-
Anti-VEGF agents confer survival advantages totumor-bearing mice
by improving cancer-associatedsystemic syndromeYuan Xuea, Piotr
Religaa, Renhai Caoa, Anker Jon Hansenb, Franco Lucchinic, Bernt
Jonesd, Yan Wue, Zhenping Zhue,Bronislaw Pytowskie, Yuxiang Liangf,
Weide Zhongf, Paolo Vezzonig,h, Björn Rozelli, and Yihai
Caoa,1
aDepartment of Microbiology, Tumor and Cell Biology, Karolinska
Institute, 171 77 Stockholm, Sweden; bBiopharmaceuticals Research,
Novo Nordisk A/S, DK-2760Malov, Copenhagen, Denmark; cCentro
Ricerche Biotecnologiche and Istituto di Microbiologia, Universita’
Cattolica del Sacro Cuore, 26100 Cremona, Italy;dDepartment of
Clinical Sciences, Faculty of Veterinary Medicine and Animal
Sciences, Swedish University of Agricultural Sciences, Uppsala,
Sweden; eImCloneSystems Incorporated, 108 Varick Street, New York,
NY 10014; fDepartment of Urology, First Hospital of Guangzhou,
People�s Republic of China; gIstituto diTecnologie Biomediche,
Consiglio Nazionale delle Ricerche, 20090 Segrate, Italy; iIstituto
Clinico Humanitas IRCCS, via Manzoni 56, Rozzano, Italy;
andhDepartment of Laboratory Medicine, Karolinska University
Hospital, Huddinge, Karolinska Institute, SE141 86 Stockholm,
Sweden
Edited by Robert Langer, Massachusetts Institute of Technology,
Cambridge, MA, and approved September 25, 2008 (received for review
August 12, 2008)
The underlying mechanism by which anti-VEGF agents prolong
cancerpatient survival is poorly understood. We show that in a
mouse tumormodel, VEGF systemically impairs functions of multiple
organs includ-ing those in the hematopoietic and endocrine systems,
leading toearly death. Anti-VEGF antibody, bevacizumab, and
anti-VEGF recep-tor 2 (VEGFR-2), but not anti-VEGFR-1, reversed
VEGF-induced cancer-associated systemic syndrome (CASS) and
prevented death in tumor-bearing mice. Surprisingly, VEGFR2
blockage improved survival byrescuing mice from CASS without
significantly compromising tumorgrowth, suggesting that
‘‘off-tumor’’ VEGF targets are more sensitivethan the tumor
vasculature to anti-VEGF drugs. Similarly, VEGF-inducedCASS
occurred in a spontaneous breast cancer mouse model overex-pressing
neu. Clinically, VEGF expression and CASS severity
positivelycorrelated in various human cancers. These findings
define novel ther-apeutic
targetsofanti-VEGFagentsandprovidemechanistic insights intothe
action of this new class of clinically available anti-VEGF cancer
drugs.
angiogenesis � antiangiogenic therapy � cancer syndrome �tumor
growth � VEGF
Although various anti-VEGF agents delivered as
monotherapydisplay significant anti-cancer effects in different
experimentaltumor models, their therapeutic efficacies in clinical
settings havebeen often evaluated as adjuvant therapies to
chemotherapeuticagents (1, 2). In contrast to mouse tumor studies
in which tumormasses are monitored, the clinically therapeutic
benefits are mainlydetermined based on prolonged survival time of
cancer patients (1,2). Intriguingly, anti-angiogenic drugs approved
on the basis of asurrogate marker of tumor size do not always
reduce mortality (3).
The underlying mechanisms by which VEGF antagonists
confersurvival advantages to cancer patients have not been fully
eluci-dated. In combinatorial therapy regimens, anti-VEGF agents
mightmodulate the efficacy of chemotherapeutic agents by
normalizationof tumor blood vessels (4). Most preclinical and
clinical studies ofanti-VEGF agents have focused on tumor
vasculature or tumorgrowth, and little is known about the systemic
effects of thesetherapeutic agents in the body. Most cancer
patients at the ad-vanced stage of disease encounter
cancer-associated systemic syn-drome (CASS), which significantly
impairs the quality of life andshortens lifespan. Clinical
manifestation of CASS includes a broadspectrum of symptoms
including defective hematopoiesis, endo-crine system, ascites, GI
track disorders, muscular and adiposeatrophy, and functional
impairment of liver, spleen, and kidney (5).Here, we report that
tumor-produced VEGF had extensivelydestructive effects on multiple
organs/tissues in mice and that ananti-VEGF receptor 2 (VEGFR-2)
agent significantly prolongedmouse life time by improving CASS. A
similar correlation betweenVEGF expression and CASS has also been
detected in patients withvarious cancers.
ResultsTumor-Derived VEGF Induced CASS in Immuno-Competent and
-Defi-cient Mice. Tumor-derived VEGF induces CASS affecting
multipletissues and organs in both immunocompetent and
immunodeficientmice. See supporting information (SI) Text and Figs.
S1–S4 fordetailed results.
To define the threshold level at which VEGF induced
CASS,different ratios of vector- and VEGF-transfected tumor cells
weremixed to create a series of in vivo tumors expressing different
levelsof VEGF in the in vivo tumors. At a serum concentration of
VEGFof 1.2 ng/ml, CASS was clearly manifested in liver, spleen,
bonemarrow (BM) and adrenal gland (Fig. 1B, Fig. S5). In contrast,
0.8ng/ml of serum VEGF did not result in any obvious CASS
pheno-types, indicating that approximately 1 ng/ml of serum VEGF
wasthe threshold level required to cause CASS in this
particularxenograft tumor model. Similar results were seen in mice
bearinganother VEGF-overexpressing tumor type, Lewis lung
carcinomatumors (Fig. S6). These findings show that the
tumor-producedVEGF affects multiple healthy organs in mice.
Vascular Phenotypes. Immunohistochemical analysis of
xenografttumor models using anti-CD31 antibody showed that blood
vesselsin the liver, spleen, BM, and adrenal cortex of VEGF
tumor-bearingmice appeared as primitive and dilated sinusoidal
vascular struc-tures, which consisted of disorganized, tortuous,
and intercon-nected vascular plexuses (Fig. 1A). Quantification
analysis showedthat although the vessel density in the spleen was
remarkablyincreased, the total vessel density in the liver was
significantlydecreased (Fig. 1 C and D). In addition to the cortex,
the adrenalmedulla developed a high density of vascular plexuses
(Fig. 1A).Consistent with structural alterations of the adrenal
cortex, theserum corticosterone level in VEGF tumor-bearing mice
wasconsequentially reduced (Fig. 1E). The reduction of
corticosteronelevels was reminiscent of hypoadrenocorticism found
in Addison’sdisease (6).
Hepatic Necrosis, Apoptosis Endothelial Cells (ECs) in the
SinusoidalBlood Vessels. The pathological hepatic changes induced
by thetumor-produced VEGF led to regional necrosis in the liver
tissue
Author contributions: Y.X., P.R., R.C., and Y.C. designed
research; Y.X., P.R., R.C., F.L., Y.L.,W.Z., and B.R. performed
research; Y.X., P.R., R.C., A.J.H., F.L., B.J., Y.W., Z.Z., B.P.,
Y.L., W.Z.,P.V., B.R., and Y.C. analyzed data; A.J.H., F.L., B.J.,
Y.W., Z.Z., B.P., P.V., and B.R. contributednew reagents/analytic
tools; and Y.X. and Y.C. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1To whom correspondence should be addressed. E-mail:
[email protected].
This article contains supporting information online at
www.pnas.org/cgi/content/full/0807967105/DCSupplemental.
© 2008 by The National Academy of Sciences of the USA
www.pnas.org�cgi�doi�10.1073�pnas.0807967105 PNAS � November 25,
2008 � vol. 105 � no. 47 � 18513–18518
MED
ICA
LSC
IEN
CES
Dow
nloa
ded
by g
uest
on
June
6, 2
021
http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=STXThttp://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF1http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF5http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF6http://www.pnas.org/cgi/content/full/0807967105/DCSupplementalhttp://www.pnas.org/cgi/content/full/0807967105/DCSupplemental
-
(Fig. S7A). Ki67, a proliferating marker, staining showed that
ECsin the hepatic sinusoidal blood vessels were actively
proliferating(Fig. S7B). There was an approximately sixfold
increase in prolif-erating ECs in the liver of VEGF tumor-bearing
mice as comparedwith the control group (Fig. S7D). TUNEL staining
showed anapproximately sixfold increase in hepatocyte death in the
VEGFgroup compared with the vector control (Fig. S7 B and E).
Thesefindings demonstrate that VEGF-induced hyperproliferation
ofECs in the sinusoidal blood vessels leads to an elevated
apoptoticrate accompanied by liver necrosis.
The increased death rate of hepatocytes might trigger an
inflam-matory response by infiltration of high numbers of
macrophages.F4/80 staining showed a �3-fold increase of macrophages
in hepatictissues and a 2-fold increase in spleen (Fig. S7 B, C,
and H).
Impairment of Liver Function. VEGF-induced hepatic tissue
damageresulted from the high rate of hepatocyte apoptosis and
necrosis;the expansion of sinusoidal blood vessels also suggested
impairmentof liver function. Indeed, measurement of serum
transaminasesshowed that levels of alanine transaminase (ALT) and
asparatetransaminase (AST) were already considerably elevated at
day 14after tumor implantation in VEGF tumor-bearing mice (Table
S1).In contrast, serum levels of other parameters reflecting
hepaticfunction such as cholesterol and albumin remained
unchangedduring the entire 14-day period, probably due to the
highly com-pensatory capacity of the remaining hepatocytes (Table
S1).
Severe Anemia in VEGF Tumor-Bearing Mice. Depletion of
hemato-poietic cells from BM suggested an anemic phenotype in
VEGFtumor-bearing mice. Gross examination of these mice revealed
asevere anemic phenotype, which manifested as considerable
pale-ness of several hairless regions of the mouse body, including
thepaws, mouth, nose, and genitals (Fig. S1C). Hematological
analysisof the peripheral blood showed a significant decrease in
hematocritin both immunocompetent and immunoincompetent mice at day
14after VEGF tumor implantation (Table S2). The level of
hemo-globin and the number of erythrocytes in the peripheral blood
weresignificantly decreased (Table S2). These results showed
thatVEGF tumor-bearing mice suffered from a severe anemia.
Inaddition, the total number of white blood cells was also
significantlydecreased, suggesting defective myelogenesis (Table
S3). Takentogether, depletion of BM hematopoietic cells and
decreasednumbers of red blood cells and white blood cells
demonstrate thattumor-produced VEGF results in severe anemia in
mice.
Extramedullary Hematopoiesis and Mobilization of BM Cells.
Hepa-tomegaly and splenomegaly, as well as infiltration of
hematopoieticcells, suggested that in VEGF-expressing tumor-bearing
mice ex-hibited active extramedullary hematopoiesis occurred in
theseorgans. Immunohistochemical analysis with a specific
anti-erythroblast antibody (Ter119) demonstrated a high density
oferythroblasts and reticulocytes in the liver and spleen tissues
ofVEGF tumor-bearing mice as compared with those of control
mice.These erythroblasts formed clusters, which appeared as
hemato-poietic islets (Fig. S7 B, C, F, and G). BM transplantation
ofsyngeneic EGFP� cells to irradiated recipient mice showed
signif-icant mobilization of GFP� BM-derived cells to the liver and
spleentissues (Fig. S8 A and B). These findings demonstrate that
activeextramedullary hematopoiesis occurs in livers and spleens of
mice withVEGF-expressing tumors, which mobilized BM cells to these
sites.
Consistent with extramedullary hematopoiesis, plasma levels
oferythropoietin (EPO) were significantly elevated in VEGF
tumor-bearing mice (Fig. S8C). Surprisingly, high levels of EPO
wereunable to initiate active BM hematopoiesis, suggesting a
defectiveresponse of BM hematopoietic cells to EPO. Although a
significantdecrease in circulating soluble VEGFR-2 was detected,
levels ofsoluble VEGFR-1, TNF-� and IL-6 were unchanged in
VEGFtumor-bearing mice (Fig. S8 D–F).
Tissue Hypoxia and Vascular Permeability. To study vascular
func-tions in CASS-affected tissues, tissue hypoxia and vascular
perme-ability were measured in various tissues using a Hypoxia
Probe kit.Hypoxic regions were unevenly distributed throughout VEGF
andvector control tumors (Fig. S9B). In contrast, the entire
hepatictissue of VEGF tumor-bearing mice was exposed to severe
hypoxia,whereas hypoxia was only detectable around a tiny area of
thecentral vein in control vector tumor-bearing mice (Fig.
S9B).Similarly, the cortex of the adrenal gland was also exposed to
severehypoxia in VEGF tumor-bearing mice and tissue hypoxia
wasundetectable in the cortex, except for low-level hypoxia in
themedulla of control vector tumor-bearing mice (Fig. S9E); BM
ofVEGF tumor-bearing mice exhibited a high level of
hypoxiathroughout the entire tissue (Fig. S9D). Interestingly, the
spleenshowed undetectable levels of hypoxia in VEGF or control
vectortumor-bearing mice (Fig. S9C). Vascular permeability was
increased intumors and livers of VEGF tumor-bearing mice (Fig.
S9F). Thesefindings suggest that VEGF induced abnormal vessels in
the affectedtissues and organs that are highly permeable and lack
appropriate bloodperfusion, although they contain a high number of
microvessels.
Expression of VEGFR-1 and VEGFR-2 in Various Organs. The
formationof aberrant sinusoidal vasculature in various organs
suggested thatVEGFRs are expressed in blood vessels in the tissues.
Immuno-histochemical analysis was performed using two specific
antibodiesagainst mouse VEGFR-1 (MF1) and VEGFR-2 (TO14). Blood
A
B
C D E
Fig. 1. Vascular alterations in various organs. (A)
Microvascular networks inliver, spleen, adrenal gland, and BM were
revealed by immunohistochemicalstaining with anti-CD31. Arrows
point to sinusoidal blood vessels. (B) Vascularnetworks in tumor,
liver, and BM from the circulating levels of 0.8 ng/ml and 1.2ng/ml
VEGF in mice were compared. (C and D) Vascular areas were
quantified bymeasuring CD31-positive signals and the mean values
are presented (� SD). (E)Blood corticosterone levels were measured
on day 14 after tumor implantation.Cx � cortex; M � medulla. (Scale
bars in A and B, 50 �m.)
18514 � www.pnas.org�cgi�doi�10.1073�pnas.0807967105 Xue et
al.
Dow
nloa
ded
by g
uest
on
June
6, 2
021
http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF7http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF7http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF7http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF7http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF7http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=ST1http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=ST1http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF1http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=ST2http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=ST2http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=ST3http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF7http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF8http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF8http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF8http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF9http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF9http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF9http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF9http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF9http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF9
-
vessels in the hepatic tissue of nontumor- and tumor-bearing
miceexpressed high levels of both VEGFR-1 and VEGFR-2 (Fig.
S10A).Although the expression patterns of VEGFR-1 and VEGFR-2 inthe
liver vasculature almost completely overlapped in nontumor-and
tumor-bearing mice, VEGFR-2 was expressed at a higher leveland in a
broader spectrum of vascular networks than VEGFR-1. Inspleen,
VEGFR-2 expression was significantly higher in VEGFtumor-bearing
mice as compared with control mice and VEGFR-1was barely detectable
in the blood vessels (Fig. S10B). WhereasVEGFR-2 was mainly
expressed in blood vessels, VEGFR-1 wasexpressed on non-ECs in the
adrenal gland of VEGF tumor-bearingmice (Fig. S10C). Notably, VEGF
induced accumulation ofVEGFR-1-positive cells in the cortex and
medulla of the adrenalgland (Fig. S10C). BM showed an overlapping
distribution ofVEGFR-1 and VEGFR-2 (Fig. S10D). Tumor blood vessels
ex-pressed high levels of both VEGFR-1 and VEGFR-2, althoughVEGF
tumors exhibited a significantly higher density of overlap-ping
VEGFR-1 and -2 positive signals. These findings providemolecular
targets of the tumor VEGF-induced CASS.
Reversal of VEGF-Induced CASS by Anti-VEGFR-2 but Not by
Anti-VEGFR-1. To investigate whether anti-VEGF agents could
reverseVEGF-induced CASS in VEGF tumor-bearing mice and to
define
receptor signaling pathways involved in the development of CASS,
twospecific neutralizing anti-mouse VEGFR-1 (MF1) and
VEGFR-2(DC101) monoclonal antibodies at a low dose (800 �g/mouse)
and ahigh dose (1600 �g/mouse) were administered to the
tumor-bearingmice. After a 12-day treatment, the VEGF-induced
tissue damage,including pathological changes in the liver, spleen,
adrenal gland, BM,anemia, and ascites, could be completely
prevented by the DC101antibody at both low and high doses (Fig. 2A
and C). The liver andspleen weights were significantly reduced and
reverted almost to
thoseofnontumor-bearingmice(Fig.3EandF).Thedilatedsinusoidalbloodvessels
in the liver appeared normal (Fig. 2C). Consistent with
thesehistological changes, hematocrit, hemoglobin, erythrocytes,
and liverfunction were all normalized by the anti-VEGFR-2
neutralizing anti-body (Tables S1–3). Surprisingly, anti-VEGFR-2 at
the effective dosefor normalization of systemic tissues and organs
did not show asignificant antitumor effect in our model (Fig. 2B).
However, anincreased dose of anti-VEGFR-2 showed remarkable
antitumor activ-ity and complete reversal of VEGF-induced CASS
(Fig. 2B, C). Incontrast to anti-VEGFR-2, treatment with
anti-VEGFR-1 at the samedose produced no effects on VEGF-induced
systemic syndrome ortumor growth, demonstrating that VEGFR-1 is not
the primary target
H
I J
K L
A
C
B
D
E F
G
Fig. 2. VEGF blocking and prolongation of survivals(A) At day 14
after treatment with MF1 and DC101, arepresentative mouse of each
group was photo-graphed. Arrows point to nose/mouth and paws.
As-terisks mark the abdomens of mice. (B) Tumor volumeswere
measured at the indicated times to determinetumor growth rates. (D)
The percentage of survivalanimals in each group is presented during
a 15-day-treatment course. (E and F) After killing of animals onday
15 after treatment, livers and spleens wereweighed and mean values
are presented. (C) At thesame time point, liver, spleen, adrenal
gland, and BMof buffer-treated, MF1-treated, and DC101-treatedmice
(n � 8/group) were stained with H&E (top foursets of images).
PA � portal area; RP � red pulp; WP �white pulp; Cx � cortex; and M
� medulla. Vascularnetworks in tumors and livers were revealed by
stain-ing with a CD31 antibody (bottom two sets of images).(Scal
bar, 50 �m.) (G) CD31 positive signals were quan-tified in tumor
tissues. (H) VEGF tumors were allowedto grow into sizes of 0.8 cm3,
followed by treatmentwith bevacizumab for 10 days. The mouth/nose
andpaws from a representative mouse of each group wasphotographed.
(I-K) Tumor growth rates, liver weight,and spleen weight were
measured. (L) The percentagesof survival animals in bevacizumab-
versus buffer-treated groups were presented during a 19-day
exper-imental period.
Xue et al. PNAS � November 25, 2008 � vol. 105 � no. 47 �
18515
MED
ICA
LSC
IEN
CES
Dow
nloa
ded
by g
uest
on
June
6, 2
021
http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF10http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF10http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF10http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF10http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF10http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=ST1
-
for CASS (Fig. 2 A–C, Tables S1–3). These findings demonstrate
thatnontumor vasculature was more susceptible than tumor
vasculature toanti-VEGF agents and that VEGFR-2-mediated signaling
is crucial forcausing the systemic damage of multiple tissues and
organs.
Improvement of Survival by Anti-VEGF Agents. Despite thefact
that theanti-VEGFR-2 neutralizing antibody remarkably prevented the
sys-temic VEGF syndrome, surprisingly, the tumor growth rate was
notaffected by this treatment. Consistent with this finding, tumor
bloodvessels were unaffected by this treatment (Fig. 2 C and G).
Strikingly,the anti-VEGFR-2 treatment at an effective dose
significantly im-proved the lifetime of VEGF tumor-bearing mice
(Fig. 2D). These datashow that an anti-VEGFR-2 neutralizing agent
at an optimally low dosecould significantly prolong the lifetime of
VEGF tumor-bearing micewithout significantly compromising tumor
growth.
Treatment with the humanized anti-VEGF antibody bevaci-zumab was
evaluated to further validate the survival advantage ofanti-VEGF
agents by improving CASS. At day 16 after tumorimplantation,
approximately 50% of nontreated VEGF-expressingtumor-bearing mice
(n � 8) died of CASS and the experiments hadto be terminated at the
endpoint determined by ethical consider-ations (tumor volume �1.5
cm3) (Fig. 2I). At 5 mg/kg, bevacizumabsignificantly delayed the
tumor growth rate (Fig. 2I). Interestingly,none of the
bevacizumab-treated mice (n � 8) died during theprolonged period of
experimentation (Fig. 2 I and L). Improvementof survival by
bevacizumab was not due to suppression of tumorgrowth because none
of the bevacizumab-treated mice died evenwhen the tumor reached the
ethically determined endpoint (volume�1.5 cm3). These findings
suggest that bevacizumab may prolongsurvival by improving CASS.
Indeed, VEGF-induced anemia andhepatosplenomegaly were
significantly improved by bevacizumab (Fig.2 H, J, and K). These
data confirmed the survival advantage of theVEGFR-2 blockage by
improving CASS, not by tumor inhibition per se.
VEGF-Induced CASS in a Spontaneous Mouse Tumor Model. To
studythe physiopathological relevance of our findings, a
spontaneous tumormodel of a transgenic mouse line overexpressing
the neu oncogeneunder the tissue-specific promoter of the mouse
mammary tumor virus(MMTVneu) was used (7). Female CD1 mice carrying
the neu onco-gene developed mammary tumors at the age of
approximately twomonthsandthe tumorsgrewtoarelatively
largesizeduring thenext twomonths. Strikingly, gross examination of
these mice showed pale paws,suggesting that MMTVneu tumor-bearing
mice suffered from anemia(Fig. 3A). Hematological analysis
confirmed the severe anemic pheno-type, showing significantly
reduced levels of hemoglobin, hematocrit,and erythrocytes in
peripheral blood (Fig. 3 G–I). Similar to theVEGF-overexpressing
xenograft tumor model, MMTVneu tumor-bearing mice also showed
hepatosplenomegaly (Fig. 3 C–E). Histolog-ical analysis
demonstrated that a high density of sinusoidal vasculaturefilled
the entire liver and the adrenal cortex tissues (Fig. 3B).
Indeed,anti-CD31 staining showed that the vasculature in the liver
and adrenalgland of MMTVneu tumor-bearing mice mainly consisted of
dilatedsinusoidal microvessels (Fig. 3B). In the spleen, margins of
white pulp(WP) and red pulp (RP) disappeared and were replaced by
expandinghematopoietic red pulp (Fig. 3B). In addition, the average
tumor-freebody weight of the MMTVneu transgenic mice was
significantly de-creased compared to that of wild-type mice (Fig.
3F). Consistent withdevelopment of CASS, the circulating VEGF level
was also significantlyelevated in MMTVneu tumor-bearing mice (Fig.
3J). Remarkably,treatment of these spontaneous tumor-bearing mice
with DC101 at thelow dose twice/wk almost completely reversed the
systemic syndromeincludinganemiaandhepatosplenomegaly
(Fig.3AandC–I).Similarly,histology of the liver, spleen and adrenal
gland, and vascular networksin theseorgansshowedthat
theywerecompletelynormalizedbyDC101treatment (Fig. 3B). It should
be noted that there was a trend of tumorgrowth inhibition by DC101
treatment, although the spontaneoustumor sizes were heterogeneous
and thus difficult to quantify. These
AB
C D
E F
G H
I J
Fig. 3. CASS in a spontaneous mouse tumor model.Spontaneous
mammary tumors developed in MMTV-neu transgenic mice at 2-month age
and mice werekilled when they reached 4 months old. One group
ofmice (n � 6) received the anti-VEGFR-2 treatment at adose of 800
�g/mouse. Paws (A), liver and spleen (C)were photographed. (B)
Liver, spleen, and adrenalgland were evaluated by H&E staining
(top three setsof images). The arrow indicates a hematopoietic
isletin the liver tissue. Arrowheads indicate dilated sinu-soidal
blood vessels. Tissue sections of liver and adre-nal gland were
stained with anti-CD31 (bottom twosets of images). CV, central
vein; RP, red pulp; WP,white pulp; Cx, cortex; M, medulla. (Scale
bars, 50 �m.)Liver weight (D), spleen weight (E), and net
bodyweight (F) were measured. Blood samples were col-lected and
hemoglobin (G), hematocrit (H), and eryth-rocytes (I) were
determined. (J) The serum levels ofVEGF in various groups of mice
were measured using asensitive ELISA.
18516 � www.pnas.org�cgi�doi�10.1073�pnas.0807967105 Xue et
al.
Dow
nloa
ded
by g
uest
on
June
6, 2
021
http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=ST1
-
findings demonstrate that a spontaneous tumor mouse model that
wasnotgeneticallypropagatedtooverexpressVEGFalsodevelopedCASS,which
was correlated with an elevated level of circulating VEGF
andreversed by an anti-VEGFR-2 agent.
Correlation of Circulating VEGF Levels with Development of CASS
inHuman Cancer Patients. To further correlate our findings with
clinicalrelevance, we analyzed blood samples derived from cancer
patients. Weused exactly the same ELISA method for our mouse
experiments tomeasure the circulating levels of VEGF.
Interestingly, we found that thecirculating VEGF level was
significantly higher in prostate and bladdercancer patients and
renal cell carcinoma (RCC) patients as comparedwith healthy
individuals (Fig. 4A). The circulating VEGF levels in thesepatients
(average of 1–1.5 ng/ml) were in a range similar to those foundin
our xenograft mouse tumor model (Fig. 1B, Fig. S5).
Interestingly,RCC patients had slightly higher VEGF levels than
prostate andbladder cancer patients. Circulating VEGF levels
correlated well withseverity of hepatomegaly, splenomegamy and
ascites (Fig. 4 B, D, andE). Consistent with this positive
correlation, liver tissues in patients withhigh VEGF levels showed
sinusoidal dilation of vascular networks andimpaired functions,
including high levels of ALT and AST, and lowlevels of albumin
(Fig. 4 C and G–I). Again, RCC patients showed asignificantly
positive correlation between VEGF level and impairmentof liver
function (Fig. 4 K–M). In contrast, hemoglobin levels
weresignificantly decreased and reversely correlated with the
circulatingVEGF levels in these cancer patients, particularly in
RCC patients (Fig.4 F and J). These findings demonstrate an
equivalent circulating VEGFlevel between human cancer patients and
VEGF tumor-bearing mice.Furthermore, VEGF levels were positively
correlated with the severityof CASS in human cancer patients.
DiscussionHere, we show that tumor-produced VEGF induces CASS by
damag-ing the structures and functions of multiple tissues and
organs. VEGF-induced CASS was manifested as severe anemia, hepatic
dysfunction
and necrosis, ascites, loss of body weight, and low serum levels
ofcorticosterone. The severity of these systemic changes was
generallywell correlatedwith thecirculatingVEGFlevel
inbothmiceandhumancancer patients. VEGF-induced CASS resembles
cancer cachexia andparaneoplastic syndromes, which manifests
functional failures of mul-tiple organs often at an advanced stage
of the malignancy (8). Cancercachexia and paraneoplastic syndrome
are the primary causes ofmortality in cancer patients. Although a
few cytokines including TNF-�and IL-6 contribute to CASS, its
underlying molecular mechanismsremain unknown (8, 9). Our present
study provides a novel mechanisticinsight into the role of
tumor-derived VEGF in the development ofCASS.
In the xenograft VEGF tumor model, we were able to determine
thethreshold level of VEGF that causes CASS by mixing
VEGF-producingand vector-transfected tumor cells in different
ratios. Similar circulatingVEGF levels were detected in various
cancer patients including RCC,prostate cancer and bladder cancer
patients. In fact, the averagecirculating VEGF level in RCC
patients was approximately 1.5 ng/ml.Intriguingly, most of these
cancer patients developed obvious CASSincluding severe anemia,
hepatosplenomegaly, and ascites. These clin-ical data correlated
well with our VEGF tumor model in mice. It isestimated that at the
time of diagnosis, the rate of CASS is �7–10% ofpatients with
malignancy and that as many as 50% of all cancer
patientsmayexperiencesuchasyndromeat sometimeduring thecourseof
theirillness (5). Similar to our present findings in mouse tumors
and humanpatients, autopsies of RCC patients revealed that �20% of
patients hadsinusoidal dilation in the liver, spleen, and adrenals
(10). The sinusoidaldilation of these organs is considered to be a
nonmetastatic tumor-specific manifestation, although the etiology
remains unclear. It shouldbe mentioned that VEGF-induced vascular
leakage might be involvedin the axis of the reactive
oxygen-rac-angiopoietin-2 pathway (11).
CASS is defined as a constellation of symptoms in
associationwith the presence of an actively growing tumor that
releases anunknown factor in excess into the circulation. The
identity of thisunknown factor has not been characterized. Our
present study with
A B
C
D E
F G H I
J K L M
Fig. 4. CASS in human cancer patients.Clinical samples were
collected from RCC,bladder and prostate cancer patients.
(A)Circulating levels of VEGF were measuredby ELISA. (C)
Histological micrographs ofH&E and CD31
immunohistochemicalstaining of livers from RCC patients at ahigh
(1.2 ng/ml) and a low (0.3 ng/ml) cir-culating VEGF level are
presented. (Scalebars, 50 �m.) The development of hepato-megaly (B)
and splenomegaly (D) were cor-related with circulating VEGF levels.
(E) Per-centage of patients with ascites wascorrelated with the
average circulatingVEGF level. Levels of blood hemoglobin (F),ALT
(G), AST (H), and albumin (I) were mea-sured and correlated with
circulating VEGFexpression levels (J–M). HI, healthy individ-uals;
PC, prostate cancer patients; BC, blad-der cancer patients; RCC,
renal cell carci-noma patients. Statistic analyses wereindicated as
in figures.
Xue et al. PNAS � November 25, 2008 � vol. 105 � no. 47 �
18517
MED
ICA
LSC
IEN
CES
Dow
nloa
ded
by g
uest
on
June
6, 2
021
http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF5
-
the VEGF-expressing tumor model in mice resembles an
equivalentsituation in cancer patients possessing advanced
VEGF-secretingtumors and demonstrates that the unknown factor is
likely to beVEGF. In addition to RCC, sinusoidal dilation of
tissues has alsobeen observed in other tumors. For example, nine
patients withdifferent cancers including Hodgkin’s disease, stomach
cancer,lymphosarcoma, and gastric carcinoma had dilation of
hepaticsinusoidal blood vessels (12). A similar systemic syndrome
is alsopresent in patients with malignant histiocytosis, pediatric
Wilm’stumor, and pancreatic cancers (13–15). The percentage of
patientswith highly dilated hepatic sinusoidal blood vessels is
probably muchhigher than that reported in the literature because
almost all caseswere encountered at autopsy, which is not routinely
performed.
Intriguingly, high VEGF levels induced CASS in both Lewis lung
andfibrosarcoma models, suggesting that the VEGF-induced
systemiceffect is independent of tumor type. Indeed, continuous
injection ofpurified VEGF protein into nontumor-bearing mice could
also causehepatosplenomegaly in mice (Fig. S11). In addition to
high VEGF-producing tumor models, a spontaneous breast cancer mouse
model,which is not genetically propagated to express VEGF, also
developeda similar systemic syndrome as manifested by severe
anemia, hepato-splenomegaly, ascites, and loss of body weight. The
circulating VEGFlevels in these spontaneous tumor-bearing mice were
lower than that ofmice with VEGF xenograft tumors, and the
development and growthrate of these spontaneous tumors were
considerably slower than forxenograft tumors. VEGF may accumulate
in various tissues and organsover a relatively long period of tumor
development. Persistent exposureof these organs to VEGF might
result in initiation of vascular growthand impairment of vascular
function. Indeed, vascular networks in liver,spleen, and adrenal
glands of spontaneous tumor-bearing mice exhib-ited a high degree
of disorganization, dilation, and tortuous architec-ture.
Importantly, anti-VEGFR-2 could completely reverse
vascularabnormalities and tissue structures in MMTVneu tumor mice.
Takentogether, this finding demonstrates that VEGF plays an
important rolein initiation, progression and maintenance of CASS in
spontaneoustumor-bearing mice.
Surprisingly, BM hematopoietic cells were virtually
completelyeradicated by VEGF in mice. Due to a lack of a sufficient
numberof hematopoietic stem cells in BM, both red blood cells and
whiteblood cells in the peripheral blood were dramatically
decreased.Development of anemia is unlikely due to the direct
inhibitoryeffect of VEGF on hematopoiesis because extramedullary
hema-topoiesis in the liver and spleen was stimulated by VEGF.
Overall, our studies demonstrate that in both xenograft and
spon-taneous tumor-bearing mice, tumor-expressed VEGF induces
CASS,which resembles cachexia and paraneoplastic syndromes in
human
cancer patients. Circulating VEGF levels correlated well with
CASSseverity in tumor-bearing mice and human cancer patients. We
suggestthat nontumor tissues are important therapeutic targets for
improve-ment in cancer patient survival. The functional and
pathologicalchanges in tissues and organs might serve as useful
noninvasive markersfor the effectiveness of anti-VEGF therapy in
improving cancer patientsurvival rates. Thus, these results provide
molecular insight into theglobal impact of tumor-produced VEGF in
cancer patients and suggestthat combinatorial therapies of
anti-VEGF agents with other drugs toimprove tissue and organ
function will produce immense benefits forcancer patients.
Experimental ProceduresAnimals, Human Materials, and Mouse Tumor
Model. All animal studies werereviewed and approved by the animal
care and use committees of the local animalboard. All human studies
were approved by the Chinese Medical Information Com-mittee.
Detailed methods and criteria of patient selection are described in
SI Text.
Tissue and Organ Collection, ELISA, and Blood Sample Analysis.
See SI Text fordetails.
Tissue Hypoxia Analysis and Vascular Permiability Assay. Tissue
hypoxia in tumortissues, liver, spleen, BM, and adrenal glands was
measured according to a standardprotocol using HypoxyprobeTM-1 Plus
kit (Chemicon). See SI Text for details.
Bone Marrow Transplantation and Tumor Implantation. See SI Text
for details.
Histological Studies, Whole-Mount Staining and Immunofluorescent
Staining.Malignant and nonmalignant paraffin-embedded tissues were
sectioned in 5 �mthickness and stained with hematoxylin-eosin
(H&E) according to our previouslydescribed methods (18).
Paraffin sections of BM tissues were stained with theanti-mouse
CD31 antibody and positive signal were developed using DAB as
thesubstrate. Whole-mount staining was performed according to
previously pub-lished methods (19). See SI Text for details.
Statistical Analysis.
Statisticalanalysiswasperformedusingthestudent’s t
testbyaMicrosoftExcelprogram.Datawerepresentedasmeansofdeterminants
(�SD)and p-values � 0.05 were considered as statistically
significant. The Kaplan-Meiersurvival curve was generated using
Statistica 5.0 (Statsoft).
ACKNOWLEDGMENTS. We thank Dr. Rolf Brekken at the University of
TexasSouthwestern Medical Center for supplying the anti-VEGFR-2
polyclonal antibody.This work was supported by the laboratory of
Y.C. through research grants from theSwedish Research Council, the
Swedish Heart and Lung Foundation, the
SwedishCancerFoundation,theKarolinskaInstituteFoundation,andtheTorstenandRagnarSöderberg’s
Foundation and by European Union Integrated Projects of
Angiotar-geting Contract 504743 (to Y.C.) and VascuPlug Contract
STRP 013811 (to Y.C.), andsupported in part by a grant from
Fondazione Cariplo (N.O.B.E.L Project) (to P.V.).
1. Hurwitz H (2004) Integrating the anti-VEGF-A humanized
monoclonal antibody bev-acizumab with chemotherapy in advanced
colorectal cancer. Clin Colorectal Cancer 4(Suppl 2):S62–S68.
2. Motzer RJ, et al. (2007) Sunitinib versus interferon alfa in
metastatic renal-cell carci-noma. N Engl J Med 356:115–124.
3. Ledford H (2008) Drug markers questioned. Nature
452:510–511.4. Jain RK (2005) Normalization of tumor vasculature:
An emerging concept in antian-
giogenic therapy. Science 307:58–62.5. Nathanson L, Hall TC
(1997) Introduction: Paraneoplastic syndromes. Semin Oncol
24:265–268.6. Vieweg WV, Reitz, RE, Weinstein RL (1973)
Addison’s disease secondary to metastatic
carcinoma: an example of adrenocortical and adrenomedullary
insuffiency. Cancer31:1240–1243.
7. Lucchini F, et al. (1992) Early and multifocal tumors in
breast, salivary, harderian andepididymal tissues developed in
MMTY-Neu transgenic mice. Cancer Lett (Shannon,Irel)
64:203–209.
8. Oliff A (1988) The role of tumor necrosis factor (cachectin)
in cachexia. Cell 54:141–142.9. Kayacan O, et al. (2006) Impact of
TNF-alpha and IL-6 levels on development of cachexia
in newly diagnosed NSCLC patients. Am J Clin Oncol
29:328–335.10. Aoyagi T, Mori I, Ueyama, Y, Tamaoki N (1989)
Sinusoidal dilatation of the liver as a
paraneoplastic manifestation of renal cell carcinoma. Hum Pathol
20:1193–1197.
11. Perry BN, et al. (2006) Pharmacologic blockade of
angiopoietin-2 is efficacious againstmodel hemangiomas in mice. J
Invest Dermatol 126:2316–2322.
12. Karasawa T, Shikata, T, Smith RD (1979) Peliosis hepatis.
Report of nine cases. ActaPathol Jpn 29:457–469.
13. Bjork O, Eklof O, Willi, U, Ahstrom L (1985) Veno-occlusive
disease and peliosis of theliver complicating the course of Wilms’
tumour. Acta Radiol Diagn (Stockh) 26:589–597.
14. Fine KD, Solano M, Polter, DE, Tillery GW (1995) Malignant
histiocytosis in a patientpresenting with hepatic dysfunction and
peliosis hepatis. Am J Gastroenterol 90:485–488.
15. Otani M, et al. (1992) An autopsy case of renal cell
carcinoma associated with extensivepeliosis hepatis. Acta Pathol
Jpn 42:62–68.
16. Eriksson A, et al. (2002) Placenta growth factor-1
antagonizes VEGF-induced angio-genesis and tumor growth by the
formation of functionally inactive PlGF-1/VEGFheterodimers. Cancer
Cell 1:99–108.
17. Sacco MG, et al. (2002) Combined effects on tumor growth and
metastasis by anti-estrogenic and antiangiogenic therapies in
MMTV-neu mice. Gene Ther 9:1338–1341.
18. Cao R, et al. (2003) Angiogenic synergism, vascular
stability and improvement ofhind-limb ischemia by a combination of
PDGF-BB and FGF-2. Nat Med 9:604–613.
19. Cao R, et al. (2006) Hepatocyte growth factor is a
lymphangiogenic factor with anindirect mechanism of action. Blood
107:3531–3536.
18518 � www.pnas.org�cgi�doi�10.1073�pnas.0807967105 Xue et
al.
Dow
nloa
ded
by g
uest
on
June
6, 2
021
http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=SF11http://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=STXThttp://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=STXThttp://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=STXThttp://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=STXThttp://www.pnas.org/cgi/data/0807967105/DCSupplemental/Supplemental_PDF#nameddest=STXT