Thyroid Study Unit, Department of Medicine, University of Chicago, Chicago, Illinois 60637
Address all correspondence and requests for reprints to: Dr. Leslie J. DeGroot, Thyroid Study Unit, Department of Medicine, University of Chicago, Chicago, Illinois 60637.
Although most thyroid cancers are cured by surgery and 131I therapy, 1020% of patients die from advanced differentiated and anaplastic tumors. A higher proportion of medullary thyroid cancers (MTC) has an adverse prognosis. Chemotherapy has been relatively ineffective to date, although there is always hope for the future. Thus, alternative approaches, such as gene therapy, are of interest, and thyroid tumors are an especially good target, for two reasons. First, certain gene promoters expressed in the tumors [thyroglobulin (TG), TSH receptor (TSH-R), and calcitonin (CT)] have no or very limited expression elsewhere, and second, if the therapy leads to destruction of all normal thyroid tissue, this would be inconsequential in contrast, for instance, to destruction of the liver or lung. So the crucial questions are whether gene therapy can be effective, and whether it is safe. We believe that the answer to both questions is yes.
Many methods are available, including 1) direct injection of an expressible DNA into tumor, 2) blocking expression of an oncogene by expression of antisense messenger ribonucleic acid, 3) transfection of biopsied host tumor cells in vitro with DNA and reinjection, 4) transduction of cells in vitro with a retrovirus expressing the desired gene and reinjection to the host, and 5) infection in vitro or in vivo of tumor cells using a nonreplicative (or replicative) adenovirus to reexpress a gene lost in dedifferentiation such as p-53, p-10, or sodium/iodide symporter (1).
A currently common strategy is to express herpes simplex virus (HSV)
thymidine kinase (tk) in the tumor, and then administer ganciclovir
(GCV), which is converted by tk into a cell-lethal metabolite. A second
common strategy is to use a virus to transduce expression in the tumor
of a cytokine that augments antitumor immunity. Interferon-,
interleukin-2 (IL-2), IL-12, granulocyte-macrophage colony-stimulating
factor (GM-CSF), and lymphotactin are among the genes employed. Other
strategies target induction of apoptosis or antiangiogenesis
factors.
Before reviewing some of the current preclinical and clinical studies, it may be appropriate to consider the problems associated with gene therapy. In general, adenovirus must be directly injected into tumor if a cytotoxic gene is employed, because when given iv, most of the virus goes to the liver and can in high dosage be fatal (in animals). When injected into a tumor mass, only a small proportion of cells is infected. Thus, it is unlikely that a single injection of a suicide gene can kill all of a tumor. However, expression of a gene that augments immunity to the tumor might lead to rejection of the entire tumor and distant tumor as well. Another benefit of the use of genes that augment immunity is that their effect may persist for months or longer.
Virally mediated expression of genes in nontarget organs after iv administration or leakage from a tumor can be a serious problem. This can be overcome by using cell-specific promoters such as the CT or TG promoters, which have very highly regulated expression but are often weaker than nonspecific promoters, such as the commonly employed cytomegalovirus (CMV) promoter. This is the method we have employed with the TG and CT promoters for treatment of thyroid and medullary thyroid tumors. Similarly, a prostate-specific antigen (PSA)-specific promoter has been used to drive gene expression in PSA-positive prostate cancer cells (2).
Targeting virus specifically to tumor would be very desirable and has been partially accomplished in some systems. Targeting can be achieved to some extent by employing an agent bound to the virus to direct it to a specific receptor on the tumor cell surface. For example, an antibody can be bound to the viral rod protein, and this antibody can be coupled with a tumor-targeting molecule, such as a growth factor or second antibody (3). In another strategy, a phage library displaying linear dodecapeptides was used for biopanning differentiated ciliated airway epithelial cells. After repeated rounds, a peptide with effective binding was recovered, and this peptide was coupled to the surface of an adenovirus using polyethylene glycol. The chemically modified vector could affect gene transfer to human airway epithelial cells (4). Many approaches to this are possible, and in the thyroid field one obvious target is the TSH receptor. Several laboratories are currently investigating targeting vectors to thyroid tumor cells bearing the TSH receptor.
Another problem involves immunity to the vector, such as adenovirus. Most adults are immune to adenovirus, the commonly used vector. However, it appears that this may prevent iv distribution of the virus (actually a benefit), but only somewhat reduces, rather than obliterating, expression in the tumor after direct injection of the vector. Bramson et al. (5) studied the effect of injecting IL-12-expressing adenovirus in animals with preexisting immunity. In naive animals, virus was found disseminating into liver 72 h after injection. Previously, immunized animals had much less expression of the virus in liver because of neutralization by serum antibodies. Expression of the transgene after direct injection of tumor was approximately 50% reduced in previously immunized animals compared with that in nonimmune mice. The data suggest that immunity could actually be advantageous in that it is not a complete block to gene transfer, but does diminish viral dissemination.
A last problem for thyroid researchers has been the lack of good animal tumor models for differentiated thyroid cancer. Fortunately, the opposite situation exists for studies on MTC, as cell lines producing tumors in both rats and mice are available.
Many other new vectors are under development, including vectors based on adeno-associated virus and canarypox (6). Amalfitano et al. (7) constructed a new adenoviral vector with deletions of early 1a region (E1a), E1b, and E3 genes. They note that use of high titers of E1-deleted vectors can result in the production of replication-competent adenovirus derived by recombination with the E1 sequences present in 293 cells used for virus production. The more extensively deleted vectors replicate well and are thought to have much less potential for generation of recombinant-competent virus.
Studies of animal models and human thyroid cells in vitro
As noted, studies of differentiated thyroid cancer treatment have been hampered by the lack of good animal models. Nevertheless, work goes on at a rapid rate in this field. With differentiated thyroid cancer cells it might be possible to direct treatment that would be entirely specific for the tumor cells or residual normal cells using a promoter expressed only in thyroid tissue. Following this approach, Braiden et al. (8) developed a retrovirus expressing HSV-tk under control of the TG promoter. In rat FRTL-5 cells, transduction by the virus plus GCV treatment caused cell death. However, in FRO cells, a human anaplastic thyroid cancer cell line that does not express TG, the viral/GCV treatment had (appropriately) no effect. When an in vitro transduced FRTC cell line (a malignant cell line developed from FRTL-5 cells), was used to produce tumors in severe combined immune deficiency mice, the tumors were rejected during GCV treatment. Nagayama et al. (9) have enhanced the function of the tk/GCV system by combining it with the Cre-LoxP system to enhance the expression of tk. They used the TG promoter to target expression.
Zhang et al. developed an adenoviral vector transducing tissue-specific expression of genes under control of the rat TG promoter (10). Using this promoter to express luciferase activity, infected FRTL-5 (thyroid) cells produced luciferase activity, whereas HEPG2, COS-1, rat and human MTC, HeLa, GH3, P98G, and CA77 cells did not, demonstrating a high degree of specificity of TG promoter function. When the adenovirus with TG promoter driving expression of tk (Ad-TG-tk) was used to transduce FRTL-5 cells, and GCV was added, more than 90% of the cells were killed, and a minimal effect was observed in the other cell lines. When a virus using the nonspecific CMV promoter to express tk (Ad-CMV-tk) was given iv to rats, and animals were treated with GCV, severe liver damage was induced using 1 x 109 to 1 x 1010 plaque-forming units (pfu) of the virus. In striking contrast, when the same amount of adenovirus expressing tk under control of the TG promoter was given iv, no adverse effect on the animals was detected, SGOT levels were equal to those in normal controls, and the liver, lungs, and kidneys showed no histological abnormality. Thus, use of the TG promoter to express a suicide gene (or a cytokine) appears to have a very beneficial effect to lower toxicity. Clearly this vector has to be given by direct injection into the tumor, but a virus expressing a cytokine can be expected to have an effect on tumors at a distance as well as the injected site.
Another strategy, employed by Moretti et al. and Nagayama et al. (11, 12) in a preclinical study, is already being tested in many clinics in phase I or II studies in patients with metastatic tumors. This treatment takes advantage of the fact that most poorly differentiated tumors have lost expression of normal p53 tumor suppressor gene as part of the process of becoming unregulated. A replicating adenovirus (or a nonreplicative form) is used to infect the tumor cells, and expression of the p53 protein leads to apoptosis of tumor cells. Apparently, expression of the gene in normal cells, which already have an intact and expressed p53, causes less, or in some cases no, harm. Nagayama et al. used a replication-defective adenovirus to reexpress p53 in four human anaplastic thyroid cell lines that do not normally express the gene. The tumor cells were destroyed in vitro and partially in vivo in nude mice, whereas cultured normal thyroid cells were more resistant. They also found that the treatment induced apoptosis of the tumor cells and sensitized two cell lines to the chemotherapeutic effect of doxorubicin both in vitro and in vivo. This approach clearly offers potential for the therapy of anaplastic cancers, with the caveat that it may be difficult to transduce all cells by direct tumor injection.
Nagayama et al. elaborated on this system during a recent presentation at the 12th International Thyroid Conference (13). To develop an adenovirus system that can replicate exclusively in wt-p53-deficient cells, they have infected the cells with two recombinant viruses. One has an expression unit of the synthetic p53-responsive promoter and the cAMP response element (Cre) recombinase gene. The other has a constitutive CA promoter and E1A gene flanked by a pair of loxP elements, followed by E1B19K gene under control of the CMV promoter. When the viruses are coinfected into cells with wild-type p53 expression, Cre would be activated, excise out the E1A gene, and switch off E1A expression, so that the second virus would not replicate. However, in cells with a mutant p53, the p53 response element would not be activated, E1A would be expressed and lead to virus replication and cell lysis. They use this system to show that the double infection caused viral replication and a cytopathic effect only in cells not expressing wild-type p53.
Scala et al. developed an adenovirus carrying the HMGI(Y) gene in an antisense orientation. This virus induced cell death in two human thyroid anaplastic carcinoma cell lines (ARO and FB-1), but not in normal thyroid cells, and also killed a variety of other tumor cell lines (14).
Although it is logical to use a thyroid-specific promoter such as TG in directing gene therapy to thyroid cancer cells, clearly some thyroid tumors lose expression of TG as they progressively undifferentiate. Chun et al. and Zeiger transfected non-TG-expressing ARO and WRO thyroid cell lines, which had low intrinsic TG, thyroid transcription factor-1, and PAX-8 gene expression. Transfection of these cells with the two genes known to be involved in thyroid cell differentiation, thyroid transcription factor-1 and PAX-8, resulted in increased function of the TG promoter. This is a model that might be important in gene therapy, because it suggests a way to reintroduce expression of a specific gene in thyroid cells (15).
Gene therapy of MTC
Progress with animal models of MTC has been more rapid. Soler et al. studied the effect of the tk/GCV system and IL-2 secretion in the Wag/Rij rat MTC model (16). These researchers used a retrovirus to transduce expression of the desired genes in vitro, and then injected the cells into the host. Presumably, such cells are nearly totally transduced by the virus during in vitro infection, in contrast to the less effective transduction when the virus is injected into the tumor in vivo. When the animals were treated with GCV, tumors were eradicated in 60% of animals. In subsequent studies, cells that had been transduced with the tk-producing retrovirus, and cells that had been transduced with an IL-2-producing virus were mixed and inoculated. The combination of the suicide gene and IL-2 gene produced superior tumor inhibition. This system is obviously complicated, as it requires in vitro transduction of the tumor cells by the retrovirus and subsequent reinjection. Lausson et al. (17) also reported effective control of this model tumor using a vector producing IL-2.
Zhang and co-workers studied adenovirally mediated treatment of a rat MTC model tumor in Wag/Rij rats. An adenovirus expressing the viral tk gene under control of the CMV promoter was used for treatment. Adenovirus (2 x 109 pfu) was injected into MTC tumors of less than 100 mm3 volume, and animals were treated with GCV ip 24 h later at a dose of 100 mg/kg/day for 7 consecutive days. In these small tumors, the viral/GCV treatment caused destruction or stabilization of the tumor. However, in larger tumors, averaging 400 mm3, the treatment was not effective (18). These tumors allow little bystander killing of neighboring cells (18). To get around this problem, the researchers developed a replication-defective adenovirus that produced the mouse IL-2 (mIL-2) gene under control of the CMV promoter. The studies were carried out in a mMTC tumor system in syngeneic BALB/c mice. Tumors less than 30 mm3 were destroyed after injection of 1 x 109 pfu of virus, and larger tumors were stabilized (19). Mice that were cured of their tumor by treatment with the Ad-CMV-mIL2 virus did not develop tumors after reinjection of wild-type MTC cells, indicating that long-term immunity developed. Histological and immunological studies revealed massive CD4+ and CD8+ T cell infiltration in the tumors. Thus, IL-2 was shown to induce effective antitumor immunity in this model system. Similar responses were found using a rat medullary tumor line (20).
To study dissemination of the virus and toxicity, adenovirus expressing lac-z under control of the CMV promoter was injected into tumors at a dose of 1 x 109 pfu. Functional virus was detected by ß-galactosidase activity in the tumor during the subsequent 5 days and in the liver on the fifth and seventh days after treatment, but not in lung, spleen, or kidney. Liver is known to be the main site at which viral particles in the blood are taken up. However, after iv administration of doses of 2 x 109 to 1 x 1010 pfu of the Ad-CMV-mIL-2 vector, there was no elevation of SGOT or SGPT. Thus, intratumoral administration of Ad-CMV-mIL-2 appears to be safe, considering the minimal evidence of liver toxicity, even after administration of 1 x 1010 pfu directly iv, which is presumed to lead to deposition of more than 95% of the virus in the liver (20).
As both the tk/GCV system and the IL-secreting virus independently have antitumor effects, it is logical to combine the two treatments. The tk/GCV suicide gene treatment is believed to act by killing transfected cells, ideally killing neighboring cells through the bystander effect, and to lead to the development of antitumor immunity through liberation of tumor antigens. The cytokine treatment leads, through local expression of cytokines, to targeting of cytotoxic T cells and natural killer cells to the area of the tumor and to activation of these cells against the tumor. To express both genes in one vector, both Soler et al. (16) and Zhang and DeGroot (21) developed an adenoviral vector secreting tk and human IL-2 under control of CMV promoter in two separate expression cassettes within the same virus. The genes were inserted into the E1 region of adenovirus type 5. Rat MTC cells infected with the virus showed high in vitro sensitivity to ganciclovir and produced large amounts of mIL-2. Two thirds of treated tumors were destroyed when this vector was injected intratumorally into established tumors. This represented an increment over the function of either the Ad-CMV-IL-2 or the Ad-CMV-tk virus used individually.
Many cytokines have been tested for antitumor activity to date. These
include tumor necrosis factor-, interferon-
, IL-2, IL-12, and
GM-CFS. IL-12 is a heterodimeric cytokine that causes proliferation of
natural killer cells and CD8+ T cells. It also
activates macrophages through lymphocyte IFN-
synthesis and
stimulates differentiation of naive CD4+ T cells
into Th1 cells. Systemic administration of recombinant IL-12 has been
shown to inhibit tumor growth and produce an antitumor response, but
causes severe dose-dependent toxicity, in animals and humans. Tahara
et al. studied the use of a retroviral vector producing
IL-12 and showed its efficacy in a murine tumor model
(22), and another group used gene gun-mediated transdermal
transfection of metastatic murine tumors with IL-12 DNA, and this
resulted in regression of established tumors (23). Zhang
and DeGroot developed an adenovector that has both the P35 and P40
subunits of mIL-12 expressed under the control of CMV promoters in
separate cassettes inserted in the E1 region of adenovirus 5. Direct
injection of the vector into established tumors using 1 x
109 pfu eradicated most tumors and produced
long-term immunity to challenge with naive MTC tumor cells at 8 weeks
after the original cure. This viral system appeared to be the most
effective system examined to date, comparing the Ad-CMV-tk,
Ad-CMV-IL-2, and Ad-CMV-IL-12 vector systems. Perhaps it shows the
greatest promise for use in human studies (24).
Other studies of IL-12 in preclinical gene therapy
Oshikawa et al. (25) used a gene gun to transcutaneously implant various combinations of plasmids expressing IL-12, pro-IL-18, and IL-1-converting enzyme. They studied the ability of the directly injected DNA to induce the production of cytokines and to cause regression of sc implanted mammary adenocarcinoma in mice. A mixture of the three complementary DNAs produced the most potent antitumor activity. Davidoff et al. (26) directly injected an adenoviral vector encoding both subunits of murine IL-12 into sc neuroblastoma tumors in mice. A nonreplicative E1- and E3-deleted virus with a CMV promoter driving the expression was used in this study. A dose of 1 x 109 pfu of the virus was injected. Seventy-two percent of mice given the specific vector had tumor regression, and 48% completely rejected their tumors. All of the mice that rejected tumor were rechallenged with unmodified tumor at a later date and rejected the tumors. Siders et al. (27) used the Ad-CMV-IL-12 virus in a murine hepatic metastatic tumor and gave the virus systemically. These animals were given 5 x 107 pfu/injection on 4 days. They found a strong infiltrate of T cells, macrophages, and neutrophils around the tumor. The virus was also effective in animals depleted of CD4+ or CD8+ cells and in severe combined immune deficiency mice, suggesting that the inhibition is mediated by nonlymphocytic effector cells, including macrophages and neutrophils, and could involve antiangiogenic chemokines. The addition of virus expressing GM-CSF was found to enhance the antitumor activity of viruses expressing tk and mIL-2 in a study by Chen et al. (28). Local expression of GM-CSF in the hepatic tumor model and prolonged mIL-2 expression generated persistent antitumor immunity and animal survival.
The overall impact of these studies is that vectors expressing IL-12 are highly effective in inducing antitumor immunity. The exact mechanism is not clearly defined and may involve factors in addition to activation of macrophages and lymphocytes. Prior immunity does not prevent the effect of direct intratumor treatment with adenovirus.
In summary, these preclinical studies in cell lines in vitro and animal models demonstrate clearly the effectiveness of the tk/GCV suicide gene strategy and the ability of immune-enhancing gene expression (especially IL-12) (29, 30) to facilitate tumor destruction in vitro. The use of cell-specific promoters greatly limits toxicity, although still requires direct tumor injection. Leakage of virus using a cell-specific promoter into the bloodstream appears to have little adverse effect.
Selected human trials with adenovirus-mediated gene therapy in patients
To date there have been no published reports of gene therapy for human thyroid cancer, but there are many ongoing studies involving other tumors. Sterman et al. (31) conducted a phase I trial of adenovirus-mediated intrapleural HSV-tk/GCV gene therapy in patients with mesothelioma. A replication-incompetent adenoviral vector containing the HSV-tk gene under control of the Rous sarcoma virus promoter-enhancer was introduced into the pleural cavity of patients with malignant mesothelioma followed by 2 weeks of systemic therapy with GCV at a dose of 5 mg/kg twice daily. Side-effects were minimal and included fever, anemia, transient liver enzyme elevations, and bullous skin eruptions as well as a temporary systemic inflammatory response in those receiving the highest dose. Strong intrapleural and intratumoral immune responses were generated. This study demonstrates that intrapleural administration of an adenoviral vector containing the HSV-tk gene is well tolerated and results in detectable gene transfer when delivered at high doses.
Klatzmann et al. (32) used an HSV-tk/GCV system in which allogeneic M11 cells were transduced by retrovirus in vitro and injected into the surgical cavity after debulking of glioblastoma. Despite extensive surgery for glioblastoma, residual tumor cells always lead to relapse. After a 7-day transduction period, GCV was administered for 14 days. Twelve patients with recurrent glioblastoma were treated without serious adverse events. Twenty-five percent of the patients survive longer than 12 months. At 4 months after treatment, 4 of 12 patients had no recurrence. One patient was still free of detectable recurrence, steroid free and independent, 2.8 yr after treatment. Thus, injections of M11 retroviral vector cells producing tk were well tolerated and were associated with significant therapeutic responses.
Stewart et al. (33) conducted a phase 1 study using an E1- and E3-deleted adenovirus encoding IL-2 (Ad-CA-IL-2), directly injected into sc deposits of melanoma and breast cancer. Local inflammation was observed at the site of injection in 60% of patients, but side-effects were otherwise minor. Incomplete local tumor regression occurred at the site of injection in 24% of patients, but no conventional clinical responses were seen. Circulating CD4+ and CD8+ T cell counts fell significantly 24 h after injection. IL-2 was detectable by ELISA in tumor biopsies. No Ad5E1 sequences were detected either before or after injection, indicating the absence of replication-competent virus. Antiadenovirus and neutralizing antibody titers were elevated 1 month after injection in all patients. This trial, therefore, confirms the safety of use of adenoviral vectors for gene delivery in humans and demonstrates successful transgene expression even in the face of preexisting immunity to adenovirus.
Palmer et al. (34) used a recombinant retrovirus to transduce expression of IL-2 in samples of patient melanoma cells. These IL-2-secreting tumor cells were then used to vaccinate the donors. Twelve patients were vaccinated sc 1, 2, or 3 times with approximately 107 irradiated, autologous, IL-2-secreting tumor cells. Treatment was well tolerated, with local reactions at 11 of 24 injection sites and minor systemic symptoms of fever and headache after 6 injections. One patient developed antitumor delayed-type hypersensitivity after the first vaccination and showed an increased response after the second vaccination. Three patients had stable disease for 715 months, and 1 had not progressed after 15+ months. Thus, patient vaccination with autologous, genetically engineered tumor cells is feasible and safe. Antitumor delayed-type hypersensitivity and cytotoxic T lymphocytes can be induced in some patients with such a vaccine.
Herman et al. (35) also studied direct in situ gene therapy for adenocarcinoma of the prostate using a replication- deficient adenovirus expressing tk and administration of GCV. Patients received injections of increasing concentrations of virus into the prostate under ultrasound guidance. GCV was then given iv for 14 days (5 mg/kg every 12 h). Eighteen patients received 1 x 108 to 1 x 1011 IU. All cultures of blood and urine specimens were negative for growth of adenovirus. One patient at the highest dose level developed spontaneously reversible thrombocytopenia and hepatotoxicity. Three patients achieved an objective response, documented by a fall in serum PSA levels by 50% or more, that was sustained for 6 weeks to 1 yr. Further trials are underway to identify the optimal distribution of vector within the prostate and to explore the safety of repeat courses of gene therapy.
In an important study, Shalev et al. (36) reviewed the treatment of 52 patients who received multiple cycles of gene therapy using an adenovirus-expressing tk by direct injection into prostate cancer. Toxicity increased from 35% up to 75%, in patients who received from 24 cycles of therapy, but all toxic events were mild and resolved completely. No additive toxicity was noted. Results for 28 patients thoroughly studied indicated a mean decrease of 44% in PSA in 43% of the patients, showing that direct injection of HSV-tk, followed by iv GCV, was safe and effective, even in multiple trials.
A problem facing gene therapy is to deliver the virus into sites away from the skin. Computer-aided tomography (CAT)-guided direct needle injection into a tumor has been studied for the treatment of non-small cell lung cancer by Kauczor et al. (37). In a prospective clinical phase I trial, six patients with non-small cell lung cancer and a mutation of the tumor suppressor gene p53 were treated by CAT-guided intratumoral gene therapy. Ten milliliters of a vector solution (replication-defective adenovirus expressing wild-type p53 cDNA) were injected under CAT guidance. The CAT-guided gene therapy was easily performed in all six patients without intervention-related complications. Besides flu-like symptoms, no significant adverse effects of gene therapy were noted. After 28 days, four of the six patients showed stable disease at the treated tumor site, whereas other tumor manifestations progressed. Computed tomography-guided injections are an easy to perform procedure for intratumoral gene therapy.
Clearly gene therapy can be applied safely and effectively to human tumors by injection into pleural or tumor cavities or by direct tumor injection. The dosage of virus indicated in these trials caused at most mild side-effects. We can anticipate application in the near future of some of the methods now in preclinical trials to human thyroid tumors that are beyond control by conventional therapy.
Received September 18, 2000.
Revised February 23, 2001.
Accepted February 28, 2001.
References