patient’s age; 4, tumour site; 5, MRI status. The different interactions are denoted by the
juxtaposition of the feature numbers. The partial association between the two feature i and j (denoted by ij) is computed by comparing the fit
(that is, evaluating the c2 difference) of the complete model that includes all the two-way interactions with that of the model that excludes the
interaction between features i and j. The marginal association between the two features i and j is computed by comparing the fit of the model that
includes all the main effects (that is, order 1) with that of the model obtained after the addition of the interaction between feature i and j only.
The (marginal or partial) P-values <0.1 identify the features and interactions included in the initial log-linear model to fit the data. The best
model is the simplest model that is able to explain the data efficiently (that is, without any significant loss of information).
Max.Lik., maximum licelihood; Partial Ass., partial association; Mrg. Ass., marginal association; ECM, extracellular matrix; TN-C, tenascin-C;
MRI, magnetic resonance imaging.
highly significant loss of information (P = 0.002). The
best (that is, the simplest) log-linear model identifying the
main significant effects able to explain the data efficiently
(without any loss of information) was then generated
from an initial model (see Table 4B). This best model consisted of only the feature interactions between ECM TN-C
expression and the patient’s age (interaction 13) or
tumour site (interaction 14), MRI status and astrocytoma
grade (interaction 25) or location (interaction 45), and
astrocytoma grade and the patient’s age (interaction 23).
These five interactions were thus necessary and sufficient
to explain the data distribution in the five-dimensional
contingency tables.
Figure 4 summarizes the information provided by this
log-linear analysis, showing the direct and indirect links
so evidenced between the features. This clarified the data
reported in the previous section by indicating the presence
of a direct and essential association between ECM TN-C
expression and the astrocytoma sites. The influence of
patient’s age also seems to be an essential factor that has
© 2007 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 34, 316–329
Tenascin-C expression in astrocytomas
325
ECM TN-C expression in the series of low-grade astrocytomas, which was also enhanced in the case of grade II
astrocytomas (see Table 5). We also observed that tumour
site did not add any prognostic information to the models
described in Table 5 (data not shown).
Discussion
Figure 4. Illustration of the direct links evidenced between the
different features analysed in the best log-linear model fitted to the
data (see Table 4 and text). Of these, the thick lines indicate the
links associated with significant partial associations, that is,
particularly required to explain the data.
to be taken into account. This contrasts with the absence
of any direct relation between MRI status and ECM TN-C
expression.
In a second log-linear analysis, we simply replaced the
ECM TN-C feature by the VE TN-C one. As seen for the
initial analysis, the best log-linear model generated on
these data was of order 2 (data not shown). In contrast,
this model did not retain any interaction involving VE
TN-C but included it as an individual feature (that is, effect
of order 1). However, it should be noted that significant
and almost significant partial associations were, respectively, observed between VE TN-C and tumour grade
(P = 0.03), and between VE TN-C and MRI status
(P = 0.05). The actual importance of these associations
has to be confirmed on a larger series. In fact, the twofeature interactions retained in the best model resulting
from the second analysis involved the other features
(tumour grade, patient’s age, tumour site and MRI status)
and confirmed the results already reported in Figure 4.
Prognostic impact of TN-C expression in
low-grade astrocytomas
We were not able to demonstrate any prognostic value for
VE TN-C. However, for ECM TN-C ,while no correlation
was observed for pilocytic astrocytomas (P = 0.33;
Figure 5A), positive ECM-TN-C expression was associated
with a higher risk of recurrence in grade II astrocytomas
(P = 0.07; Figure 5B).
In order to test the prognostic contribution of this
marker in the presence of standard clinical variables,
multivariate Cox regression analyses were performed. The
results confirmed an independent prognostic value for
Tenascin-C antibodies are presented as promising agents
in the design of treatment protocols for different types of
solid cancers, including gliomas [27–30]. As explained by
Brack et al. [27], this approach is motivated by the fact
that antigens preferentially expressed in modified tumour
ECM are ideal targets for tumour-targeting applications.
However, this type of application requires a detailed
knowledge of the distribution of the antigen expression in
normal and tumoral tissue, taking into account the high
level of heterogeneity encountered in cancers.
While TN-C expression in the normal brain was rarely
detected in the first studies [10,31–33], most authors now
describe TN-C expression as being restricted to the ECM. It
should be noted that only limited information is available
regarding normal brain tissues, and for a few studies peritumoral tissue was used as a point of normal reference by
different authors [15, 32–34]. This approach was avoided
in the present study because of the impossibility of
excluding the TN-C secreted by astrocytoma cells migrating in peritumoral areas (or by normal cells, such as
endothelial ones, reacting in the tumour environment).
The discrepancy in the data on the TN-C expression in the
normal brain could also be explained by the great complexity of the CNS. In fact, very limited data are available
on the actual distribution of TN-C expression across the
different structures of the normal brain. As previously
shown by a few authors [18,19], the TN-C expression in
the normal brain is heterogeneous and increases in the
white matter as compared with the grey matter. Our study
reveals that this heterogeneity is greater than supposed
and is related to specific tissue sites. Indeed, we observed
very few, if any, cases of TN-C expression in the cerebellar
white matter, although a readily discernible amount of
TN-C is present in the cerebral white matter.
In the case of astrocytomas, it is clear that TN-C expression is heavily increased in glioblastomas as compared
with normal tissue [15,17,18,34,35]. All the authors also
agree that TN-C expression in ECM increases from low- to
high-grade astrocytomas [15,17,18,34]. However, in the
matter of TN-C involvement in astrocytoma aggressive-
© 2007 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 34, 316–329
326
C. Maris et al.
1.0
A
Cumulative Proportion of
Recurrence-free patients
p = 0.33
0.8
0.6
negative ECM TN-C
0.4
0.2
positive ECM TN-C
0.0
0
50
100
150
200
250
300
350
400
Time (months)
Cumulative Proportion of
Recurrence-free patients
1.0
B
0.8
negative ECM TN-C
0.6
p = 0.07
0.4
positive ECM TN-C
0.2
0.0
0
50
100
150
Time (months)
200
250
300
Figure 5. Kaplan–Meier curves evidencing the relationships between the ECM expression of TN-C (positive/negative) and the
recurrence-free survival periods for pilocytic astrocytomas (A) and grade II astrocytomas (B). The dots symbolize the recurrences and the
crosses represent the relapse-free cases respectively. ECM, extracellular matrix; TN-C, tenascin-C.
ness, conflicting data have been reported in the literature,
leading different authors to come to opposing conclusions
concerning the roles played by TN-C in astrocytoma cell
migration, that is, a stimulating effect [14,15] vs. an
inhibiting one [16]. Similarly, in vitro investigations,
which performed migration assays, reported either a
stimulating effect [5,11] or an inhibiting one
[8,15,21,36].
In view of these conflicting results, we focused our
analysis on pilocytic and grade II astrocytomas, because
these two histological entities are associated with very
different invasive abilities [37]. As recently shown, these
two entities have different molecular profiles with respect
to adhesion-, ECM- and invasion-related genes [38]. In
the present series, neuroimaging evaluation revealed that
while a large majority of the grade II cases seemed to be
invasive, about half of the pilocytic cases displayed illcircumscribed aspects.
Our monovariate analyses revealed that the ECM TN-C
expression in the pilocytic and/or grade II astrocytomas
© 2007 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 34, 316–329
Tenascin-C expression in astrocytomas
327
Table 5. Cox regression analysis
Model/P-value
Variable
b
P-value
All pilocytic and grade II cases
P = 0.00004
ECM TN-C
Grade
Age
Surgery
Adjuvant treatment
0.94
0.73
-0.04
-1.24
0.74
0.02
0.08
0.005
0.03
0.05
Grade II cases
P = 0.002
ECM TN-C
Age
Surgery
Adjuvant treatment
1.93
-0.06
-0.70
0.58
0.006
0.002
0.37
0.26
The ‘Model/P-value’ indicates the overall level of significance of the model. Except ‘Age’, which is a quantitative variable, all the others are
binary. ECM TN-C distinguishes between negative and positive expression, grade between pilocytic and grade II astrocytomas, surgery between
total and subtotal, and adjuvant treatment between absence and presence. The equation at the basis of the Cox Regression model is an
exponential function of a linear combination of the variables considered, where b indicates the coefficient of each variable in the linear
combination. The associated P-value is a measure of the level of significance of the contribution of each variable to the model (and leads to the
conclusion that b is significantly different from zero). If P < 0.05, the feature is associated with a significant prognostic value independently of
the other parameters taken into account.
ECM, extracellular matrix; TN-C, tenascin-C.
varied according to the tumours’ sites, the patients’ ages
and the MRI status. Multivariate log-linear analyses were
thus required to evaluate the actual interdependence
between the different features and to identify which
factors actually are related to the ECM TN-C expression in
the astrocytomas. This approach has two main advantages. The first is that it provides a systematic approach to
the analysis of complex multidimensional tables, while
the second is that it enables the relative importance of the
different effects (such as feature interactions) to be judged.
This helped us to clarify that ECM TN-C expression in
low-grade astrocytomas is indirectly related to the MRI
infiltration status and directly related to the tumours’
location. It is important to note that this model characterizes low-grade astrocytomas, which are well known for
their occupation of both cerebellar and cerebral sites in
contrast to high-grade astrocytomas, which are more frequently located in the cerebral hemispheres.
According to our hypothesis, the lack of TN-C in the
normal cerebellar ECM in contrast to its strong expression
in the normal cerebral ECM leads to very different tumoral
ECMs. Indeed, we have shown that a greater number of
both supratentorial pilocytic and grade II astrocytomas
exhibit ECM TN-C expression as compared with infratentorial ones. It is known that high-grade astrocytoma cells
are able to secrete TN-C and thus strongly modify the ECM
composition [6]. Our data suggest that low-grade astrocytomas do not have the same ability, and this may well
reflect on cell migration.
While future research should confirm these results on a
larger series, our data suggest that VE TN-C could be associated with the MRI status, especially in the case of pilocytic astrocytomas. We also agree completely with Zagzag
et al. [15], who emphasized that pilocytic astrocytomas
display an increase in perivascular TN-C enhancement,
and especially around hyperplastic vessels. As reported in
the literature, endothelial cells are able to secrete TN-C
[39], and the detection of TN-C in the vascular areas of
astrocytomas suggests a functional role played by TN-C in
angiogenesis [15,17,18] and, more particularly, in the
migration of endothelial cells [36]. The literature shows
that both glioma angiogenesis and invasion are invasive
processes sharing common mechanisms of regulation
that can be simultaneously inhibited by naturally occurring factors [15,17,18]. This may (at least partly) explain
the association between VE TN-C and the MRI status
suggested by our data. However, we did not succeed in
showing any prognostic value associated with VE TN-C for
which the conflicting results reported in the literature
concerned small series that did not include pilocytic astrocytomas [15,17,18]. In contrast, ECM TN-C expression
was associated with an independent prognostic value, particularly in the case of grade II astrocytomas. This agrees
with another result associating ECM TN-C immunopositivity with shorter survival for patients with glioblastomas
(grade IV) [34].
In conclusion, similarly to normal brain, low-grade
astrocytomas exhibit variations in TN-C expression with
© 2007 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 34, 316–329
328
C. Maris et al.
sites, and this expression is associated with an independent prognostic value in terms of recurrence. However,
additional information could be provided by the currently
lacking analysis of different TN-C isoforms, especially in
the case of normal brain and low-grade astrocytomas. To
the best of our knowledge, only one study mentions the
absence of large TN-C isoforms in these two categories of
tissue as opposed to their presence in high-grade astrocytomas [34]. In fact, the scant availability of commercially
available antibodies against alternatively spliced domains
now limits immunohistochemical evaluation.
Acknowledgements
We are grateful to Ms Nathalie Watteau for secretarial
support and Ms Blair Jenkins for help in preparing the
manuscript. This work was carried out with the support of
grants awarded by the Fonds Yvonne Boël (Brussels,
Belgium).
References
1 Kleihues P, Cavenee WKC (Eds). WHO Classification of
Tumours. Pathology and Genetics. Tumours of the Nervous
System. Lyon: IARC Press, 2000
2 Kleihues P, Louis DN, Scheithauer BW, Rorke LB, Reifenberger G, Burger PC, Cavenee WK. The WHO classification of tumors of the nervous system. J Neuropathol Exp
Neurol 2002; 61: 215–25;discussion 26–9
3 Perilongo G, Garre ML, Giangaspero F. Low-grade
gliomas and leptomeningeal dissemination: a poorly
understood phenomenon. Childs Nerv Syst 2003; 19:
197–203
4 Buschmann U, Gers B, Hildebrandt G. Pilocytic astrocytomas with leptomeningeal dissemination: biological
behavior, clinical course, and therapeutical options.
Childs Nerv Syst 2003; 19: 298–304
5 Gladson CL. The extracellular matrix of gliomas: modulation of cell function. J Neuropathol Exp Neurol 1999; 58:
1029–40
6 Zamecnik J. The extracellular space and matrix of
gliomas. Acta neuropathol 2005; 110: 435–42
7 Rauch U. Extracellular matrix components associated
with remodeling processes in brain. Cell Mol Life Sci 2004;
61: 2031–45
8 Bellail AC, Hunter SB, Brat DJ, Tan C, Van Meir EG.
Microregional extracellular matrix heterogeneity in brain
modulates glioma cell invasion. Int J Biochem Cell Biol
2004; 36: 1046–69
9 Sobel RA. The extracellular matrix in multiple sclerosis
lesions. J Neuropathol Exp Neurol 1998; 57: 205–17
10 Joester A, Faissner A. The structure and function of tenascins in the nervous system. Matrix Biol 2001; 20: 13–22
11 Jones PL, Jones FS. Tenascin-C in development and
disease: gene regulation and cell function. Matrix Biol
2000; 19: 581–96
12 Jones FS, Jones PL. The tenascin family of ECM glycoproteins: structure, function, and regulation during embryonic development and tissue remodeling. Dev Dyn 2000;
218: 235–59
13 Bourdon MA, Ruoslahti E. Tenascin mediates cell attachment through an RGD-dependent receptor. J Cell Biol
1989; 108: 1149–55
14 Friedlander DR, Zagzag D, Shiff B, Cohen H, Allen JC,
Kelly PJ, Grumet M. Migration of brain tumor cells on
extracellular matrix proteins in vitro correlates with
tumor type and grade and involves alphaV and beta1
integrins. Cancer Res 1996; 56: 1939–47
15 Zagzag D, Friedlander DR, Miller DC, Dosik J, Cangiarella
J, Kostianovsky M, Cohen H, Grumet M, Greco MA. Tenascin expression in astrocytomas correlates with angiogenesis. Cancer Res 1995; 55: 907–14
16 Giese A, Loo MA, Rief MD, Tran N, Berens ME. Substrates
for astrocytoma invasion. Neurosurgery 1995; 37: 294–
301; discussion 301–2
17 Herold-Mende C, Mueller MM, Bonsanto MM, Schmitt
HP, Kunze S, Steiner HH. Clinical impact and functional
aspects of tenascin-C expression during glioma progression. Int J Cancer 2002; 98: 362–9
18 Leins A, Riva P, Lindstedt R, Davidoff MS, Mehraein P,
Weis S. Expression of tenascin-C in various human brain
tumors and its relevance for survival in patients with
astrocytoma. Cancer 2003; 98: 2430–9
19 Tews DS. Adhesive and invasive features in gliomas.
Pathol Res Pract 2000; 196: 701–11
20 Kulla A, Liigant A, Piirsoo A, Rippin G, Asser T. Tenascin
expression patterns and cells of monocyte lineage:
relationship in human gliomas. Mod Pathol 2000; 13:
56–67
21 Giese A, Loo MA, Norman SA, Treasurywala S, Berens
ME. Contrasting migratory response of astrocytoma cells
to tenascin mediated by different integrins. J Cell Sci
1996; 109: 2161–8
22 D’Haene N, Maris C, Sandras F, Dehou MF, Remmelink
M, Decaestecker C, Salmon I. The differential expression
of Galectin-1 and Galectin-3 in normal lymphoid tissue
and non-Hodgkin’s and Hodgkin’s lymphomas. Int J
Immunopathol Pharmacol 2005; 18: 431–43
23 Mathieu A, Saal I, Vuckovic A, Ransy V, Vereerstraten P,
Kaltner H, Gabius HJ, Kiss R, Decaestecker C, Salmon I,
Remmelink M. Nuclear galectin-3 expression is an independent predictive factor of recurrence for adenocarcinoma and squamous cell carcinoma of the lung. Mod
Pathol 2005; 18: 1264–71
24 Bieche I, Parfait B, Le Doussal V, Olivi M, Rio MC,
Lidereau R, Vidaud M. Identification of CGA as a novel
estrogen receptor-responsive gene in breast cancer: an
© 2007 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 34, 316–329
Tenascin-C expression in astrocytomas
25
26
27
28
29
30
31
32
outstanding candidate marker to predict the response to
endocrine therapy. Cancer Res 2001; 61: 1652–8
Agresti A. Categorical Data Analysis. Hoboken: John
Wiley, 1990
Lakhan VC, Lavalle PD. Use of loglinear models to assess
factors influencing concern for the natural environment.
Environ Manage 2002; 30: 77–87
Brack SS, Silacci M, Birchler M, Neri D. Tumor-targeting
properties of novel antibodies specific to the large isoform
of tenascin-C. Clin Cancer Res 2006; 12: 3200–8
De Santis R, Albertoni C, Petronzelli F, Campo S,
D’Alessio V, Rosi A, Anastasi AM, Lindstedt R, Caroni N,
Arseni B, Chiodi P, Verdoliva A, Cassani G, Chinol M,
Paganelli G, Carminati P. Low and high tenascinexpressing tumors are efficiently targeted by ST2146
monoclonal antibody. Clin Cancer Res 2006; 12: 2191–6
Petronzelli F, Pelliccia A, Anastasi AM, D’Alessio V,
Albertoni C, Rosi A, Leoni B, De Angelis C, Paganelli G,
Palombo G, Dani M, Carminati P, De Santis R. Improved
tumor targeting by combined use of two antitenascin
antibodies. Clin Cancer Res 2005; 11: 7137s–45s
Akabani G, Reardon DA, Coleman RE, Wong TZ, Metzler
SD, Bowsher JE, Barboriak DP, Provenzale JM, Greer KL,
DeLong D, Friedman HS, Friedman AH, Zhao XG, Pegram
CN, McLendon RE, Bigner DD, Zalutsky MR. Dosimetry
and radiographic analysis of 131I-labeled anti-tenascin
81C6 murine monoclonal antibody in newly diagnosed
patients with malignant gliomas: a phase II study. J Nucl
Med 2005; 46: 1042–51
Higuchi M, Ohnishi T, Arita N, Hiraga S, Hayakawa T.
Expression of tenascin in human gliomas: its relation to
histological malignancy, tumor dedifferentiation and
angiogenesis. Acta Neuropathol 1993; 85: 481–7
Erickson HP. Tenascin-C, tenascin-R and tenascin-X: a
family of talented proteins in search of functions. Curr
Opin Cell Biol 1993; 5: 869–76
329
33 Bourdon MA, Wikstrand CJ, Furthmayr H, Matthews TJ,
Bigner DD. Human glioma-mesenchymal extracellular
matrix antigen defined by monoclonal antibody. Cancer
Res 1983; 43: 2796–805
34 Carnemolla B, Castellani P, Ponassi M, Borsi L, Urbini S,
Nicolo G, Dorcaratto A, Viale G, Winter G, Neri D, Zardi L.
Identification of a glioblastoma-associated tenascin-C
isoform by a high affinity recombinant antibody. Am J
Pathol 1999; 154: 1345–52
35 Natali PG, Nicotra MR, Bigotti A, Botti C, Castellani P,
Risso AM, Zardi L. Comparative analysis of the expression
of the extracellular matrix protein tenascin in normal
human fetal, adult and tumor tissues. Int J Cancer 1991;
47: 811–16
36 Zagzag D, Shiff B, Jallo GI, Greco MA, Blanco C, Cohen H,
Hukin J, Allen JC, Friedlander DR. Tenascin-C promotes
microvascular cell migration and phosphorylation of
focal adhesion kinase. Cancer Res 2002; 62: 2660–8
37 Kachra Z, Beaulieu E, Delbecchi L, Mousseau N, Berthelet
F, Moumdjian R, Del Maestro R, Beliveau R. Expression of
matrix metalloproteinases and their inhibitors in human
brain tumors. Clin Exp Metastasis 1999; 17: 555–66
38 Rorive S, Maris C, Debeir O, Sandras F, Vidaud M, Bieche
I, Salmon I, Decaestecker C. Exploring the distinctive biological characteristics of pilocytic and low-grade diffuse
astrocytomas using microarray gene expression profiles.
J Neuropathol Exp Neurol 2006; 65: 794–807
39 Zagzag D, Friedlander DR, Dosik J, Chikramane S, Chan
W, Greco MA, Allen JC, Dorovini-Zis K, Grumet M.
Tenascin-C expression by angiogenic vessels in human
astrocytomas and by human brain endothelial cells in
vitro. Cancer Res 1996; 56: 182–9
© 2007 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 34, 316–329
Received 25 July 2007
Accepted 27 July 2007