Aberrant angiogenic characteristics of human brain arteriovenous malformation endothelial cells.

Abstract

To identify and characterize the phenotypic and functional differences of endothelial cells derived from cerebral arteriovenous malformations (AVM), as compared with endothelial cells derived from a normal brain.
Isolated AVM brain endothelial cells and control brain endothelial cells were evaluated immunohistochemically for expression of the endothelial cell markers von Willebrand factor and CD31, as well as angiogenic factors including vascular endothelial growth factor A, interleukin-8, and endothelin-1. Vascular endothelial growth factor receptors 1 and 2 were also evaluated using immunohistochemistry techniques. Functional assays evaluated cell proliferation, cytokine production, tubule formation, and cell migration using the modified Boyden chamber technique.
Endothelial cells derived from AVMs expressed high levels of vascular endothelial growth factor A and significantly overexpressed the vascular endothelial growth factor receptors 1 and 2 (P < 0.05), as compared with control endothelial cells. In addition, comparison to control brain endothelial cells demonstrated that AVM brain endothelial cells proliferated faster, migrated more quickly, and produced aberrant tubule-like structures.
Endothelial cells derived from cerebral AVMs are highly activated cells overexpressing proangiogenic growth factors and exhibiting abnormal functions consistent with highly activated endothelial cells.

Full text

Page 1

NEUROSURGERY
VOLUME 64 | NUMBER 1 | JANUARY 2009 | 139
CLINICOPATHOLOGICAL STUDY
Mark N. Jabbour, M.D.
Department of Pathology,
Keck School of Medicine,
University of Southern California,
Los Angeles, California
James B. Elder, M.D.
Department of Neurological Surgery,
Keck School of Medicine,
University of Southern California,
Los Angeles, California
Christian G. Samuelson, M.D.
Department of Neurosurgery,
University of Texas,
San Antonio, Texas
Shabnam Khashabi, B.S.
Department of Pathology,
Keck School of Medicine,
University of Southern California,
Los Angeles, California
Florence M. Hofman, Ph.D.
Department of Pathology,
Keck School of Medicine,
University of Southern California,
Los Angeles, California
Steven L. Giannotta, M.D.
Department of Neurological Surgery,
Keck School of Medicine,
University of Southern California,
Los Angeles, California
Charles Y. Liu, M.D., Ph.D.
Department of Neurological Surgery,
Keck School of Medicine,
University of Southern California,
Los Angeles, California, and
Division of Chemistry and
Chemical Engineering,
California Institute of Technology,
Pasadena, California
Reprint requests:
Charles Y. Liu, M.D., Ph.D.,
Department of Neurological Surgery,
Keck School of Medicine,
University of Southern California,
1200 N. State Street, Suite 5046,
Los Angeles, CA 90033.
Email: cliu@usc.edu
Received, March 11, 2008.
Accepted, July 11, 2008.
B
mains a considerable challenge facing neuro-
surgeons. This vascular pathology usually
becomes apparent before the age of 40 years
and affects up to 0.1% of the population in the
United States (1, 8). AVMs can present with
seizure disorders or neurological deficits from
steal phenomenon, or they can be found inci-
dentally. Intracranial hemorrhage may also be
a clinical presentation of AVMs, resulting in a
high risk of death and disability (1). In fact,
rupture of these malformations causes approx-
imately 2% of all strokes and an even higher
rain arteriovenous malformations (AVM)
are a tangle of abnormally formed blood
vessels, the management of which re -
percentage of intracerebral hemorrhage in
young patients (1, 3).
The goals of treatment for AVMs include alle-
viating mass effect and intracranial hyperten-
sion associated with rupture, preventing future
ruptures, alleviating steal phenomenon, and
treating associated epilepsy. The modern treat-
ment of brain AVMs often involves multimodal
therapy, including microsurgery, stereotactic
radiosurgery, and endovascular procedures.
Each of these modalities carries significant risk,
depending on the location of the lesion in rela-
tion to nearby eloquent cortical structures, deep
subcortical structures, and neurovascular struc-
tures (12, 16, 20). In addition, some patients
ABERRANT ANGIOGENIC CHARACTERISTICS OF
HUMAN BRAIN ARTERIOVENOUS MALFORMATION
ENDOTHELIAL CELLS
OBJECTIVE:To identify and characterize the phenotypic and functional differences of
endothelial cells derived from cerebral arteriovenous malformations (AVM), as com-
pared with endothelial cells derived from a normal brain.
METHODS: Isolated AVM brain endothelial cells and control brain endothelial cells
were evaluated immunohistochemically for expression of the endothelial cell markers
von Willebrand factor and CD31, as well as angiogenic factors including vascular
endothelial growth factor A, interleukin-8, and endothelin-1. Vascular endothelial
growth factor receptors 1 and 2 were also evaluated using immunohistochemistry tech-
niques. Functional assays evaluated cell proliferation, cytokine production, tubule for-
mation, and cell migration using the modified Boyden chamber technique.
RESULTS: Endothelial cells derived from AVMs expressed high levels of vascular
endothelial growth factor A and significantly overexpressed the vascular endothelial
growth factor receptors 1 and 2 (P ? 0.05), as compared with control endothelial cells.
In addition, comparison to control brain endothelial cells demonstrated that AVM brain
endothelial cells proliferated faster, migrated more quickly, and produced aberrant
tubule-like structures.
CONCLUSION: Endothelial cells derived from cerebral AVMs are highly activated cells
overexpressing proangiogenic growth factors and exhibiting abnormal functions con-
sistent with highly activated endothelial cells.
KEY WORDS: Angiogenesis, Angiogenic markers, Arteriovenous malformations, Brain vasculature
Neurosurgery 64:139–148, 2009
DOI: 10.1227/01.NEU.0000334417.56742.24
www.neurosurgery- online.com
ABBREVIATIONS: AVM, arteriovenous malformation; BEC, brain endothelial cell; ELISA, enzyme-linked
immunosorbent assay; ET-1, endothelin-1; FCS, fetal calf serum; IL-8, interleukin-8; VEGF, vascular
endothelial growth factor; VEGFR, vascular endothelial growth factor receptor; vWF, von Willebrand factor

Page 2

140 | VOLUME 64 | NUMBER 1 | JANUARY 2009
www.neurosurgery-online.com
JABBOUR ET AL.
panel of molecular markers. We show here that endothelial
cells derived from cerebral AVMs produce angiogenic factors,
such as VEGF and endothelin-1 (ET-1), that are not normally
produced by the quiescent brain vasculature. Furthermore,
these endothelial cells are highly activated and exhibit func-
tional abnormalities in migration and tubule formation consis-
tent with cells in active angiogenesis.
MATERIALS AND METHODS
Endothelial Cell Cultures
Tissues were obtained and handled in accordance with the Keck
School of Medicine, University of Southern California, Institutional
Review Board’s guidelines. Tissues from 6 patients who had undergone
surgical excision of their AVMs were collected. Control endothelial
cells were derived from surgical specimens of 3 patients who under-
went temporal lobectomies for epilepsy associated with mesial tempo-
ral sclerosis. Clinical and demographic data for all patients are listed in
Table 1. Endothelial cells were isolated from all surgical specimens as
described previously (31). Briefly, tissues were gently homogenized.
Microvessels were isolated on a 15% sucrose gradient and enyzmati-
cally dissociated with collagenase/dispase. For both AVM BECs and
control BECs, distinctions based on type of vessel such as vein, artery,
capillary, and nidus or caliber of vessel were not made. All endothelial
cells, regardless of their caliber or vessel of origin, were included in the
culture associated with their respective sample. Endothelial cells were
cultured in Gibco RPMI 1640 medium (Invitrogen Corp., Carlsbad,
CA), 100 ng/mL endothelial cell growth supplement (Upstate Bio -
technologies, Rochester, NY), 2 mmol/L L-glutamine (Invitrogen
Corp.), 10 mmol/L N-2-hydroxyethylpiperazine-N?-2-ethanesulfonic
acid (Invitrogen Corp.), 24 mmol/L sodium bicarbonate (Invitrogen
Corp.), 300 U heparin (Sigma-Aldrich, St. Louis, MO), 1% peni-
cillin/streptomycin (Invitrogen Corp.), and 10% fetal calf serum (FCS)
(Omega Scientific, Inc., Tarzana, CA). The purity of control BECs and
AVM BECs was confirmed by immunostaining for specific endothelial
cell markers: CD31 (platelet/endothelial cell adhesion molecule-1)
(Santa Cruz Biotechnology, Inc., Santa Cruz, CA), von Willebrand
diagnosed with AVMs are considered to have lesions that are
inoperable or untreatable with current treatment modalities (27).
Indeed, the microsurgical treatment of brain AVMs is consid-
ered to represent one of the highest technical challenges facing
neurosurgeons. The optimal treatment for patients with AVMs
remains controversial, and patients with unruptured AVMs are
the subject of an ongoing prospective study called A Random -
ized Trial of Unruptured Brain AVMs (11). Clinical studies such
as this will help optimize therapeutic strategies for patients with
AVMs with the use of existing treatment options. In the future,
however, less invasive methods such as advanced endovascular
techniques and pharmacological therapy are likely to play
increasing roles in the treatment of patients with AVMs because
of the potential for improved risk and side effect profiles asso-
ciated with these treatments. There are currently no pharmaco-
logical treatments, although recent in vitro work has begun to
explore possibilities in this area (14).
AVMs are histopathologically characterized by the presence
of a nidus, a conglomerate of tortuous vessels lacking capillar-
ies, an arterial feeder, and a draining vein (8). There has been
considerable interest in characterizing AVMs, including the
abnormalities in angiogenesis that lead to their formation.
Previous research characterizing these lesions demonstrated
increased expression of angiogenic factors, such as vascular
endothelial growth factor (VEGF), at the protein and messenger
ribonucleic acid levels (10, 25). However, these studies were
usually performed on whole tissues rather than purified, iso-
lated vessels. To understand the etiology of this disease and
more specifically characterize this vascular abnormality, the
present experiments were undertaken to study isolated AVM
endothelial cells derived from surgical specimens and to exam-
ine the phenotypic and functional characteristics of these cells
as compared with normal brain endothelial cells (BEC). To
accomplish this, we isolated pure populations of endothelial
cells as described and verified the purity of these cells using a
aAVM, arteriovenous malformation; L, left; ICH, intracerebral hemorrhage; R, right; IVH, intraventricular hemorrhage; BEC, brain endothelial cell; NA, not applicable.
bPatient demographics of the tissue samples studied. Note that all patients were embolized before surgical resection, and none underwent radiosurgery.
TABLE 1. Clinical characteristics of patients involved in this studya, b
Tissue sample
and patient no.
Age (y)/
sex
Presenting symptoms
Location of
pathology
Indication for surgery
Previous
embolization
Previous
radiosurgery
AVM
1
2
3
4
5
6
BEC
1
2
3
26/F
23/M
21/M
68/M
42/F
16/F
Headache
Headache
Seizures
Neurological deficit
Headache
Incidental
L cerebellum
R occipital lobe
R parietal lobe
L cerebellum
L occipital lobe
R parietal lobe
ICH
IVH, ICH
IVH
ICH
IVH
Family decision
Yes
Yes
Yes
Yes
Yes
Yes
No
No
No
No
No
No
23/F
65/M
32/F
Seizures
Seizures
Seizures
R temporal lobe
L temporal lobe
L temporal lobe
Mesial temporal sclerosis
Mesial temporal sclerosis
Mesial temporal sclerosis
NA
NA
NA
NA
NA
NA

Page 3

NEUROSURGERY
VOLUME 64 | NUMBER 1 | JANUARY 2009 | 141
CHARACTERISTICS OF ARTERIOVENOUS MALFORMATION ENDOTHELIAL CELLS
Factor (vWF) (Dako North America, Inc., Carpinteria, CA), and CD105/
endoglin (Santa Cruz Biotechnology, Inc.). Control BEC and AVM BEC
cultures were grown on 1% gelatin-coated surfaces, and only cells from
Passages 3 to 6 were used in experiments. Digital images were taken
using a Sony DSC-P9 camera (Sony Corp., Tokyo, Japan) and Nikon
TMS microscope (objective lens, ?10 magnification; 0.25 numerical
aperture) (Nikon Corp., Tokyo, Japan) at room temperature.
Immunohistochemistry and Analysis
Specimens were immunostained as described previously (17). Briefly,
cytocentrifuge cell preparations were fixed in acetone, blocked with 5%
goat serum, and incubated overnight with the following commercially
available primary antibodies: goat anti-human CD31 (1:100) (Santa Cruz
Biotechnology, Inc.); mouse anti-human glial fibrillary acidic protein
(1:100) (Dako North America, Inc.), rabbit anti-human vWF (1:600), mouse
anti-human CD11b (1:100); rabbit anti-human ET-1 (1:1000) (Bachem, San
Carlos, CA); anti-human VEGF-A (1:100) (Chemicon International,
Temecula, CA); mouse anti-human VEGF receptor (VEGFR) 1/2 (1:100)
(Chemicon International); and goat anti-human interleukin-8 (IL-8) (R&D
Systems, Inc., Minneapolis, MN). The slides were then incubated with the
appropriate secondary reagent, either biotinylated goat anti-rabbit or
horse anti-mouse antibody (Vector Laboratories, Inc., Burlingame, CA) for
45 minutes, treated with the avidin biotin peroxidase complex (Vector
Laboratories, Inc.) for 30 minutes, then treated with aminoethyl carbazol
substrate (Vector Laboratories, Inc.) for 10 minutes. Samples were coun-
terstained with hematoxylin for 1.5 minutes. The red precipitate denoted
positive staining.
Enzyme-linked Immunosorbent Assay
Cells were seeded, at a density of 105cells per 30-mm dish, in media
containing 1% FCS for 72 hours. Supernatants were then collected and
analyzed for VEGF-A, IL-8, and ET-1 using commercially available
enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems, Inc.)
according to the manufacturer’s protocol.
Cell Migration Assay
Migration assays were performed using transwell chambers as
described previously (6). Briefly, 8-μm-pore size, 6.5-mm-diameter
polyethylene terephthalate filters (BD BioCoat; BD Biosciences, San Jose,
CA) were coated with 0.5% gelatin on the top and bottom by immersion
for at least 2 hours and air-dried overnight. Subconfluent cultures of
endothelial cells were harvested using a trypsin/ethylenediamine tetra-
acetic acid (0.05%) solution (Invitrogen Corp.); 5 ? 104cells were then
resuspended in RPMI medium supplemented with 1% FCS and placed
on top of the filter in the top chamber (100 (l/well). The chemoattrac-
tant, IL-8, was placed in the lower compartment (600 (l/well).
Transwell plates were then incubated at 37?C in 5% CO2for 6 hours. At
the termination of the experiments, media were removed from inserts,
and the remaining cells on the upper side of the chamber were
removed with a wet cotton swab. The filters were then stained with
Diff-Quick staining solution (Siemens Healthcare Diagnostics, Inc.,
Deerfield, IL) according to the manufacturer’s protocol. Cells that
migrated through to the underside of the membrane were counted
under ?20 power magnification. Ten fields per filter were counted. All
experiments were performed in triplicate.
Proliferation Assay
Next, 5 ? 104 endothelial cells in RPMI medium with 10% FCS were
plated in 6-well plates and incubated for 24, 48, and 72 hours. Then,
cells were trypsinized and counted using the trypan blue dye exclusion
technique. The experiments were performed in triplicate for each time
point. Results were expressed as total number of cells retrieved at each
time point.
Tubule Formation Assay
Endothelial cells were trypsinized and resuspended in reduced
serum media, as described above. Resuspended cells (2 ? 104cells in
200 μL media) were added to growth factor-reduced Matrigel (BD
Biosciences)-coated 96-well plates and incubated for 12 and 24 hours at
37?C. Tubule formation was quantitated using NIH Image J (National
Institutes of Health, Bethesda, MD).
Statistical Analysis
All experiments were performed in triplicate. Analysis of variance
was used for multigroup comparisons, and the post-hoc Tukey test
was used for testing statistical significance (P ? 0.05). Two-group com-
parisons were analyzed using the paired t test. Quantitative data are
reported as mean ? standard error of the mean. Statistical analyses
were performed using standard database software.
RESULTS
Characterization of Endothelial Cells Derived
from AVM BECs
Pathological examinations of AVM tissues demonstrated
unusual vascular morphology and density. Immunostaining
demonstrated that more than 99% of AVM BECs and control
BECs expressed typical endothelial cell markers, such as vWF
(Fig. 1, A and B), CD31 (Fig. 1, D and E), and CD105 (data not
shown). Cultured AVM BECs and control BECs were negative
for both glial fibrillary acidic protein, a marker of astrocytes, and
CD11b, a marker of monocytes/macrophages/microglia (data
not shown). Two glioma cell lines (U87 and LN229) served as
negative controls for the endothelial cell markers vWF and
CD31 (Fig. 1, C and F). Morphological differences were appar-
ent when cultures of AVM BECs were grown to confluency.
These cells lacked the typical cobblestone-like appearance asso-
ciated with cultured BECs (Fig. 1G). Thus, AVM BECs displayed
marked morphological distinctions from control BECs.
Expression of Angiogenic Factors in AVM BECs
Analysis of immunocytochemistry experiments showed that
AVM BECs expressed higher levels of intracellular VEGF-A as
compared with control BECs (Fig. 2A). Secreted VEGF was also
evaluated using the ELISA method by testing the supernatants
from cell cultures. Results from these experiments indicated
that AVM BECs secreted approximately twice the amount of
VEGF-A as compared with control BECs (2059.5 ? 350.2 versus
1038.9 ? 225.5 pg/mL, respectively; P ? 0.05) (Fig. 3A).
Stimulation of 105BECs with tumor necrosis factor served as
the positive control. Comparisons of VEGFR expression in
AVM BECs and control BECs showed that AVM BECs dis-
played elevated VEGFR 1 and diminished VEGFR 2 as com-
pared with control BECs (Fig. 2, B and C).
Immunostaining with anti-ET-1 antibody revealed that AVM
BECs exhibited qualitatively higher ET-1 levels as compared

Page 4

142 | VOLUME 64 | NUMBER 1 | JANUARY 2009
www.neurosurgery-online.com
JABBOUR ET AL.
with control BECs (Fig. 4, A and B). ELISA results demon-
strated higher ET-1 production by AVM BECs as compared
with control BECs (39.1 ? 0.8 versus 25 ? 1.7 pg/mL; P ? 0.5)
(Fig. 3B). Analysis of IL-8 immunostaining showed that neither
AVM BECs nor control BECs expressed appreciable amounts of
this proangiogenic cytokine compared with glioma cells, which
served as the positive control (Fig. 4, C–E). ELISA results also
demonstrated much lower IL-8 production by AVM BECs
(83.3 ? 3.3 pg/mL) and control BECs (64 ? 3.1 pg/mL) as
compared with glioma cells (746.7 ? 13.3 pg/mL; P ? 0.05, as
compared with both AVM BECs and control BECs). Differences
between AVM BECs and control BECs were also statistically
significant (P ? 0.05) (Fig. 3C).
Functional Analysis of AVM BECs
Evaluation of the cell proliferation assay showed that AVM
BECs demonstrated a higher number of cells at each time point
as compared with control BECs. There were 63% more cells at
24 hours, 60% more cells at 48 hours, and 50% more cells at
FIGURE 2. Immunostaining showing expression of vascular endothelial
growth factor A (VEGF-A) and VEGF receptors (VEGFR) in AVM BECs
and control BECs. Cerebral AVM BECs and control BECs were immunos-
tained for the proangiogenic growth factor VEGF-A, as well as receptors
VEGFR 1 and 2. A and B, VEGF-A, qualitatively elevated in AVM BECs
as compared with control BECs. C and D, VEGFR 1, qualitatively
expressed at higher levels in AVM BECs as compared with control BECs.
E and F, VEGFR 2 expression, lower in AVM-BECs as compared with
control BECs. Arrows denote positive staining. Scale bars, 100 µm.
AB
CD
EF
FIGURE 1. Immunostaining showing characterization of cell surface mark-
ers on arteriovenous malformation (AVM)-derived endothelial cells. A–F,
AVM brain endothelial cells (BEC), control BECs, and glioma cell lines were
immunostained for the endothelial-specific markers von Willebrand Factor
(vWF) (A–C, respectively) and CD31 (D–F, respectively). Both AVM BECs
and control BECs were more than 98% positive for vWF and CD31, whereas
glioma cells were negative for both. G and H, using light microscopy (?20
magnification), the cobblestone appearance normally associated with
endothelial cells was not apparent in AVM BEC cultures (G), but was
apparent in control BEC cultures (H). Original magnification, ?200 (A, B,
F–H), ?100 (C), and ?400 (D, E).
AB
CD
EF
GH

Page 5

NEUROSURGERY
VOLUME 64 | NUMBER 1 | JANUARY 2009 | 143
CHARACTERISTICS OF ARTERIOVENOUS MALFORMATION ENDOTHELIAL CELLS
72 hours (Fig. 5A). Cell migration assays using transwell cham-
bers revealed that AVM BEC migration was significantly greater
than migration of control BECs (47 ? 7 versus 23 ? 4 cells/
high-power field; P ? 0.05) (Fig. 5B). When both cell types were
stimulated with the chemokine IL-8 (10 ng/mL), there was
enhanced migration of control BECs (50 ? 2 cells/high-power
field; P ? 0.05) as expected, and no enhancement in migration of
AVM BECs (42 ? 4 cells/high-power field).
Results of the tubule formation assay demonstrated that
AVM BECs exhibited fewer tubules compared with control
BECs and that these tubules were comparatively aberrant (Fig.
6). Evaluation of tubule lengths demonstrated that AVM BEC
tubules were significantly shorter compared with control BEC
tubules (4009 ? 1673 versus 9612 ? 2368 μm; P ? 0.05). Thus,
the AVM BECs formed fewer, shorter, and more irregular
tubules compared with control BECs.
DISCUSSION
The results presented in this study demonstrate that AVM
BECs are “activated” endothelial cells compared with control
BECs in terms of proangiogenic cytokine secretion and proan-
giogenic receptor expression. VEGF is a critical mediator of
angiogenesis (4). This growth factor stimulates different func-
tions of endothelial cells, including differentiation, survival,
proliferation, and migration (7, 32). The increased VEGF-A
secretion by AVM BECs observed in our experiments is consis-
tent with recent research reporting increased tissue levels of
VEGF-A in AVM specimens (15, 25), although these previous
studies were performed on whole tissues composed of a mix-
ture of cells rather than isolated endothelial cells. Thus, we
have identified at least 1 source of VEGF as being the endothe-
lial cells.
Regarding the levels of VEGFRs 1 and 2 on AVM BECs as
compared with control BECs, conflicting reports exist.
Increased levels of these receptors in AVM tissue were previ-
FIGURE 3. Graphs showing results of enzyme-linked
immunosorbent assays (ELISA), which were used to
evaluate supernatants of cell cultures of AVM BECs
and control BECs. A–C, supernatants of AVM BEC
cultures contained higher concentrations of VEGF-A
(A), endothelin-1 (ET-1) (B), and interleukin-8 (IL-8)
(C), as compared with supernatants of control BEC
cultures. Results were statistically significant for each
cytokine. Error bars display standard error of the
mean (SEM).
A
B
C
AB
CD
FIGURE 4. Immunostaining show-
ing expression of ET-1 on AVM
BECs and control BECs. A and B,
cell preparations of AVM BECs (A)
and control BECs (B) immunos-
tained for ET-1. AVM BECs exhib-
ited elevated expression of ET-1, as
compared with control BECs.
Control BEC and AVM BEC cul-
ture preparations were stained with
the antibody to IL-8. C–E, both AVM BECs (C) and control BECs (D) were
negative, whereas glioma cells (E) were positive. Original magnification,
?400 (A) and ?100 (B–D).
E

Page 6

144 | VOLUME 64 | NUMBER 1 | JANUARY 2009
www.neurosurgery-online.com
JABBOUR ET AL.
ously observed in the vasculature of AVMs using whole tissue
blocks (28). However, another study reported a decreased level
of both receptors when comparing AVM vessels with normal
brain tissue (13). A number of factors, including tissue hetero-
geneity, low sample numbers, and experimental design, may
contribute to the discrepancy between these reports. Our
results showing increased VEGFR 1 and decreased VEGFR 2 in
AVM BECs are, therefore, not consistent with either report.
Thus, although VEGF and its receptors may play a role in the
pathogenesis of AVMs, further investigations are required to
definitively characterize these processes.
In addition to VEGF-A, other proangiogenic factors such as
ET-1 and IL-8 may be important in AVM pathogenesis. ET-1 is
a proangiogenic factor responsible for endothelial cell migra-
tion and proliferation, as well as a potent vasoconstrictor asso-
ciated with hypertension and cerebral vasospasm after experi-
mental subarachnoid hemorrhage (24, 29). The results reported
above show that ET-1 was overexpressed in AVM BECs com-
pared with control BECs. These results fit with previous
research, which indicated that AVM endothelial cells are contin-
uously stimulated, as well as other work associating ET-1 with
AVMs (2). Although the expression of IL-8 was not appreciably
different in AVM BECs compared with control BECs on the
basis of immunohistochemistry, the ELISA results showed
increased production of this cytokine in AVM BECs. It is not
clear whether this increase in secretion of IL-8 in AVM endothe-
lial cells is biologically significant. Overall, the presented data
demonstrate that AVM BECs produced abnormally high levels
of proangiogenic factors compared with control BECs.
The angiogenic process requires endothelial cell proliferation,
migration, and tubule formation, and this study analyzed these
activities in AVM BECs. The in vitro functional studies demon-
strated that AVM BECs proliferated and migrated faster than
control cells and formed fewer tube-like structures. Furthermore,
the tube-like structures formed by AVM BECs were much more
irregular in configuration as compared with the tubules formed
by control BECs, indicating decreased tubular organization in
AVM BECs. These features are reminiscent of the highly acti-
vated, proangiogenic endothelial cells associated with tumor-
derived endothelial cells (5). However, unlike tumors, these
apparent abnormalities appear to be a result of factors within the
AVM vasculature rather than a tumor environment. Overall,
AVM BECs displayed aberrant behaviors in functional assays as
compared with control BECs. Future work will continue to char-
acterize these differences, and, in concert with further molecular
and genetic characterizations, may ultimately yield targets for
the first pharmacological therapy for AVMs. Characterization of
AVM BECs may also provide information useful in the opti-
mization of endovascular or radiosurgical techniques.
Certain limitations of this study preclude more definitive con-
clusions. The low number of surgical specimens is attributable
to the recent infrequency of open craniotomy for AVMs at our
institution. Currently, only a small percentage of patients who
present with AVMs undergo surgical resection. This low num-
ber precludes more powerful statistical analysis and prevents
statistically valid interpretation of quantified immunostaining
results. The low number of patients also prevents comparisons
between AVM groups, such as between patients with unrup-
tured (n ? 1) and ruptured (n ? 5) AVMs. Despite this limita-
FIGURE 6. Comparison of tubule formation by AVM BECs and control
BECs. Both cell types were evaluated using the modified 3-dimensional
tubule formation assay. After 24 hours, tubule lengths were analyzed and
calculated using Image J software (National Institutes of Health, Bethesda,
MD). A and B, tubules formed by AVM BECs (A) were shorter and more
irregular compared with those formed by control BECs (B).
FIGURE 5. Graphs showing cell proliferation, cell
migration, and influence of IL-8. A, AVM BECs
demonstrating a higher rate of cell proliferation as com-
pared with control BECs. The numbers of endothelial
cells in AVM BEC and control BEC cultures were
evaluated every 24 hours for 3 consecutive days. AVM
BECs proliferated more rapidly than control BECs. B,
AVM BECs demonstrated a faster rate of migration as
compared with control BECs. AVM BECs and control
BECs, untreated or treated with IL-8 (10 ng/mL), were
allowed to migrate through a Boyden chamber filter
for 6 hours and were then counted on the underside of
the filter. Ten different regions of the filters were eval-
uated for each experiment. Error bars display SEM.
AB
A
B

Page 7

tion, comparisons between AVM BECs and control BECs in
terms of angiogenic factors (cytokines and receptors) and func-
tional activities clearly indicate a distinct phenotypic difference.
The use of temporal lobectomies from epilepsy patients as
the control tissue is another limitation of this study. The
resected brain tissue in these patients cannot be considered
truly “normal” because of the presence of an epileptogenic
focus and other possible alterations in the brain parenchyma,
such as hypoplasia and gliosis. Although the possibility exists
that endothelial cells from vessels in these samples may not be
perfectly representative of “normal” BECs, they are likely the
closest approximation available and have been used by us and
others in the field of vascular biology (19, 22, 26).
Another limitation of this study is that experiments were
conducted after exposure of AVM BECs and control BECs to
cell culture conditions. The process of culturing cells in vitro
may cause changes in the cells being evaluated, and compar-
isons of results between 2 cell populations grown in vitro must
always mention this caveat. However, these kinds of mechanis-
tic analyses can only be performed on cells in vitro. We are
currently exploring in vivo disease models to use for future
studies. Passaging cells and exposing cells to the artificial cul-
ture environments may alter cell characteristics. For this reason,
we correlated the data obtained from in vitro studies to those
of the AVM vasculature in vivo and showed that results from
in vitro experiments were consistent with the ex vivo tissue
blocks. Based on this information, primary human endothelial
cell cultures are the appropriate in vitro model. Cell cultures in
Passages 3 to 6 were used, because, by Passage 8, morpholog-
ical differences in the endothelial cells were observed.
The novel approach in this study is the use of isolated endo -
thelial cells derived from cerebral AVM tissues in characterizing
their functions. Over the years, several studies have analyzed
angiogenic factors and receptors in brain AVMs, both at the pro-
tein and messenger ribonucleic acid levels (10, 25). However,
these in situ studies used tissue specimens containing a hetero-
geneous mixture of cells, which had significant limitations
because of the difficulty in identifying and distinguishing suben-
dothelial from endothelial staining structures in a background of
extensive gliosis that is typically associated with brain AVMs.
Therefore, the advantages of using isolated AVM BECs are the
clear identification of the nature and site of growth factor pro-
duction, as well as the ability to test the functional capacity of
these cells as compared with the normal vasculature. Such a
study is crucial for understanding the disease process and iden-
tifying a treatment for inoperable cases of cerebral AVMs.
An explanation for the differences in characteristics observed
between AVM BECs and control BECs is beyond the scope of
the present study, but it merits some general speculation with
an eye toward future work. These studies will include character-
ization of flow dynamics, cell surface markers, cytokine secre-
tion, and genetic differences among both cell types. Elevated
shear stress as a result of malformed vasculature has previously
been implicated as a factor contributing to remodeling of the
cerebral vasculature in the setting of an AVM (23). Future work
involving in vitro assessment of the effects of shear stress may
NEUROSURGERY
VOLUME 64 | NUMBER 1 | JANUARY 2009 | 145
CHARACTERISTICS OF ARTERIOVENOUS MALFORMATION ENDOTHELIAL CELLS
further clarify the role and mechanism of this effect. Larger
studies may have the statistical power to evaluate differences
based on vessel type, such as feeding artery, nidus, and drain-
ing vein. Overall, with this accumulation of data, identification
of factors associated with the pathogenesis of AVMs may
become clearer. At this point, speculation regarding the cause of
the observed differences could guide further research.
The optimal treatment for patients diagnosed with AVMs
remains a topic of controversy and is currently the subject of
a prospective clinical trial (11, 30). Recent work indicates that
the natural history risk of patients with unruptured AVMs is
lower than that for patients who undergo intervention for
unruptured AVMs and patients who present with ruptured
AVMs. This is attributed to the risks involved with the stan-
dard treatment options for AVMs, which include open surgi-
cal resection, endovascular embolization, and radiation ther-
apy (12, 16, 18, 21).
In the future, treatments for complex intracranial pathologies
such as AVMs will continue to use less invasive techniques,
such as endovascular procedures, and open craniotomies will
play a decreasing role. Pharmacotherapy could offer a noninva-
sive treatment alternative if an appropriate target and effective
agent could be identified. For example, recent research has pro-
posed tetracylines for preventing pathological vascular remod-
eling and suppression of matrix metalloproteinase 9 with doxy-
cycline (9, 14). Further characterization of the differences
be tween AVM BECs and control BECs may yield additional
insight into factors important for AVM pathogenesis and assist
scientists in designing innovative treatments. Investigation of
AVM BECs in terms of their angiogenic properties, as described
in this study, may further these goals toward noninvasive man-
agement by elucidating unique properties that could be tar-
geted with pharmacological agents. Differentiating properties
unique to the endothelium of AVMs may provide a specific
target for therapy, thereby avoiding unwanted side effects on
normal endothelial cells.
CONCLUSION
This study demonstrated that human brain AVM endothelial
cells can be successfully isolated from surgical specimens.
Furthermore, AVM BECs expressed angiogenic factors and
demonstrated functional characteristics of activated cells. It
remains unclear whether this is because of inherent differences
in the cell populations or, rather, a reactive phenomenon
because of abnormal flow and ischemia. More studies are
required to further elucidate these questions. An improved
understanding of these issues will provide greater insights for
the treatment of AVMs, which continue to present some of the
greatest challenges in neurosurgery. Despite advances in micro-
surgical technique, radiosurgery strategies, and endovascular
methodologies, these current treatment modalities carry signif-
icant risks. It may be possible to take advantage of the knowl-
edge of the angiogenesis process in brain AVMs to normalize
these vessels, thereby making them more stable and thus resist-
ant to rupture. Alternatively, these AVM vessels may become

Page 8

146 | VOLUME 64 | NUMBER 1 | JANUARY 2009
www.neurosurgery-online.com
JABBOUR ET AL.
sensitized to radiosurgery, making AVMs more treatable.
Insights into abnormal angiogenesis evident in brain AVMs may
also lead to understanding of various brain pathologies, includ-
ing neoplastic and cerebrovascular disorders.
Disclosures
These studies were supported by the Anspach Foundation (Charles Y. Liu,
M.D., Ph.D.). The authors have no personal financial gain or institutional inter-
est in any of the drugs, materials, or devices described in this article.
REFERENCES
1. Anonymous: Arteriovenous malformations of the brain in adults. N Engl J
Med 340:1812–1818, 1999.
2. Boscolo E, Pavesi G, Zampieri P, Conconi MT, Calore C, Scienza R, Parnigotto
PP, Folin M: Endothelial cells from human cerebral aneurysm and arteriove-
nous malformation release ET-1 in response to vessel rupture. Int J Mol Med
18:813–819, 2006.
3. Brown RD Jr, Flemming KD, Meyer FB, Cloft HJ, Pollock BE, Link ML:
Natural history, evaluation, and management of intracranial vascular malfor-
mations. Mayo Clin Proc 80:269–281, 2005.
4. Carmeliet P: VEGF as a key mediator of angiogenesis in cancer. Oncology
69:4–10, 2005.
5. Charalambous C, Chen TC, Hofman FM: Characteristics of tumor-associated
endothelial cells derived from glioblastoma multiforme. Neurosurg Focus
20:E22, 2006.
6. Charalambous C, Pen LB, Su YS, Milan J, Chen TC, Hofman FM: Interleukin-
8 differentially regulates migration of tumor-associated and normal human
brain endothelial cells. Cancer Res 65:10347–10354, 2005.
7. Drake CJ, LaRue A, Ferrara N, Little CD: VEGF regulates cell behavior dur-
ing vasculogenesis. Dev Biol 224:178–188, 2000.
8. Fleetwood IG, Steinberg GK: Arteriovenous malformations. Lancet 359:863–
873, 2002.
9. Frenzel T, Lee CZ, Kim H, Quinnine NJ, Hashimoto T, Lawton MT, Guglielmo
BJ, McCulloch CE, Young WL: Feasibility of minocycline and doxycycline use
as potential vasculostatic therapy for brain vascular malformations: Pilot
study of adverse events and tolerance. Cerebrovasc Dis 25:157–163, 2008.
10. Gault J, Sarin H, Awadallah NA, Shenkar R, Awad IA: Pathobiology of
human cerebrovascular malformations: Basic mechanisms and clinical rele-
vance. Neurosurgery 55:1–17, 2004.
11. Hartmann A, Mast H, Choi JH, Stapf C, Mohr JP: Treatment of arteriovenous
malformations of the brain. Curr Neurol Neurosci Rep 7:28–34, 2007.
12. Hartmann A, Pile-Spellman J, Stapf C, Sciacca RR, Faulstich A, Mohr JP,
Schumacher HC, Mast H: Risk of endovascular treatment of brain arteriove-
nous malformations. Stroke 33:1816–1820, 2002.
13. Hashimoto T, Emala CW, Joshi S, Mesa-Tejada R, Quick CM, Feng L, Libow
A, Marchuk DA, Young WL: Abnormal pattern of Tie-2 and vascular endothe-
lial growth factor receptor expression in human cerebral arteriovenous mal-
formations. Neurosurgery 47:910–919, 2000.
14. Hashimoto T, Matsumoto MM, Li JF, Lawton MT, Young WL; University of
California, San Francisco, BAVM Study Group: Suppression of MMP-9 by
doxycycline in brain arteriovenous malformations. BMC Neurol 5:1, 2005.
15. Hashimoto T, Wu Y, Lawton MT, Yang GY, Barbaro NM, Young WL:
Coexpression of angiogenic factors in brain arteriovenous malformations.
Neurosurgery 56:1058–1065, 2005.
16. Haw CS, terBrugge K, Willinsky R, Tomlinson G: Complications of emboliza-
tion of arteriovenous malformations of the brain. J Neurosurg 104:226–232,
2006.
17. Hofman FM: Immunohistochemistry, in Coligan JE, Kruisbeek AM, Margulies
DH, Shevach EM, Strober W (eds): Current Protocols in Immunology. New
York, John Wiley & Sons, 1996, pp 5.8.1–5.8.23.
18. Jayaraman MV, Marcellus ML, Hamilton S, Do HM, Campbell D, Chang SD,
Steinberg GK, Marks MP: Neurologic complications of arteriovenous malfor-
mation embolization using liquid embolic agents. AJNR Am J Neuroradiol
29:242–246, 2008.
19. Kaplanski G, Fabrigoule M, Boulay V, Dinarello CA, Bongrand P, Kaplanski
S, Farnarier C: Thrombin induces endothelial type II activation in vitro: IL-1
and TNF-α-independent IL-8 secretion and E-selectin expression. J Immunol
158:5435–5441, 1997.
20. Lawton MT, Du R, Tran MN, Achrol AS, McCulloch CE, Johnston SC,
Quinnine NJ, Young WL: Effect of presenting hemorrhage on outcome after
microsurgical resection of brain arteriovenous malformations. Neurosurgery
56:485–493, 2005.
21. Moreno-Jiménez S, Celis MA, Lárraga-Gutiérrez JM, de Jesús Suárez-Campos
J, García-Garduño A, Hernández-Bojórquez M: Intracranial arteriovenous
malformations treated with linear accelerator-based conformal radiosurgery:
Clinical outcome and prediction of obliteration. Surg Neurol 67:487–492,
2007.
22. Petzelbauer P, Watson CA, Pfau SE, Pober JS: IL-8 and angiogenesis: Evidence
that human endothelial cells lack receptors and do not respond to IL-8 in
vitro. Cytokine 7:267–272, 1995.
23. Rossitti S, Svendsen P: Shear stress in cerebral arteries supplying arteriove-
nous malformations. Acta Neurochir (Wien) 137:138–145, 1995.
24. Salani D, Taraboletti G, Rosanò L, Di Castro V, Borsotti P, Giavazzi R, Bagnato
A: Endothelin-1 induces an angiogenic phenotype in cultured endothelial
cells and stimulates neovascularization in vivo. Am J Pathol 157:1703–1711,
2000.
25. Shenkar R, Elliott JP, Diener K, Gault J, Hu LJ, Cohrs RJ, Phang T, Hunter L,
Breeze RE, Awad IA: Differential gene expression in human cerebrovascular
malformations. Neurosurgery 52:465–478, 2003.
26. Söderquist B, Källman J, Holmberg H, Vikerfors T, Kihlström E: Secretion of
IL-6, IL-8 and G-CSF by human endothelial cells in vitro in response to
Staphylococcus aureus and staphylococcal exotoxins. APMIS 106:1157–1164,
1998.
27. Spetzler RF, Martin NA: A proposed grading system for arteriovenous mal-
formations. 1986. J Neurosurg 108:186–193, 2008.
28. Uranishi R, Baev NI, Ng PY, Kim JH, Awad IA: Expression of endothelial cell
angiogenesis receptors in human cerebrovascular malformations. Neurosurgery
48:359–368, 2001.
29. Vijay A, Santhanam R, Katusic ZS: Genetic modification of cerebral arterial
wall: Implications for prevention and treatment of cerebral vasospasm.
Neurol Res 28:759–768, 2006.
30. Wedderburn CJ, van Beijnum J, Bhattacharya JJ, Counsell CE, Papanastassiou
V, Ritchie V, Roberts RC, Sellar RJ, Warlow CP, Al-Shahi Salman R; SIVMS
Collaborators: Outcome after interventional or conservative management of
unruptured brain arteriovenous malformations: A prospective, population-
based cohort study. Lancet Neurol 7:223–230, 2008.
31. Wu Z, Hofman FM, Zlokovic BV: A simple method for isolation and character-
ization of mouse brain microvascular endothelial cells. J Neurosci Methods
130:53–63, 2003.
32. Zadeh G, Guha A: Angiogenesis in nervous system disorders. Neurosurgery
53:1362–1376, 2003.
Acknowledgments
We thank Ligaya Pen, B.Sc., for her assistance with the cell cultures. We thank
the University of Southern California Doheny Eye Institute for generous access
to the core microscope laboratory. We thank Thomas C. Chen, M.D., of the
Department of Neurological Surgery, Keck School of Medicine, University of
Southern California, for his contribution and access to the laboratory.
COMMENTS
J
and normal brain. The authors isolated target cells and evaluated them
immunohistochemically for expression of von Willebrand factor, CD31,
vascular endothelial growth factor A (VEGF-A), interleukin-8, endothe-
lin-1, and vascular endothelial growth factor receptors 1 and 2 (VEGF-
R1/R2). Functional assays were used to evaluate cell proliferation,
abbour et al. investigated the phenotypic and functional differences
between endothelial cells from arteriovenous malformations (AVMs)

Page 9

cytokine production, tubule formation, and cell migration. Results
showed that AVM endothelial cells expressed high levels of VEGF-A
and significantly overexpressed VEGF-R1/R2 as compared with con-
trol endothelial cells. Moreover, AVM endothelial cells proliferated
faster, migrated more quickly, and produced aberrant tubule-like struc-
tures. On the basis of these findings, the authors concluded that AVM
endothelial cells are highly activated cells that overexpress proangio-
genic growth factors and exhibit abnormal functions.
This study confirms what was already largely suspected. By identi-
fying specific markers, however, the authors have generated molecular
targets to aid in lesion diagnosis and treatment modalities. As the cur-
rent management of asymptomatic AVM patients remains controver-
sial, future investigations should focus on correlating these cell mark-
ers with lesion natural history and patient outcome. The question
remains which patients with unruptured AVMs should receive inter-
vention. Current literature cannot answer this question definitively,
but this study is a step in the proper direction by further elucidating
AVM pathophysiology.
Ricardo J. Komotar
E. Sander Connolly, Jr.
New York, New York
C
of this study is straightforward. The authors isolated endothelial cells
of cerebral AVMs and immunohistochemically evaluated the expres-
sion of cell markers VEGF, VEGF receptors, interleukin-8, and endothe-
lin-1. They functionally analyzed cell proliferation and migration,
cytokine production, and tubule formation, comparing them with
endothelial cells from normal brain. They found that endothelial cells
from cerebral AVMs were highly activated. Because of the small sam-
ple size, they did not analyze any differences between AVMs with and
without hemorrhage, or between patients of younger age and older
age. Effects of prior embolization on changes in cell markers and
cytokine production also need to be evaluated. Although the findings
in this study do not seem to be specific to cerebral AVMs, regional dif-
ferences in AVMs (feeder, nidus, drainer) may be interesting topics for
further study.
The most important point is that cultured cells may change their
inherent characteristics after isolation and preparation, and immuno-
histochemical findings in tissue samples analyzed immediately after
resection may be helpful in discussing protein expression. This kind of
approach may provide a new avenue to treatment of inoperable and
challenging AVMs in the future.
erebral AVMs are reported to change in size dynamically and to
recur after total extirpation, particularly in childhood. The concept
Kazuhiko Nozaki
Otsu, Japan
T
normal endothelia of brain vessels. This “proangiogenic” phenotype of
BAVM-derived endothelial cells is likely related to the formation,
growth, and rupture of the lesions.
Unfortunately, the authors could not distinguish from which part of
the BAVM the endothelial cells were derived; an interesting question is
whether the “proangiogenic” endothelia are restricted to the nidus of
the BAVM or can be found in the whole lesion. Furthermore, compar-
isons between endothelial cells from different parts of the BAVM (e.g.,
nidus versus draining veins) would be interesting.
It is hoped that further studies will dissect the molecular aberra-
tions that cause the more active phenotype of BAVM endothelial cells.
his study elegantly demonstrates a functional and phenotypic dif-
ference between the endothelial cells of brain AVMs (BAVMs) and
NEUROSURGERY
VOLUME 64 | NUMBER 1 | JANUARY 2009 | 147
CHARACTERISTICS OF ARTERIOVENOUS MALFORMATION ENDOTHELIAL CELLS
Identification of specific molecular aberrations behind the more active
and proangiogenic phenotype of BAVMs could open the door for the
design of pharmaceutical therapy that could inhibit the growth and
rupture of BAVMs or even cause the lesions to regress.
Juhana Frösen
Mika Niemelä
Juha Hernesniemi
Helsinki, Finland
J
secretion, cell proliferation and migration, and tubule formation in
vitro in comparison to human brain capillary endothelium from epilep-
tic temporal lobe. Compared with normal specimens, the endothelium
cultured from AVMs showed different morphology, expressed and
secreted VEGF and VEGF-R1 and -R2, proliferated and migrated more
rapidly, and tended to form aberrant tubules.
Endothelia are a protean cell type. They are highly adaptable to the
local environment and can rapidly express a different phenotype in
response to both intra- and extravascular substances and other cell
types. They are highly participatory in inflammation and can modulate
critical responses to injury (e.g., intimal hyperplasia, blood-brain bar-
rier disruption, etc. [2]). In part, it is this phenotypic adaptability that
makes endothelium difficult to study. As the authors point out, even
multiple passages can “erase” a phenotypic characteristic as the
endothelia adapt to the new culture environment. The same issue per-
tains to findings in the current experiment. Activation of the “prolifer-
ative” endothelial phenotype, as observed in these experiments, can
occur in response to many stimuli, particularly increased shear stress
(1). Additionally, endothelium in an AVM is removed from cellular
and autocrine influences in adjacent brain. Thus, it is not clear whether
the phenotype observed in BAVM endothelium is an adaptation to the
high-flow AVM environment or is related to the pathogenesis of AVM
pathology. Until we better understand the pathogenesis of AVMs, it
will be difficult to answer this question.
abbour et al. carefully analyzed human BAVM endothelial cells
derived from surgical specimens, examining protein expression and
Marc R. Mayberg
Seattle, Washington
1. Desai SY, Marroni M, Cucullo L, Krizanac-Bengez L, Mayberg MR, Hossain
MT, Grant GG, Janigro D: Mechanisms of endothelial survival under shear
stress. Endothelium 9:89–102, 2002.
2. Krizanac-Bengez L, Kapural M, Parkinson F, Cucullo L, Hossain M, Mayberg
MR, Janigro D: Effects of transient loss of shear stress on blood-brain barrier
endothelium: Role of nitric oxide and IL-6. Brain Res 977:239–246, 2003.
I
epilepsy specimens, which presumably have normal vessels. They then
isolate endothelial cells from the epilepsy patients and the AVM
patients and evaluate them for phenotypic and functional differences.
Other studies have been done of tissues from AVMs, but this is a study
of the isolated endothelial cells. The authors state that their cells show
a purity of at least 99% by endothelial markers and that they are neg-
ative for typical markers of astrocytes, macrophages, and microglia.
The authors find that the isolated endothelial cells express higher lev-
els of VEGF-A and the receptors VEGF-R1/R2. They do feel that the
endothelial cells derived from the AVMs also are morphologically dis-
tinct with regard to their appearance on confluent plates. The authors
also imply that endothelin-1 appeared to be at higher levels in the
n this article, the authors report on tissues from 6 patients with
human BAVMs and compare them with tissues derived from

Page 10

AVM cells, but not interleukin-8 cytokine. In addition, on assays, the
cells themselves seemed to proliferate at a faster rate, migrate more
quickly, and produce aberrant tubules.
The study suggests that isolated cultures and cultured cells from
AVMs are different from cells cultured from normal brain. When eval-
uating the results of such methodology, one must take into account the
potential of stimulation of function or primary regulation of functions
by the isolation and culture process itself. In addition, we have no bet-
ter control than that of the epilepsy brains, but we must caution that
these may not be truly normal cells. This speculation, which the
authors do mention, should be expanded upon. The question of
whether these differences are an inherent function of the cells of AVMs,
suggesting genetic modification, which has been present from the ori-
gin of the AVM, or whether they are a particular response to the flow
characteristics of the AVM should also be developed. If the latter is true,
flow itself may be a clinical target. Further explanation and discussion
of the importance of these distinctions will shed light on these results.
Robert J. Dempsey
Madison, Wisconsin
JABBOUR ET AL.
148 | VOLUME 64 | NUMBER 1 | JANUARY 2009
www.neurosurgery-online.com
Blastocyst. Credit: Wellcome Library, London and Sandel MJ: Embryo ethics—The moral logic of stem-
cell research. N Engl J Med 351:207–209, 2004. See Apuzzo, p 1, and Farin et al., pp 15–39.