Virally-directed Fluorescent Imaging (VFI) Improves Diagnostic Sensitivity in the Detection of Minimal Residual Disease Following Potentially Curative Cytoreductive Surgery

Abstract

Background

Completeness of cytoreduction is an independent prognostic factor following cure-intended surgery for peritoneal carcinomatosis (PC). Intraoperative detection of the minimal residual disease may aid in achieving complete cytoreduction. NV1066, a genetically-engineered herpes simplex virus carrying the transgene for green fluorescent protein (GFP), selectively infects cancer cells. NV1066-infected cancer cells express GFP that can be detected by fluorescence laparoscopy. We sought to determine the feasibility of Virally-directed Fluorescent Imaging (VFI) in the intraoperative detection of minimal residual disease following cytoreductive surgery.

Methods

Human cancer cell lines OCUM-2MD3 (gastric) and JMN (malignant Mesothelioma) were infected with NV1066 at MOIs (multiplicity of infection; ratio of viral particles to cancer cells) of 0.01, 0.1 and 1.0. Viral infectivity was determined by flow cytometry for GFP and cytotoxicity was determined by a colorimetric assay. PC was established in mice by injection of OCUM cells into the peritoneal cavity. Forty-eight hours following intraperitoneal injection of NV1066, two experienced surgeons resected all visible disease and identified mice free of disease. Five independent observers examined these mice by bright-field and fluorescent laparoscopy and documented residual disease as per the peritoneal cancer index. Selective expression of GFP in tumor tissue was evaluated by histology and PCR for the viral gene ICP0.

Results

In vitro, NV1066 infected, expressed GFP, and killed both cell lines at all MOIs. GFP signal was detected as early as 4-6 hours following infection. GFP signal intensity of infected cells was significantly higher than the autofluorescence of normal cells (230 – 670 -logs). In vivo, macroscopically undetectable tumor nodules by gross examination and conventional bright-field laparoscopy were identified by GFP fluorescence. Following resection, 8 of 13 mice thought to be free of disease were found to have residual disease as identified by green fluorescence (mean number of observations: 5 range: 1-9). Residual disease was most frequently observed in the retroperitoneum, pelvis, peritoneal surface, and liver (inter-observer agreement 99%). Specificity of NV1066 infection to tumor nodules was confirmed by immunohistochemistry and by PCR for viral gene.

Conclusion

We have demonstrated that virally-directed fluorescent imaging (VFI), a novel molecular imaging technology, can be used for real-time visualization of minimal residual disease following cytoreductive surgery and can improve the completeness of cure-intended resection.

Introduction

The prognosis for patients with peritoneal carcinomatosis from gastrointestinal malignancies is dismal. Recent efforts combining aggressive cytoreduction of intraperitoneal disease with perioperative chemotherapy have demonstrated improved overall survival and quality of life in these patients(1;2). Among the most important prognostic factors determining survival after cytoreduction is completeness of resection. Unfortunately, the ability to identify macro- or microscopic residual tumor deposits at the time of cytoreductive surgery is limited and, as such, disease recurs in the majority of patients. Methods that enhances intraoperative detection of minimal residual disease may improve completeness of cytoreduction and patient outcomes(3).

Oncolytic herpes viruses are replication-competent, attenuated herpes simplex type-1 viruses (HSV) that selectively infect cancer cells, sparing normal cells. Their therapeutic efficacy in experimental models of gastrointestinal cancer has been demonstrated in previous publications(4;5). One such virus is NV1066, a genetically-engineered oncolytic HSV strain that carries the transgene for marker protein, enhanced green fluorescent protein (GFP). Cancer cells infected with NV1066 constitutively express GFP that can be detected using fluorescent imaging. Advantages of such a marker protein have been demonstrated in several in vitro experiments. Recent development in our laboratory of a fluorescent laparoscopic system, in combination with the advent of NV1066, has led to the discovery of novel applications of in vivo cancer imaging.

We sought to evaluate the ability of NV1066-mediated cancer cell-specific expression of GFP to enhance the intraoperative detection of minimal residual disease following cytoreductive surgery using Virally-directed Fluorescent Imaging (VFI).

Materials and Methods

Cells

The human gastric cancer cell line OCUM- 2MD3 and the malignant mesothelioma cell line JMN were studied. OCUM-2MD3 cells were a generous gift from Dr. Masakazu Yashiro (Osaka City University Medical School, Japan) and were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with high glucose, 2 mM L-glutamine, and 0.5 mM sodium pyruvate, 100 U/ml penicillin, 100 mg/ml streptomycin, and 10% fetal calf serum. JMN cells were a generous gift of Dr. Francis Sirotnak (Memorial Sloan-Kettering Cancer Center, New York) and were grown in RPMI 1640 media supplemented with 100 U/ml penicillin, 100 mg/ml streptomycin, and 10% fetal calf serum. Vero cells (American Type Culture Collection, Rockville, MD) were grown in minimum essential medium supplemented with 100 U/ml penicillin, 100 mg/ml streptomycin, and 10% fetal calf serum. Cells were maintained in a 5% CO₂ humidified incubator at 37°C.

Virus

NV1066 is a replication-competent oncolytic herpes simplex type-1 viral strain whose construction has been previously described in detail(6). Briefly, NV1066 was derived from the wild-type HSV-1 virus (F strain) via deletions in the viral virulence genes ICP0, ICP4, and γ₁34.5. These deletions attenuate the virus conferring selectivity for infection of cancer cells and render the virus safe for use in humans. In addition, the transgene for enhanced green fluorescent protein (GFP) was inserted into the deleted region under the control of a constitutively expressed cytomegalovirus (CMV) promoter. Viral stocks were propagated on Vero cells and titered by standard plaque assay. NV1066 was obtained from MediGene, Inc. (Martinsried, Germany).

Detection of GFP expression by fluorescent microscopy

Monolayer cultures of OCUM-2MD3 cells were incubated at 37°C and infected with NV1066 at an MOI of 1.0. A Zeiss LSM 510 confocal laser scanning microscope (Carl Zeiss, Inc., Germany) and the MetaMorph Imaging System (Downingtown, PA) were used to visualize GFP expressing cancer cells hourly after infection. GFP expression was identified after placement of specific excitation and emission filters to detect GFP. The Retiga EX digital CCD camera (Qimaging, Burnaby, British Columbia) was used for image-capture.

Detection of GFP expression by flow cytometry

Cultured cells (5 × 10⁴) were plated in 6-well flat-bottom assay plates (Becton Dickinson, Franklin Lakes, NJ) in 2 ml of media. After overnight incubation at 37°C, cells were infected with NV1066 at MOIs (multiplicity of infection; ratio of the number of viral particles to the number of tumor cells) of 0.01, 0.1, or 1.0 in 100 μl of phosphate-buffered saline (PBS). Untreated cells served as a negative control. Daily after infection, cells were harvested with 0.25% trypsin in 0.02% EDTA, combined with the supernatant fraction, centrifuged, washed with PBS, and resuspended in 100 μl of PBS. Five μl of 7-amino-actinomycin (7-AAD; BD Pharmingen, San Diego, CA) was added as an exclusion dye for cell viability. Data for GFP expression from 1 × 10⁴ cells was acquired on a FACS Calibur unit equipped with Cell Quest software (Becton Dickinson, San Jose, CA). Results are reported as percent of live cells expressing GFP. Additionally, GFP-positive cells were sorted using the MoFlo High-Performance Cell Sorter (DakoCytomation, Carpinteria, CA) and stained with rabbit anti-HSV-1 polyclonal antibody (Biogenex, San Ramon, CA) to confirm viral infection of green cells. A biotinylated secondary antibody was added and visualized with streptavidin-labeled horseradish peroxidase and chromogen solutions. Each experiment was repeated a minimum of three times.

Cytotoxicity assay

Cytotoxicity assays were performed by plating 2 × 10⁴ cells in 24-well plates in 1 ml of media. After overnight incubation at 37°C, cells were infected with NV1066 diluted in 100 μL media at MOIs of 0.01, 0.1, and 1.0. On days 3 to 7 after infection, cells were lysed with 1.35% Triton-X solution to release intracellular lactate dehydrogenase (LDH). LDH was then quantified with a Cytotox 96 non-radioactive cytotoxicity assay (Promega, Madison, WI) that measures the conversion of a tetrazolium salt into a red formazan product. Absorbance was measured at 450 nm with a microplate reader (EL321e, Bio-Tek Instruments, Winooski, VT). Results are expressed as the surviving percentage of cells as determined by the measured absorbance of each sample relative to control, untreated cells. All samples were tested, and experiments were replicated, in triplicate.

Animal model of peritoneal carcinomatosis

All animal procedures were performed under the guidelines approved by the Memorial Sloan-Kettering Institutional Animal Care and Use Committee. Eight to ten week-old athymic mice (National Cancer Institute, Bethesda, MD) were housed in a temperature- and light-controlled animal facility. Food and water were permitted ad libitum. Animals were anesthetized with inhalational methoxyflurane for all experimental manipulations and were sacrificed by CO₂ inhalation at the termination of the experiment.

Peritoneal carcinomatosis (PC) was established by injection of 1 × 10⁷ OCUM-2MD3 cells suspended in 500 μl PBS into the peritoneal cavity of athymic mice (n = 24). Twenty-one days following implantation of tumor cells, animals were treated with a single intraperitoneal injection of 1 × 10⁷ plaque forming units (PFU) of NV1066 in 100 μl PBS (n=21). Three animals with PC were treated with an intraperitoneal injection of 100 μl PBS and three additional animals without PC were treated with a single intraperitoneal injection of 1 × 10⁷ PFU of NV1066 to serve as negative controls. Forty-eight hours following viral administration, laparotomy was performed on all animals by two experienced surgeons with intent to resect all intraabdominal disease. Partial or complete resection of organs was performed if deemed necessary to achieve complete resection.

In vivo fluorescent imaging

Immediately following resection, the peritoneal cavities of all animals were systematically examined by both bright-field and fluorescent laparoscopy. We use a laparoscopic system developed in concert with Olympus America, Inc. (Scientific Equipment Division, Melville, NY) that images in both bright-field and fluorescent modes permitting the detection of GFP. The light source is derived from the Olympus Visera CLV-U40 model (Olympus America, Inc., Melville NY) adapted with an interchangable excitation filter set at 470 ± 20 nm to accommodate the minor excitation peak of GFP at 475 nm and an emission filter fixed at 500 nm to accommodate the emission peak of GFP at 509 nm. The camera processor was an Olympus Visera OTVS7V with an emission filter set at 510 nm(5). A control button incorporated directly into the camera head enables rapid exchange between bright-field and fluorescent modes. GFP images were taken with minimal background illumination to illustrate the surrounding organs.

Using this system, each mouse was individually and systematically examined by five independent, blinded observers. Each observer examined 12 predetermined anatomic areas (liver, right sub-diaphragm, stomach, spleen, left sub-diaphragm, intestine, mesentery, retroperitoneum, left kidney, right kidney, pelvis, and peritoneal surface) in a systematic fashion for the presence of residual disease as determined by the presence or absence of green fluorescence. These anatomic areas were designed on the peritoneal cancer index developed by Sugarbaker(7). Investigators recorded the presence (“Yes”) or absence (“No”) of green fluorescence on a data sheet. Animals and data sheets were matched and coded with random numbers known only to a single investigator who was neither a surgeon nor an observer. Animals in which grossly evident disease remained following resection were excluded from the study. This experiment was repeated twice. Animals with disease treated by PBS and animals without disease treated by NV1066 (negative controls) were also observed in a similar fashion as described above.

Histological confirmation of residual microscopic disease

Following complete examination of each animal by all observers, intraabdominal tissue biopsies, areas that are both green and not green, were fixed in 10% phosphate-buffered formalin and embedded in paraffin for histologic analysis. Serial 8 μm sections of all tissue blocks were cut and stained with hematoxylin and eosin (H&E) to assess for the presence of tumor. Additional slides were stained with rabbit anti-HSV-1 polyclonal antibody (Ready-to-Use, Biogenex, San Ramon, CA) to detect the presence of virus in harvested tissues. A biotinylated secondary antibody was added and visualized with streptavidin-labeled horseradish peroxidase and chromogen solutions (Super Sensitive Ready-to-Use Detection System, Biogenex). Counterstaining with Harris hematoxylin was performed.

Confirmation of viral specificity to residual microscopic disease by real-time polymerase chain reaction (PCR)

Additional random intra-abdominal tissue samples were harvested and snap-frozen in liquid nitrogen. Genomic DNA was isolated using standard protocols (Wizard Genomic DNA Isolation Kit, Promega, Madison, WI). Real-time quantitative PCR was performed using an ABI Prism 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA). Forward (5′-ATGTTTCCCGTCTGGTCCAC-3′) and reverse (5′-CCCTGTCGCCTTACGTGAA-3′) primers and a dual-labeled fluorescent TaqMan probe (5′-FAM-CCCCGTCTCCATGTCCAGGATGG-TAMRA-3′) were designed to amplify and detect the 111-bp fragment of the herpes simplex virus immediate-early gene ICP0. Forward (5′-CGCCTACCACATCCAAGGAA-3′) and reverse (5′-GCTGGAATTACCGCGGCT-3′) primers and a dual-labeled fluorescent TaqMan probe (5′-VIC-TGCTGGCACCAGCTTGCCCTC-TAMRA-3′) were also designed for the 87-bp coding sequence of 18s rRNA to normalize to the amount of total DNA present. The PCR conditions were as follows: stage 1: 50°C for 2 minutes; stage 2: 95°C for 10 minutes; stage 3 (35 cycles): 95°C for 15 seconds and 60°C for 1 minute; and stage 4, 25°C soak.

Results

NV1066 infects, expresses GFP, and kills cancer cells

In vitro, NV1066 infected, expressed GFP, and was cytotoxic to both cell lines at all MOIs (Figure 1). Flow cytometry for GFP was performed to demonstrate infectivity and expression of GFP in NV1066-treated OCUM-2MD3 and JMN cells (Figure 1A and C). Six days following infection of OCUM-2MD3 cells at MOIs of 0.1 and 1.0, 100% (± 0%) of cells expressed GFP (p <0.01) (Figure 1A). Even at a low MOI of 0.01, 86% (± 3%) cells expressed GFP six days after infection (p <0.01). Similar results were obtained with JMN cells (Figure 1C).

Figure 1.

NV1066 infects, expresses GFP, and kills cancer cells. Monolayer cultures of OCUM-2MD3 human gastric cancer cells and JMN human malignant mesothelioma were incubated at 37°C and infected with NV1066 at MOIs of 0.01 (open triangle), 0.1 (circle), 1.0 (square). Cells were harvested daily after infection and analyzed by either flow cytometry for GFP expression or LDH cytotoxicity assay to determine cell kill. Expression of GFP following infection of OCUM-2MD3 cells is plotted as % of live cells expressing GFP and is shown in Panel A. Cytotoxicity of OCUM-2MD3 cells after infection is plotted as % of live cells remaining compared to untreated control cells and is shown in Panel B. Similar results are observed with JMN cells (Panels C and D). MOI: multiplicity of infection, ratio of viral particles to tumor cells, GFP: green fluorescent protein.

To examine the oncolytic efficacy of NV1066, dose-dependent cytotoxicity assays were performed against OCUM-2MD3 and JMN cells (Figure 1B and D). NV1066 demonstrated dose-dependent cytotoxicity against both cell lines. Seven days following infection of OCUM-2MD3 cells at an MOI of 1.0, 100% (± 1%) of cells were killed (p <0.01) (Figure 1B). At 10-fold lower MOIs of 0.1 and 0.01, 97% (± 1%) and 79% (± 4%) of cells were killed 7 days following infection, respectively (p < 0.01). Similar results were seen with JMN cells (Figure 1D).

GFP can be detected within infected cancer cells within hours of infection

GFP signal was detected as early as 4-6 hours following infection (Figure 2A). GFP signal intensity of infected cells was significantly higher than the autofluorescence of normal cells (230-670 logs). Immunohistochemistry for herpes viral antigen confirms localization of GFP in herpes virus infected cancer cells (Figure 2B and C). All the green cells that were sorted by flow sorter were stained by HSV-1 antibody, confirming that the green fluorescence is indeed due to NV1066 infection. Similarly, non-green cells were not stained with HSV-1 antibody.

Figure 2.

GFP can be detected within infected cancer cells within hours of infection. Monolayer cultures of OCUM-2MD3 human gastric cancer cells were infected with NV1066 at an MOI of 1.0. Within 4-6 hours of infection, GFP is detected by fluorescent microscopy (Panel A, magnification 10x). Immunohistochemical staining for herpes viral antigen demonstrates the presence of intracellular herpes viral antigen (Panel B, magnification 40x). Digital fluorescent overlay demonstrates production of GFP in herpes virus infected cells (Panel C, magnification 40x). MOI: multiplicity of infection, ratio of viral particles to tumor cells, GFP: green fluorescent protein.

NV1066-mediated GFP expression enables the detection of minimal residual disease following complete resection

In vivo, macroscopically undetectable tumor nodules by gross inspection and bright-field laparoscopy were readily identified by fluorescent laparoscopy due to green fluorescence. Representative images are shown in Figure 3.

Figure 3.

NV1066-mediated GFP expression enables the detection of minimal residual disease following complete resection. A murine model of peritoneal carcinomatosis was established by injection of 1 × 10⁷ OCUM-2MD3 cells into the peritoneal cavity of athymic mice. Twenty-one days later, animals were treated with a single intraperitoneal injection of 1 × 10⁷ PFU of NV1066. Forty-eight hours following viral administration, laparotomy with complete cytoreduction was performed. The peritoneal cavities of all animals were then systematically examined by both bright-field and fluorescent laparoscopy for the detection of residual disease. Overlay images are created by digital superimposition of the fluorescent image over the bright-field image. Representative images are shown. Panels A and B demonstrate residual subdiaphragmatic and sub-hepatic tumor, respectively, that were not detected by bright-field examination. Panels C and D demonstrate mesenteric and small intestinal serosal tumor deposits that were not identified by routine bright-field laparoscopy. Panel E demonstrates a 1 mm tumor deposit on the peritoneal surface that was missed by all 5 observers using bright-field laparoscopy but identified by all five observers with fluorescent laparoscopy. Panel F shows residual pelvic disease that was only identified by fluorescent laparoscopy. PFU – plaque forming units.

A maximum of 780 observations were possible (13 mice × 12 anatomic sites × 5 observers). Only 757 observations were recorded, however, as completely resected organs were excluded from the analysis. Residual tumor, as identified by green fluorescence during laparoscopy, was detected in 8 of 13 mice. The mean number of green observations per animal for these eight mice was 5.25 (range, 1-9). Table I shows the number of green observations for each of the twelve anatomic sites, aggregated over all mice and observers. Minimal residual disease as identified by green fluorescence was most frequently observed in the retroperitoneum, pelvis, peritoneal surface, and liver. Minimal residual disease was least commonly observed in the mesentery, stomach, and spleen. No areas of green fluorescence were identified in the negative control animals. Analysis of inter-observer agreement among the five observers revealed agreement in 99% of observations.

Table I.

Frequency of enhanced identification of minimal residual disease green fluorescence at each of twelve anatomic sites

Anatomic Site	Number of bright-field observations	Number of times minimal residual disease was identified by green fluorescence*	Percentage (%)**
Retroperitoneum	65	35	54
Pelvis	65	34	52
Peritoneal surface	65	25	38
Liver	65	25	38
Left sub-diaphragmatic	55	20	36
Intestine	45	15	33
Left kidney	60	15	25
Right kidney	65	15	23
Right sub-diaphragmatic	50	10	20
Mesentery	40	6	15
Stomach	40	5	13
Spleen	40	5	13

^*Data is aggregated over all mice and observers.

^**Percentages are calculated as the number of green observations per total number of observations at each anatomic site. The following sites were missing both bright-field and green fluorescence observations as the organs underwent near-complete resection due to macroscopic disease {left sub-diaphragm (10), intestine (20), left kidney (5), right sub-diaphragmatic (15), mesentery (25), stomach (25) and spleen (25)}.

Confirmation of residual microscopic disease

Biopsies of tissues that were identified as green during fluorescent laparoscopy were confirmed to harbor tumor via routine H&E histological analysis. Furthermore, immunohistochemical staining for herpes viral antigen demonstrated that NV1066 localized to tumor deposits in a highly specific manner. No tumor or virus was detected in non-green tissue biopsies that were analyzed. Representative sections are shown in Figure 4 PCR amplification of the viral gene ICP0 was used to further confirm the absence of NV1066 in non-tumor bearing (non-green) tissues. ICP0 was not detected in any normal tissue biopsy analyzed confirming the high specificity for viral infection of cancer cells, sparing normal cells (data not shown).

Figure 4.

Histological confirmation of residual microscopic metastatic disease. Intraabdominal tissue biopsies were harvested, fixed in formalin, and embedded in paraffin. Serial 8 μm sections of tissue blocks were cut and stained with hematoxylin and eosin (H&E) to assess for the presence of tumor. Immunohistochemistry was performed using the HSV-1 polyclonal antibody to detect for the presence of NV1066. Representative sections are shown. Hematoxylin and eosin staining demonstrates tumor deposits (T) on the diaphragm (M) that were missed by bright-field laparoscopy but identified via fluorescent laparoscopy (panel A). Panel B confirms the presence of virus in the tumor by immunohistochemistry and further demonstrates the specificity of virus for tumor cells. Panel C (H&E) confirms the presence of tumor in a sub-hepatic deposit that was missed by bright-field laparoscopy and panel D (IHC) confirms the presence of virus in the tumor. Magnification – 40x – panels A,B; magnification 20x – panels C,D. HSV – herpes simplex virus. H&E – hematoxylin and eosin. IHC – immunohistochemistry. T – Tumor; M –muscle of the diaphragmatic; L – Liver.

Discussion

Peritoneal surface involvement occurs in as many as 20-30% of patients with gastric, colon, appendiceal, and pancreatic cancers(8). Median survival in these patients with peritoneal disease is usually less than one year(9). Moreover, recurrent bowel obstructions, malignant ascites, fistulization, and pain result in diminished quality of life. Aggressive loco-regional treatment of these patients combining cytoreductive surgery with perioperative intraperitoneal chemotherapy aims to improve patient survival and quality of life(7;8). Cytoreduction entails resection of all visible tumor and stripping of all peritoneal surfaces that contain metastatic nodules. Visceral peritoneal involvement often requires concomitant resection of intraabdominal organs including the stomach, small intestine, and colorectum. In addition, perioperative intraperitoneal chemotherapy is administered to sterilize residual microscopic disease.

Such aggressive loco-regional treatment has resulted in improved survival and quality of life as reported in several publications. In one such study, Verwaal et al. randomized 117 patients with peritoneal carcinomatosis due to colorectal cancer to receive either standard systemic chemotherapy with or without palliative surgery, or aggressive cytoreduction combined with hyperthermic intraperitoneal chemotherapy (HIPEC)(3). Their reported 5-year survival of 19% exceeded historical controls, whose 5-year survival approached 0%. Similar favorable results using this approach have also been reported in patients with carcinomatosis secondary to gastric cancer, pseudomyxoma peritonei, and non-gastrointestinal malignancies including peritoneal mesothelioma and sarcomatosis(8;10;11). Among the most important prognostic factors following cytoreduction is completeness of resection. In the series by Verwaal et al., median survival in a subgroup of patients in which complete cytoreduction was achieved was 42.9 months, compared to 17.4 months in patients with minimal residual disease, defined as residual macroscopic tumor ≤ 2.5 mm following cytoreduction(3). Similarly, the completeness of cytoreduction score developed by the Sugarbaker group, an assessment made by the operating surgeon of the extent of residual disease following cytoreduction, has been shown to be a major prognostic indicator in patients with peritoneal surface malignancies(7).

The ability to detect small macroscopic residual peritoneal disease is largely limited by the lack of contrast difference between tumor nodules and surrounding normal tissues. Moreover, microscopic nodules are beyond the limits of detection of the unaided eye. Technology improving the intraoperative detection of residual peritoneal disease would facilitate complete cytoreduction and improve survival in these patients.

We demonstrate that virally-directed fluorescent imaging (VFI) utilizing NV1066-mediated tumor cell-specific production of GFP enhances the detection of minimal residual disease following cytoreduction in a murine model of peritoneal carcinomatosis. The green fluorescence emitted by NV1066 is 230-670 logs greater than background autofluorescence, allowing clear discrimination between tumor and normal tissues. The mean fluorescent intensity of enhanced GFP expressed by NV1066 is sixfold greater than non-enhanced GFP and it matures four-times more rapidly reducing lag time from infection to detection of green fluorescence. Furthermore, enhanced GFP has specific emission and excitation wavelengths (475/509 nm) – eliminating autofluorescent interference. Following a single intraperitoneal injection of NV1066 and presumed complete cytoreduction by two experienced surgeons, 8 of 13 mice were found to have at least one site of residual disease using VFI. In our study, residual disease was identified most commonly in the retroperitoneum, pelvis, peritoneal surface, and liver. These sites correspond to patterns of recurrence in published human series(12;13).

In addition to providing a diagnostic benefit, the use of oncolytic herpes viruses in this setting may offer a therapeutic effect. The therapeutic efficacy of NV1066 in a murine model of peritoneal carcinomatosis secondary to gastric cancer has already been demonstrated by work in our laboratory(5). Dose-dependent reduction of peritoneal weights in a murine model of peritoneal carcinomatosis was observed following intraperitoneal injection of NV1066. Furthermore, we have also shown that oncolytic herpes viruses interact synergistically when used in combination with chemotherapy(14). In this study, mice with gastric carcinomatosis treated with combination viral therapy and Mitomycin C had reduced tumor burdens compared to either treatment alone.

Finally, the specificity of NV1066 for cancer cells in this study further confirms observations from previous work in our laboratory(4;15). Immunohistologic and PCR analyses of viral presence following administration confirms specific uptake of virus in tumor tissue, sparing normal tissue. As such, expected toxicity from these agents is minimal, as demonstrated in numerous studies using NV1066 in murine models(4;5;15). Furthermore, the safety of related oncolytic herpes simplex virus strains in humans has been established in phase I clinical trials both from our group and others(16;17).

Conclusions

Promising survival benefits are being realized for patients with peritoneal carcinomatosis from gastrointestinal malignancies following an aggressive multimodal therapeutic approach combining cytoreductive surgery and perioperative chemotherapy. Treatment failure in these patients is frequently the result of unresected minimal residual disease. We demonstrate that virally-directed fluorescent imaging (VFI) utilizing NV1066-mediated tumor cell-specific production of GFP enhances the detection of minimal residual disease following cytoreduction of peritoneal carcinomatosis. Furthermore, NV1066 in addition to being diagnostic is therapeutic and can be combined with chemotherapy, an approach recommended in the loco-regional treatment of these patients with advanced disease.