Use of an oncolytic virus secreting GM-CSF as combined oncolytic and immunotherapy for treatment of colorectal and hepatic adenocarcinomas

Use of an oncolytic virus secretingGM-CSF as combined oncolytic andimmunotherapy for treatment ofcolorectal and hepatic adenocarcinomasSandeep Malhotra, MD, Teresa Kim, MD, Jonathan Zager, MD, Joseph Bennett, MD,Michael Ebright, MD, Michael D’Angelica, MD, and Yuman Fong, MD, New York, NY

Background. Oncolytic cancer therapy using herpes simplex viruses (HSV) that have direct tumoricidaleffects and cancer immunotherapy using the cytokine granulocyte-macrophage colony-stimulating factor(GM-CSF) have each been effective in preclinical testing. NV1034 is a multimutated oncolytic HSVcarrying the gene for murine GM-CSF that attempts to combine these 2 anticancer strategies. Thepurpose of this study was to compare NV1034 to NV1023, the parent HSV mutants lacking GM-CSF,to determine if such combined oncolytic and immunotherapy using a single vector has advantages overoncolytic therapy alone.Methods. Expression GM-CSF in vitro did not alter the infectivity, cytotoxicity, or replication ofNV1034 compared to the noncytokine-secreting control. Tumors infected with NV1034 producedGM-CSF in picogram quantities. In vivo efficacy of the viruses against murine colorectal carcinomaCT26 and murine hepatoma Hepa l-6 was then tested in subcutaneous tumors in syngeneic Balb/cand C57 L/J mice, respectively. In these immune-competent models, NV1034 and NV1023 eachdemonstrated potent antitumor activity.Results. Treatment with NV1034 had significantly better antitumor effect compared to treatment withNV1023. Furthermore, there was no difference in the antitumor efficacy of these viruses in micedepleted of CD4� and CD8� T lymphocytes.Conclusions. Viral vectors combining oncolytic and immunotherapy are promising agents in treatmentof colorectal carcinoma and hepatoma. (Surgery 2007;141:520-9.)

From the Department of Surgery, Memorial Sloan-Kettering Cancer Center, New York, NY

Replication-competent herpes simplex viruses(HSV) have been shown to have efficacy in models ofneurologic1,2 and nonneurologic malignancies in-cluding those of the breast, prostate, stomach, co-lon, lung, and head and neck.3-6 The herpes virusesthat are being tested for oncolytic therapy havebeen engineered to attenuate their neurovirulence

Supported in part by grants RO1 CA 75416, RO1 CA 72632, andRO1 RO1CA/DK80982 (Y.F.) from the National Institutes ofHealth; Grant MBC-99366 (Y.F.) from the American CancerSociety; a grant from the Lustgarten Foundation, and the ByrneFoundation.

Accepted for publication October 28, 2006.

Reprint requests: Yuman Fong, MD, Department of Surgery,Hepatobiliary Division, Memorial Sloan-Kettering Cancer Cen-ter, 1275 York Avenue, New York, New York 10021. E-mail:[email protected].

0039-6060/$ - see front matter

© 2007 Mosby, Inc. All rights reserved.

doi:10.1016/j.surg.2006.10.010

520 SURGERY

while maintaining their natural ability to infect andlyse tumor cells.

Another therapeutic strategy against tumorshas been to enhance host antitumor immunity.Investigators have attempted to produce tumor vac-cines by altering tumor cells ex vivo to secreteimmunostimulatory cytokines and reinfusing suchcells to elicit a host antitumor immune response.7,8

The development of efficient viral vectors for genetransfer9,10 has improved upon this strategy by allow-ing in vivo transduction of tumor cells with genescoding for immunostimulatory molecules. Intratu-moral injection of vectors carrying the genes forvarious cytokines can result in efficient gene trans-fer and in sufficient cytokine production in prox-imity to tumor to result in antitumor efficacy.11,12

Investigators have attempted to combine onco-lytic and immunotherapy to boost the antitumoreffect of oncolytic therapy. It has been demon-strated that the antitumor actions of an intratu-
moral injection using a cytotoxic virus was greatly

nder

Surgery Malhotra et al 521Volume 141, Number 4

enhanced by the addition of a replication-defectiveHSV vector encoding interleukin-12, and the com-bination resulted in local and systemic antitumorimmunity.13 Recently, replication-competent onco-lytic HSV viruses carrying cytokine genes have dem-onstrated benefit when injected directly intoexperimental squamous cell cancers of the headand neck.14 The goal of the current study was todetermine if a replication-competent HSV can actboth as an oncolytic agent and as a vehicle forgranulocyte-macrophage colony-stimulating factor(GM-CSF) gene transfer to produce antitumor ef-fects against colorectal and hepatocellular tumorsin a murine model.

GM-CSF is a cytokine that induces myeloid precur-sor cells to proliferate and differentiate into neutro-phils, monocytes, macrophages, and eosinophils.GM-CSF also recruits and stimulates dendritic cells.15

In experiments studying tumor vaccines, GM-CSFwas the most potent of a number of cytokines,adhesion molecules, and other immunostimulatorymolecules for the induction of specific and long-lasting antitumor immunity.16,17 Vaccination withirradiated B16 melanoma cells engineered to se-crete murine GM-CSF stimulated potent, specific,and long-lasting antitumor immunity.7 Murine tu-mor cells infected ex vivo or in vivo with disabled

Fig 1. Structure of f-strain of wild-type HSV-1, NV10shown here schematically (not to scale) in the prototsequences (UL and US), each bound by inverted regenes. The locations of the following genes are sho�34.5 (the major HSV-1 neurovirulence gene),junction of NV1023 contains 5.2 kb of HSV-2 DNcontrol of the �47 promoter between nucleotidesaccession no. X14112). NV1034 has a similar strumacrophage colony-stimulating factor (GM-CSF) u

infectious single-cycle herpes viral vectors ex-

pressing GM-CSF might function effectively inpreventing cancer formation and treating estab-lished tumors.18,19

MATERIAL AND METHODSCells and cell culture. The murine colorectal

carcinoma cell line CT26, murine hepatoma cellline Hepa 1-6, and the murine melanoma cell lineB16 were used in this study. All cells lines wereobtained from American Type Culture Collection(Rockville, Md). CT26 cells were maintained inRoswell Park Memorial Institute (RPMI) mediumsupplemented with 10% fetal calf serum (FCS),100 �g/ml penicillin, 100 �g/ml streptomycin,and 5 mM nonessential amino acids. Hepa 1-6 andB16 cells were maintained in Dulbecco modifiedEagle high glucose medium (DMEM) supple-mented with 10% FCS, 100 �g/ml penicillin, and100 �g/ml streptomycin.

Viruses. Structures of the 2 viruses used in thisstudy (NV1023 and NV1034) are shown in Fig 1.NV1023 is derived from R7020, which is a previ-ously described attenuated replication-competentvirus based on the HSV-1 strain-F.20 R7020 has 1copy of the �134.5 neurovirulence gene deleted.Also deleted are the UL24, the UL56 genes, andthe L/S junction.20 The viral thymidine kinase

d NV1034. The wild-type HSV-1 genomic structure,entation, consists of long (L) and short (S) uniqueegions (RL and RS). HSV-1 encodes more than 80reference: thymidine kinase (TK), UL24, UL5/6,e immediate-early genes �4 and �47. The L/S

tched box) and carries the E. coli lacZ gene under8 to 146007 of HSV-1 at the �47 locus (GenBankto NV1023 and carries the murine granulocyte-

control of a hybrid �4-TK promoter.

23, anype oripeat rwn forand thA (ha14467cture

gene has been moved and placed under the �4

522 Malhotra et al SurgeryApril 2007

promoter. NV1023 differs from R7020 in thatthe deleted endogenous thymidine kinase (TK)gene and the adjoining UL24 promoter regionhave been repaired. In addition, the E. coli lacZmarker gene has been inserted into the ICP47locus. NV1034 is a derivative of NV1023 in which amurine GM-CSF cDNA has been inserted. Thesynthesis of these viruses has been described previ-ously.14 NV1023 and NV1034 were propagated onVero cells and titered by standard plaque assay.

In vitro infectivity and cytotoxicity. The ability ofNV1034 to infect CT26 or Hepa 1-6 cells was as-sessed. Cells were plated 5 � 104 cells/well in 12-well plates (Costar, Corning Inc, Corning, NY) andwere infected with NV1034 at multiplicities of in-fection (MOIs) ranging from .01 to 5.0. Controlwells were treated with phosphate-buffered saline(PBS). To assess infectivity, cells were fixed in 1%glutaraldehyde at 24-hour intervals for 6 days andstained with X-Gal (Gibco-BRL, Grand Island, NY)to detect �-galactoside, which is the product of thelacZ marker gene. The number of stained (blue)cells/high-power field (HPF) and the total numberof cells/HPF were counted, and the percentage ofinfected cells was calculated. Cell viability was as-sessed by counting live cells at 24-hour intervalspostinfection using trypan blue exclusion.

In vitro viral proliferation. Viral growth assayswere performed to compare the ability of CT26 andHepa 1-6 cells to support the replication of NV1034or NV1023. Cells were infected with NV1034 orNV1023 at an MOI of 1.0. Cells and supernatantswere harvested at 6 hours and daily postinfectionfor 6 days. Three cycles of freeze-thaw lysis wereperformed and viral titers were determined by stan-dard plaque assay on Vero cells.

In vitro GM-CSF production. The ability of theNV1034 to produce GM-CSF when it infects CT26and Hepa 1-6 cells was assessed. Cells were theninfected with NV1034 or NV1023 at MOIs of 0.1,1.0, or 5.0. Supernatants were collected daily, andconcentration of GM-CSF was determined by en-zyme-linked immunosorbent assay (ELISA; R&DSystems, Minneapolis, Minn).

Animal studies. Four- to 6-week–old male Balb/cmice (Taconic, Germantown, NY) or C57 L/J mice(Jackson Laboratories, Ann Arbor, Me) were used.All animal studies were approved by the MemorialSloan-Kettering Institutional Animal Care and UseCommittee and performed under strict guidelines.Animals were anesthetized with an intraperitoneal(ip) injection of ketamine (70 mg/kg) and xylazine(10 mg/kg) for experimental procedures.

Establishment of subcutaneous tumors. First,
1 � 105 CT26 cells or 5 � 105 Hepa 1-6 cells were
injected subcutaneously into the flanks of Balb/cmice or C57L/J mice, respectively. Palpable tumors3- to 5-mm in diameter developed reliably in 14days after tumor cell inoculation for CT26 and in 7days for Hepa 1-6.

In vivo viral infection and GM-CSF production.This experiment was conducted to assess in vivoviral infection and GM-CSF production by NV1034.Subcutaneous CT26 tumors were induced in theflanks of Balb/c mice as described above. Tumorswere injected with 5 � 107 plaque-forming units(PFU) of NV1034 or NV1023 and were harvestedon days 1, 2, 3, and 7 postinjection. Half the tumorwas immediately frozen in liquid nitrogen and laterhomogenized in PBS for assessment for GM-CSFexpression by ELISA (R&D Systems). The otherhalf of the tumor was embedded in Tissue-Tek(Electron Microscope Services, Hatfield, Pa) priorto cutting into 8 �m frozen sections for subsequenthistochemical staining with Xgal (GibcoBRL) todetect marker gene �-galactosidase (lacZ) expres-sion as an indicator of viral infection.

Treatment of subcutaneous flank tumors withvirus. The efficacy of NV1034 to treat CT26 tumorsin immune-competent mice was compared to thatof NV1023. Palpable CT26 flank tumors (diameter,3 to 5 mm) were treated with intratumoral injec-tion of 1 � 106 PFU (low dose), 1 � 107 PFU(medium dose), or 5 � 107 PFU (high dose) ofNV1034, NV1023, or PBS (n � 6 to 7/group).These experiments were then repeated to ensureconsistency. Hepa 1-6 tumors were treated with 1 �105 PFU, 1 � 106 PFU, or 5 � 106 PFU of theseviruses in similar experiments. The longest andshortest diameters of the tumors were measuredover 14 days after treatment, and the tumor volumein each case was calculated using the followingformula:

Tumor volume � 4 ⁄ 3(Pi)(DL1 ⁄ 2)(DS1 ⁄ 2)(DS1 ⁄ 2)

where DL is the longest diameter and DS is theshortest diameter. Mean tumor volume in eachgroup was compared using the Student t test.

Tumor cell rechallenge after excision of flanktumor. This experiment was conducted to assessthe ability of viral therapy to induce antitumorimmunity and to protect the mice against futuregrowth of tumor of the same type. CT26 flanktumors treated as stated above were excised after3 weeks. The opposite flank subsequently was sub-cutaneously injected with 1 � 106 CT26 cells. Inaddition, 8 naive mice were also injected subcuta-neously with 1 � 106 CT26 cells in the flank as
controls. The mice were followed for 40 days for


the development of any flank tumors. A similarexperiment also was performed in animals withHepa 1-6 flank tumors. Mice that did not developtumor after the second tumor cell injection, werethen injected with 1 � 106 B16 melanoma cellssubcutaneously into the same flank and followed todetermine if the protection against tumor develop-ment was tumor type-specific. Results were com-pared using the Fisher exact test.

Effect of CD4� and CD8� T lymphocyte deple-tion on antitumor efficacy of virus. Both CD4�and CD8� T lymphocytes were depleted simulta-neously by ip injection of monoclonal antibodies(0.2 mg of each antibody was injected ip on days 0,1, and 2) derived from the hybridomas GK1.5 and53-6.72. Depletion was confirmed by fluorescent-activated cell sorter (FACS) analysis of spleens onday 6. The depletion was maintained with ip injec-tion of 0.5 mg of each antibody once a week there-after. CT26 flank tumors were induced in thedepleted mice (n � 49) as described above. Oncethe flank tumors were palpable (diameter, 3 to 5mm), they were treated with intratumoral injectionof 1 � 107 PFU of NV1034, NV1023, or PBS (n �7/group). PBS-treated animals were also depletedof CD4� and CD8� T lymphocytes. Tumor vol-umes were measured, and the mean tumor volumeswere compared using the Student t test. After exci-sion of the primary tumor, the opposite flank wasreinoculated with 1 � 106 CT26 cells. The micewere followed for development of tumor in theopposite flank. A similar experiment also was con-ducted using syngeneic Hepa1-6 cells in C57 L/Jmice.

Statistical analysis. All continuous variables arepresented as mean � standard error of the mean(SEM). The statistical test used in each experimentis mentioned in the section.

RESULTSIn vitro infectivity assay. The first objective

was to determine the infectivity of NV1034 for colo-rectal or hepatocellular adenocarcinomas. None ofthe control wells treated with PBS showed any bluestaining with Xgal (Gibco-BRL). The percentage ofCT26 cells expressing the lacZ gene as a marker ofNV1034 infection at the various time points isshown in the Table. It was observed that NV1034 atan MOI of 1.0 infected 100% cells by day 5. Thesecretion of GM-CSF by the tumor cells did notalter the infectivity of the virus.

In vitro cytotoxicity. These studies were per-formed to determine the oncolytic efficacy ofNV1034 compared to its parent virus NV1023. It
demonstrated a 100% CT26 cell kill by NV1034 and
NV1023 with an MOI of 1.0 at 7 days postinjection.At an MOI of 0.1, NV1034 killed 92% of the CT26cells, whereas NV1023 killed 89%. The viable cellsremaining at various time points (as a percent ofcontrol) are demonstrated in Fig 2, A1 and B1.Hepa 1-6 cells demonstrated a similar sensitivity tothese viral agents (Fig 2, A2 and B2). The addi-tional secretion of GM-CSF by NV1034 did notdecrease the in vitro cytotoxicity of the virus.

In vitro viral proliferation. Plaque assays wereperformed on cells and supernatants to determinethe ability of CT26 and Hepa 1-6 cells to supportthe replication of these viruses. Viral titers weredetermined over 6 days for CT26 cells and 7 daysfor Hepa 1-6 cells. Both cell lines supported thereplication of all viruses tested (Fig 3).

In vitro GM-CSF production. The ability ofNV1034 to produce GM-CSF when it infects tumorcells was then tested by performing ELISA analysison supernatants collected from tumor cells in-fected with either virus. NV1034 produced pico-gram amounts of GM-CSF in both the cell linestested (Fig 3, C). Treatment of tumor cells byNV1023 did not result in production of any GM-CSF. In the CT26 cell line, infection of the cellswith an MOI of 5.0 resulted in a cumulative GM-CSF plateau of only 819 pg/ml media by 5 dayspostinfection, as most of the tumor cells were lysedby the oncolytic effect of the virus early in the assay.Infection at an MOI of 0.1 on the other handproduced 1810 � 150 pg/ml of GM-CSF by day 7and had not peaked during the time period as-sayed. Similar results were seen for the Hepa 1-6cells (Fig 3, D).

In vivo infectivity and GM-CSF production. Fro-zen sections of tumor injected with NV1034 and

Table. Infection of CT26 cells in vitro withvarious MOIs of NV1034

Postinfection(days) Control MOI 0.01 MOI 0.1 MOI 1.0

0 0 0 0 01 0 0 8 172 0 0 10 233 0 0 27 464 0 0 53 895 0 0 79 1006 0 29 100 100

MOI, Multiplicity of infection.CT26 cells were exposed to various MOIs of NV1034 and X-Gal stainingat various time points determined the percentage of lacZ expressing cells.MOIs of 0.1 and 1.0 infected all the cells by day 6, but an MOI of 0.01infected only about 30% of the cells in the same time period.

stained with Xgal (Gibco-BRL) demonstrated viral

Y-axis, viable cells (% of control).


Fig 3. Viral replication (A) and GMCSF production (B) in CT26 or Hepa 1-6. For viral replicationexperiments, tumor cells were infected with NV1023 (square) or NV1034 (diamond) at an MOI of 1. Cells andsupernatants were collected periodically and titered by a standard plaque assay (mean � standard deviation[SD]). X-axis, time postinfection (days); Y-axis, viral titer in PFUs. For GM-CSF production, tumor cells wereinfected with NV1034 at an MOI of 5.0 (cross), MOI of 1.0 (triangle), or MOI of 0.1 (square). Treatment withmedia only (diamonds) produced no GM-CSF (Malhotra et al22). X- axis, time postinfection (days); Y-axis,cumulative GM-CSF (picograms/ml). MOI, Multiplicity of infection; PFU, plaque-forming units; GM-CSF,

Fig 2. 1A, In vitro cytotoxicity of NV1023 in CT26 cells; 1B, In vitro cytotoxicity of NV1034 in CT26 cells;2A, In vitro cytotoxicity of NV1023 in Hepa 1- 6 cells; 2B, In vitro cytotoxicity of NV1034 in Hepa 1- 6 cells.Tumor cells were infected with each virus at varying multiplicities of infection (MOIs) (diamond, 0.1;rectangle, 1.0; triangle, 5.0), and the proportion of viable cells at different time points was determined bycounting using trypan blue exclusion (mean � standard deviation [SD]). X-axis, time postinfection (days);

granulocyte-macrophage colony-stimulating factor.


lacZ gene expression throughout the first 7 daysassayed (Fig 4). Maximum expression was seen ondays 2 and 3. ELISA analysis performed on tumorsfor in vivo murine GM-CSF production detected noGM-CSF in the tumors treated with NV1023. Fortumors treated with NV1034, the GM-CSF produc-tion (in pg/gm tumor tissue) on days 1, 2, 3, and 7was 2400 � 700, 2000 � 600, 1000 � 200, and 440� 370, respectively.

Treatment of subcutaneous flank tumors withvirus. Subcutaneous injection of 1 � 105 CT26 cellin the flank of 4- to 6-week–old Balb/c mice reli-ably produced tumors 3 to 5 mm in diameter in 2weeks. When treated with a single intratumoralinjection of 1 � 106 PFU of virus, the tumor volumeon day 14 posttreatment was 150 � 70 mm3 in theNV1034 group (P � .01 vs control), 270 � 120 mm3

in the NV1023 group (P � .05 vs control), and700 � 200 mm3 in the control group (Fig 5, B).When treated with a single intratumoral injectionof 1 � 107 PFU of virus, the tumor volume on day14 posttreatment was 37 � 23 mm3 in the NV1034group (P � .01 vs control), 210 � 70 mm3 in theNV1023 group (P � .05 vs control), and 710 � 203mm3 in the control group (Fig 5, A). The antitu-mor effect of NV1034 was significantly better thanthat of NV1023 (P � .05) (Figs 5, A and B). Similarresults were observed when subcutaneous Hepa 1-6flank tumors were treated with a single intratu-moral injection. When treated with a single intra-tumoral injection of 1 � 106 PFU of virus, thetumor volume on day 14 posttreatment was 0.5 �0.5 mm3 in the NV1034 group (P � .01 vs control),12 � 3 mm3 in the NV1023 group (P � .05 vs

Fig 4. In vivo infection of CT26 tumors with NV1034.CT26 flank tumors were injected with 1 � 107 PFU ofNV1034. Tumors were harvested on days 1, 2, 3, and 7,and 8�m-frozen sections stained with Xgal solution. LacZexpression (blue cells) could be detected as early as day1,peaked on days 2 and 3 but could still be detected on day7. PFU, plaque-forming units.

control), and 290 � 90 mm3 in the control group

(Fig 5, B). When treated with a single intratumoralinjection of 1 � 105 PFU of virus, the tumor volumeon day 14 posttreatment was 2 � 2 mm3 in theNV1034 group (P � .01 vs control), 72 � 38 mm3

in the NV1023 group (P � .05 vs control), and280 � 80 mm3 in the control group (Fig 5, B). Athighest doses (5 � 107 for CT-26, and 5 � 106 forHepa 1-6), the effects of NV1034 were indistin-guishable from those of NV1023, which was likelydue to the dominant oncolytic effects at highdoses.

We observed that all oncolytic viruses reducedtumor burden compared to PBS-treated controls inboth tumor models. In addition, the cytokine-secret-ing NV1034 significantly reduced tumor burden com-pared to the non–cytokine-secreting NV1023 virus inboth the murine colorectal and the hepatoma tumormodels.

Effect of CD4� and CD8� T lymphocyte deple-tion on antitumor effects of virus. This experimentwas conducted to demonstrate the effect of simul-taneous depletion of both CD4� and CD8� Tlymphocyte depletion on the antitumor effects ofNV1034 and NV1023. When 3- to 5-mm CT26 flanktumors were treated with 1 � 107 PFU of virus inthe setting of T-cell depletion, the tumor volume(in mm3) on day 14 posttreatment was as follows:Controls, 1010 � 110; NV1034, 680 � 100; andNV1023, 730 � 120 (Fig 6, A). Only NV1034 re-duced tumor burden significantly (P � .05). How-ever, the antitumor effects of NV1023 and NV1034were similar.

A similar experiment using 1 � 106 PFU of virusalso was conducted with Hepa 1-6 flank tumors inC57 L/J mice. The tumor volume (mm3) on day 10posttreatment was as follows: Control, 860 � 97;NV1023, 420 � 40; and NV1034, 324 � 39 (Fig 6,B). Both NV1023 and NV1034 reduced tumor bur-den significantly compared to control (P � .05).However there was no difference between the ef-fects of NV1023 and NV1034 (P � not significant[ns]). Thus, the cytokine-secreting virus NV1034has superior antitumor effects only in mice with anintact cellular immune system, and it is mediatedvia CD4� or CD8� T lymphocytes.

Tumor cell rechallenge. An experiment was per-formed to determine whether treatment of primarytumors provided any protection against a tumorrechallenge of the same tumor type. CT26 flanktumors treated with NV1023, NV1034, or PBS as inthe previous experiments were resected and theopposite flank challenged with 1 � 106 CT26. Eightnaive animals also were injected with tumor cells. Itwas observed that all the naive mice developed
flank tumors. A total of 80% (4/5) in the control


group (treated with PBS) developed flank tumors,whereas only 20% (1/5) in the groups treated withNV1023 or NV1034 developed tumors when fol-lowed for 40 days. All mice depleted of CD4� andCD8� T lymphocytes developed tumors when re-challenged after resection of the primary treatedtumor.

This was repeated for Hepa 1-6–bearing ani-mals. When C57 L/J mice with Hepa 1-6 tumorswere rechallenged in similar fashion, all the naiveand PBS control mice developed tumors within 7days. A total of 50% (4/8) of mice treated withNV1023 (P � ns) and 25% (2/8) treated withNV1034 (P � .01) also developed tumors within 7days. To determine whether the protection againsttumor rechallenge was specific for the tumor type,the animals that did not develop tumors after re-challenge with Hepa 1-6 cells were injected withsyngeneic B16 melanoma cells subcutaneously into

Fig 5. In vivo effects of tumor treatment with NV1dose intratumoral viral therapy of CT26 (A) and Hand C57 L/J mice, respectively. In CT26 tumors, a s107 PFU or 1 � 106 significantly reduced tumorcompared to PBS-injected controls (P � .05; 2-tailsignificant reduction in tumor volume comparedNV1023 (P � .05; paired 2-tailed t test, day 28). BNV1023 or NV1034 at 1 � 106 PFU or 1 � 105 signifito PBS-injected controls (P � .05; 2-tailed t test). Ireduction in tumor volume compared to those t(P � .04; 2-tailed t test). X-axis, time (days); Y-axisPFU, plaque-forming units.

the flank. All animals inoculated with B16 devel-

oped tumors. Treatment with the cytokine-produc-ing virus thus protects against future tumorchallenge, and this protection is tumor specific.

DISCUSSIONIn this study, we compared the antitumor effi-

cacy of a GM-CSF–producing oncolytic virusNV1034 to that of the oncolytic virus NV1023 in amurine model of colorectal carcinoma and hepa-toma. NV1023 is an attenuated, replication-compe-tent HSV-1 derived from R7020, a virus originallydesigned as a vaccine against both HSV-1 andHSV-2 infections.20 R7020 is an oncolytic virus thathas demonstrated efficacy in the treatment of avariety of tumors in murine models.4,21,22 It is cur-rently undergoing phase I clinical trials for meta-static liver tumors. Compared to the parent R7020virus, NV1023 and NV1034 have had their endog-enous HSV thymidine kinase and UL24 genes re-

square), NV1034 (cross), or PBS (diamond). Single--6 (B) flank tumors in immunocompetent Balb/cntratumoral injection of NV1023 or NV1034 at 1 �e (mean � standard error of the mean [SEM])st). In addition, the NV1034-treated group had ase treated with the non–cytokine-expressing virusepa 1-6 tumors, a single intratumoral injection ofreduced tumor volume (mean � SEM) compared

ition, the NV1034-treated group had a significantwith the non–cytokine-expressing virus NV1023

or volume (mm3). PBS, phosphate-buffered saline

023 (epa 1

ingle ivolumed t teto tho, In Hcantlyn addreated, tum

stored and contain the lacZ marker gene within the


ICP47 locus. ICP47 is involved in the virus’s abilityto inhibit MHC class I peptide expression in hu-mans by the infected cell.23 The interruption ofICP47 by the lacZ gene therefore may potentiallyincrease MHC class I antigen expression and theo-retically make the tumor cell more recognizable bycirculating immune cells.

NV1034 differs from NV1023 by the insertion ofthe murine GM-CSF gene adjacent to the HSV-2fragment. These studies demonstrate NV1023 and

Fig 6. Intratumoral herpes viral therapy for CT26 (A)and Hepa 1-6 (B) flank tumors in mice depleted ofCD4� and CD8� T lymphocytes. Established flank tu-mors in depleted animals (n � 6 to 8/group) weretreated with a single intratumoral dose of NV1023(square), NV1034 (triangle), or PBS (diamond). In bothmodels, the 2 viruses demonstrated similar efficacy inreducing tumor burden (mean tumor volume � stan-dard error of the mean [SEM]) compared to controls (P� .05; 2-tailed t test). There was no difference in theantitumor efficacy between the 2 viruses (P � ns; 2-tailedt test). X-axis, time (days); Y-axis, tumor volume (mm3).PBS, phosphate-buffered saline; ns, not significant.

NV1034 to be potent oncolytic viruses. Further-

more, the addition of the GM-CSF gene did notalter the in vitro infectivity, replication, or directcytotoxicity of NV1034 compared to the non–cyto-kine-secreting control NV1023. The GM-CSF geneis under the control of a hybrid �4 enhancer-TKpromoter. The current study demonstrates thatboth NV1023 and NV1034 have similar infectiveand replicative abilities in CT26 and Hepa 1-6 cellsin vitro. When administered in vivo by intratumoralinjection, both viruses decreased tumor volume sig-nificantly compared with saline-treated controls. Inaddition, GM-CSF–secreting NV1034 demonstratedan enhanced antitumor efficacy in vivo, likelythrough CD4� and CD8� T cell-mediated im-mune mechanisms.

GM-CSF is a cytokine that induces myeloid pre-cursor cells to proliferate and differentiate intoneutrophils, monocytes, macrophages, and eosino-phils. GM-CSF also recruits and stimulates den-dritic cells.24 In experiments studying tumorvaccines, GM-CSF was the most potent of a numberof cytokines, adhesion molecules, and other immu-nostimulatory molecules for the induction of spe-cific and long-lasting antitumor immunity.16,17

Vaccination with irradiated B16 melanoma cellsengineered to secrete murine GM-CSF stimulatespotent, specific, and long-lasting antitumor immu-nity.7 Murine tumor cells infected ex vivo or in vivowith disabled infectious single-cycle herpes viralvectors expressing GM-CSF may function effectivelyin preventing cancer formation and in treatingestablished tumors.18,19 In the current study, theGM-CSF–producing NV1034 produced a signifi-cant reduction in tumor burden compared to thepurely oncolytic virus NV1023. This demonstratedthat local cytokine production could add to thealready potent antitumor efficacy of oncolytic vi-ruses.

Previous studies have demonstrated that evennon–cytokine-expressing oncolytic HSV (G207)can induce antitumor immune responses.24 Resultsfrom the current study confirms previous observa-tions that oncolytic viral therapy alone inducessome immunity against tumor. Further enhancingthis anticancer immune response during oncolytictherapy has great appeal, because during the timerequired to generate an immune system, the onco-lytic properties of the HSV provide a direct andimmediate means of controlling tumor growth.Some previous attempts to combine use of onco-lytic viruses and cytokine immunotherapy had usedcombined delivery of oncolytic viruses along withdelivery of separate replication incompetent ampli-con vectors for cytokine gene transfer.13,25 The use
of a single vector for oncolysis and immune stimu-


lation greatly simplifies application. The currentstudy demonstrates that such single vector multi-modality therapy is feasible, safe, and efficacious.

The ability of NV1034 to induce a potent, spe-cific, and lasting immune response against theCT26 and Hepa 1-6 tumors was tested by reinocu-lating the animals with tumor cells after surgicallyexcising the residual tumor after viral therapy.Treatment with NV1023 protected some of themice against tumor growth in the CT-26 model.CT26 is a relatively immunogenic tumor, and treat-ment of tumors with saline alone protected 40% ofthe mice against rechallenge. Experiments there-fore were repeated in the Hepa 1-6 tumor modelthat is much less immunogenic. The results indi-cate that addition of GM-CSF enhances the antitu-mor immunity of oncolytic therapy. Furthermore,this protection was tumor specific because all themice that did not develop tumor with Hepa 1-6rechallenge developed tumors when inoculatedwith B16 melanoma cells. When virus (NV1023 orNV1034) is injected into the tumor in high doses,the oncolytic effect is so strong that the effect ofimmunostimulation by the GM-CSF is not notice-able. However, when the tumor burden is high,host toxicity of the virus limits the dose that can beadministered.

Immunity also may play a detrimental role inoncolytic viral therapy. Because 90% of adults carryantibodies to HSV type 1,26 the host immune re-sponse theoretically can decrease efficacy of suchviral therapy. However, experimental data indicatethat treatment using HSV in mice preimmunized toproduce cellular and humoral immunity against HSVstill can have biologic and antitumor effects.22,27

What is not known is whether use of additional im-munostimulatory agents such as GM-CSF will en-hance further the antiviral response and decreaseviral antitumor efficacy.

Five-year survival after surgical resection of colo-rectal metastases to the liver or hepatocellular car-cinoma is only 23% to 38%28,29 and 40% to 50%,respectively.30 Local and distant recurrences aftersurgical therapy for colorectal and hepatocellularcancer occur as a result of microscopic disease thatwas not detectable at the time of resection.31 Anumber of conventional strategies including che-motherapy and radiation therapy have had little orno impact on improving recurrence rates and sur-vival.32-38 Most patients have distribution of diseasethat is not surgically treatable, and survival for thesepatients, even with our best palliative chemother-apy regimens, is usually measured in months.39,40

There is a need to develop novel forms of therapy
that can minimize these recurrences after resection
and to improve results of palliative treatments. Thecurrent data are encouraging for investigation ofcombined oncolytic and immunotherapy in thetherapy of these tumors.

The authors thank Brian Horsburgh, PhD, and Medi-gene, Inc, for constructing and providing us with theNV1034 virus.

REFERENCES1. Mineta T, Rabkin SD, Yazaki T, Hunter WD, Martuza RL.

Attenuated multi-mutated herpes simplex virus-1 for thetreatment of malignant gliomas. Nat Med 1995;1:938-43.

2. Yazaki T, Manz HJ, Rabkin SD, Martuza RL. Treatment ofhuman malignant meningiomas by G207, a replication-com-petent multimutated herpes simplex virus 1. Cancer Res1995;55:4752-6.

3. Walker JR, McGeagh KG, Sundaresan P, Jorgensen TJ,Rabkin SD, Martuza RL. Local and systemic therapy ofhuman prostate adenocarcinoma with the conditionally rep-licating herpes simplex virus vector G207. Hum Gene Ther1999;10:2237-43.

4. Bennett JJ, Kooby DA, Delman K, McAuliffe P, HaltermanMW, Federoff H, et al. Antitumor efficacy of regional onco-lytic viral therapy for peritoneally disseminated cancer. JMol Med 2000;78:166-74.

5. Kooby DA, Carew JF, Halterman MW, Mack JE, Bertino JR,Blumgart LH, et al. Oncolytic viral therapy for human colo-rectal cancer and liver metastases using a multi-mutatedherpes simplex virus type-1 (G207). FASEB Journal 1999;13:1325-34.

6. Carew JF, Kooby DA, Halterman MW, Federoff HJ, Fong Y.Selective infection and cytolysis of human head and necksquamous cell carcinoma with sparing of normal mucosa bya cytotoxic herpes simplex virus type 1 (G207). Hum GeneTher 1999;10:1599-606.

7. Dranoff G, Jaffee E, Lazenby A, Golumbek P, Levitsky H,Brose K, et al. Vaccination with irradiated tumor cells engi-neered to secrete murine granulocyte-macrophage colony-stimulating factor stimulates potent, specific, and long-lasting anti-tumor immunity. Proc Natl Acad Sci U S A1993;90:3539-43.

8. Kemeny N, Gonen M, Sullivan D, Schwartz L, Benedetti F,Saltz L, et al. Phase I study of hepatic arterial infusion offloxuridine and dexamethasone with systemic irinotecan forunresectable hepatic metastases from colorectal cancer.J Clin Oncol 2001;19:2687-95.

9. Andreansky S, He B, van Cott J, McGhee J, Markert JM,Gillespie GY, et al. Treatment of intracranial gliomas inimmunocompetent mice using herpes simplex viruses thatexpress murine interleukins. Gene Ther 1998;5:121-30.

10. Bramson JL, Hitt M, Addison CL, Muller WJ, Gauldie J,Graham FL. Direct intratumoral injection of an adenovirusexpressing interleukin- 12 induces regression and long-last-ing immunity that is associated with highly localized expres-sion of interleukin-12. Hum Gene Ther 1996;7:1995-2002.

11. D’Angelica M, Karpoff H, Halterman M, Ellis J, Klimstra D,Edelstein D, et al. In vivo interleukin-2 gene therapy ofestablished tumors with herpes simplex amplicon vectors.Cancer Immunol Immunother 1999;47:265-71.

12. D’Angelica M, Tung C, Allen P, Halterman M, Delman K,Delohery T, et al. Herpes simplex virus (HSV)-mediatedICAM-1 gene transfer abrogates tumorigenicity and inducesanti-tumor immunity. Mol Med 1999;5:606-16.

13. Toda M, Martuza RL, Kojima H, Rabkin SD. In situ cancer


vaccination: an IL-12 defective vector/replication-competentherpes simplex virus combination induces local and systemicantitumor activity. J Immunol 1998;160:4457-64.

14. Wong RJ, Patel SG, Kim S, DeMatteo RP, Malhotra S, Ben-nett JJ, et al. Cytokine gene transfer enhances herpes onco-lytic therapy in murine squamous cell carcinoma. HumGene Ther 2001;12:253-65.

15. Culver KW, Ram Z, Wallbridge S, Ishii H, Oldfield EH,Blaese RM. In vivo gene transfer with retroviral vector-producer cells for treatment of experimental brain tumors.Science 1992;256:1550-2.

16. Dranoff G, Mulligan RC. Activities of granulocyte-macroph-age colony-stimulating factor revealed by gene transfer andgene knockout studies. Stem Cells 1994;12(Suppl 1):173-82.

17. Kim SH, Carew JF, Kooby DA, Shields J, Entwisle C, Patel S,et al. Combination gene therapy using multiple immuno-modulatory genes transferred by a defective infectious sin-gle-cycle herpes virus in squamous cell cancer. Cancer GeneTher 2000;7:1279-85.

18. Todryk S, McLean C, Ali S, Entwistle C, Boursnell M, Rees R,et al. Disabled infectious single-cycle herpes simplex virus asan oncolytic vector for immunotherapy of colorectal cancer.Hum Gene Ther 1999;10:2757-68.

19. Ali SA, McLean CS, Boursnell MEG, Martin G, Holmes CL,Reeder S, et al. Preclinical evaluation of “whole” cell vac-cines for prophylaxis and therapy using a disabled infectioussingle cycle-herpes simplex virus vector to transduce cyto-kine genes. Cancer Res 2000;60:1663-70.

20. Meignier B, Longnecker R, Roizman B. In vivo behavior ofgenetically engineered herpes simplex viruses R7017 andR7020: Construction and evaluation in rodents. J Infect Dis1999;158:602-14.

21. Bradley JD, Kataoka Y, Advani S, Chung SM, Arani RB,Gillespie GY, et al. Ionizing radiation improves survival inmice bearing intracranial high- grade gliomas injected withgenetically modified herpes simplex virus. Clin Cancer Res1999;5:1517-22.

22. Delman KA, Bennett JJ, Zager JS, Burt BM, McAuliffe PF,Petrowsky H, et al. Effects of preexisting immunity on theresponse to herpes simplex-based oncolytic viral therapy.Hum Gene Ther 2000;11:2465-72.

23. Jugovic P, Hill AM, Tomazin R, Ploegh H, Johnson DC.Inhibition of major histocompatibility complex class I anti-gen presentation in pig and primate cells by herpes simplexvirus type 1 and 2 ICP47. J Virol 1998;72:5076-84.

24. Todo T, Rabkin SD, Sundaresan P, Wu A, Meehan KR,Herscowitz HB, et al. Systemic antitumor immunity in ex-perimental brain tumor therapy using a multimutated, rep-lication-competent herpes simplex virus. Hum Gene Ther1999;10:2741-55.

25. Carew JF, Kooby DA, Halterman MW, Kim SH, Federoff HJ,Fong YM. A novel approach to cancer therapy using anoncolytic herpes virus to package amplicons containing cy-tokine genes. Mol Ther 2001;4:250-6.

26. Whitley RJ, Kimberlin DW, Roizman B. Herpes simplex
viruses. Clin Infect Dis 1998;26:541-53.
27. Herrlinger U, Kramm CM, Aboody-Guterman KS, Silver JS,Ikeda K, Johnston, et al. Pre-existing herpes simplex virus 1(HSV-1) immunity decreases, but does not abolish, genetransfer to experimental brain tumors by a HSV-1 vector.Gene Ther 1998;5:809-19.

28. Fong Y, Cohen AM, Fortner JG, Enker WE, Turnbull AD,Coit DG, et al. Liver resection for colorectal metastases.J Clin Oncol 1997;15:938-46.

29. Weber SM, Jarnagin WR, DeMatteo RP, Blumgart LH, FongY. Survival after resection of multiple hepatic colorectalmetastases. Ann Surg Oncol 2000;7:643-50.

30. Fong Y, Sun RL, Jarnagin W, Blumgart LH. An analysis of412 cases of hepatocellular carcinoma at a Western center.Ann Surg 1999;229:790-9.

31. Panis Y, Ribeiro J, Chretien Y, Nordlinger B. Dormant livermetastases: an experimental study. Br J Surg 1992;79:221-3.

32. Ono T, Nagasue N, Kohno H, Hayashi T, Uchida M, YukayaH, et al. Adjuvant chemotherapy with epirubicin and car-mofur after radical resection of hepatocellular carcinoma:a prospective randomized study. Semin Oncol 1997;24(2 Suppl 6):S6.

33. Kohno H, Nagasue N, Hayashi T, Yamanoi A, Uchida M,Ono T, et al. Postoperative adjuvant chemotherapy afterradical hepatic resection for hepatocellular carcinoma(HCC). Hepatogastroenterology 1996;43:1405-9.

34. Izumi R, Shimizu K, Iyobe T, Ii T, Yagi M, Matsui O, et al.Postoperative adjuvant arterial infusion of lipiodol contain-ing anticancer drugs in patients with hepatocellular carci-noma. Hepatology 1994;20:295-301.

35. Wu CC, Ho YZ, Ho WL, Wu TC, Liu TJ, P’eng FK. Preop-erative transcatheter arterial chemoembolization for resect-able large hepatocellular carcinoma. A reappraisal. Br JSurg 1995;82:122-6.

36. Whalen GF, Bird I, Tanski W, Russell JC, Clive J. Laparo-scopic cholecystectomy does not demonstrably decrease sur-vival of patients with serendipitously treated gallbladdercancer. J Am Coll Surg 2001;192:189-95.

37. Lai EC, Lo CM, Fan ST, Liu CL, Wong J. Postoperativeadjuvant chemotherapy after curative resection of hepato-cellular carcinoma: a randomized controlled trial. ArchSurg 1998;133:183-8.

38. Fong Y, Jarnagin W, Blumgart LH. Gallbladder cancer:comparison of patients presenting initially for definitiveoperation with those presenting after prior noncurative in-tervention. Ann Surg 2000;232:557-69.

39. Kemeny N, Fong Y. Treatment of liver metastases. In: Hol-land JF, Frei E, Bast RC, Kufe DW, Morton DL, Weichsel-baum RR, editors. Cancer Medicine. 4th ed. Baltimore:Williams & Wilkins; 1997. p. 1939-54.

40. Fong Y, Kemeny N, Lawrence TS. Cancer of the liver andbiliary tree. In: DeVita VT, Hellman S, Rosenberg SA, edi-tors. Cancer: principles and practice of oncology. 6th ed.Philadelphia: Lippincott Williams and Wilkins; 2001. p.
1162-203.

Documents

Use of an oncolytic virus secreting GM-CSF as combined oncolytic and immunotherapy for treatment of colorectal and hepatic adenocarcinomas