J Neurosurg 97:441–449, 2002

Gene therapy for meningioma: improved gene delivery

with targeted adenoviruses

CLEMENS M. F. DIRVEN, M.D., PH.D., JACQUES GRILL, M.D.,

MARTINE L. M. LAMFERS, PH.D., PAUL VAN DER VALK, M.D., PH.D.,
ANGELIQUE M. LEONHART, M.S., VICTOR W. VAN BEUSECHEM, PH.D.,
HIDDE J. HAISMA, PH.D., HERBERT M. PINEDO, M.D., PH.D., DAVID T. CURIEL, M.D.,
W. PETER VANDERTOP, M.D., PH.D., AND WINALD R. GERRITSEN, M.D., PH.D.
Departments of Neurosurgery and Pathology, and Division of Gene Therapy, Department of Medical
Oncology, Vrije Universiteit Medical Center, Amsterdam; Department of Therapeutic Gene
Modulation, Univeristy Center for Pharmacology, University of Groningen, the Netherlands;
Department of Pediatrics, Gustave Roussy Institute, Villejuif, France; and The Gene Therapy Center,
Division of Human Gene Therapy, Departments of Medicine, Pathology, and Surgery, University of
Alabama at Birmingham, Alabama

Object. Due to their surgical inaccessibility or aggressive behavior, some meningiomas cannot be cured with cur-
rent treatment strategies. Gene therapy is an emerging strategy for the treatment of brain tumors, which the authors in-
vestigated to determine whether adenoviruses could be used for gene transfer in meningioma cells.
Methods. The presence of the high-affinity Coxsackievirus and adenovirus receptor (CAR) for adenovirus type 5,
as well as endothelial growth factor receptor (EGFR) and alphav integrins (ITGAVs), were analyzed in primary tumors
by using immunohistochemical studies and in primary meningioma cell cultures by using fluorescence-activated cell
sorting. Targeting of adenoviruses to EGFR was achieved using bispecific antibodies, whereas targeting of adenovi-
ruses to the ITGAVs was accomplished by insertion of an RGD (arginine-glycine-aspartic acid) motif in the adenovi-
rus fiber HI loop. Gene transfer efficiency of untargeted and targeted vectors was compared in primary cell cultures
and in spheroids derived from patients’ resected tumor material.
The presence of CARs was observed in all tumors and in all but one of the derived primary meningioma cells. The
higher expression of EGFRs and ITGAVs indicated that these receptors could be used as alternative targets to redirect
the adenoviruses. Redirection of adenoviruses to the EGFRs or integrins enhanced gene transfer threefold (range two–
sevenfold) for EGFRs in primary meningioma cells and ninefold (range three–23-fold) for integrins (p = 0.002, analy-
sis of variance). The effect of adenovirus targeting was confirmed in spheroids composed of primary meningioma cells.
Conclusions. Gene transfer with adenoviruses targeted to tumor-specific receptors is very effective in primary me-
ningioma cells and spheroids. These vectors are promising agents for gene therapy of meningiomas.

KEY WORDS • meningioma • gene therapy • endothelial growth factor receptor •

integrin • adenovirus • Coxsackievirus and adenovirus receptor

account for approximately 20% of all es.2,28 Therefore, an alternative treatment is required for pa-

M
ENINGIOMAS
primary intracranial tumors.6 They are mostly be-
nign lesions that can be cured by total surgical re-
section, although 10% exhibit an aggressive course even
after macroscopically complete removal. In patients older
than 60 years, surgical treatment has been shown to result
in significant mortality and morbidity rates in 15% of cas-
Abbreviations used in this paper: ANOVA = analysis of variance;
CAR = Coxsackievirus and adenovirus receptor; CHO = Chinese
hamster ovary; CMV = cytomegalovirus; DMEM = Dulbecco mod-
ified Eagle medium; EGFR = endothelial growth factor receptor;
FACS = fluorescence-activated cell sorting; FCS = fetal calf serum;
IgG = immunoglobulin G; ITGAV = alpha v integrin; ITGAVB3 =
alpha v beta 3 integrin; ITGAVB5 = alpha v beta 5 integrin; mAb =
monoclonal antibody; PBS = phosphate-buffered saline; SD = stan-
dard deviation; vps = viral particles.
tients with either a malignant or a recurrent benign menin-
gioma that is surgically inaccessible or carries too high a
surgical risk.
Radiation therapy is effective for both malignant and be-
nign meningiomas.15,50 In more than half of patients, howev-
er, the tumor will recur within 5 years, and almost half of
those with atypical and malignant tumors will die within 10
years after diagnosis.10,22 In patients with residual or recur-
ring benign tumors, there is increasing concern about radi-
ation-related side effects that may occur even with highly
accurate therapies such as radiosurgery.33 Chemotherapy
and hormonal therapy have not acquired a place in the stan-
dard treatment of recurring meningiomas, despite inciden-
tal reports of tumor responses and stable disease after treat-
ment with these modalities.9,12,27,34,35,43,44,46

J. Neurosurg. / Volume 97 / August, 2002 441


Gene therapy, especially with adenoviral vectors, has de-
veloped as a promising option for treating tumors. Infection
of a large number of tumor cells is crucial for the success
of this type of therapy. Clinical trials in malignant gliomas
have shown that the vectors used so far were unable to pro-
duce efficient transgene delivery into tumor cells, which ex-
plains the lack of significant therapeutic effect.39,40 Our pres-
ent research is focused on the development of new vectors
that are targeted toward tumor cells in such a way that gene
transfer is enhanced in tumor cells and diminished in nor-
mal ones. To this end, an advantage is realized from knowl-
edge of the basic mechanisms for adenovirus infection.
Human adenovirus serotype 5 entry is a two-step process
involving high-affinity binding of the fiber of the virus to
the CAR, followed by interactions of the penton-base pro-
tein with ITGAVs for internalization.4,26 Targeting strate-
gies based on immunological principles use bispecific mol-
ecules that bind to the adenoviral capsid on one side and to
a cellular receptor on the other.13,20,32,51 Such molecules re-
direct the infection to a pathway independent of the cell
membrane receptor, CAR, by making a molecular bridge
between the virus and the tumor cell. Genetic targeting is
achieved by direct modifications of the adenoviral genome,
that is, insertion of the DNA sequence encoding for a pep-
tide with specific binding affinity. Only a few attempts with
this second approach have been successful, generally by in-
sertion of the sequence of small peptides into the fiber knob
sequence.26,49
Because low CAR expression has been shown to limit
gene transfer with adenoviruses in many tumor types,1,29 we
undertook this study to define the efficacy of adenovirus for
transferring genes in primary meningioma cells and sphe-
roids and to evaluate which targeting strategies can improve
gene transfer in these settings.
Materials and Methods
Histological Studies
The histological features of the tumors were reviewed by one of
the authors (P.v.d.V.) and classified according to the revised World
Health Organization system.25,38 The number of mitoses per 10 hpfs
was assessed and meningiomas were classified as malignant if this
number was at least 20 or atypical if it was between four and 20. Ad-
ditional immunohistochemical studies (most often epithelial mem-
brane antigen) were performed if required.
Primary Meningioma Cell Cultures and Spheroids
We used fresh material collected during brain tumor surgery in
adult patients, after informed consent had been obtained.
Primary meningioma cell cultures were obtained after mechanical
dissociation, according to the method described by Darling.11 The
cells were cultured in DMEM or the Ham F-12 medium supplement-
ed with 10% FCS and antibiotic agents. Nonglial origin was con-
firmed by morphological findings and by the absence of staining
with the anti–glial fibrillary acidic protein mAb clone 6F2. To pre-
vent endothelial cell contamination, large blood vessels were sep-
arated from the tumor material. All the experiments on these cells
were completed before the fifth passage, usually done between the
second and third passage.
Organotypic multicellular meningioma spheroids were grown
from small explants of fresh human tumor in 48-well plates coated
with agarose (one spheroid per well), according to the method origi-
nally described by Bjerkvig, et al.,5 for glioma. After checking them
for viability by morphological appearance and trypan blue exclusion,
we used spheroids of similar diameters (400–500 m) for gene trans-
fer experiments.
442
C. M. F. Dirven, et al.
Cell Lines
The U373MG (anaplastic astrocytoma) and the human glioma-
derived cell line U118MG (glioblastoma multiforme) were grown in
DMEM supplemented with 10% FCS and antibiotic agents.
Recombinant Adenoviruses
A recombinant E1-deleted adenovirus expressing the luciferase
reporter gene under the CMV promoter, AdCMVLuc, was used. The
vector Ad5LucRGD, which expresses luciferase under the same
CMV promoter and contains an integrin-targeting peptide (arginine-
glycine-aspartic acid, the RGD-4C peptide) in the HI loop of the
fiber knob, was generated as described previously.14 The presence of
the RGD binding motif in the HI loop was confirmed by polymerase
chain reaction and restriction enzyme analysis. The functionality of
the RGD binding motif was verified by the ability of the vector to en-
hance gene transfer in a CAR-negative but ITGAV-positive cell line
(U118MG). Recombinant adenoviruses were propagated on the per-
missive 293 cell line and purified using CsCl gradient banding. Sub-
sequently, they were titered in parallel: in vps by using the OD260
method and in plaque-forming units by using end-point titration on
293 cells. The titers were 3.3  1012 vps/ml and 4.1  1012 vps/ml
for AdCMVLuc and Ad5LucRGD, respectively. The ratio of infec-
tious vps/total vps was 1:4 and 1:10 for AdCMVLuc and Ad5Luc-
RGD, respectively. When the two viruses were compared, the same
number of vps was used.
Immunohistochemical Reagents
The mouse anti-CAR mAb (RmcB4) was prepared as ascites fluid.
The anti-ITGAVB5 mAb clone P1F6 and the anti-ITGAVB3 mAb
clone LM609 were used. A supernatant of the 425 hybridoma47 cul-
ture was used as a source for anti-EGFR mAb.
A new mouse mAb against firefly luciferase (LUC-Y) was used to
detect luciferase expression in infected cells (details of this mAb will
be published elsewhere).
Flow Cytometry
Cultured cells were trypsinized, washed with PBS, and centri-
fuged. Half a million cells were incubated with the first antibody for
1 hour on ice; the antibody had been diluted in PBS containing 0.1%
bovine serum albumin. Concentrations of the primary antibody were
10 g/ml, 2.5 g/ml, and 10 g/ml for RmcB, P1F6, and LM609,
respectively. For the 425 mAb, undiluted supernatant of the hybrido-
ma was used at a concentration of 1 g/ml of IgG-1, according to en-
zyme-linked immunosorbent assay. The irrelevant primary antibody
used, at a concentration of 10 g/ml, was mouse mAb IgG-1 323A3
directed against an epithelial marker not expressed on glioma cells.
Subsequently, the cells were washed and incubated with fluorescein
isothiocyanate–conjugated rabbit anti–mouse antibody for 30 min-
utes on ice and in the dark. After washing in PBS, the cells were re-
suspended in 500 l of PBS containing 1% formaldehyde. Analysis
was performed using an FACS apparatus. The human glioma–de-
rived cell line U373MG served as a positive control for CAR, EGFR,
ITGAVB3, and ITGAVB5 because high positivity has been shown
previously for these four antigens, and U118MG served as a nega-
tive control for CAR.32 The fluorescence intensity of each sample
was quantified as the ratio between the median fluorescence inten-
sities of the stained sample and of the irrelevant antibody control.
These ratios were compared with those of the positive control cell
line, U373MG, and expressed as a percentage of the latter value.
Immunohistochemical Studies
Immunohistochemical studies were performed on 4-m frozen
sections fixed with acetone. We used a frozen pellet of 293 cells and
normal human liver as positive controls for CAR. Normal human
skin was used as a positive control for EGFR and normal human ton-
sil as a positive control for ITGAVB3 and ITGAVB5. Nonspecific
antibody binding was blocked by incubation with 2% normal rabbit
serum. Tumor slides and positive controls were incubated for 1 hour
at room temperature in blocking buffer (1% bovine serum albumin
in PBS) with the same mouse mAbs that were used for cytometry.
Concentrations were 1 g/ml for 425 and 10 g/ml and for RmcB,
J. Neurosurg. / Volume 97 / August, 2002
Gene transfer in meningiomas with adenoviruses

LM609, and P1F6, respectively. Two negative controls were used TABLE 1
for each staining; these controls were treated either with blocking Clinical and pathological characteristics
buffer only or with normal mouse IgG-1. Specimens were then of the meningiomas cultured*
washed in PBS and incubated with biotinylated rabbit anti–mouse
IgG antibody diluted 1:500 in PBS for 30 minutes. The second anti- Case Age Histological Mitoses/
body binding was visualized with a peroxidase-conjugated streptavi- No. (yrs) Tumor Location Finding 10 hpf
din avidin–biotin complex kit diluted 1:200 in PBS for 1 hour.
VU-16 68 convexity (rt frontal) benign 1
Immunocytochemical Studies VU-18 63 tentorial (posterior fossa) benign 0
For immunocytochemical studies, infected cells were fixed direct- VU-22 42 skull base (parasellar) benign 0
ly in the plates with methanol/acetone (1:1). After washing in PBS, VU-29 36 skull base (petrosal) benign 0
the fixed cells were incubated with 5 g/ml purified LUC-Y anti- VU-36 72 skull base (sphenoidal) malignant 20
body diluted in diluent for 1 hour at room temperature. After wash- VU-38 28 cervical (patient w/ NF2) benign 0
ing in PBS, the cells were then incubated with the second antibody, VU-41 44 convexity (lt frontal) atypical 5
10 g/ml goat anti–mouse Ig antibodies conjugated to peroxidase, VU-44 62 skull base (temporal fossa) atypical 8
for 30 minutes at room temperature. After a subsequent wash, the red * NF2 = neurofibromatosis Type 2.
conversion of chromogen -amino-9-ethylcarbazole was used to de-
tect the presence of LUC-Y bound to the luciferase protein in the
cells. Cells were counterstained with hematoxylin and eosin.
Bergh at the Department of Radiobiology, Vrije Universiteit Med-
Bispecific Single-Chain Antibody Constructs ical Center, and Dr. J. T. Douglas at the Gene Therapy Center,
University of Alabama. The AdCMVLuc was provided by Dr. R.
Bispecific single-chain antibody constructs were made as de- D. Gerard at the University of Texas Southwestern Medical Cen-
scribed previously.20 The 425-S11 fusion protein recognizes the ter, Dallas, TX. The mouse anti-CAR mAb (RmcB) was obtained
EGFR on one side and the fiber knot on the other; it also contains the from Dr. R. L. Crowell at Hahnemann University, Philadelphia,
secretion signal from IgG-1. PA. The FCS was purchased from Life Technologies, Grand Island,
An undiluted supernatant of CHO cells, stably transfected with the NY. The Ham F-12 medium was supplied by Invitrogen, Carlsbad,
plasmid encoding for the bispecific single-chain antibody construct CA. The mAb clone P1F6 was acquired from Life Technologies,
425-S11, was used for the targeting experiments. The cells were Breda, The Netherlands, and the mAb clone LM609 was purchased
cultured in a cultivation system. Concentrations of bispecific sin- from Chemicon, Temecula, CA. A supernatant of the 425 hybrido-
gle-chain antibody constructs were determined using enzyme-linked ma culture was supplied by the American Type Culture Collection,
immunosorbent assays. The optimal ratio between the CHO cell Rockville, MD.
supernatant and vps was determined by targeting experiments on The fluorescein isothiocyanate–conjugated rabbit anti–mouse an-
U118MG. A single batch of supernatant from transfected CHO cells tibody was acquired from Dako A/S, Glostrup, Denmark, as were
was used for all experiments. the biotinylated rabbit anti–mouse IgG antibody, the avidin–bio-
tin complex kit, the diluent, the goat anti–mouse Ig antibodies, and
Gene Transfer Assays the chromogen 3-amino-9-ethylcarbazole. The FACScan device
To assess adenoviral infection in cells, 105 cells/well were plated was purchased from Becton-Dickinson, Erembodegem-Aalst, Bel-
in a 24-well plate in triplicate and incubated overnight in 1 ml of cul- gium. The cell cultivation system (model CL350) was obtained
ture medium to allow adherence. Before infection, 108 vps of adeno- from INTEGRA Biosciences AG, Wallisellen, Switzerland. The
virus were incubated with 25 l of the 425-S11 containing CHO su- luciferase chemiluminescence assay system was purchased from
pernatant for 30 minutes at room temperature. Next, the mixture was Promega, Madison, WI. The luminometer (model LB 9507) was
diluted in DMEM containing 2.5% FCS to a concentration of 5  acquired from EG&G Berthold, Bad Wildbad, Germany.
107 vps/ml, and 200 l was added to each well, that is, 100 vps/cell.
The cells were incubated at 37˚C in 5% CO2 for 1 hour, washed with
PBS, and then supplemented with 1 ml of DMEM containing 10%
FCS. Twenty-four hours after infection, the cells were assayed for Results
luciferase expression.
To assess adenoviral infection in spheroids, each one was plated
in a separate well of a 96-well plate coated with agarose, in 150 l Clinicopathological Correlation
of DMEM with 10% FCS, then 107 vps of the conjugated viruses or Primary meningiomas resected between May 2000 and
of the control viruses, diluted in 50 l of DMEM containing 2.5% October 2000 in the neurosurgery department of the Vrije
FCS, were added to the wells. The luciferase assay was performed
after 24 hours of incubation at 37˚C in 5% CO2. Universiteit Medical Center were used for this study. One
tumor was malignant, two were atypical, and five were
Luciferase Assay benign according to the new World Health Organization
To measure luciferase activity in the cells, we used the luciferase criteria (Table 1). Half of the lesions were located on the
chemiluminescent assay system. After removing the culture medi- skull base.
um, lysis buffer was added to the wells and the whole plate was snap-
frozen on dry ice. After thawing, the luciferase activity was mea- Expression of CAR and the ITGAVs
sured during 10 seconds immediately after initiation of the light
reaction in a luminometer. Values were normalized per number of The expression of CAR, ITGAVB3, and ITGAVB5 was
viable cells by using trypan blue exclusion. analyzed using immunohistochemical studies performed in
To measure luciferase activity in the spheroids, they were re- the frozen tumor specimens and by FACS. We were able to
moved from the wells, rinsed with PBS, and resuspended individu- obtain matched samples for immunohistochemical studies
ally in 100 l of lysis buffer. After three cycles of freezing and thaw-
ing, and vortexing to ensure complete lysis of the spheroids, the and FACS in six of the eight tumors.
luciferase measurement was performed using 25 l of the sample. The CAR was present in all meningiomas, although its
Values were expressed per spheroid. level of expression was quite variable. In the tumors that
had a high level of CAR expression, staining was found in
Sources of Supplies and Equipment the cytoplasm, whereas in ones with lower levels, only peri-
The U373MG and U118MG cell lines were supplied by J. Van der nuclear staining could be detected (Fig. 1a and e). In all

J. Neurosurg. / Volume 97 / August, 2002 443


FIG. 1. Photomicrographs of sections obtained in primary me-
ningiomas for immunohistochemical studies. Immunohistochem-
ical studies were conducted in frozen material from the primary
C. M. F. Dirven, et al.
FIG. 2. Graph showing results of adenovirus-mediated luciferase
gene transfer in primary meningioma cell cultures relative to CAR
expression on FACS. Early passages of the primary meningioma
cell cultures were infected with AdCMVLuc for 1 hour, and lu-
ciferase expression was measured 24 hours after infection. The
amount of luciferase activity (mean  SD), which measures the
efficacy of gene transfer, is correlated with the expression of CAR.
The black square represents U118MG glioma cells, which were
negative for CAR. The black diamond represents U373MG glioma
cells, which were highly positive for CAR. The white circles repre-
sent the meningioma samples. There is a clear trend showing that
the efficacy of gene transfer is correlated with the amount of CAR
present on the cells. RLUs = relative light units.
tumor by using mouse mAbs as primary antibody and a 33-di-
aminobenzidine detection method with a biotin–streptavidin am-
plification. The tumor designated VU-18 is a benign meningioma
and the one labeled VU-41 is a malignant meningioma. a and e:
Expression of CAR as detected with RmcB mAb. b and f: Ex-
pression of EGFR as detected with 425 mAb. c and g: Expression
of ITGAVB3 as detected with 1976 mAb. d and h: Expression of
ITGAVB5 as detected with the P1F6 mAb.
cases but one, CAR expression was maintained in the short-
term cultures of the tumor cells, according to FACS mea-
surements.
The ITGAVB5s were highly expressed in all tumors (Fig.
1d and h) and in all cell cultures. In contrast, ITGAVB5 ex-
pression was less pronounced (Fig. 1c and g) and varied be-
tween tumor samples and cultured cells (Table 2). Tumor
vessels always stained strongly positive for these two in-
tegrins.
As an alternative receptor for adenovirus retargeting,
EGFR expression in tumors and in cultured cells was ana-
lyzed. Significant amounts of EGFR were expressed both in
vivo and in vitro in all cases (Fig. 1b and f). This expression
had a homogeneous pattern in six of seven tumors stud-
ied, including one malignant, one atypical, and four benign
meningiomas.

TABLE 2
Expression of target antigens in primary meningioma cell cultures (FACS) and in the corresponding tumors (IHC)*
ITGAVB3 ITGAVB5
CAR (RmcB) EGFR (425) (1976) (P1F6)
Case
No. IHC† FACS‡ IHC FACS IHC FACS IHC FACS

VU-16
10 25–50 50 50–75
1 50–75 50 75

VU-18
10 NS 10–50 50–75
10 25–50 50 75
VU-22 10–50 25–50 10–50 25–50
1 25–50 50 75
VU-29 50 50–75 50 25–50 50 NS 50 50–75
VU-36
10 50–75 50 75
10 75 50 75
VU-41 50 25–50 50 50–75 10–50 NS 10–50 25–50
* The mouse mAbs in parentheses were the primary antibodies. Abbreviations: IHC = immunohistochemistry; NS = fluorescence
shift not significant.
† See Immunohistochemical Studies for details about the detection method. Values represent the percentage of tumor cells that stained
positively.
‡ See Flow Cytometry for details about detection methods. Values represent the percentage of fluorescence compared with that in the
positive control cell line.

444 J. Neurosurg. / Volume 97 / August, 2002

Gene transfer in meningiomas with adenoviruses
On immunohistochemical studies of the normal brain ad-
jacent to the tumor, which was investigated in two cases,
CAR was also expressed. Only the vessels were weakly
positive for ITGAVs. Isolated microglial cells were occa-
sionally positive for EGFR (data not shown). We can thus
conclude that the principal known receptors for adenovirus
(CAR and ITGAVs) are present on meningioma cells. The
significantly higher expression of ITGAVs and EGFR in
meningiomas compared with normal brain tissue indicates
that these receptors could be used for tumor-specific target-
ing strategies.
Efficiency of Adenovirus-Mediated Gene Transfer in
Primary Meningioma Cells
The infectivity of primary meningioma cells was tested
with an adenovirus expressing the luciferase gene, Ad-
CMVLuc. Twenty-four hours after infection with the ade-
novirus at a multiplicity of infection of 100 vps/cell, lucifer-
ase activity was measured with a chemiluminescence assay.
Results of experiments performed in triplicate are depict-
ed in Fig. 2. In all cases but one, the gene transfer efficacy,
as measured by the transgene expression, was in the range
of the positive control cell line, U373MG, which expresses
high levels of CAR and ITGAVs and is easy to infect with
adenovirus.32 All samples showed infectivity far above that
of the CAR-negative cell line, U118MG, which is known to
be refractory to adenovirus infection. There was a positive
correlation between the amount of CAR present in the cells
and the efficiency of adenovirus-mediated gene transfer
(R = 0.734, p = 0.05). The expression level of ITGAVB3
and ITGAVB5s did not correlate with gene transfer effi-
ciency.
Higher Transduction Efficiency on Primary Meningioma
Cells With Adenoviruses Targeted to EGFR or ITGAVs
Than With Untargeted Adenoviruses
Adenoviruses were targeted either to the EGFR or
ITGAVs. The bispecific antibody blocked most binding
sites of the fiber and redirected the virus to the EGFR. In-
sertion of the RGD sequence in the HI loop of the fiber
results in a new adenovirus that can still bind to its native
receptor and to the ITGAVB3s and ITGAVB5s (Fig. 3).
Figure 4 shows a representative example, in which signifi-
cant gene transfer enhancement was observed in the tar-
geted viruses. In the primary meningioma cells obtained in
the patient in Case VU-18, cells that express low levels of
the native adenoviral receptor CAR, the increase in gene
transfer was 7.5-fold with EGFR targeting and 23.6-fold
with integrin targeting (Fig. 4 upper; p = 0.0001 according
to ANOVA, with significant differences between the three
types of vectors, as shown by the Scheffé post hoc analy-
sis). Indeed, in these cells, which were relatively refractory
to adenoviral infection, gene transfer could be restored to
levels similar to those of the meningioma cells expressing
higher levels of the CAR by using targeted adenoviruses.
The number of infected cells, as measured with an immuno-
histochemical assay for the luciferase protein, was marked-
ly increased when the targeted vectors were used (Fig. 4
lower). Percentages of positive cells were less than 1, 15,
and 30% for AdCMVLuc, AdCMVLuc 425-S11, and
Ad5LucRGD, respectively.
Figure 5 shows the results for all eight primary menin-
J. Neurosurg. / Volume 97 / August, 2002
FIG. 3. Schematic drawing showing targeted adenoviral vectors
used for gene transfer studies; AdCMVLuc is a replication-defec-
tive virus expressing the luciferase gene under the CMV promoter.
It has a native tropism and infects cells through the CAR. Although
Ad5LucRGD has the same expression cassette for luciferase, in the
HI loop of the fiber knob, an integrin-binding motif, RGD-4C, has
been inserted. In this virus, the CAR-binding site of the fiber knob
is intact. Consequently, Ad5LucRGD can infect cells after binding
to CAR, or to the ITGAVs independently of CAR. On the other
hand, AdCMVLuc can be complexed with bispecific single-chain
miniantibodies (425-S11) that bind on one side to the fiber knob
and block its interaction with CAR and on the other side to the
EGFR. With the experimental condition we used, however, it is not
possible to saturate the virus completely with bispecific single-
chain miniantibodies. Consequently, some degree of CAR binding
is still possible with this type of virus.
gioma cell cultures studied. The median gene transfer en-
hancement with immunological targeting toward the EGFR
was threefold (range two–sevenfold). The median enhance-
ment with genetic targeting toward ITGAVs was ninefold
(range three–23-fold). Targeting, as performed in these ex-
periments, always resulted in marked enhancement of gene
transfer (p = 0.002, ANOVA). The Scheffé post hoc com-
parison showed that the differences in gene transfer were
statistically significant between AdCMVLuc and Ad5Luc-
RGD, and also between AdCMVLuc 425-S11 and Ad-
5LucRGD.
Combined Targeting Approaches for Heterogeneous
Meningioma Tumor Transduction
Because the expression of the target antigens is not al-
ways homogeneous in tumors, we sought to determine if
the combination of the two targeting approaches would be
feasible. The Ad5LucRGD targeted to the EGFR consists
of a mosaic virus with two different types of fiber knobs,
some redirected toward the EGFR, and some retargeted
toward the ITGAVs. It was expected that such a virus would
have increased potency to transduce tumors that express
heterogeneous levels of target antigens.
To construct a relevant model for human tumors, primary
organotypic spheroids were made from meningioma spec-
imens according to the method described by Bjerkvig, et al.5
These spheroids, which are directly derived from prima-
ry tumors, usually conserve the architecture of the tumor as
445

FIG. 4. Results of targeted gene transfer into meningioma cells
obtained in the patient in Case VU-18. The VU-18 meningioma
cell culture was used as a model to study gene transfer with the tar-
geted vectors described in Fig. 3. These cells expressed the lowest
levels of CAR and were the most difficult to infect with untar-
geted vectors (see Fig. 2). Upper: Bar graph showing results of
gene transfer: 105 cells were infected with 100 vps/cell of the three
different vectors for 1 hour, and luciferase activity was measured
24 hours later. The mean values  SD are given (triplicate ex-
periment). The numbers over the bars refer to the percentage of
gene transfer enhancement compared with the untargeted vectors.
Lower: Photomicrographs showing cells infected with the same
conditions as described earlier; they were stained for luciferase ex-
pression by using a new mouse mAb against luciferase, LUC-Y, 24
hours after infection (a). With untargeted vectors, positive cells
were rare (
1%, b) whereas with targeted vectors, infected cells
were clearly more abundant ( 10%, c), the staining being more
intense with Ad5LucRGD vectors.
well as the expression of several tumor antigens.24,48 The
spheroids were cultured and infected individually in a 96-
well plate coated with agarose at a concentration of 5  107
vps/ml of the four different viral preparations. Gene trans-
fer with targeted vectors compared with the nontargeted
AdCMVLuc showed a 1.7-fold increase with EGFR-target-
ed AdCMVLuc, a threefold increase with Ad5LucRGD,
and a 30-fold increase with EGFR-targeted Ad5LucRGD
(Fig. 6; p = 0.0022, according to ANOVA). The Scheffé
post hoc comparison was significant between EGFR target-
ed Ad5LucRGD and AdCMVLuc with or without EGFR
targeting. This confirms that adenoviruses with multiple li-
gands expressed on the capsid have increased efficiency for
infection of tumors comprising a heterogeneous cell popu-
lation.
Discussion
Gene therapy is an emerging therapeutic option for brain
tumors that cannot be cured using the current therapies. Al-
though the early clinical trials in malignant gliomas have
been disappointing, they have clearly indicated that poor
446
C. M. F. Dirven, et al.
FIG. 5. Graph showing gene transfer with targeted adenoviruses
in primary meningioma cells. Gene transfer of luciferase was done
with the three types of vectors described in Fig. 3. Infections were
performed as described in Fig. 4. Each circle represents the mean
luciferase activity in the cell culture derived from one patient, and
the bars represent the median of the whole group. The EGFR-tar-
geted vectors and integrin-targeted vectors infected the tumor cells
much more efficiently, threefold and ninefold more than the con-
trols, respectively.
gene transfer into the tumor cells was the principal cause
of failure of gene therapy.39,40 This critical issue there-
fore needs to be addressed before embarking on new trials.
In the case of meningiomas, the preclinical studies of gene
therapy are very few, and none has evaluated the gene trans-
fer efficiency precisely.
In this study, we systematically identified the presence of
the CAR and integrins, as well as gene transfer efficiency of
adenoviruses in primary human meningioma material. Our
results show that the high-affinity primary receptor for ade-
noviruses, CAR, and the secondary receptors that promote
virus internalization, ITGAVB3 and ITGAVB5, are both
present, not only in primary tumor material, but also in pri-
mary cell cultures. The absence of major phenotypic chang-
es between the primary tumor and the early passages of the
cells in in vitro cultures allow us to consider that, in the case
of gene transfer experiments with adenoviruses, short-term
cultures of meningioma cells represent a relevant in vitro
model. The relative phenotypic stability of meningioma
cells in culture has recently been reported also for mesen-
chymal and epithelial antigens.36
Because of the presence of the primary and secondary
receptors for adenovirus, it was expected that gene trans-
fer would be efficient with those vectors. Indeed, luciferase
gene transfer in primary meningioma cells by adenovirus
with native tropism was very effective, usually in the range
of the cell lines with the highest susceptibility to adenovirus
infection. This situation is unique among tumors, because
many rarely express CAR and are therefore difficult to
transduce with adenoviruses.18,32
In two previous studies, it has been shown that meningi-
oma cells were quite susceptible to recombinant adenovi-
rus, indicating the presence of CAR and integrins in menin-
gioma cells.23,45 It is expected, however, that intratumoral
J. Neurosurg. / Volume 97 / August, 2002
Gene transfer in meningiomas with adenoviruses
adenoviral gene therapy for these tumors will face the same
problems as are encountered in gliomas, that is, low selec-
tivity because of high CAR expression in surrounding nor-
mal brain cells.16 Therefore, new strategies that target the
virus to meningioma cells are needed to increase the se-
lectivity and eventually the efficacy of gene transfer. We
have shown that targeting strategies can, on the one hand,
increase gene transfer to tumor cells, and on the other, make
adenoviral infection more selective, thereby sparing nor-
mal cells.18,21 Integrins play a central role in cell adhesion to
the extracellular matrix and also in cell migration, signal
transduction, and induction of angiogenesis by tumors. As
shown in our study and in the literature, the cellular recep-
tors ITGAVB3 and ITGAVB5 are highly expressed in me-
ningioma cells and in meningioma vasculature, but not in
microvessels surrounding normal dura or in normal arach-
noidal cells.3,17 In addition, the ligand for ITGAVB3s, fibro-
nectin, is also highly expressed in the extracellullar matrix
of meningiomas.37 This differentiated integrin expression,
which was high in the tumor cells and vessels and low in the
normal surrounding cells, led us to examine the role of in-
tegrins as possible targets for improved adenoviral trans-
gene delivery. Using the integrin-targeted Ad5LucRGD
vector, a significant increase in transgene delivery to me-
ningioma cells was obtained compared with the untarget-
ed vector (three–23-fold improvement, median ninefold).
We found that, the lower the CAR expression, the higher
the gain observed with RGD-targeted vectors.
Alternatively, adenoviruses can be redirected toward a
tumor-specific receptor by bispecific antibodies. Because
EGFR expression has been regularly shown in meningio-
mas and is absent from normal brain, in our series of tumors
we analyzed whether EGFR could serve to redirect the ade-
noviruses.7,8,30
All the meningioma specimens analyzed, fresh tumor
material as well as primary cell cultures, revealed EGFR
expression. Expression of EGFR in normal brain cells was
almost absent, except in a few microglial cells. Infection of
primary meningioma cell cultures with the EGFR-target-
ed adenovirus resulted in an increase of transgene delivery
by a factor of three (range two–seven) compared with un-
targeted adenovirus. This increase might be explained by
a larger number of EGFRs present on the cell membrane
compared with CAR, as is indicated by our FACS data,
which shows a higher labeling ratio for EGFR than for
CAR in each tumor. The single-chain antibody construct
that binds to the fiber knob blocks the knob’s interaction
with CAR. In theory, incubating the adenovirus with 425-
S11 conjugates should ablate native tropism of the virus. In
practice, however, it is very difficult to saturate completely
all the CAR-binding sites of the adenovirus with bispecific
antibodies. For this reason, although native tropism can be
tremendously reduced in this setting, some residual CAR
binding can be observed. A further improvement may be
possible if mutant adenoviruses that do not bind to CAR
anymore could be targeted with a similar approach.
From our transfection experiments it can be concluded
that RGD targeting yields higher levels of transgene deliv-
ery than EGFR targeting. Theoretically the combination of
these two targeting strategies would redirect viral infec-
tion toward the EGFR and the integrins, thereby increasing
the chances for the adenovirus to find a receptor for infec-
tion. Indeed, when Ad5LucRGD was incubated with 425-
J. Neurosurg. / Volume 97 / August, 2002
FIG. 6. Bar graph showing gene transfer efficiency with target-
ed adenoviruses, in meningioma organotypic spheroids. Organo-
typic multicellular spheroids derived from the meningioma of the
patient in Case VU-18 were grown on agarose-coated wells in a
48-well plate. After 2 weeks in culture, eight spheroids per group
were infected with 107 vps of each of the four types of viruses:
AdCMVLuc (untargeted control vector); AdCMVLuc 425-S11
(vector targeted to the EGFR); Ad5LucRGD (vector targeted to
the ITGAVs); and Ad5LucRGD 425-S11 (vector targeted to
the EGFR and the ITGAVs simultaneously). Infections were per-
formed as described in Fig. 4. The mean amount of gene trans-
fer  SD is indicated by the numbers over the bars.
S11 bispecific single-chain antibody constructs, a further
increase in gene transfer was obtained, resulting in a 30-fold
increase of transgene delivery into meningioma spheroids.
The question arises, whether our approach would also
be of value in vivo, meaning that the targeting strategies
we have described will enhance transgene delivery to the
tumor, diminish liver toxicity, and spare the normal brain
cells. Using a similar strategy with adapter molecules, in vi-
vo targeted gene transfer has been shown in an ovarian can-
cer model and in the lung endothelium.41,42 Furthermore,
it has also been shown previously that fibroblast growth
factor receptor–targeted adenovirus demonstrates efficient
liver detargeting after intravenous administration, thereby
mitigating hepatic toxicity of the treatment.19 Concerning
the risk of unwanted gene delivery to normal brain cells,
this will be dependent on the presence of the targeted recep-
tor in these cells. Because the receptors targeted by our
strategies are not highly expressed in the brain, whereas the
CAR is, one would expect to invert the unfavorable thera-
peutic ratio with normal adenoviruses toward a more favor-
able one with targeted adenoviruses. To explore adequate
tumor targeting and brain detargeting, one would need to
develop an orthotopic human xenograft meningioma mod-
el. Such a model did not exist until recently, when Mc-
Cutcheon, et al.,31 described their xenograft model in nude
mice. For intraarterial administration in the clinic, however,
one could take advantage of the distinct vascularization of
meningiomas that arise in the external carotid artery circu-
lation. Theoretically, the targeting strategies we used would
then be of benefit, because we showed that meningioma en-
dothelial cells express EGFR and ITGAVs. It is very unlike-
ly that the use of bispecific antibodies in combination with
the adenovirus, or the insertion of the small RGD motif in
the fiber, will alter the capability of the virus to enter the tu-
mor after intravenous injection, because the size of the anti-
body is only a fraction of the viral size.
447

Conclusions
Adenoviruses are good vector candidates for gene trans-
fer into meningiomas. These tumors express all the nec-
essary receptors for adenovirus infection and are readily
infected with unmodified vectors. In addition, targeting
strategies based on the approaches we have experimented
with successfully in gliomas18 proved also to be effective
in enhancing gene transfer in meningiomas. We expect
that these adenoviruses targeted to tumor-specific receptors
would increase the therapeutic ratio in vivo; gene transfer
can be increased in tumor cells specifically, while sparing
normal cells from the potential toxicity of the virus and its
transgene. In meningiomas specifically, more than a 10-fold
improvement of gene transfer was observed, compared
with an untargeted gene transfer, which was already very
efficient compared with other tumor types. The feasibility
and the potential for clinical application was shown in rele-
vant in vitro models (primary tumor cells and spheroids).
This represents an important step toward the successful im-
plementation of the potential advantages of gene therapy in
the treatment of meningiomas.
Acknowledgment
Dr. Van Beusechem is a research fellow of the Royal Dutch Acad-
emy of Arts and Sciences. Drs. Dirven and Grill each contributed
equally to this work.
References
1. Asaoka K, Tada M, Sawamura Y, et al: Dependence of efficient
adenoviral gene delivery in malignant glioma cells on the ex-
pression levels of the Coxsackievirus and adenovirus receptor. J
Neurosurg 92:1002–1008, 2000
2. Awad IA, Kalfas I, Hahn JF, et al: Intracranial meningiomas in
the aged: surgical outcome in the era of computed tomography.
Neurosurgery 24:557–560, 1989
3. Bello L, Zhang J, Nikas DC, et al: (v)3 and (v)5 integrin ex-
pression in meningiomas. Neurosurgery 47:1185–1195, 2000
4. Bergelson JM, Cunningham JA, Droguett G, et al: Isolation of a
common receptor for Coxsackie B viruses and adenoviruses 2
and 5. Science 275:1320–1323, 1997
5. Bjerkvig R, Tonnesen A, Laerum OD, et al: Multicellular tumor
spheroids from human gliomas maintained in organ culture. J
Neurosurg 72:463–475, 1990
6. Black PM: Meningiomas. Neurosurgery 32:643–657, 1993
7. Camby I, Nagy N, Rombaut K, et al: Influence of epidermal
growth factor and gastrin on the cell proliferation of human me-
ningiomas versus astrocytic tumors maintained as ex vivo tissue
cultures. Neuropeptides 31:217–225, 1997
8. Carroll RS, Black PM, Zhang J, et al: Expression and activation
of epidermal growth factor receptors in meningiomas. J Neuro-
surg 87:315–323, 1997
9. Chamberlain MC: Adjuvant combined modality therapy for ma-
lignant meningiomas. J Neurosurg 84:733–736, 1996
10. Coke CC, Corn BW, Werner-Wasik M, et al: Atypical and malig-
nant meningiomas: an outcome report of seventeen cases. J Neu-
rooncol 39:65–70, 1998
11. Darling JL: The in vitro biology of brain tumors, in Thomas DGT
(ed): Neuro-Oncology: Primary Malignant Brain Tumours.
London: 1990, pp 1–25
12. Davis C: Surgical and non-surgical treatment of symptomatic in-
tracranial meningiomas. Br J Neurosurg 9:295–302, 1995
13. Dmitriev I, Kashentseva E, Rogers BE, et al: Ectodomain of cox-
sackievirus and adenovirus receptor genetically fused to epider-
mal growth factor mediates adenovirus targeting to epidermal
448
C. M. F. Dirven, et al.
growth factor receptor-positive cells. J Virol 74:6875–6884,
2000
14. Dmitriev I, Krasnykh V, Miller CR, et al: An adenovirus vector
with genetically modified fibers demonstrates expanded tropism
via utilization of a coxsackievirus and adenovirus receptor-inde-
pendent cell entry mechanism. J Virol 72:9706–9713, 1998
15. Dziuk TW, Woo S, Butler EB, et al: Malignant meningioma: an
indication for initial aggressive surgery and adjuvant radiothera-
py. J Neurooncol 37:177–188, 1998
16. Fechner H, Haack A, Wang H, et al: Expression of coxsackie ade-
novirus receptor and v-integrin does not correlate with ade-
novector targeting in vivo indicating anatomical vector barriers.
Gene Ther 6:1520–1535, 1999
17. Gladson CL, Cheresh DA: Glioblastoma expression of vitro-
nectin and the  v  3 integrin. Adhesion mechanism for trans-
formed glial cells. J Clin Invest 88:1924–1932, 1991
18. Grill J, Van Beusechem VW, Van Der Valk P, et al: Combined
targeting of adenoviruses to integrins and epidermal growth fac-
tor receptors increases gene transfer into primary glioma cells
and spheroids. Clin Cancer Res 7:641–650, 2001
19. Gu DL, Gonzalez AM, Printz MA, et al: Fibroblast growth factor
2 retargeted adenovirus has redirected cellular tropism: evidence
for reduced toxicity and enhanced antitumor activity in mice.
Cancer Res 59:2608–2614, 1999
20. Haisma HJ, Grill J, Curiel DT, et al: Targeting of adenoviral vec-
tors through a bispecific single-chain antibody. Cancer Gene
Ther 7:901–904, 2000
21. Haisma HJ, Pinedo HM, Rijswijk A, et al: Tumor-specific gene
transfer via an adenoviral vector targeted to the pan-carcinoma
antigen EpCAM. Gene Ther 6:1469–1474, 1999
22. Hug EB, Devries A, Thornton AF, et al: Management of atypical
and malignant meningiomas: role of high-dose, 3D-conformal
radiation therapy. J Neurooncol 48:151–160, 2000
23. Ikeda K, Saeki Y, Gonzalez-Agosti C, et al: Inhibition of NF2-
negative and NF2-positive primary human meningioma cell pro-
liferation by overexpression of merlin due to vector-mediated
gene transfer. J Neurosurg 91:85–92, 1999
24. Kaaijk P, Troost D, Das PK, et al: Long-term culture of organo-
typic multicellular glioma spheroids: a good culture model for
studying gliomas. Neuropathol Appl Neurobiol 21:386–391,
1995
25. Kleihues P, Burger PC, Scheithauer BW: The new WHO classi-
fication of brain tumors. Brain Pathol 3:255–268, 1993
26. Krasnykh VN, Douglas JT, van Beusechem VW: Genetic target-
ing of adenoviral vectors. Mol Ther 1:391–405, 2000
27. Lamberts SW, Tanghe HL, Avezaat CJ, et al: Mifepristone (RU
486) treatment of meningiomas. J Neurol Neurosurg Psychi-
atry 55:486–490, 1992
28. Lang DA, Neil-Dwyer G, Garfield J: Outcome after complex
neurosurgery: the caregiver’s burden is forgotten. J Neurosurg
91:359–363, 1999
29. Li Y, Pong RC, Bergelson JM, et al: Loss of adenoviral receptor
expression in human bladder cancer cells: a potential impact on
the efficacy of gene therapy. Cancer Res 59:325–330, 1999
30. Maenpaa A, Kovanen PE, Paetau A, et al: Lymphocyte adhesion
molecule ligands and extracellular matrix proteins in gliomas and
normal brain: expression of VCAM-1 in gliomas. Acta Neuro-
pathol 94:216–225, 1997
31. McCutcheon IE, Friend KE, Gerdes TM, et al: Intracranial injec-
tion of human meningioma cells in athymic mice: an orthotopic
model for meningioma growth. J Neurosurg 92:306–314, 2000
32. Miller CR, Buchsbaum DJ, Reynolds PN, et al: Differential sus-
ceptibility of primary and established human glioma cells to ade-
novirus infection: targeting via the epidermal growth factor re-
ceptor achieves fiber receptor-independent gene transfer. Cancer
Res 58:5738–5748, 1998
33. Morita A, Coffey RJ, Foote RL, et al: Risk of injury to cranial
nerves after gamma knife radiosurgery for skull base meningi-
omas: experience in 88 patients. J Neurosurg 90:42–49, 1999
J. Neurosurg. / Volume 97 / August, 2002
Gene transfer in meningiomas with adenoviruses
34. Newton HB, Slivka MA, Stevens C: Hydroxyurea chemother-
apy for unresectable or residual meningioma. J Neurooncol 49:
165–170, 2001
35. Oura S, Sakurai T, Yoshimura G, et al: Regression of a presumed
meningioma with the antiestrogen agent mepitiostane. Case re-
port. J Neurosurg 93:132–135, 2000
36. Pallini R, Casalbore P, Mercanti D, et al: Phenotypic change of
human cultured meningioma cells. J Neurooncol 49:9–17, 2000
37. Pau A, Bruzzone L, Dorcaratto A, et al: Accumulation of onco-
fetal fibronectin in the vessels of anaplastic meningiomas. J Neu-
rol Neurosurg Psychiatry 64:412–413, 1998
38. Perry A, Scheithauer BW, Stafford SL, et al: “Malignancy” in
meningiomas: a clinicopathologic study of 116 patients, with
grading implications. Cancer 85:2046–2056, 1999
39. Puumalainen AM, Vapalahti M, Agrawal RS, et al: Beta-galac-
tosidase gene transfer to human malignant glioma in vivo using
replication-deficient retroviruses and adenoviruses. Hum Gene
Ther 9:1769–1774, 1998
40. Rainov NG: A phase III clinical evaluation of herpes simplex vi-
rus type 1 thymidine kinase and ganciclovir gene therapy as an
adjuvant to surgical resection and radiation in adults with previ-
ously untreated glioblastoma multiforme. Hum Gene Ther 11:
2389–2401, 2000
41. Rancourt C, Rogers BE, Sosnowski BA, et al: Basic fibroblast
growth factor enhancement of adenovirus-mediated delivery of
the herpes simplex virus thymidine kinase gene results in aug-
mented therapeutic benefit in a murine model of ovarian cancer.
Clin Cancer Res 4:2455–2461, 1998
42. Reynolds PN, Zinn KR, Gavrilyuk VD, et al: A targetable, in-
jectable adenoviral vector for selective gene delivery to pulmo-
nary endothelium in vivo. Mol Ther 2:562–578, 2000
43. Schrell UM, Rittig MG, Koch U, et al: Hydroxyurea for treatment
of unresectable meningiomas. Lancet 348:888–889, 1996
44. Sharif S, Brennan P, Rawluk D: Non-surgical treatment of menin-
gioma: a case report and review. Br J Neurosurg 12:369–372,
1998
J. Neurosurg. / Volume 97 / August, 2002
45. Shu J, Lee JH, Harwalkar JA, et al: Adenovirus-mediated gene
transfer of dominant negative Ha-Ras inhibits proliferation of pri-
mary meningioma cells. Neurosurgery 44:579–587, 1999
46. Stewart DJ, Dahrouge S, Wee M, et al: Intraarterial cisplatin plus
intravenous doxorubicin for inoperable recurrent meningiomas. J
Neurooncol 24:189–194, 1995
47. Takahashi H, Herlyn D, Atkinson B, et al: Radioimmunodetec-
tion of human glioma xenografts by monoclonal antibody to epi-
dermal growth factor receptor. Cancer Res 47:3847–3850, 1987
48. Tonn JC, Ott MM, Bouterfa H, et al: Inverse correlation of cell
proliferation and expression of progesterone receptors in tumor
spheroids and monolayer cultures of human meningiomas. Neu-
rosurgery 41:1152–1159, 1997
49. van Beusechem VW, van Rijswijk AL, van Es HH, et al: Re-
combinant adenovirus vectors with knobless fibers for targeted
gene transfer. Gene Ther 7:1940–1946, 2000
50. Wara WM, Sheline GE, Newman H, et al: Radiation therapy of
meningiomas. AJR 123:453–458, 1975
51. Watkins SJ, Mesyanzhinov VV, Kurochkina LP, et al: The ‘ade-
nobody’ approach to viral targeting: specific and enhanced ade-
noviral gene delivery. Gene Ther 4:1004–1012, 1997
Manuscript received July 23, 2001.
Accepted in final form March 29, 2002.
This study was supported by a grant to Dr. Grill from the Fonda-
tion de France/Federation Nationale des Centres de Lutte Contre le
Cancer, the 1997 Spinoza Award to Dr. Pinedo from the Nether-
lands Organization for Scientific Research, and Grant Nos. R01
CA74242 from the National Institutes of Health, and P50 CA89019
and P50 CA83591 from the United States Department of Defense to
Dr. Curiel.
Address reprint requests to: Clemens M. F. Dirven, M.D., Ph.D.,
Department of Neurosurgery, Vrije Universiteit Medical Center,
De Boelelaan 1117, 1007 MB Amsterdam, The Netherlands. email:
c.dirven@vumc.nl.
449