3452
`
`Long-term progression and therapeutic response of
`visceral metastatic disease non-invasively monitored
`in mouse urine using B-human choriogonadotropin
`secreting tumor cell lines
`
`Giulio Francia,1 Urban Emmenegger,1
`Christina R. Lee,1 Yuval Shaked,1
`Christopher Folkins,1'2 Miriam Mossoba,1
`Jeffrey A. Medin,2 Shan Man,1 Zhenping Zhu,3
`Larry Witte,3 and Robert S. Kerbell'2
`
`1Molecular and Cellular Biology Research, Sunnybrook Health
`Sciences Centre, and 2Department of Medical Biophysics,
`University of Toronto, Toronto, Ontario, Canada; and 3|mC|one
`Systems, lnc., New York, New York
`
`Abstract
`
`the use of mouse models of metastatic
`Historically,
`disease to evaluate anticancer therapies has been ham-
`pered because of difficulties in detection and quantifica-
`tion of such lesions without sacrificing the mice, which
`in turn may also be dictated by institutional or ethical
`guidelines. Advancements in imaging technologies have
`begun to change this situation. A new method to non-
`invasively measure tumor burden, as yet untested to mon-
`itor spontaneous metastases, is the use of transplanted
`tumors expressing secretable human B-chorionic gonado-
`tropin (B-hCGI that can be measured in urine. We describe
`examples of B-hCG—transfected tumor cell
`lines for
`evaluating the effect of different therapies on metastatic
`disease, which in some cases involved monitoring tumor
`growth for >100 days. We used B-hCG—tagged mouse
`B16 melanoma and erbB-Z/Her-Z—expressing human
`breast cancer MDA-MB-231 models, and drug treatments
`included metronomic low—dose cyclophosphamide chemo-
`therapy with or without a vascular endothelial growth
`factor receptor 2—targeting antibody (DC101) or trastu-
`
`Received 2/27/08; revised 5/12/08; accepted 5/18/08.
`Grant support: The Ontario Institute for Cancer Research, NIH USA
`(CA-41233; RS. Kerbel), the National Cancer Institute of Canada,
`and a sponsored research agreement with |mC|one Systems Inc. RSK is a
`Tier l Canada Research Chair in Tumor Biology, Angiogenesis and
`Antiangiogenic Therapy. Trastuzumab was generously provided by
`Genentech, lnc., South San Francisco.
`The costs of publication of this article were defrayed in part by the
`payment of page charges. This article must therefore be hereby marked
`advertisement in accordance with 18 U.S.C. Section 1734 solely to
`indicate this fact.
`Requests for reprints: Robert S. Kerbel, Molecular and Cellular Biology
`Research, Sunnybrook Health Sciences Centre, S-217, 2075 Bayview
`Avenue, Toronto, Ontario, Canada M4N 3M5. Phone: 416-480-5711;
`Fax: 416—480-5884. E-mail: robert.kerbel@sri.utoronto.ca
`Copyright © 2008 American Association for Cancer Research.
`doi:10.1158/1535-7163.MCT-08«0200
`
`zumab, the erbB-2/Her-2—targeting antibody. Both exper-
`imental and spontaneous metastasis models were studied;
`in the latter case, an increase in urine B-hCG always
`foreshadowed the development of lung, liver, brain, and
`kidney metastases. Metastatic disease was unrespon-
`sive to DC101 or trastuzumab monotherapy treatment, as
`assessed by B-hCG levels. Our results also suggest that
`B-hCG levels may be set as an end point for metastasis
`studies, circumventing guidelines, which have often
`hampered the use of advanced disease models. Collec-
`tively, our data indicates that B-hCG is an effective
`noninvasive preclinical marker for the long term monitoring
`of untreated or treated metastatic disease. [Mol Cancer
`Ther 2008;7(10):3452—9l
`
`Introduction
`
`The decision to evaluate a new anticancer drug or treatment
`in phase I and II clinical
`trials is usually preceded by
`preclinical therapy studies using tumor-bearing mice. The
`models used often involve transplanted ”primary” tumors
`grown so or in orthotopic organ sites (1, 2) but also
`spontaneously arising primary tumors arising in genetically
`engineered mouse models of cancer (3). In contrast, patients
`in phase I and II trials typically have advanced high-volume
`metastatic disease. This difference is likely one significant
`factor for the frequent failure to reproduce encouraging
`therapeutic results in preclinical models when the same
`therapies are evaluated in the clinical setting, as bulky
`metastatic disease is usually more difficult to treat than
`microscopic disease, in addition to the fact that the organ
`environment in which a metastatic lesion resides, e.g., the
`brain, can also limit therapeutic efficacy (4, 5). Given the
`dominance of (advanced) metastatic disease as the target for
`most cancer therapies, it is surprising to note the rarity of
`preclinical studies, which involve initiation of treatment at
`this stage of tumor progression (4). Instead, studies of
`metastatic cancer therapy often involve treatment of low-
`volume microscopic disease, e.g.,
`initiating treatment
`within a few days after iv injection of tumor cells (4). In
`part, this has been due to the inherent difficulties of detect—
`ing metastases, monitoring their response to therapies over
`time, and quantifying overall tumor burden. However, recent
`developments in practical noninvasive imaging meth—
`ods such as whole body optical (bioilluminescent) imaging
`of luciferase—tagged tumor cells (6), or analysis of (tumor
`derived) DNA in plasma (7), has stimulated increased
`interest in preclinical therapy studies of metastatic disease.
`A complementary non—invasive molecular method of
`detecting tumor burden involves the use of secreted
`
`Downloaded from mct.aacrjourna|s.org on December 6, 2017. © 2008 American Association for Cancer Hospira V. Genentech
`D """"""“
`|PR2017-00737
`
`Mol Cancer Ther 2008;7(10). October 2008
`Genentech 2079
`
`Genentech 2079
`Hospira v. Genentech
`IPR2017-00737
`
`

`

`Molecular Cancer Therapeutics 3453
`
`tumor-associated cell marker protein. The use of prostate-
`specific antigen levels in the blood of men with recurring
`prostate cancer and reductions of prostate-specific antigen
`levels after therapy are examples of this approach. In this
`regard, Shih et al. (8) developed a method that permits a
`marker to be used in a similar fashion for any transplanted
`tumor. The cDNA for the B-subunit of human chorionic
`gonadotropin (B-hCG) is transfected into tumor cell lines,
`which are then injected into recipient mice. Secreted B—hCG
`(9) is then measured in the urine as a surrogate marker of
`tumor growth, tumor burden, and response to therapy (8).
`Others and we successfully used this approach to detect
`relative primary or ascites tumor burden (8, 10, 11) but not
`for metastatic disease. Here, we report that monitoring of
`B—hCG is also a reliable and robust surrogate marker for
`monitoring the progression and response of metastatic
`disease, even over extended periods of observation.
`
`Materials and Methods
`Cell Lines
`
`the
`line MDA-MB-231,
`The human breast cancer cell
`derived erbB2 transfectant 231-H2N (12), the human cancer
`cell line MDA-MB-435, and murine 316 melanoma cells
`were grown in DMEM. Murine EMT-6 breast cancer cells
`were grown in Waymouth’s medium. All medium was
`supplemented with 10% FCS and 2 mmol/ L L-glutamine.
`Human prostate cancer cells PCB and B-hCG—transfected
`variants, and human MDA-MB—435hCG—transfected var-
`iants (10), as well as HT—29 human colorectal carcinoma
`and green fluorescent protein (GFP)-transfected variants
`(11) were cultured as previously described.
`Plasmid Transfection
`
`The full coding B-hCG sequence in the pCIneo vector, as
`described by Shih et a1. (8), was excised by XhoI and XbaI
`digestion and cloned into the corresponding region of the
`multiple cloning site of the pIRES plasmid (a gift from
`Steve Hobbs, Institute for Cancer Research, London, U.K.),
`thus generating the pIRES.hCG vector. Human 231-H2N
`cells, and murine EMT-6 breast cancer and B16 melanoma
`
`cells were transfected with the pIRES.hCG vector using
`Lipofectamine 2000 (Invitrogen) according to the manufac-
`turer’s instructions. After transfection, stable B-hCG—
`expressing variants (H2N.hCG, EMT-6.hCG, and
`Bl6.hCG, respectively) were obtained by puromycin
`selection. Human MDA-MB-231 cells were transfected
`
`with B-hCG. pCIneo (8) using Lipofectamine 2000; two high
`B-hCG—expressing clones were isolated, following from
`G418 selection, and subsequently combined to generate the
`MDA-MB-231.hCGneo variant line.
`B-hCG Measurements
`To assay B-hCG in tissue culture medium and in the
`mouse urine, we used the commercially available Free [3—
`hCG ELISA kit from OMEGA DIAGNOSTICS Ltd., which
`allows for quantitative determination of B—hCG. Results
`obtained were consistent with those obtained from inde-
`pendent assay kits from UBI. United Biotech, Inc., or from
`ALPHA DIAGNOSTIC INTERNATIONAL, as well as from
`
`Mol Cancer Ther 2008.7 (10). October 2008
`
`Sunnybrook Hospital Biochemistry services. Urine B-hCG
`levels were normalized by concomitant measurement of
`urine creatinine levels (using QuantiChrom TM Creatinine
`assay kit from BioAssay Systems) as detailed by Shih et al.
`(8). Urine was collected by placing mice in empty, aerated,
`tip boxes for 2 to 3 h. Typically, 0.1 to 0.4 mL of urine would
`be collected per mouse in this manner, of which (depend-
`ing on the tumor burden) 1 to 50 uL would be used for
`B—hCG detection and, 4 uL would be used for Creatinine
`measurement. In some experiments, before the mice were
`sacrificed, blood was collected by cardiac puncture into
`heparinized tubes and placed on ice. Plasma was then
`isolated by centrifugation and used to measure plasma B-
`hCG levels.
`
`Experimental Metastasis Assays
`Experimental metastases were generated by injection of
`cells into the lateral tail vein (13) of 6- to 8-wk-old female
`mice. B16F1.hCG cells were washed and resuspended in
`PBS to 2.5 x 106 cells/mL, and C57Bl/6 mice (obtained
`from Charles River Canada) were each injected with
`500,000 cells. Similarly, human cancer cell lines (one million
`cells per 200 uL) were injected iv. into female CB17 SCID or
`nude mice as indicated. Mice were sacrificed when any
`one of the following criteria were observed as follows:
`cachexia (defined by >15% weight loss), moribund state
`(lethargy and reduced mobility), or observable lymph node
`metastases.
`
`Spontaneous (Orthotopic) Metastasis Assays
`Human HZNhCG (Her-2—positive) cells were injected
`orthotopically (2 X 106 cells in a 50-uL volume, as fully
`detailed and described in du Manoir et al.; ref. 12) into the
`inguinal mammary fat pad of female C817 severe com—
`bined immunodeficient
`(SCID). Tumors were removed
`when they reached an average size of 500 m3, and the
`mice were thereafter monitored for B-hCG readings and
`body weight.
`Derivation of Metastatically Competent Sublines
`Upon autopsy, the mammary fat pad was inspected to
`confirm the absence of a primary tumor recurrence, and
`thereafter, any tissues found to have evidence of metastatic
`disease were rapidly excised and placed in serum-free
`medium at 4°C. Tissues were washed in ice cold PBS and
`
`briefly iced and mixed 1:1 with sterile 2X digestion mixture
`(Collagenase 3 at 4 mg/mL, Hyaluronidase at 2 mg/mL,
`and Collagenase IV at 2 mg/mL in PBS), and placed at
`37°C for 30 min with occasional swirling. The digested
`tissue was then washed twice in PBS, passed through
`autoclaved gauze to remove undigested debris, and plated
`in growth medium containing Pennicillin/Streptomycin
`and Fungizone.
`Treatment Regimens
`Trastuzumab (Herceptin) was provided by Genentech
`and administered twice weekly at 20 mg/kg i.p., as
`described (12). Metronomic administration (20 mg/kg/d)
`of cyclophosphamide was carried out by addition of the
`drug to the drinking water as originally described (10), and
`later modified (14) by the addition of an upfront i.p. bolus
`dose (150 mg/ kg). DC101 is an antiangiogenic endothelial
`
`
`F)
`Downloaded from metaacrjoumalsorg on December
`
`AVAL‘6, 2017. © 2008 American Association for Cancer
`
`

`

`3454 hCG to Monitor Metastases
`
`growth factor receptor-2 (VEGFR-Z) blocking antibody
`obtained from Imclone, and administered i.p. at 800 pg
`per mouse twice weekly as previously detailed (10).
`Western Blotting
`Cells were washed in PBS and lysed in ice cold Lysis
`Buffer [10 mmol/L Tris (pH 7.5), 137 mmol/L NaCL,
`100 mmol/L NaF, 10% glycerol, 1% NP40, 2 mmol/ L
`Na3VO4,
`1 mmol/ L phenyl methylsulfonylfluoride,
`10 ug/mL aprotinin, and 10 ug/mL leupeptin]. Lysates
`were placed on ice for 10 min and then cleared by
`centrifugation. Protein concentration was measured using
`the Bio-Rad Protein Assay Dye Reagent (Bio-Rad Labora-
`tories). Samples were run on 8% SDS-PAGE (30 pg per lane)
`and then transferred onto Immobilon-P membrane (Milli-
`pore; Canada Ltd.). Membranes were blocked in (excess)
`10% milk in TBST for 1 h at room temperature and then
`probed with primary antibody (at 1:1,000 dilution) over-
`night at 4°C. Membranes were then washed in TBST and
`probed with 1:5,000 dilution of anti-rabbit IgG—horseradish
`peroxidase or anti-mouse IgG—horseradish peroxidase
`(Promega). Detection was carried out using enhanced
`chemiluminescence reagent
`(Amersham Biosciences).
`HERZ/ErbBZ antibody was from Cell Signaling Tech—
`nologies, and anti—B—actin was obtained from Sigma
`Aldrich Canada.
`
`Results and Discussion
`
`B-hCG Monitoring of B16 Melanoma Experimental
`Metastases in Mice Undergoing Metronomic Adminis-
`tration of Cyclophosphamide and Antiangiogenic Drug
`Treatment Regimens
`Previously, we have reported the antitumor effects of
`metronomic cyclophosphamide combined with the anti-
`VEGFRZ antibody DC101 in several experimental primary
`tumor models (10). To test if such a regimen was also
`effective against “artificial” (experimental) Bl6 melanoma
`metastases, Bl6.hCG cells (Table 1) were injected i.v. into
`
`Table 1. Derived hCG expressing cell lines
`
`syngeneic mice. On day 9 after injection, after the first
`confirmed positive detection of B-hCG (indicating estab-
`lished metastases), the mice were randomized into four
`groups. These were treated with control PBS i.p., or
`administered a bolus plus oral
`low-dose metronomic
`cyclophosphamide regimen (as described in ref. 14), or
`given DC101 i.p.
`(15). The fourth group received the
`combination of bolus dose plus metronomic cyclophospha-
`mide together with DC101 administration. As shown in
`Fig. 1A, DC101 on its own had no appreciable effect on
`urine B-hCG levels compared with controls. Surprisingly,
`the bolus plus low-dose cyclophosphamide alone group
`and the combination group showed a (delayed) super
`imposable B-hCG curve throughout the first 12 days of
`therapy (up to day 21; Fig. 1B), and only thereafter did the
`combination group show a slower increase in urine B-hCG
`levels (Fig. 1A). These results suggest that the effect of
`DC101 in the setting of established metastatic disease may
`be more important in blunting tumor repopulation as the
`therapy starts to fail, rather than contributing to an initial
`antitumor cytotoxic effect of this treatment. Figure 1C
`shows the corresponding survival curve (generated as mice
`had to be sacrificed due to cachexia), showing increased
`survival in the combination group compared with bolus
`dose plus metronomic cyclophosphamide, which is con—
`cordant with the observed BrhCG data. Figure 1D shows
`values for individual mouse urine sample analysis (on days
`8 and 14) that are consistent with values for the pooled
`urine (shown in Fig. 1A and B).
`B-hCG Analysis of Experimental Metastasis of
`Human Tumor Cells
`
`We next attempted to monitor experimental metastasis of
`B-hCG—tagged human tumor cells. To evaluate the kinetics
`of urine B-hCG readouts in experimental metastasis assays
`of human tumor cells, MDA435.hCG cells were generated
`and injected i.v.
`into female nude mice. Monitoring
`continued for 150 days thereafter, which revealed negative
`[3-th readings for the first 40 days after injection followed
`
`Parent line
`
`Derived line
`
`Notes
`
`MDA231 (human breast)
`
`MDA231.hCGneo
`
`231H2N (Her2+ve)
`
`HZN. hCG
`
`HT29 (human colon)
`
`MDA-MB—435
`
`PC3 (human prostate)
`EMT-6 (mouse breast)
`
`231H2N. hCG
`
`H2N.hCG.met1
`
`HT29.hCG (10, 11) HT—GFPhCG
`
`MDA435.hCGneo
`PC3.hCGneo (10, 13) PC3.hCGneo.VEGF
`EMT-6. hCG
`
`Bl6F1 (mouse melanoma)
`
`Bl6F1. hCG
`
`NOTE: Numbers in parentheses refer to the relevant references.
`
`Lung and kidney metastasis observed
`after iv. injection
`Spontaneous metastasis to liver,
`lung, and bone
`Spontaneous metastasis to brain, lung,
`bone and kidney
`Metastatic after iv. injection in SCID mice
`but poorly metastatic in nude mice
`
`Highly metastatic after iv. injection
`but poorly metastatic from orthotopic
`primary tumors
`Refer to above EMT—6.hCG notes
`
`Mol Cancer Ther 2008;7 (10). October 2008
`
`F) AAAAAA L.
`Downloaded from mct.aacrjournals.org on December 6, 2017. © 2008 American Association for Cancer
`
`

`

`2000
`
`1500
`
`.L OOO
`
`500
`
`
`
`hCG(mIU/mgCreatinine)
`
`0
`
`.54.;053%000
`
`
`
`hCG(mIU/mgCreatinine) 8
`
`
`
`
`Figure 1. Growth behavior of A
`B16.hCG metastases under metro-
`nomic-based regimens. BlBhCG
`cells were generated by transfection
`of leES»thol/Xbal)B~hCG followed
`by puromycin selection, and injected
`i.v.
`(5 x 105cells in 200 uL)
`into
`female C57BL/6 mice. Treatment
`began on day 9 (arrow), after B—
`hCG detection in the mouse urine.
`Treatments involved bolus plus low-
`dose metronomic cyclophosphamide
`(BD + LDM CTX; bolus was 150 mg/
`kg on days 9 and 30;
`low-dose
`metronomic cyclophosphamide was
`20 mg/kg/d in the drinking water)
`alone or combined with DC101 (anti- B
`VEGFR2 Ab; 800 pg per mouse
`twice per week, i.p.). Controls were
`given PBS (i.p.). B-hCG levels (in
`mIU/mg of Creatinine) in the urine
`was used to monitor growth of
`metastases—data are shown from
`pooled urine from each group shown
`in (A, and with the y axis magnified
`in B—this was done to highlight the
`overlap of the curves in the first
`20 d). C, survival in this experiment.
`D,
`individual mouse urine B-hCG
`levels as measured on days 8 and
`14. Using a log-rank test analysis for
`the survival data shown in C, bolus Di 3 100
`plus low-dose metronomic cyclo-
`,5
`phosphamide showed a significant
`5 50
`increase in survival compared with
`§
`controls (P = 0.0047), as did bolus
`('3
`plus low-dose metronomic cyclo-
`a)
`phosphamide plus DC101 (P =
`E 40
`0.002). DC101 alone did not signif—
`2
`icantly affect survival.
`5. 2n
`0.i:
`
`Molecular Cancer Therapeutics 3455
`
`-- BD+LDM CTX+DC101 (n 4)
`_._ BD+LDM CTX (m4)
`Control (n: =5)
`—'- DC101 (11-5)
`
`9 - indicates
`start of
`treatment
`
`1.
`
`
`
`Percentsurvival
`
`0
`
`+10
`
`20
`
`30
`days
`
`40
`
`so
`
`100
`
`85
`
`r——-
`
`L
`
`BD+LDM crx
`
`l
`
`Creatinine)
`
`6
`
`a
`
`10
`
`12
`
`14
`
`16
`
`days
`
`50
`
`.A OO
`
`-- —- —-
`
`
`
`hCGlmlU/mgCreatinine) 3E
`
`O
`
`
`
`.<..
`
`
`
`hCG(mlU/mgCreatinine)
`
`hCG(mlU/mg
`
`by a gradual appearance of detectable B-hCG (between
`40 and 100 days after injection; Fig. 2A). Positive B—hCG
`readings were eventually observed in 6 of 18 mice injected.
`Most importantly, over the course of 160 days of this
`experiment, no mouse with a negative urine B-hCG reading
`ever developed metastatic disease. All
`the mice that
`developed detectable B-hCG levels had to be sacrificed
`breast cancer cells transfected with B-hCG (for which initial
`eventually (due to cachexia), and all these mice had visible
`B-hCG detection took some 50 days; Fig. 2A), or with
`HT20.GFP.hCG colon cancer cells (for which 1 month was
`metastatic disease at autopsy. Furthermore, in all cases
`where B-hCG became detectable, levels could be monitored
`necessary for B-hCG detection in SCID mice but >100 days
`
`
`for 4 to 8 weeks before any other or additional indications
`that the mouse had metastatic disease. Therefore, urine
`B-hCG measurement permitted early, ”asymptomatic”
`detection of human tumor experimental metastases
`(Fig. 2A), growth of which could thereafter be monitored.
`Similar observations were made with human MDA-MB-231
`
`Mol Cancer Ther 2008;7 (10). October 2008
`
`D nnnnnn l.
`Downloaded from mct.aacrjouma|s.org on December 6, 2017. © 2008 American Association for Cancer
`
`

`

`3456 hCG to Monitor Metastases
`
`A
`
`MDA-MB-43S.hCG
`
`MDA-MB-435JlCG
`
`y
`/
`I
`
`I
`
`'
`
`_
`
`ii:
`
`I
`
`i
`guru-"I
`so
`140
`days
`
`+9“,
`+m§
`1m:
`+4.9
`*“7
`-I— m9
`A r1112
`
`32.5
`E 275
`g
`‘
`3 2“
`a 1::
`-
`5
`1
`.
`5
`75
`8
`‘
`25-
`‘
`
`1'
`
`'
`-
`-'
`'
`I i
`
`.
`-
`i
`.
`I-
`I“
`fleas-o
`
`41::
`so
`140
`days
`
`Ml)A-.‘\lB~231.h(‘(;
`
`MDA-MB-435.hCG - mouse weights
`
`§ 307.6
`z 2575-
`g 2075
`{5‘ 157.5
`- 107.5-
`5
`E
`8 67.6-
`‘
`75
`
`32.5
`
`g 27.5
`‘5 22.5'
`5
`g 17.6
`g 12.5
`g 7.5
`2 2.5
`
`R
`2» 3‘
`5
`E
`5-;
`E
`g 2‘
`
`_.- M31
`4-6— M82
`
`.
`
`'I
`
`”*l—-"—-'--1-—l_'
`15
`4o
`65
`90
`115
`days
`
`B
`
`HT-GFP-hCG iv in SCID mice
`
`ii
`
`|
`A 250
`g
`E
`2“?
`'6 150
`
`
`
`E
`i 100
`E 50
`o
`u
`.E
`0
`
`iii
`
`29
`
`38
`
`41
`
`4a
`
`days
`
`Plasma levels of hCG in mice injected
`
`+qu
`4, M
`+ml
`+"""
`+m7
`
`
`
`WV"""~‘"‘I""_"“"T""
`40
`90
`140
`days
`
`HT-GFP-hCG i.v. in nude mice
`
`-HTZ
`m HT4
`
`-
`
`124
`
`-
`
`.-
`.-
`140
`
`129
`days
`
`A
`g
`g
`§
`(3
`
`S
`i
`E
`o
`o
`.C
`
`
`i.v. with HT-GFP—IICG at time of death
`
`Figure 2. A, examples of B-hCG readouts in the urine of individual mice to monitor experimental metastasis of human MDA-MB-231.hCGneo (data for
`M81 and MBZ) or MDA-MBv435.hCGneo (all other data points shown) in female nude mice. Cells were injected i.v. (1 X 106 cells in 200 (1L), and urine was
`collected and assayed for B-hCG every 1 to 2 wk. Top left, data for mice (n = 18; m1—m18) injected with MDA—MB-435.hCGneo cells, of which 6 mice
`eventually developed metastasis [the remaining mice did not develop metastasis or detectable B-hCG levels, and are collectively presented as a negative
`population (pop) in the figure] as initially indicated by detectable B-hCG levels in the mouse urine; the same data are shown on the top right with a
`magnified Y-axis. Bottom right, examples of mouse weights for the corresponding mice injected with MDA-MB-435.hCGneo that developed metastasis; B-
`hCG detection provided early indication of the establishment of growing metastases whose relative growth could thereafter be monitored for 4 wk or longer
`in the absence of other indicators of metastatic disease such as weight loss or cachexia. Bottom left, examples of nude mice injected with MDA-MB-
`231 .hCGneo. B, examples of B-hCG readouts from (i) SCID mice (H82, H83, and H84) or (ii) nude mice (HT2 and HT4) injected i.v. with HT-GFP-hCG cells
`(1 X 106cells in 200 pL)—note that for nude mice, >100 d of monitoring were necessary before B-hCG levels became detectable in the mouse urine. The
`results were confirmed by (iii) B-hCG measurement in plasma obtained at the time the mice were sacrificed and by autopsy; iv, examples of visible lung
`metastases of HT-GFP-hCG cells.
`
`D nnnnnn L‘
`Downloaded from mctaacrjournals‘org on December 6, 2017. © 2008 American Association for Cancer
`
`Mol Cancer Ther 2008;7 (1 0). October 2008
`
`

`

`Molecular Cancer Therapeutics 3457
`
`in nude mice; Fig. 2B). In HT20.GFP.hCG experiments,
`plasma was collected at the time the experiments were
`terminated, and used as a source of independent measure-
`ment of B-hCG levels (Fig. 23).
`
`Analysis of a Spontaneously Metastatic Variant of
`MDA-MB—231 Engineered to Overexpress erbB-Z/Her-Z
`We next used a ErbB2/HER—2—transduced MDA-MB-231
`variant, termed 231-H2N. The 231—H2N line, which we
`
`A
`
`2x10‘5 H2N.hCG cells in mfp
`
`—>2x106 H2N.hCGmet1 cells in mfp
`
`
`
`,“o
`_
`"‘ d
`
`i
`
`6 weeks
`‘ -.‘
`Prima
`tumor (500mm3)
`tidy'filsurgical resection
`
`4 months 1
`
`.‘L a 1
`ZsJ/d
`1 Liver and Lung
`metastases excrsed
`and adapted to
`[Issue “11mm
`
`H2N.hCG.met1 cells
`
`.‘l'
`.l
`' ‘u - «I
`
`1
`
`6 weeks
`g - .‘
`._...‘
`
`1-4 months i
`
`Prim
`
`tumor (500mm3)
`surgical resection
`
`,1 0.-
`L.../’
`+/-Trastuzumab ip
`1 5
`th )
`( ' mon S
`Urine hCG analysis
`
`fl
`0
`+ Saline
`
`1
`
`1|
`nub
`
`2|
`
`.
`
`In
`
`
`
`hCGlmlU/muCreatininol
`
`
`
`
`
`M20[WinsCranium.)
`
`Figure 3. A, schematic of the deriva-
`tion of spontaneously metastatic variants
`of MDA-MB-231 engineered to overex-
`press erbBZ. H2N.hCG cells were gener—
`ated by transfection of
`leES-(Xhol/
`XbaIm-hCG followed by puromycin selec-
`tion, and injected orthotopically into the
`inguinal mammary fat pad of female SCID
`mice. Primary tumors were removed
`when they reached an average size of
`500 mma. Metastases eventually resulting
`from this experiment were adapted to
`grow in tissue culture and combined to
`form the H2N.hCG.met1 population.
`H2N.hCG.met1 cells were then injected
`into the inguinal mammary fat pad of
`female SCID mice and, after removal of
`the primary tumor, some mice were used
`to test the effect of trastuzumab mono-
`therapy on the growth of metastases. B,
`B-hCG readouts in female SCID mice with
`H2N.hCG.met1 metastases treated with
`saline (top) or trastuzumab (bottom, given
`twice weekly at 20 mg/kg i.p.) treatment
`initiated 2 wk after B-hCG was first
`detectecd in the urine.
`i, B-hCG values in
`the scale 0 to 250 mlU/mg Creatinine, and
`no difference in the B—hCG curves between
`the two treatment groups;
`ii, B-hCG
`values in the full range detected in this
`experiment (0 — 4,000 mlU/mg Creatinine);
`i/‘i, same data on a log plot of B-hCG
`values. C, examples of H2N.hCG.met1
`metastases to the (i)
`lymph nodes,
`(ii)
`brain, and (iii)
`lungs.
`iv, Western blot
`analysis for Her—2 in cells adapted to tissue
`culture from H2N.hCGmet1 metastases
`(met) to different tissues in mice under
`no treatment, or saline or trastuzumab
`monotherapy as indicated. H2N cells (used
`as a positive control) were included to-
`gether MDA-MB-231 (parental cells as
`negative control) and the H2N.hCG trans-
`fectants. B-actin was used as a loading
`control.
`
`Mol Cancer Ther 2008;7 (10). October 2008
`
`3 W
`g woo
`
`III E
`IE
`
`g
`5
`g
`‘
`
`..
`i
`3
`
`..
`H
`7
`4
`l
`* Saline M
`
`‘
`
`21
`
`ll
`
`EE
`
`s
`
`g
`g
`g,
`2
`
`D
`
`4
`'
`
`7
`
`34
`
`2r
`
`2!
`
`
`u + Trastuzumab
`
`
`___ B-actin
`’4— Trastuzumab —>!4_Saline _, I
`treated
`treated
`
`
`
`6, 2017. © 2008 American Association for Cancer
`AAAAAA l"
`D
`Downloaded from metaacrjournalsorg on December
`
`

`

`3458 hCG to Monitor Metastases
`
`previously described (12), was chosen for analysis for a
`number of reasons. Previously, we used this line to show
`the efficacy of combining the anti—Her—2 antibody trastu-
`zumab (Herceptin) with metronomic low-dose cyclophos-
`phamide to suppress the growth of orthotopically
`implanted 231-H2N primary tumors (12). The derivation
`of a metastatic variant of 231-H2N would therefore allow
`
`us to test if the same therapy is effective when treating
`metastatic disease. In addition, a metastatic model respon-
`sive to trastuzumab whose growth kinetics could be
`monitored (i.e., via B-hCG readings) would allow us to
`study conditions where resistance to trastuzumab-based
`regimens eventually develops. This, in turn, would permit
`proactive initiation of second-line therapies (12) when the
`first—line therapy begins to fail.
`H2N.hCG cells were implanted in the mammary fat pad
`of female SCID mice. Primary tumors were then removed
`when they reached an average size of 500 m3, and urine
`B-hCG dropped to undetectable levels. It took between
`1 and 4 months after surgery for mice to develop
`macroscopic metastatic disease (Fig. 3A), and this occurred
`with 3 of 5 mice over a monitoring period of 8 months.
`Upon autopsy, metastases were found in the lungs and
`liver, from which lines were generated and adapted to
`tissue culture. The lines were then mixed (to maximize the
`heterogeneity of these metastatic variants), thus generating
`the H2N.hCG.met1 population (Fig. 3A), and injected
`orthotopically into 10 female SCID mice. Once again, pri-
`mary tumors were removed when they reached 500 mm3
`(see Fig. 3A). In this second round of in viva selection,- two
`mice were found to have positive urine [s-hCG readings
`immediately after surgical removal of the primary tumors
`(presumably because they already had established metas-
`tases). For the other 8 mice, it took from 1 to 4 more months
`for B—hCG to become detectable in urine, and eventually all
`mice were found to have detectable B-hCG levels. As each
`of these 8 mice became B-hCG positive, it was placed into
`one of two groups. As shown in Fig. 3, 2 weeks after B—hCG
`detection, each mouse was randomized into a control-
`treated (PBS i.p.) group or a trastuzumab treatment group.
`Under these conditions, the growth of metastases (as solely
`determined by B-hCG readings) was found to be similar
`between trastuzumab treatment and control
`(Fig. 3B),
`suggesting trastuzumab monotherapy is ineffective in this
`setting of advanced metastatic disease. This stands in
`contrast to what we reported with respect to treatment of
`primary tumors using the parent (231-H2N) tumor line, i.e.,
`an initial response to trastuzumab monotherapy lasting ~ 3
`to 5 weeks followed by the emergence of resistance (12). We
`did however note that at the time the mice had to be
`
`sacrificed (due to cachexia), B—hCG levels were orders of
`magnitude higher in the trastuzumab-treated group than
`for controls (Fig. SB). This observation suggests that
`trastuzumab monotherapy allowed the mice to tolerate a
`much higher tumor cell burden, the significance of which
`remains to be determined. Furthermore, some sublines
`from metastases in the trastuzumab treatment group
`showed reduced expression of erbBZ (Fig. 3C), as had
`
`previously been noted for some 231—H2N tumors treated
`with trastuzumab (12). As detailed above, 2 of the 10 mice
`were found to have high levels of B—hCG immediately after
`surgery. These were excluded from the therapy experiment
`and had to be sacrificed (due to cachexia) 3 weeks after
`surgery;
`individual
`tissues were minced and placed in
`tissue culture. One week later, analysis of the tissue culture
`medium for B-hCG confirmed the presence of metastasis to
`the lungs, bone marrow, and brain. These results show that
`a spontaneous metastasis model was generated, capable of
`being monitored, which will be of considerable use for
`combination therapy experiments.
`In summary, we have described examples of therapeutic
`strategies applied to different metastasis models using
`urine B-hCG levels to monitor progression of disease and
`response to therapy. In all the models tested, B—hCG proved
`to be a stable indicator of disease burden. Because of this
`
`stability, another application of the B—hCG monitoring
`system could be the study of emerging drug resistance to
`established therapeutic regimens. For example, using the
`H2N tumor model, we recently showed that primary
`tumors could be induced to regress using a combination
`of trastuzumab plus metronomic cyclophosphamide. This
`effect could be followed for up to 2 months, after which the
`tumors started to regrow while still under therapy. When
`such resistance emerged, tumor growth could be inhibited
`by the addition of anti—vascular endothelial growth factor
`antibody (12). To test if such an approach is also effective
`against metastatic disease, it would be ideal to use the B-
`hCG system as a means to detect
`the emergence of
`resistance (i.e., as a sudden upsurge in urine B-hCG levels)
`to the first-line therapy—at which point, second—line
`therapies could be proactively initiated.
`One additional feature of the B—hCG methodology should
`be stressed. After initial experiments, we noted that for each
`tumor model, there was an upper end B—hCG value, which
`would routinely predict that the mice would need to be
`sacrificed within a week, even in the absence of any
`independent indicator of metastatic disease (e.g., cachexia).
`Such empirically determined B-hCG levels could therefore
`be set as an end point for survival studies where
`institutional, ethical, or other guidelines dictate that
`symptoms associated with advanced metastatic such as
`cachexia are not permissible or unacceptable. The availabil-
`ity of the lines described in this study should also facilitate
`the experimental analysis of different aspects of the biology
`and therapy of metastatic disease, including the testing of
`new strategies against drug-resistant disseminated disease.
`
`Disclosure of Potential Conflicts of Interest
`Z. Zhu and L. Witte: employees of lmClone Systems, Inc.; R.S. Kerbel:
`consultant for lmClone Systems, Inc. and Genentech/Roche. The authors
`hereby declare that there are no actual, potential, or apparent conflict of
`interest with regard to publication of this manuscript.
`
`Acknowledgments
`We thank Cassandra Chen for her excellent secretarial assistance, Bert
`Vogelstein for the PC|~hCG plasmid, Steve Hobbs for the leES plasmid,
`and Ian Hart for invaluable suggestions.
`
`Mol Cancer Ther 2008;7 (1 0). October 2008
`
`Downloaded from mct.aacrjoumals.org on December 6, 2017. © 2008 American Association for Cancer
`
`

`

`Molecular Cancer Therapeutics
`
`3459
`
`References
`
`1. Fidler IJ. Orthotopic implantation of human colon carcinomas into nude
`mice provides a valuable model for the biology and therapy of metastasis.
`Cancer Metastasis Rev 1991 ;10:229—43.
`2. Hoffman RM. Orthotopic metastatic (MetaMouse) models for discovery
`and development of novel chemotherapy. Methods Mol Med 2005;111:
`297 — 322.
`
`3. Becher OJ, Holland EC. Genetically engineered models have advan-
`tages over xenografts for preclinical studies. Cancer Res 2006;66:
`3355—8, discussion.
`4. Kerbel RS. Human tumor xenografts as predictive preclinical models for
`anticancer drug activity in humans: better than co