Review article: gene therapy, recent developments and future prospects in gastrointestinal oncology


  • This uncommissioned review article was subject to full peer-review.

Dr Y. Touchefeu, Section of Cell and Molecular Biology, Targeted Therapy Laboratory, The Institute of Cancer Research, 237 Fulham Road, London SW3 6JB, UK. E-mail:


Aliment Pharmacol Ther 2010; 32: 953–968


Background  Gene therapy consists of the introduction of genetic material into cells for a therapeutic purpose. A wide range of gene therapy vectors have been developed and used for applications in gastrointestinal oncology.

Aim  To review recent developments and published clinical trials concerning the application of gene therapy in the treatment of liver, colon and pancreatic cancers.

Methods  Search of the literature published in English using the PubMed database.

Results  A large variety of therapeutic genes are under investigation, such as tumour suppressor, suicide, antiangiogenesis, inflammatory cytokine and micro-RNA genes. Recent progress concerns new vectors, such as oncolytic viruses, and the synergy between viral gene therapy, chemotherapy and radiation therapy. As evidence of these basic developments, recently published phase I and II clinical trials, using both single agents and combination strategies, in adjuvant or advanced disease settings, have shown encouraging results and good safety records.

Conclusions  Cancer gene therapy is not yet indicated in clinical practice. However, basic and clinical advances have been reported and gene therapy is a promising, new therapeutic approach for the treatment of gastrointestinal tumours.


Despite recent progress, there is an important need for new therapeutic agents in gastrointestinal oncology. Gene therapy consists of the introduction of genetic material into cells for a therapeutic purpose. The initial aim of the field of gene therapy was the correction of inherited genetic diseases by providing to the targeted cells a functional copy of the deficient gene responsible for the disease. However, gene therapy can also be applied to the treatment of acquired diseases, in particular cancers. Indeed, more than 1500 gene therapy clinical trials have been recorded worldwide and about two-thirds of these concern cancers ( In gastrointestinal oncology, most approaches have targeted liver, colon and pancreatic tumours, and this review will focus on recent knowledge concerning the use of gene therapy in the treatment of these cancers.

Literature search

A literature search was performed using the PubMed database, limited to documents published in the English language. Searches were not date-restricted. Search terms included free text words and combinations of the following Mesh terms: [Digestive system neoplasms], [Colorectal Neoplasms], [Liver Neoplasms], [Pancreatic Neoplasms], [Gene Therapy], [Genetic Vectors] and [Clinical trial]. The ‘Gene Therapy Clinical Trial Worldwide’ website was also used to search for clinical trials.

Principles of gene therapy: selection of a gene, a vector and a management strategy

Selection of the gene

Mutant gene correction.  The principles of gene selection strategies are illustrated in Figure 1. In the case of inherited monogenic diseases, the aim of gene therapy is to transfer and express the defective gene. The situation is more complex in cancer gene therapy because cancer most often results from sequential genetic and epigenetic alterations, affecting oncogenes, tumour-suppressor genes and microRNAs. One gene therapy approach is, thus, to restore tumour-suppressor gene expression or to inhibit oncogene expression. About 11% of transferred genes in gene therapy clinical trials are tumour-suppressor genes and many trials have been performed in cancer gene therapy using the p53 gene, mostly including patients with lung or head and neck cancers (

Figure 1.

 Principles of gene therapy. (a) Gene re-expression. The vector carries a wild-type version of a mutant gene into the tumour. (b) Suicide therapy. A combination of the systemic administration of a nontoxic prodrug and the tumour-specific delivery of the prodrug-activating enzyme gene. (c) Immune therapy. The vector carries a gene encoding a specific immunogenic tumour antigen (1), a co-stimulatory signalling molecule (2) or an inflammatory cytokine (3), leading to immune stimulation. (d) Oncolytic viruses. Oncolytic viruses specifically replicate in, and kill, tumour cells. (e) Therapeutic RNA interference. The RNAi (siRNA) or the RNAi gene (shRNA) is delivered into the cell. After maturation and incorporation in the RNA-induced silencing complex (RISC), RNAi can bind an oncogene RNA, leading to oncogene repression and cell death.

Suicide genes.  The aim of suicide gene therapy is to enable, selectively, the transfected cell to transform a prodrug into a toxic metabolite, resulting in cell death. The most widely described suicide gene is the herpes simplex virus thymidine kinase (HSV-tk) gene. HSV-tk can phosphorylate ganciclovir, which is a poor substrate for mammalian thymidine kinases. Ganciclovir can therefore be transformed into ganciclovir triphosphate, which is cytotoxic to the transfected cell, resulting in cell death.1 This cell death can also affect neighbouring cells that do not express HSV-tk. This phenomenon is called a local bystander effect, as opposed to a bystander effect that can be observed in distant, nontransduced tumour sites.2 This distant bystander effect involves the immune system.

Immunotherapy.  Cancer immunotherapy has been developed to stimulate immune response against tumour cells. Gene therapy can be used to transfer genes into tumour cells to render them more highly immunogenic. Gene transfer of tumour-specific antigens, co-stimulatory molecules and/or inflammatory cytokines has been assessed. Tumour-associated antigens can be recognized by T lymphocytes. These antigens can be derived from oncogenic viruses (Epstein–Barr virus, human papillomavirus) or can be self-antigens. These self-antigens can be overexpressed antigens or antigens that are altered by virtue of a gene mutation or a post-translational modification. They can also be onco-foetal antigens, such as 5T4 antigen and carcinoembryonic antigen (CEA).3 Immunization against a specific antigen can induce cellular and/or humoral immune responses. T-cell activation requires not only the interaction between major histocompatibility molecules bearing a specific peptide and the T-cell receptor, but also non-antigen-specific co-stimulatory activation by interaction of molecules expressed on the T-cell and the antigen presenting cells (such as interactions between CD28 and CTLA-4 expressed on the T-cell and B7 expressed on the antigen presenting cells). Finally, vectors encoding inflammatory molecule genes [such as interleukin-2, interleukin-12, TNF-α, interferon-γ, granulocyte macrophage colony-stimulating factor (GM-CSF)] have been engineered. In terms of the site of gene transfer, immunization can be carried out in situ in the tumour or at a distant site.4

RNA interference.  RNA interference is a promising new therapeutic approach for many diseases, including cancer. MicroRNAs (miRNA) are short, endogenous ∼22 nucleotide RNAs. They can regulate gene expression at the post-transcriptional level by binding to the 3′-untranslated region of the target mRNA, resulting in either mRNA degradation or inhibition of translation.5, 6 They act as fine-regulators of the proteome.7 MiRNAs can have oncogenic activities when they are upregulated and target tumour suppressor genes.8 miRNAs can also have a tumour suppressor potential.9 Furthermore, miRNAs can modify the response to therapeutic agents. For example, the Let-7 family of miRNAs can modify the response of cancer cells to radiation therapy.10 The endogenous RNA interference pathway has been exploited to develop other RNA interference molecules: synthetic, exogenous, double-stranded, short, interfering RNA (siRNA) and vectors expressing short hairpin RNAs (shRNA). As opposed to siRNAs, shRNAs are synthesized in the nucleus and use maturation pathways similar to the miRNA maturation pathways.11 As cancer cells present many well-known RNA interference targets, there are multiple opportunities for therapeutic gene silencing in oncology.12

The choice of the vector.  The effectiveness of gene therapy is highly dependent on the efficacy of gene transfer. The vector has to be safe, has to protect the genetic material from degradation in the extracellular environment, and must release the genetic material to the target cell.

Viral vectors.  Recombinant viruses have been shown to be efficient for gene transfer both in vitro and in vivo. Many viruses such as adenoviruses, adeno-associated viruses (AAVs), retroviruses, lentiviruses, herpes viruses, poxviruses, measles viruses, simian virus 40 recombinant (SV40r) and vesicular stomatitis viruses have been proposed for preclinical studies (Table 1). Other viruses are used as virotherapy, rather than gene therapy agents, because they cannot be genetically modified (reoviruses) or because they have not been modified with the insertion of a therapeutic transgene. Data and clinical trials related to these viruses are not reported in this review. Viruses can be either replication-defective or replication-competent. To engineer replication-defective viruses, one or several genes required for viral replication can be deleted. Vectors that are able to replicate selectively in cancer cells would be of great interest in oncology. A new strategy involving so-called ‘oncolytic’ viruses has been proposed. In this context, the aim is to generate a pathogen for the tumour, which is unable to propagate in, and to harm, normal tissues. Two main characteristics of oncolytic viruses are (i) they replicate selectively in cancer cells and have self-amplification properties; and (ii) they have cancer-cell-specific toxicity.13 Some viruses, such as reoviruses, have natural, inherent tumour selectivity.14, 15 From a general point of view, oncolytic viruses can be engineered in three ways. (i) Viruses can be deleted for genes that are necessary for their replication in normal cells, but not in cancer cells. (ii) Another possibility is to engineer a virus with genes under the control of a tumour-specific promoter. (iii) Viruses can also be engineered to bind to specific tumour antigens on the cell surface.13, 16 In the first two approaches, the aim is to restrict the replication of the viruses, whereas the third strategy aims at selective infection of cancer cells. However, viral vectors have many limitations. There is a limiting immunogenicity and there are safety considerations, including a possible insertional mutagenesis for retroviruses and lentiviruses.

Table 1.   Some advantages and disadvantages of different gene therapy vectors
Viral vectors
 AdenovirusLarge transgene insert capacity
Biologically safe
High immunogenicity
 Herpes simplex virusLarge transgene insert capacity
Availability of anti-herpetic drugs
 Adeno-associated virusesStable transgene expression
Reduced immunogenicity
Small transgene insert size capacity
Need for helper viruses during manufacture
 RetrovirusStable transgene expression
Only infects dividing cells
Small transgene insert size capacity
Possible insertional mutagenesis (integration)
 LentivirusInfects both dividing and nondividing cellsPossible insertional mutagenesis (integration)
 Vaccinia virusLong history of safe human use
Large transgene insert capacity
High immunogenicity
Productive infection in immune suppressed patients
Replication in skin lesions (eczema, psoriasis)
 Vesicular stomatitis virusSelective replication-competence in cells with defective interferon response (tumour cells)High immunogenicity
Animal pathogen (safety/environmental concerns)
 Measles virusLong history of safe human use (Edmonston vaccine strain)Most adults are immune
Wild-type virus is immunosuppressive
Rare measles-like illness with vaccine strains
Nonviral vectors
 Naked DNA‘Easy’ engineering
Low immunogenicity
Rapid clearance, low transfection efficiency in vivo
 Synthetic vectorsLarge gene carrying capacity
Nanoparticles can accumulate into tumours as a result of the enhanced permeability and retention effect
Cationic liposomes have an inflammatory toxicity and a low transfection efficiency in vivo

Nonviral vectors

Naked DNA.  Naked or plasmid DNA has a rapid clearance and a low cellular uptake. As a consequence, the main application of naked DNA injection is in vaccination and immunotherapy strategies. Electroporation involves the generation of a transient increase of the permeability of cell membranes using electric pulses and is more efficient than a simple injection of naked DNA. Electroporation increases the transfer efficiency of drugs, DNA or RNA into the cells, in vitro and in vivo in the case of intratumoural injections.17–19 However, the safety and efficiency of this procedure need to be assessed in humans.

Synthetic vectors.  Cationic liposomes comprise a lipid bilayer membrane. They are amphiphilic molecules and can interact with negatively charged DNA, hereby making it compact and protecting it from circulating endonucleases. The structure generated upon incubation of DNA with liposome is called a lipoplex. Cationic lipoplexes can bind to negatively charged cell membranes and induce cellular uptake. However, in vivo transfer efficiency remains low.20

Blood vessels in tumours can have gaps between endothelial cells, up to 800 nm in size. Nanoparticles can extravasate through these gaps and accumulate selectively in tumours. This enhanced permeability and retention (EPR) effect could explain the tumour selectivity of several gene therapy vectors.21 Nanoparticles are small-sized vectors that can be lipoplexes, other polymers, or metal- and ceramic-based. They can be used as gene therapy vectors and can accumulate in tumours as a result of the EPR effect.22, 23

Biological nonviral vectors.  Bacteria can be used as gene therapy vectors. They can be used as gene transfer vectors or protein delivery systems when the therapeutic transgene is already expressed in the bacterium.24 In addition, many mammalian cell types can be used as carriers of gene therapy vectors.25 Haematological and progenitor mesenchymal cells are promising agents for use in cancer gene therapy. For example, CIK (cytokine-induced killer) cells can be obtained from mouse splenocytes or human peripheral blood, after in-vitro stimulation and culture with interferon-γ, interleukin-2 and anti-CD3 antibodies. These cells naturally target tumours and have cytotoxic effects. They have been combined with the oncolytic vaccinia virus and a synergistic antitumour effect has been observed in mouse models.26

Choice of strategy: in vivo vs. ex vivo.  The ex vivo strategy consists of collecting a cell type or its precursor from the subject, culturing and transducing the cells with the vector and then reintroducing the genetically modified cells into the subject. The in vivo strategy consists of direct injection of the vector into the subject (either intratumourally or systemically). The ex vivo strategy allows control of the transduction and limitation of dissemination of the vector. The in vivo strategy does not require a cell culture step, but cell targeting and transduction efficiency are more difficult to control.

Gene therapy in gastrointestinal oncology: clinical trials and future prospects

Liver tumours

Clinical trials.  Based on preclinical data, several clinical phase I and phase II trials concerning gene therapy of liver tumours have been published (Table 2).27–34 Re-expression of tumour-suppressor genes has been evaluated in vitro, in vivo and in clinical trials, mainly with vectors encoding wild-type p53. In a recently published, phase II trial, 46 patients with unresectable hepatocellular carcinoma were randomized to either group 1 (multiple hepatic arterial injections of 5-FU and a recombinant adenovirus encoding the p53 gene, after transcatheter arterial chemoembolization (TACE)) or group 2 (TACE alone). Both treatments were safe. Partial response or stable disease was reported for 69.5% of subjects in group 1 and 65.2% of subjects in group 2. Times to progression were 9.6 months in group 1 and 8.3 months in group 2. Overall survival times were 12.8 months in group 1 and 10.4 months in group 2.34

Table 2.   Clinical trials concerning gene therapy in primary liver tumours
StudyClinical phaseNo. patientsClinical settingVector and geneTreatment protocolComments
  1. PR, partial response; SD, stable disease; PD, progressive disease; RFS, recurrence free survival; OS, overall survival.

Habib et al.27II Randomized10Hepatocellular carcinomaE1B-deleted adenovirusIntratumoural ethanol (group 1) or adenovirus (group 2) injectionsTumour response assessment: 2 SD and 3 PD in group 1, 1 PR and 4 PD in group 2
Makower et al.28II19Advanced or metastatic hepatobiliary cancerONYX-015 adenovirusIntratumoral injectionsGood safety records with two grade 4 toxicities likely related to disease progression
Sixteen patients assessed for response: 1 PR, 12 SD, 3 PD
Palmer et al.29I18Resectable primary or metastatic liver cancersAdenovirus encoding nitroreductase (prodrug activating enzyme)Intratumoural injection, 3–8 days before hepatic resectionGood safety records
Transgene expression in the tumour
Sangro et al.30I21Advanced liver, colorectal and pancreatic cancersIL-12-encoding adenovirusIntratumoural injectionsGood safety records
Ninety-nine patients assessed for response at day 30: 1 PR, 10 SD, 8 PD
Mazzolini et al.31I17Nine patients with primary liver tumour, 5 with colorectal cancer, 3 with pancreatic cancersAutologous dendritic cells transfected with an adenovirus encoding the IL-12 geneIntratumoural injectionsGood safety records
11 patients assessed for response: 1 PR, 2 SD, 8 PD
Li et al.32II Randomized45Hepatocellular carcinoma, adjuvant therapy after liver transplantationHSV-tk-encoding adenovirusGroup 1: Liver transplantation + post-operative systemic epirubicin injections
Group 2: Same procedure + post-operative peritoneal injections of HSV viruses and systemic ganciclovir injections
RFS and OS at 2 years respectively: 9.1% and 19.9% (Group 1), 43.5% and 69.6% (Group 2)
Park et al.33I14Advanced primary or metastatic liver cancersOncolytic vaccinia virus, JX-594Intratumoural injectionsGood safety records
10 patients assessed for response: 3 PR, 6 SD, 1 PD
Tian et al.34II Randomized46Advanced hepatocellular carcinoma, in association with transcatheter arterial chemoembolization (TACE)Adenovirus encoding the p53 geneGroup 1: TACE alone
Group 2: adenovirus + 5-FU hepatic infusions
OS: 10.4 months (group 1) vs. 12.8 months (Group 2)

Other gene therapy trials have considered suicide therapy strategies. A recent, phase II clinical trial compared patients with hepatocellular carcinoma and tumours bigger than 5 cm, treated by liver transplantation, with or without adenoviral-mediated HSV-tk therapy. During transplantation surgery, adenoviruses encoding HSV-tk were injected into the peritoneal cavity around the liver and ganciclovir therapy was started 24 h after surgery and continued for 10 days afterwards. Forty-five patients were included in the trial, 22 received liver transplantation only and 23 received liver transplantation and gene therapy. Recurrence-free survival and overall survival were significantly higher in the gene therapy group, particularly in patients without vascular invasion before treatment.32

A phase I trial included 21 patients (nine patients with primary liver cancers, five with metastatic colorectal cancers and seven with pancreatic cancer) in an immunotherapy approach. Intratumoural injections of an adenovirus encoding the interleukin 12 gene were well tolerated, and a partial, objective remission was observed in one patient with hepatocellular carcinoma.30 Another phase I trial included 17 patients (nine patients with primary liver cancers, five with colorectal cancers and three with pancreatic cancer). Patients were treated with intratumoural injections of autologous dendritic cells transfected with an adenovirus-encoding IL-12 gene. The treatment was well tolerated. It was associated with a marked increase of infiltrating CD8+ T lymphocytes in three of 11 tumour biopsies analysed. Eleven patients could be assessed for response. One partial response was observed in a patient with pancreatic carcinoma. Stable disease was observed in two patients and progression in eight patients.31

Oncolytic viruses have also been evaluated in clinical trials. In a phase II study, 19 patients with advanced or metastatic hepatobiliary cancers (tumours of the liver, gall bladder, or bile duct) received intratumoural injections of ONYX-015, a first-generation, replication-selective adenovirus. Sixteen patients could be evaluated for response. One patient had a partial response, one patient had prolonged disease stabilization (49 weeks), and eight patients had a biological response with a decrease in tumour markers of more than 50%.28 Recent clinical studies with other therapeutic viruses have been reported, highlighting the interest in oncolytic viruses for cancer gene therapy. A phase I trial, including 14 patients with refractory primary or metastatic liver tumours, evaluated the maximum tolerated dose of intratumoural injection of JX-594. JX-594 is a targeted, oncolytic poxvirus (vaccinia virus) modified by disruption of the thymidine kinase gene and insertion of the human GM-CSF. Patients received a mean of 3.4 cycles of intratumoural injections. Four patients received only one cycle (two because of toxicity and two because of self-withdrawal from the study) and were not analysed for tumour response. Among the 10 patients analysed, three showed a partial response, six had stable disease and one showed progression of disease after CT-scan examination. Hyperbilirubinaemia was a limiting factor in patients treated with the highest dose. Flu-like symptoms were the most frequent adverse event. Of particular interest, JX-594 was found in peripheral blood after the initial injection and delayed viraemia was detected several days afterwards. JX-594 was found in non-injected tumours, suggesting that the virus could target disseminated tumours after systemic injection.33 Another study included three patients with concomitant hepatocellular carcinoma and hepatitis B virus infection (HBV) with active viral replication. JX-594 intratumoural injections were associated with tumour response and a decrease in HBV DNA levels.35

Perspectives from preclinical models.  New strategies aiming at improving the rate of transduction of cells have been developed for gene re-expression strategies, with interesting preclinical results. For example, an adenoviral vector encoding a fusion protein, p53-VP22, has been engineered. VP22 is a HSV type 1 tegument protein, exhibiting the capacity to be exported from the cell to neighbouring cells. Some VP22 fusion proteins, such as p53, can retain this capacity. In a model of heterotopic, subcutaneous hepatocellular carcinoma xenograft in mice, intratumoural injections of VP22-p53 adenoviruses were more efficient than those of p53-alone adenoviruses, leading to partial tumour regression.36 The most widely described suicide gene is the HSV-tk gene. However, other suicide gene strategies have been developed. For example, an adenoviral vector encoding carboxypeptidase G2 (CPG2) has been engineered. CPG2 is an enzyme that catalyses the conversion of alkylating prodrugs such as ZD2767P into cytotoxic agents. Tumour regression was observed in a model of subcutaneous hepatocellular carcinoma xenograft, after intratumoural injection of the vector and treatment with the prodrug, ZD2767P.37 Other immune therapy strategies have also been evaluated in vitro. Adenoviruses encoding CD40 ligand were injected into orthotopic hepatocellular carcinomas in rats. The treatment caused a tumour response, with 10 of 23 treated animals showing a complete response and nine animals a partial response, and an improved survival of treated rats. The treatment induced lymphocytic infiltration of the tumour, and the tumour response was dependent on CD8+ T cells.38 In a model of alpha-fetoprotein-expressing hepatocellular carcinoma spontaneously developing in mice (BW7756 tumours), the combination of alpha-fetoprotein vaccination (by injection of expression plasmids into muscles) and an adenoviral-mediated chemokine IP-10 and interleukin-12 expression (intratumoural injection) resulted in a synergistic effect with improved survival.39

The sodium-iodide symporter (NIS) gene encodes a membrane glycoprotein that mediates active iodide transport across the basolateral membrane of thyroid follicular cells. NIS can drive the uptake of radioisotopes (123Iodide, 124Iodide, 125Iodide, 131Iodide, 188Rhenium, 211Astatine) into the cells.40 A viral-guided NIS gene therapy is an innovative approach of radio-iodine therapy. Such an approach was assessed in a rat model of hepatocarcinoma (diethylnitrosamine-treated rats). Animals were treated by intraportal injections of an adenovirus encoding rat NIS. Long-term retention of radioiodine was demonstrated after viral transduction, both in treated, tumour-bearing rats and in healthy, treated rats, but not in control, nontreated rats. After intratumoural injection of the virus, intraperitoneal injections of 131I were associated with inhibition of tumour growth and prevention of the formation of new tumour nodules around the injected nodule. Furthermore, intraperitoneal injection of the virus followed by 131I therapy was associated with improved survival of diethylnitrosamine-treated rats.41

New strategies include antiangiogenic gene therapy and therapeutic RNAi. Antiangiogenic therapies are currently well developed and widely used in oncology. Endostatin is an inhibitor of angiogenesis and tumour growth.42 An adenovirus encoding human endostatin was tested in a model of subcutaneous hepatocellular carcinoma xenograft in mice and demonstrated an antitumour effect.43 Pigment epithelium-derived factor (PEDF) is another recently discovered, antiangiogenic factor. In a model of heterotopic, subcutaneous hepatocellular carcinoma xenograft in mice, intratumoural injections of a plasmid expressing PEDF resulted in a significant reduction of tumour volume compared with controls.44 Kallistatin is another example of new antiangiogenic inhibitors. Intraportal injections of a recombinant AAV encoding kallistatin were performed in hepatic and subcutaneous hepatocarcinoma murine models. Treatment with the recombinant virus resulted in significant tumour growth inhibition, as compared with treatment with empty AAV.45

There has been recent interest in the RNAi strategy. The pituitary tumour-transforming gene 1 (PTTG1) is an oncogene, which is frequently expressed in human liver cancers. An adenovirus encoding an RNAi targeting PTTG1 (Ad.PTTG1-siRNA) has been engineered. In vitro, the virus could deplete PTTG1 in hepatoma cells, resulting in apoptosis. Furthermore, intratumoural injections of the virus led to inhibition of tumour growth in a xenograft tumour in mice.46 Another recent study concerns RNAi cancer gene therapy in a mouse model of hepatocellular carcinoma (tet-O-MYC; LAP-tTa). Hepatocellular carcinoma cells exhibit a downregulation of the expression of miR-26a in human and mouse liver tumours. This miRNA plays a role in cell cycle arrest. Systemic administration by tail vein injection of an adeno-associated virus encoding miR-26a resulted in the inhibition of cancer proliferation, with six of eight treated mice exhibiting only small tumours or no tumour and six of eight mice treated with control virus developing a fulminant disease.47

Colorectal cancer

Clinical trials.  Clinical trials concerning the gene therapy of metastatic colorectal cancers have also been published (Table 3).48–55 Re-expression of tumour-suppressor genes has been evaluated in a phase I trial including 17 patients with advanced cancers, among them five patients with colorectal cancer and one patient with pancreatic cancer. Patients were treated with escalating doses of an adenovirus encoding a wild-type p53 gene (Ad5CMV-p53). The treatment was administered intravenously, daily for 3 consecutive days, every 28 days. There were no objective responses; one patient with metastatic and progressive colorectal carcinoma had a 10-month period of stable disease. The virus was detectable in plasma 14 and 28 days after treatment in the majority of patients treated at the highest dose. Seven patients had tumour biopsies before and after treatment. Six of these patients had undetectable Ad5CMV-p53 expression before treatment that became detectable after treatment. This study showed that this treatment was feasible, with no major toxicity.53

Table 3.   Clinical trials concerning gene therapy in metastatic colorectal tumours
StudyClinical phaseNo. patientsClinical settingVector and geneTreatment protocolComments
  1. PR, partial response; SD, stable disease; PD, progressive disease.

Rubin et al.48I15Advanced colorectal cancersLiposomal vector, HLA-B7 gene ransferIntratumoural injectionsGood safety record
Feasible DNA transfection with gene expression
Sung et al.49I16Advanced colorectal cancersHSV-tk-encoding adenovirusIntratumoural injections into liver metastases, followed by systemic ganciclovir administrationGood safety record
Tumour response of injected tumour: 11 SD, 5 PD
Reid et al.50I11Advanced colorectal cancersONYX-015 adenovirusInfusions of ONYX-015 into the hepatic artery, in combination with intravenous 5-FU and leucovorinGood safety record, no dose-limiting toxicity
Van der Burg et al.51I/II15Advanced colorectal cancersp53-encoding canarypoxvirusIntravenous injectionsGood safety record
Induction of p53-specific antibodies and a specific T-cell response with the highest dose of virus
Reid et al.52II27Gastrointestinal carcinoma metastatic to the liverONYX-015 adenovirusInfusions of ONYX-015 into the hepatic artery, in combination with intravenous 5-FU and leucovorinThree patients with 5-FU refractory disease had a 30–50% tumour regression
Tolcher et al.53I17Advanced cancers, including five colorectal and one pancreatic cancersAdenovirus encoding wild-type p53 gene (Ad5CMV-p53)Intravenous injectionsFever, nausea, vomiting and fatigue were common but rarely severe
Tumour response assessment: no objective response, 1 SD
Harrop et al.54I/II22Advanced colorectal cancers5T4-encoding vaccinia virusIntramuscular or intradermal injectionsGood safety record
Five patients had stable disease for a period ranging from 3 to 18 months
Kemeny et al.55I12Advanced colorectal cancersNV1020 oncolytic herpes simplex virusNV1020 injection into the hepatic arteryGood safety record with fever, rigours and headache as most common adverse effects
Tumour response assessment at day 28: 2 PR, 7 SD, 3PD

A phase I clinical trial including 16 patients with hepatic metastases from colorectal carcinomas assessed the safety of intratumoural injections of an adenovirus encoding the HSV-tk gene. Injections were followed by systemic ganciclovir administration. The treatment was safe. However, there was no complete or partial tumour response.49

A p53-specific immunity can be observed in advanced cancers. In a phase I/II, dose-escalation trial, 15 patients with advanced colorectal cancers were treated with systemic injections of a poxvirus encoding a wild-type p53 gene, aiming to induce p53-specific immunity. The treatment could induce p53-specific antibodies and, in two of five patients treated with the highest dose of virus, a specific T-cell response.51 TROVAX is a modified vaccinia virus encoding the tumour antigen, 5T4, aiming to produce an anti-5T4 immune response. Trovax is used in several cancer gene therapy protocols. A phase I/II trial was carried out in 22 patients with metastatic colorectal cancer. Repeated intramuscular injections of TROVAX were safe and induced an anti-5T4 cellular response. Five of 22 patients had stable disease for a period ranging from 3 to 18 months.54 A 5T4-specific cellular or antibody response was still detectable when given alongside a chemotherapy regimen.56

A phase II trial included 27 patients with colorectal cancer liver metastases. Infusions of ONYX-015 into the hepatic artery, in combination with 5-FU, were well tolerated with fever, rigours and fatigue as the most common side effects. The antitumoural effect of the virus could not be assessed in patients who had not previously received this chemotherapy regimen. Interestingly, three patients with 5-FU-refractory disease had a 30–50% regression of the tumour mass.52 Another phase I, dose-escalation trial included 12 patients with liver colorectal cancer metastases. All patients were refractory to a first line chemotherapy. NV1020 is an oncolytic HSV type 1. Patients had a single NV1020 injection into the hepatic artery at four different doses (three patients in each group). NV1020 administration was well tolerated with fever, rigours and headache as the most common side effects. Antitumour activity was assessed by CT-scan at day 28, even if this was a toxicity study. Two patients had a 39% and a 20% reduction respectively, in tumour size; seven patients showed stable disease and three a progressive disease.55 Floxuridin chemotherapy was administered after day 28 by an intrahepatic arterial pump, in association with intravenous irinotecan. All 12 patients had a partial response to subsequent chemotherapy.57

Perspectives from preclinical models.  In a recent study, a p53 gene re-expression strategy was improved in a combination of viral-directed therapies. An AAV vector expressing the kringle 1 domain of the human hepatocyte growth factor (AAV-HGFK1) and an adenovirus expressing p53 were combined in a murine model of colorectal cancer xenograft. The combination treatment increased the survival time, inducing apoptosis, necrosis and suppressing angiogenesis, as compared with single agent therapies. In vitro, this antiangiogenic effect was related to an inhibitory effect on endothelial cell migration.58

The HSV-tk/ganciclovir suicide strategy could also be improved. An adenovirus encoding the monocyte chemoattractant protein-1 (MCP-1) gene has been engineered. In a model of subcutaneous colorectal cell line tumours in immunocompetent mice, intratumoural injections of both MCP-1-encoding adenovirus and of HSV-tk-encoding adenovirus reduced tumour growth significantly more than HSV-tk adenovirus alone.59 Another suicide system has been developed in a murine model of colorectal cancer. A retrovirus encoding the cytosine deaminase (CD) gene was engineered. CD converts the nontoxic 5-fluorocytosine into toxic 5-fluorouracil. In a model of orthotopic liver metastasis (CT26 cells) in mice, the viral vector was administered into the portal vein via intrasplenic injection. 5-fluorocytosine was administered into the peritoneal cavity. All mice in the control groups had tumour progression, but tumour growth was significantly inhibited in the treated mice.60

Chemo-gene and radiation-gene therapies are the combinations of gene therapy with chemotherapy or radiation therapy respectively. Gene therapy can sensitize colon cancer cells to chemotherapy in vitro.61, 62 CD converts 5-fluorocytosine into the toxic metabolite, 5-fluorouracil. Intratumoural injection of free plasmids expressing CD in liver metastasis rat models, in association with an oral treatment with 5-fluorocytosine, resulted in a systemic antitumour effect.63 Flt3 Ligand (Flt3L) is an activator of dendritic cells and NK cells. Intratumoural injections of an adenovirus encoding Flt3L produced a synergistic effect in association with 5-FU therapy in subcutaneous models of hepatocarcinoma and colon cancer and induced long-term immunity against further parental cancer cell injections.64 In a recent study, cultured human colorectal cancer cells and tumour xenografts were infected with an engineered adenovirus expressing the NIS gene. Tumour xenografts concentrated 99mTechnetium, allowing quantitative imaging by SPECT/CT. NIS expression reached a peak 48 h after intratumoural injection. Administration of a single dose of virus combined with a single 131I dose 48 h after virus injection induced tumour regression with an additional effect.65

Recently, the effect of miRNA targeting c-myc was assessed in vitro and in vivo. In vitro, miRNA inhibited c-myc expression, reduced HT 29 cell proliferation and induced apoptosis. In vivo, intratumoural injections of miRNA were performed in a model of subcutaneous HT 29 tumour in athymic nude mice. Treatment significantly reduced tumour volume.66 In another study, a replication-deficient adenovirus expressing p53 and an miRNA targeting the cyclin-dependent kinase inhibitor p 21 (Adp53/mIR-p21) have been engineered. In tumour xenografts in nude mice, intratumoural injections resulted in a significant decrease in tumour volume, as compared with a control virus expressing p53, but not mIR-p21. In vitro, Adp53/mIR-p21 increased apoptosis as compared with the control virus, but also increased the sensitivity of cancer cells to doxorubicin.67

Pancreatic cancer

Clinical trials.  Clinical studies have also been carried out in pancreatic cancer (Table 4).68–71 A phase I study assessed the safety and side effects of a subcutaneous vaccination regimen using two viruses (vaccinia virus and fowlpox virus) expressing CEA and mucin-1 with three co-stimulatory molecules (B7.1, ICAM-1 and LFA-3), in association with GM-CSF therapy, in 10 patients with advanced pancreatic cancer. The most common side effects were injection site reactions, fatigue, headache and nausea. Survival was significantly better in patients who developed anti-CEA- or anti-MUC-1-specific immune responses; however, these are preliminary data.70 A phase III study in metastatic pancreatic cancer has been carried out (US 635 phase III trial), but results have not yet been published.

Table 4.   Clinical trials concerning gene therapy in pancreatic cancer
StudyClinical phaseNo. patientsClinical settingVector and geneTreatment protocolComments
  1. PR, partial response; SD, stable disease.

Mulvihill et al.68I23Locally advanced cancerONYX-015 intratumoural injectionsCT-guided injections (n = 22 patients) or intraoperative injections (n = 1) into pancreatic primary tumours every 4 weeks until tumour progressionGood safety record, no clinically significant pancreatitis
No detectable viral replication
No objective response documented
Hecht et al.69I/II21Locally advanced or metastatic cancerONYX-015 intratumoural injectionsEndoscopic ultrasound-guided ONYX-015 injections on days 1, 5, 8, 15, 36, 43, 50 and 57, in association with intravenous gemcitabine therapy on days 36, 43, 50 and 57Two duodenal perforations when transduodenal approach was used. Two instances of sepsis before institution of systematic antibiotic prophylaxis
No clinically significant pancreatitis
No objective response on day 35
After combined treatment: 2 PR and 8 SD of the targeted lesion
Kaufman et al.70I10Locally advanced or metastatic cancerSubcutaneous vaccination with vaccinia virus and fowlpox virus, both expressing tumour antigens and co-stimulatory moleculesSubcutaneous vaccination, in association with GM-CSF therapySite reaction, fatigue, headache and nausea as most-common side effects
Significantly better survival in patients developing anti-CEA or anti-1-specific immune responses
Galanis et al.71I12Gemcitabine-refractory, metastatic pancreatic cancerIntravenous injections of Rexin-G, a nonreplicative, retroviral vector expressing cyclin G1 geneRepeated intravenous injectionsGood safety record
No evidence of clinical anti-tumour activity

A phase I/II study showed the feasibility and tolerability of endoscopic ultrasound-guided, repeated administrations of the ONYX-015 virus into carcinomas of the pancreas.69 An interim analysis of the PACT study, a multicentre, randomized, controlled, phase III clinical trial, was presented at the American Society of Clinical Oncology meeting 2009. The study included 330 patients with advanced pancreatic cancers, with two arms of treatment. The control arm received standard, 5-week chemoradiation therapy, the experimental arm received chemo-radiation plus weekly, intratumoural injections of TNF-erade (Genvec, Inc., Gaithersburg, Maryland, USA), an adenovirus encoding the TNF-α gene. Both arms received adjuvant gemcitabine with the option of erlotinib. An interim analysis of overall survival was conducted after the 92nd death (one-third of total expected events) had occurred. The analysis concluded that TNF-erade appeared to be safe and well-tolerated. The overall survival interim analysis indicated a trend in favour of TNF-erade therapy, with a late effect on Kaplan Meyer curves.72

Perspectives from preclinical models.  Transduction efficiency is a major concern in gene therapy. A recently published study showed high transduction efficiency for gene transfer in cells derived from pancreatic cancers, using a lentiviral vector, both in vitro (from 68% to 98%, depending on the cell line and the multiplicity of infection) and in vivo. In tumour xenografts in mice, intratumoural injections of a lentivirus encoding human interferon-β resulted in a significant inhibition of apoptosis.73

Peritoneal dissemination is a poor prognostic event in pancreatic cancer. Intraperitoneal vector injections may therefore be of great interest. Intraperitoneal injection of retrovirus encoding the HSV-tk gene has been effective in treating nude mice with a model of metastatic pancreatic cancer disseminated to the peritoneal cavity.74 The K-ras oncogene is mutated in more than 80% of ductal pancreatic adenocarcinomas.75 An adenovirus expressing K-ras antisense RNA can induce apoptosis in vitro in human pancreatic cancer cells. Intraperitoneal injections of this adenovirus inhibited peritoneal dissemination in hamster pancreatic cancer models.76

We will focus on the promising approach of chemotherapy and radiation gene therapy. Adenoviruses encoding p53 genes showed a synergistic antitumour effect with 5-FU therapy, in vitro and in vivo, in heterotopic models in rats.77 A recombinant adenovirus encoding deoxycytidine kinase and uridine monophosphate kinase (enzymes converting gemcitabine into toxic metabolites) sensitized pancreatic cancer cells to gemcitabine, in vitro and in an orthotopic tumour hamster model.78 In a subcutaneous model of pancreatic carcinoma in mice, a single, intravenous injection of the vaccinia GLV-1h68 virus, an oncolytic, engineered poxvirus, was able to reverse the course of tumour growth. The efficacy of viral therapy was enhanced by repeated injections of gemcitabine or cisplatin. Seven of eight mice treated with the virus and cisplatin had complete tumour regression at the end of the observation period, whereas one of eight mice treated with the virus alone had complete tumour regression. The viral therapy was safe, with a high tumour selectivity of the virus after systemic administration.79 Another study demonstrated that an oncolytic virus was able to synergize with gemcitabine to kill pancreatic cancer cells in vitro and in vivo by a synergistic effect on apoptotic pathways.80 In another study, cultured human pancreatic cancer cells and tumour xenografts were infected with an engineered measles virus expressing the NIS gene. Intratumoural injections resulted in a significant reduction in tumour volume and increased survival time of the treated mice compared with the control mice. Tumour xenografts also concentrated radioiodine, allowing quantitative imaging with 123I micro-SPECT/CT.81 Pancreatic cancer is therefore a candidate for specific, viral-driven delivery of therapeutic radioisotopes.


The safety profile of gene therapy is a major concern. Recently published trials have shown good safety records. The most commonly reported side effects related to treatments involving a viral vector were fever, headache, fatigue, nausea and vomiting. Intratumoural injections were used in most studies. In trials using replication-competent agents, viruses were found in non-injected tumours and regression of both injected and distant disease occurred. This effect could be related to targeting of disseminated tumours by the vector and to local and systemic immune responses. Thus, intratumoural injections could be efficient, even for the treatment of disseminated tumours. However, lesions can be difficult, or impossible, to inject, and repeated intratumoural injections using invasive procedures could be a limiting factor for this approach. Systemic administration has also shown a good safety record and the development of oncolytic viruses that selectively target tumour cells makes this mode of administration of great interest. These data regarding oncolytic viruses may pave the way for more widespread clinical applications of this approach.

Gene therapy will need to be positioned in the context of existing cancer therapy strategies. Good safety records could make gene therapy a good candidate in all therapeutic settings (neo-adjuvant, adjuvant and advanced disease), probably in combination therapies. Combining gene therapy with chemotherapy or radiation therapy may be very interesting, as gene therapy can act as a chemo- sensitizer or radio-sensitizer, and because chemotherapy and radiation therapy can improve gene transfer efficiency and gene expression.82–86

Gene therapy is not yet suitable for use in routine clinical practice, mainly because of insufficient levels of gene delivery in patients. However, promising preclinical data include the recent development of new therapeutic targets, new treatments such as miRNA and new vectors and, in particular, vectors that are more highly selective with greater therapeutic potential. Clinical trials are ongoing in gastrointestinal cancers, as well as in other tumours such as melanoma, brain, head and neck, prostate or ovarian cancers. Even if safety profiles, drug-resistance and efficacy have to be elucidated further, cancer gene therapy may become a new weapon in anti-cancer strategies.


Declaration of personal interests: KJ Harrington has received unrestricted educational grants in support of laboratory and clinical research from Oncolytics Biotech Inc. and Genelux GmbH. JP Galmiche has served as a speaker and a consultant for AstraZeneca and Given Imaging. Declaration of funding interests: Y Touchefeu has received a doctoral research grant from the Institut National du Cancer. Part of this work was undertaken in The Royal Marsden NHS Foundation Trust that receives a proportion of its funding from the NHS Executive; the views expressed in this publication are those of the authors and not necessarily those of the NHS Executive. This work was supported by the Institute of Cancer Research, and Cancer Research UK Section of Radiotherapy (CRUK) Grant Number C46/A10588. We acknowledge NHS funding to the NIHR Biomedical Research Centre. We also acknowledge support from the Institut national de la santé et de la recherche médicale (INSERM).