Interferon-β gene therapy improves survival in an immunocompetent mouse model of carcinomatosis

Interferon-b gene therapy improves survival in an immunocompetent mouse model of carcinomatosis Samantha K. Hendren, MD, Indira Prabakaran, MSc, Donald G. Buerk, PhD, Giorgos Karakousis, MD, Michael Feldman, MD, PhD, Francis Spitz, MD, Chandrakala Menon, PhD, and Douglas L. Fraker, MD, Philadelphia, Pa

Background. Interferon-b (IFNb) has multiple antitumor effects; however, its use has been limited by its short half-life in vivo. This limitation may be overcome by IFNb gene therapy. We evaluated adenovirusIFNb therapy in an immunocompetent mouse model of carcinomatosis. Methods. Mice that were treated intraperitoneally 5 days after tumor (mouse ovarian teratoma) inoculation with an adenoviral vector that contains the mouse IFNb gene (Ad-IFNb), control adenoviral vector or saline solution. Mice were monitored for multiple outcome measures and toxicity. To determine the mechanism of antitumor effect, flow cytometry of ascites fluid was performed to differentiate immune cell populations. Nitric oxide in ascites fluid was measured with an electrochemical microsensor. Results. Tumor burden was decreased and survival was prolonged (P < .001) in the Ad-IFNb group after a single treatment of 3.3 3 108 plaque-forming units, with acceptable toxicity. By f low cytometry, an increase in the proportion of natural killer cells (from less than 2% of the gated population to more than 8%; P = .024) and an increase in macrophages were seen in the treated animals. Although there was a trend toward increased levels of nitric oxide in Ad-IFNb treatment groups, it was not statistically significant. Conclusion. IFNb gene therapy results in decreased tumor burden and improved survival in an aggressive, immunocompetent mouse model of carcinomatosis. This therapy warrants further evaluation as a treatment for disseminated peritoneal cancer. (Surgery 2004;135:427-36.) From the Departments of Surgery, Physiology and Bioengineering, and Pathology, University of Pennsylvania, Philadelphia, Pa

CARCINOMATOSIS, WHICH IS THE DISSEMINATED intraperitoneal spread of cancer, is a clinical entity with no curative treatment options and considerable morbidity for patients. Current experimental therapies focus on surgical resection of the gross tumor, combined with surface treatments such as hyperthermic intraperitoneal chemotherapy or photodynamic therapy.1,2 These treatment approaches have been designed with the recognition that microscopic tumor potentially contaminates every peritoneal surface. Although these experimental protocols have shown some limited success, most patients experience recurrence and succumb to the

Supported in part by the Georgene S. Harmelin Endowment Fund and National Institutes of Health grant CA-16520. Accepted for publication August 4, 2003. Reprint requests: Douglas L. Fraker, MD, Department of Surgery, 3400 Spruce St, 4 Silverstein, Philadelphia, PA 19104. 0039-6060/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.surg.2003.08.015

disease. Novel treatments for this pattern of cancer dissemination are required. Theoretically, carcinomatosis is a disease that is uniquely well-suited for adenoviral gene therapy. First, after surgical debulking, the residual disease in the peritoneum is of microscopic thickness; therefore, limited penetration of the adenoviral vector is not a problem in this disease entity, although it has been a limitation in solid tumors. Second, because the cancer is limited to peritoneal surfaces, the vector could be administered intraperitoneally, which avoids the toxicity of intravascular administration of adenovirus. Interferon-b (IFNb) is a pleiotropic cytokine with multiple biologic effects.3,4 It is a type I interferon, which shares sequence homology and cell-surface receptor specificity with IFNa. In several preclinical tumor models, antiproliferative, immunomodulatory, and antiangiogenic effects of IFNb have been demonstrated.5-10 Unfortunately, the clinical application of IFNb protein in human cancer patients has been disappointing, possibly because of the short in vivo half-life of the protein. SURGERY 427

428 Hendren et al

This obstacle theoretically can be overcome by the delivery of the IFNb gene directly into tumor cells or host tissues that then produce the protein continuously. IFNb gene therapy with retroviral, liposomal, or adenoviral vectors has been described in several preclinical tumor models.7,8,11,12 The mechanism of the IFNb antitumor effect appears to vary depending on the method of administration and the tumor model that is used. Various immune cell populations have been shown to be involved in specific model systems, including natural killer (NK) cells, macrophages, and T lymphocytes.7-10,12 Many of these studies have been performed in athymic nude mice, in which the immune background is abnormal. We believe that it is important to study this therapy in immunocompetent animals, due to the need to control for nonspecific immune activation because of the adenoviral vector and the incompletely understood immune mechanisms of the IFNb antitumor effect. In this study, an immunocompetent preclinical model of carcinomatosis was used to study the effect of intraperitoneal adenovirus-IFNb (AdIFNb) gene therapy. Significantly improved survival in the treatment group was demonstrated.

MATERIAL AND METHODS Preclinical tumor model. The mouse ovarian teratoma (MOT) cell line is an ovarian cancer that spontaneously arose in the immunocompetent C3H mouse strain.13 It is a well-characterized model that is passaged serially in the mouse peritoneal cavity.14-16 After intraperitoneal inoculation, it causes progressive tumor cell–laden ascites and the late formation of peritoneal surface nodules; tumor-bearing mice die of their tumor burden in 20 to 25 days. The tumor cells grow poorly in vitro. The tumor-bearing animals were the gift of Dr Angelo Russo at the National Cancer Institute (Bethesda, Md). With the use of an established protocol,17 tumor cells were passaged serially in the peritoneal cavity of C3HeB/FeJ mice (Jackson Laboratories, Bar Harbor, Me). Briefly, percutaneous paracentesis of tumor-bearing mice was followed by the washing and counting of the tumor cells; 2 3 105 tumor cells per mouse were resuspended in 0.25 mL of Ham’s F-12 medium with 10% fetal bovine serum and injected intraperitoneally into naive 6- to 10week-old, female C3HeB/FeJ mice. All animal experiments were conducted with the approval of the Institutional Animal Care and Use Committee of the University of Pennsylvania.

Surgery April 2004

Adenoviral constructs. The complementary DNA for mouse IFNb was inserted into a replication-deficient adenovirus under the control of the cytomegalovirus promoter (Ad-IFNb). A similar virus that contained the beta-galactosidase gene was constructed (Ad-LacZ) to measure transduction efficiency in this model system and to serve as an adenoviral control treatment. The viruses were constructed in the Vector Core of the Institute for Human Gene Therapy of the University of Pennsylvania Health System. In vivo transduction of tumor cells. Ad-LacZ was injected at various doses intraperitoneally into tumor-bearing mice as a control vector to measure the efficacy and extent of tumor cell transduction. Tumor cells were removed from the peritoneal cavity at various time points after virus administration for LacZ staining. This was performed in the following manner: Tumor cells were plated in 6-well plates and allowed to adhere for several hours before being washed and fixed with .05% gluteraldehyde. Cells were washed and incubated overnight in staining solution that contained 1 mg/mL 5-bromo-4-chloro-3-indolyl-b-D-galactoside (X-gal) in dimethyl-formamide, 5 mmol/L K3Fe(CN)6, 5 mmol/L K4Fe(CN)6.3H2O, and 1 mmol/L MgCl2 in phosphate-buffered saline solution (PBS). The percentage of cells with dark blue cytoplasmic staining was recorded as the average of 3 high-powered fields. With the use of this technique, a time course of in vivo tumor cell transduction at the treatment dose (3.3 3 108 plaque-forming units [PFU]) was determined. Tumor cells were removed from the ascites fluid at 1, 3, and 5 days after treatment, and the tumor cells were stained to identify the proportion of transduced cells. Days 1 through 7 after viral administration, several additional mice were killed, and their livers were frozen in optimal cutting temperature compound, (OCT), cryosectioned, and LacZ-stained, as described earlier. Enzyme-linked immunosorbent assay (ELISA) for mouse IFNb protein expression. At various time points after virus treatment, mouse ascites fluid and serum were collected for the measurement of mouse IFNb protein expression. Peritoneal fluid was centrifuged at 190g, and the supernatant was harvested and stored at ÿ808C until ELISA was performed. Serum was collected by retro-orbital blood collection of anesthetized mice. The blood was allowed to separate at room temperature, and the serum was stored at ÿ808C until it was analyzed. ELISA was performed by standard techniques. Briefly, 96-well plates were coated overnight at 48C

Surgery Volume 135, Number 4

with a monoclonal antibody against mouse IFNb (Yamasa Shoyu Co, LTD, Choshi, Japan). After being blocked with milk, protein standards, serum samples, or ascites supernatant samples were added in duplicate. Anti-mouse IFNb rabbit serum was added, and the samples were developed with a color change reaction (BD-Pharmingen, San Diego, Calif) that was measured by a microtiter plate reader at 450 nm. Mouse IFNb protein standards and anti-mouse IFNb rabbit serum were a gift from Biogen, Inc (Cambridge, Mass). In vivo treatment and toxicity experiments. Mice were injected intraperitoneally with 2 3 105 tumor cells on day 0. On day 5, Ad-IFNb or AdLacZ at a dose of 3.3 3 108 PFU or an equivalent volume of PBS was injected. Mouse weight gain (a surrogate marker for ascites tumor burden) and survival were measured carefully. Tumor cells per milliliter of ascites were counted with the use of a hemocytometer at the time of death. A separate group of animals was killed at 1, 3, and 5 days after virus treatment, and the ascites tumor cell concentration was measured to quantify early treatment responses. In separate experiments, non–tumor-bearing mice were treated with Ad-IFNb, Ad-LacZ, or PBS intraperitoneally to determine the toxicity of the adenoviral vectors. The doses that were used were 3.3 3 108 PFU (the survival treatment dose), 2.0 3 108 PFU, or 1 3 108 PFU of Ad-IFNb or Ad-LacZ (n = 5 per vector per dose). Mice were followed for clinical evidence of toxicity. Those animals that were alive and well at 69 days were killed and autopsied for evidence of toxicity. In addition to gross anatomic examination of each mouse, the livers were cut into 2 pieces for microscopic evaluation. One portion was fixed in 10% buffered formalin, paraffin-embedded, and stained with hematoxylin and eosin for histologic examination by a pathologist. The other portion was frozen in OCT compound (Tissue-Tek, Sakura Finetek USA, Inc, Torrance, Calif), cryosectioned, and stained with oil-red-O for evidence of fatty infiltration. Flow cytometry of ascites fluid to identify immune cell populations. At various time points after virus treatment, animals were killed, the entire volume of ascites fluid was collected, and all cells were isolated from the ascites fluid by centrifugation at 190g. Spleens were also isolated and Dounce-homogenized to serve as an immune cell positive control. Ascites and spleen cells were washed, counted, and resuspended in flow cytometry buffer (PBS with 1% fetal bovine serum and 5mmol/L EDTA). One million cells were used for each flow cytometry sample. Most of the ascites

Hendren et al 429

Fig 1. MOT cell line tumor model. A, An ascites–tumorbearing mouse approximately 20 days after intraperitoneal tumor inoculation beside a naı¨ve mouse. B, Gross peritoneal tumor nodules present at the time of death from untreated MOT cell lines.

cells that were visualized under the microscope were tumor cells by size and histologic appearance. Purified rat anti-mouse CD16/CD32 (Fc-c III/II receptor) monoclonal antibody (0.1 lg/mL; BD Pharmingen, San Diego, Calif) was added to all samples to block nonspecific antibody binding to Fc receptors. Specific fluorescent-conjugated antibodies to cell surface receptors of mouse immune cells were then bound to the cell suspension (5 lg/ mL). Cells were fixed in cold 2% paraformaldehyde overnight and analyzed by flow cytometry. The anti-mouse antibodies used (all from BD Pharmingen) were CD3-FITC, CD4-PE, CD8-APC, Pan-NK-PE, B220-PE, and Mac1-FITC. Isotype control antibodies conjugated to the fluorescent markers FITC, PE, and/or APC were used as negative controls for every ascites sample. To

430 Hendren et al

Surgery April 2004

Fig 2. Adenoviral transduction efficiency in vivo. Photomicrographs are shown of MOT cells 3 days after the Ad-LacZ treatment (A) and of the mouse liver sections (B) on days 1, 3, and 7 after Ad-LacZ treatment. Dark staining, which indicates viral transduction, is present in tumor cells and rarely along the liver capsule. C, The mean percentage of tumor cells were transduced at various time points after intraperitoneal Ad-LacZ treatment at the dose of 3.3 3 108 PFU.

analyze the data, the total immune cell population was selected by creating a ‘‘gate’’ with splenocytes as a positive control. Tumor cells were much larger than immune cells and were excluded by size. CD3/CD4 positive cells were considered to be CD4+ T lymphocytes; CD3/CD8 positive cells were considered CD8+ T lymphocytes; B220 positive cells were considered B lymphocytes; Mac1 positive cells were considered macrophages, and Pan-NK positive cells were considered to be NK cells. The relative proportions of various immune cell types were determined by ‘‘quadrant statistics’’ analysis. The proportion of cells in the quadrant for the isotype negative control was subtracted from the positive proportion for each analysis to control for nonspecific antibody binding. Results are expressed as a proportion of the gated immune cell population. Measurement of nitric oxide (NO) in ascites fluid after treatment. A sensitive recessed-probe microelectrode was used for the direct measure-

ment of NO in ascites fluid in vivo after treatment. The microelectrode was inserted into the ascites fluid of a live, anesthetized mouse through a small abdominal incision. The technical details of the technique that was used have been described elsewhere.18,19 Briefly, nafion-polymer–coated, recessed, gold electrochemical microsensors (tip diameter 5-15 lm) measured polarographic currents at the oxidation potential for NO (+850 mV). Multiple readings were compared with a negative control, namely PBS at 378C. Signals were acquired by a computer with 12-bit resolution at 1-Hz sampling rates, and NO signals were low-pass filtered through an analog circuit with a 5-Hz cutoff. Calibrations for the NO microsensors were made at 0 and 1800 parts per million NO gas before and after every experiment. Statistical analysis. Statistical analysis was performed using GraphPad InStat version 3.06 for Windows (GraphPad Software., San Diego, Calif), except for survival, which was analyzed with the

Surgery Volume 135, Number 4

Hendren et al 431

Statistica software program (Stat Soft Inc, Tulsa, Okla). In the bar graphs, data are presented generally as the mean ± standard error of the mean (SEM). Two-way comparisons were made with the unpaired t- test, except in the case of significantly different standard deviations, in which case the Mann-Whitney U test was used. KaplanMeier survival curves were compared by the logrank test with the use of the Statistica software program. A probability value of .05 or less was considered significant. RESULTS Figure 1 shows the gross appearance of tumorbearing animals alive and after autopsy. A treatment dose of 3.3 3 108 PFU (approximately equal to 5 3 1010 viral particles) was selected on the basis of preliminary experiments that demonstrated early toxicity at a higher dose (data not shown). The determination of the susceptibility of MOT cell line and host organs to adenoviral infection was necessary before gene therapy could be undertaken. Figure 2 shows the positive viral transduction of MOT cells. Selective localization of adenovirus to the liver that results in toxicity has been documented.20-22 Therefore, the degree of liver transduction after intraperitoneal injection of adenoviral vector was determined. Very minimal Ad-LacZ positive staining of the liver capsule occurred and was completely absent in the liver parenchyma, even at a dose twice that used in survival experiments at days 1 through 7 after the treatment (representative photomicrograph, Fig 2). At 1, 3, and 5 days after virus administration, the mean percentage of transduced tumor cells in the peritoneal fluid was 33.6% (n = 3; values, 27.6%, 42.3%, 30.8%, respectively), 25.4% (n = 3; values, 15.3%, 21.8%, 39.2%, respectively), and 22.8% (n = 2; values, 19.5%, 26.2%, respectively; Fig 2). At higher viral doses of up to 1 3 1011 particles, in vivo and in vitro tumor cell transduction efficiency as high as 70% at 3 days was observed. However, as mentioned earlier, toxicity was increased (data not shown). Results of the IFN ELISA are shown in Fig 3. The ascites fluid of Ad-IFNb–treated animals at 1, 2, 3, and 5 days after treatment contained mean IFNb protein levels of 13,034, 12,748, 10,080, and 2910 pg/mL, respectively. Mouse IFNb was undetectable in the serum samples. None of the PBS or Ad-LacZ ascites or serum samples contained detectable levels of mouse IFNb protein. Two of 10 Ad-IFNb animals had no detectable mouse IFNb protein in the ascites fluid; these animals were excluded from analysis. Ascites tumor–bearing

Fig 3. ELISA for IFNb protein production in vivo. High levels of mouse IFNb protein were measured in the ascites of Ad-IFNb–treated animals. Ascites samples from PBS and Ad-LacZ–treated animals and serum samples from all groups had undetectable levels of IFNb protein (n = 3 in all groups).

animals were treated with a single intraperitoneal treatment of Ad-IFNb (3.3 3 108 PFU) 5 days after tumor inoculation. At this time point, mice had tumor-laden ascites with no gross peritoneal surface implants. Three days after treatment, a mean of 2.54 3 107 (±1.2 3 107) tumor cells/mL were present in the Ad-IFNb group, compared with 6.85 3 107(±0.72 3 107) cells/mL in the PBS group and 6.67 3 107 (±1.8 3 107) cells/mL in the AdLacZ group (P = .03; Fig 4). An apparent decrease in the rate of ascites accumulation was also seen in the treatment group compared with controls, as measured by total body weight gain (n = 6 per group; Fig 5). Most importantly, the Ad-IFNb treatment group had an improved survival over the control groups (P < .001; Fig 6), with the median survival increasing from 20 days to approximately 30 days. No antitumor response was seen in the Ad-LacZ— treated animals with the use of any of these measures, which eliminated the possibility that the antitumor response that was seen was a nonspecific immune effect of the adenoviral vector. In a separate experiment, a higher dose of adenoviral vector (3.0 3 109 PFU) was used in an attempt to maximize response and possibly cure the treated animals. However, no improvement in antitumor effect was observed, although toxicity appeared to be increased, with a larger number of early post-treatment deaths (data not shown). To further understand the reason that Ad-IFNb– treated animals eventually died after an initial treatment response, all animals were autopsied at the time of death, and the ascites fluid was

432 Hendren et al

Fig 4. Tumor burden. Five days after tumor inoculation, mice were treated with intraperitoneal Ad-IFNb (n = 7 mice), Ad-LacZ (n = 10 mice), or PBS (n = 7 mice). The mean number of tumor cells per milliliter of ascites is shown at the early time points of 1, 3, and 5 days after treatment.

examined for the presence of tumor cells. Tumor cells were found to be present at the time of death, but at a smaller concentration in the Ad-IFNb treatment group compared with controls, as shown in Fig 6 (P = .02 vs Ad-LacZ). It is interesting to note that none of the Ad-IFNb–treated animals that responded to the therapy with prolonged survival had any peritoneal tumor implants at the time of death, although viable MOT cells were present in the ascites fluid. This finding suggests an inability to form tumor nodules, possibly because of an antiangiogenic effect, although this remains to be proven. At autopsy, animals were also examined for evidence of toxicity, and several virus-treated animals had a grossly rounded or macronodular liver at the time of death. However, this was not a consistent finding. Furthermore, microscopic examination of livers with hematoxylin and eosin and oil-red-O staining revealed no evidence of cirrhosis, hepatitis, fatty infiltration, or other abnormality to explain the animals’ deaths. Most animals showed gross evidence of pancreatic enlargement and bloody ascites, but these characteristics were present in treatment and control animals alike and were not seen in the non–tumor-bearing toxicity study. In an effort to explain the late deaths of our treatment-responsive animals despite a low tumor burden, mice were treated with various doses of AdIFNb or Ad-LacZ (n = 5 animals per group per dose) and followed for evidence of toxicity. The animals were alive and well at 69 days after infection and were killed and autopsied. In addition to a full gross evaluation of the animal at autopsy, the

Surgery April 2004

liver was examined histologically, given the documented potential for liver toxicity with adenoviral vectors.20-22 At higher doses, distinct effects were seen that differed between the 2 viral vectors. At the highest dose of Ad-IFNb (3.3 3 108 PFU), 3 of 5 animals had extensive adhesions of the bowel, which resulted in the death of 1 animal from a clinical small-bowel obstruction. This complication was not observed at the lower doses of Ad-IFNb or in the AdLacZ group. In contrast, in the AdLacZ group, 4 of 10 animals in the higher dose groups had gross abnormalities of the liver, predominantly a macronodular pattern. Three of 4 such animals were alive and well at the time of death. One such animal died at 42 days after viral treatment, although the cause of death is not known definitively. Microscopically, these livers had no evidence of cirrhosis, hepatitis, fatty infiltration, or other abnormality by hematoxylin and eosin staining of formalin-fixed, paraffin-embedded sections and oil-red-O staining of frozen sections. Grossly normal livers from Ad-IFNb– and PBS-treated animals were also examined microscopically with the same techniques, and no abnormalities were observed. The mechanism of the IFNb antitumor effect is multifactorial. In various tumor models, immunologic effects, antiangiogenic effects, and apoptosis have been shown to be important.5-10 As described earlier, no tumor nodules developed in Ad-IFNb–treated animals; therefore, the study of possible antiangiogenic effects is not feasible. Furthermore, TUNEL staining of MOT cells 1, 2, 3, or 5 days after treatment revealed no significant apoptosis, despite several repeated experiments (data not shown). Immunologic effects were therefore the focus of our mechanistic studies. Although immune mechanisms are important in the antitumor effect of IFNb, various immune cell types have been implicated in the mediation of the effect.7-9,10,12 To determine whether Ad-IFNb therapy induced a particular immune cell type to infiltrate the ascites fluid in our tumor model, we performed flow cytometry analysis of the tumor ascites to identify the proportional representation of several candidate immune cell types: CD4 T lymphocytes, CD8 T lymphocytes, B lymphocytes, macrophages, and NK cells. There was a higher proportion of NK cells in the ascites fluid of the AdIFNb group compared with controls at 3 days (P = .02) and 5 days (P = .02) after treatment (Fig 7). Differences were also seen in the macrophage (Fig 7) and CD4 T-lymphocyte populations (data not shown). Comparing Ad-IFNb–treated animals with controls, the proportion of macrophages was

Surgery Volume 135, Number 4

Fig 5. Weight gain over time of the animals in 1 representative experiment (n = 6 mice per group).

higher at 5 days (P = .002). The proportion of CD4 T cells was lower in the Ad-IFNb group compared with controls (data not shown). The production of NO by IFNb-stimulated macrophages in the tumor environment has been proposed as a mechanism of the antitumor effect.9,10 We were interested in determining the role of NO in our model system, given the increase in macrophages in treated animals. With microsensor electrodes that measure polarographic currents at the oxidation potential of NO, it is possible to measure NO directly in biologic systems.18,19 This method is an improvement over the commonly used technique of assaying for the byproducts of NO metabolism (nitrate and nitrite). Ascites fluid NO levels were measured in anesthetized animals through a small abdominal incision 5 days after treatment. The mean NO levels were always greater in the Ad-IFNb–treated animals (Fig 7), although this did not reach statistical significance.

CONCLUSION IFNb has great potential for the treatment of cancers because of its antiproliferative, immunomodulatory and antiangiogenic activities;23,24 however the clinical experience with IFN protein therapy has been disappointing because of rapid protein clearance and systemic toxicities. These limitations may be overcome by IFNb gene therapy. The present study has demonstrated a significant therapeutic effect of Ad-IFNb gene therapy in a unique, immunocompetent preclinical model that mimics human peritoneal carcinomatosis. Tumor burden and survival were improved significantly, although the animals were not cured of disease. Our work and the work of other inves-

Hendren et al 433

tigators have shown that the antitumor effect may involve a variety of immune cells, including T lymphocytes, macrophages, and NK cells.7-10,12 Our study is in agreement with other published findings that show differences in the proportion of NK cells, macrophages, and T lymphocytes in the ascites fluid of treated animals, particularly with the proportion of NK cells. To our knowledge, we are the second group to use Ad-IFNb gene therapy in an immunocompetent model of peritoneal carcinomatosis. Recently, Odaka et al25,26 demonstrated an excellent tumor response in immunocompetent mice with the use of a malignant mesothelioma model of carcinomatosis and a similar Ad-IFNb treatment vector. In their model, approximately 90% of animals achieved long-term survival. CD8 T cells were shown to be essential to the antitumor effect, although the role of NK cells was not found to be significant. The mechanism of the IFNb antitumor effect appears to depend, in part, on the preclinical model that is used, given the variable published findings of different investigators in studies that involve different model systems. Preclinical trials of novel therapies must demonstrate that the therapy is effective in multiple tumor models before trials in human subjects are initiated. Our work is complementary to that of the previous authors, in that a therapeutic effect has been shown in a different preclinical tumor model. Furthermore, this work demonstrates that the ability of the therapy to cure animals of disease is variable, with long-term survival only in the mesothelioma model.25,26 Xie et al9 and Xu et al10 have shown that IFNb treatment is associated with the increased expression of inducible NO synthase in macrophages and the production of NO. In these studies, NO was measured in vitro with an assay that detects nitrite as a byproduct of NO metabolism. We used a unique method for direct measurement of NO using a recessed microelectrode,18,19 and a trend toward greater NO levels was observed in vivo in our animal model after Ad-IFNb gene therapy, although it was not statistically significant. An unanswered question is the cause of eventual death in treated animals after an initial therapeutic response. It is possible that toxicity from the adenoviral vector or from the IFNb protein plays a role. However, in our non—tumor-bearing toxicity study, very few deaths were seen. A combination of sublethal tumor burden and sublethal adenoviral toxicity may result in the eventual deaths. There are several reasons for enthusiasm regarding this approach in the difficult clinical

434 Hendren et al

Surgery April 2004

Fig 6. Survival and total tumor cell counts at time of death. A, The Kaplan-Meier survival curve of all treated animals is shown. B, The tumor burden of treated animals is shown at the time of death (expressed as tumor cells per milliliter of ascites).

problem of peritoneal carcinomatosis. The therapy would most likely be combined with surgical debulking of tumor; therefore, the viral vector could be delivered intraperitoneally at surgery. Local administration of viral vector would be expected to avoid the potential toxicity of intravascular administration of adenovirus. Further, this approach provides a mechanism for the elimination of microscopic residual disease, even in the presence of chemotherapeutic drug resistance. Because the IFNb mechanism of action is multifactorial, the therapy may be efficacious even in the setting of multiple previous chemotherapy treatments. It must be noted, however, that although adenoviruses can be produced in very high titers and can facilitate transfer of large DNA inserts (7-8 Kb) and can infect nondividing cells and produce transient high level expression of the transgene, a major potential drawback of all adenovirusmediated gene therapy efforts that are currently in use is the activation of the host non-specific immune system. Also, neutralizing antibodies may

develop after initial administration of the vector limiting effective readministration. In conclusion, we have shown a significant antitumor effect of intraperitoneal Ad-IFNb gene therapy in an aggressive, immunocompetent mouse model of carcinomatosis. The data suggest that multiple immune cell types mediate the antitumor response. Ad-IFNb gene therapy warrants further study as a treatment for peritoneal carcinomatosis. We thank Allison Kemner for assistance with the animal experiments, Dr Rachel Rapaport Kelz for advice regarding statistical analysis, the Institute for Human Gene Therapy of the University of Pennsylvania for their work in constructing adenoviral vectors, Dr Angelo Russo at the National Cancer Institute for providing our MOT tumor cells, and Biogen, Inc, for providing antibodies and recombinant protein for our ELISA experiments. REFERENCES 1. Loggie BW, Fleming RA, McQuellon RP, Russell GB, Geisinger KR. Cytoreductive surgery with intraperitoneal hyperthermic chemotherapy for disseminated peritoneal

Surgery Volume 135, Number 4

Hendren et al 435

Fig 7. Flow cytometry analysis of immune cells in the ascites fluid and NO measurements in vivo. Fluorescent-conjugated antibody staining for individual cell types included NK cells (A) and macrophages (B). The proportional representation of both cell types in the ascites fluid was shown to be increased significantly after Ad-IFNb treatment. C, A microelectrode method for measuring NO was used to measure NO levels in the tumor microenvironment in vivo.

2.

3.

4.

5.

6.

7.

8.

cancer of gastrointestinal origin. Am Surgeon 2000;66: 561-8. Hendren SH, Hahn SM, Spitz FR, Bauer TW, Rubin SC, Zhu T, et al. Phase II trial of debulking surgery and photodynamic therapy for disseminated intraperitoneal tumors. Ann Surg Oncol 2001;8:65-71. Fish EN, Uddin S, Korkmaz M, Majchrzak B, Druker BJ, Platanias LC. Activation of a CrkL-Stat5 signaling complex by Type I interferons. J Biol Chem 1999;274:571-3. Der SD, Zhou A, Williams BRG, Silverman RH. Identification of genes differentially regulated by interferon a, b, or using oligonucleotide arrays. Proc Natl Acad Sci USA 1998; 95:15623-8. Albini A, Marchisone C, DelGrosso F, Benelli R, Masiello L, Tacchetti C, et al. Inhibition of angiogenesis and vascular tumor growth by interferon-producing cells: a gene therapy approach. Am J Pathol 2000;156:1381-93. Dong Z, Greene G, Pattaway C, Dinney CP, Eue I, Lu W, et al. Suppression of angiogenesis, tumorigenicity, and metastasis by human prostate cancer cells engineered to produce interferon-b. Cancer Res 1999;59:872-9. Lu W, Fidler IJ, Dong Z. Eradication of primary murine fibrosarcomas and induction of systemic immunity by adenovirus-mediated interferon-b gene therapy. Cancer Res 1999;59:5202-8. Qin XQ, Tao N, Dergay A, Moy P, Fawell S, Davis A, et al. Interferon-b gene therapy inhibits tumor formation and causes regression of established tumors in immune-deficient mice. Proc Natl Acad Sci USA 1998;95:14411-6.

9. Xie K, Bielenberg D, Huang S, Xu L, Salas T, Juang SH, et al. Abrogation of tumorigenicity and metastasis of murine and human tumor cells by transfection with the murine IFN-b gene: possible role of nitric oxide. Clin Cancer Res 1997; 3:2283-94. 10. Xu L, Xie K, Fidler IJ. Therapy of human ovarian cancer by transfection with the murine interferon b gene: role of macrophage-inducible nitric oxide synthase. Hum Gene Ther 1998;9:2699-708. 11. Dong Z, Juang SH, Kumar R, Eue I, Xie K, Bielenberg D, et al. Suppression of tumorigenicity and metastasis in murine UV-2237 fibrosarcoma cells by infection with a retroviral vector harboring the interferon-beta gene. Cancer Immunol Immun 1998;46:137-46. 12. Natsume A, Mizuno M, Ryuke Y, Yoshida J. Antitumor effect and cellular immunity activation by murine interferon-beta gene transfer against intracerebral glioma in mouse. Gene Ther 1999;6:1626-33. 13. Fekete E, Ferrigno MA. Studies on a transplantable teratoma of the mouse. Cancer Res 1952;12:438-43. 14. Berek JS, Cantrell JL, Lichtenstein AK, Hacker NF, Knox RM, Nieberg RK, et al. Immunotherapy with biochemically dissociated fractions of Propionibacterium acnes in a murine ovarian cancer model. Cancer Res 1984;44:1871-5. 15. Vanhaelen CPJ, Fisher RI. Requirements for successful immunotherapy and chemoimmunotherapy of a murine model of ovarian cancer. Cancer Res 1981;41:980-3. 16. Vanhaelen CPJ, Fisher RI, Appella E, Ramanathan L. Lack of histocompatibility antigens on a murine ovarian teratocarcinoma. Cancer Res 1981;41:3186-91.

436 Hendren et al

17. Tochner Z, Mitchell JB, Harrington FS, Smith P, Russo DT, Russo A. Treatment of murine intraperitoneal ovarian ascitic tumor with hematoporphyrin derivative and laser light. Cancer Res 1985;45:2983-7. 18. Buerk DG, Riva CE. Vasomotion and spontaneous low frequency oscillations in blood flow and nitric oxide in cat optic nerve head. Microvasc Res 1998;55:103-12. 19. Buerk DG, Atochin DN, Riva CE. Simultaneous tissue PO2, nitric oxide and laser Doppler blood flow measurements during neuronal activation of optic nerve. In: Hudetz A, Bruley D, editors. Oxygen transport to tissue XX, advances in experimental medicine and biology. New York: Plenum Press; 1998. p. 159-64. 20. Gallo-Penn AM, Shirley PS, Andrews JL, Tinlin S, Webster S, Cameron C, et al. Systemic delivery of an adenoviral vector encoding canine factor VIII results in short-term phenotypic correction, inhibitor development, and biphasic liver toxicity in hemophilia A dogs. Blood 2001;97:107-13. 21. Printz MA, Gonzalez AM, Cunningham M, Gu DL, Ong M, Pierce GF, et al. Fibroblast growth factor 2-retargeted adenoviral vectors exhibit a modified biolocalization pattern and display reduced toxicity relative to native adenoviral vectors. Hum Gene Ther 2000;11:191-204.

Surgery April 2004

22. Yee D, McGuire SE, Brunner N, Kozelsky TW, Allred DC, Chen SH, et al. Adenovirus-mediated gene transfer of herpes simplex virus thymidine kinase in an ascites model of human breast cancer. Hum Gene Ther 1996;7:1251-7. 23. Cao G, Su J, Lu W, Zhang F, Zhao G, Marteralli D, et al. Adenovirus-mediated interferon-beta gene therapy suppresses growth and metastasis of human prostate cancer in nude mice. Cancer Gene Ther 2001;8:497-505. 24. Izawa JI, Sweeney P, Perrotte P, Kedar D, Dong Z, Slaton JW, et al. Inhibition of tumorigenicity and metastasis of human bladder cancer growing in athymic mice by interferon-beta gene therapy results partially from various antiangiogenic affects including endothelial cell apoptosis. Clin Cancer Res 2002;8:1258-70. 25. Odaka M, Sterman DH, Wiewrodt R, Zhang Y, Keifer M, Amin KM, et al. Eradication of intraperitoneal and distant tumor by adenovirus-mediated interferon-beta gene therapy is attributable to induction of systemic immunity. Cancer Res 2001;61:6201-12. 26. Odaka M, Rainer W, DeLong PA, Tanaka T, Zhang Y, Kaiser LR, et al. Analysis of the immunologic response generated by AdIFN-b during successful intraperitoneal tumor gene therapy. Mol Ther 2002;6:210-8.