As the field of anticancer gene therapy edges closer to the mainstream, so we learn more of its current limitations as well as the potential for future success. With the technologies available today, it is simply not possible to achieve correction of every single genetic abnormality in every malignant cell. Hence most approaches investigated so far have involved a cytotoxic strategy, aiming to selectively destroy tumours while avoiding “collateral damage” to normal tissues.1

The success or failure of such cytotoxic gene therapy is influenced by two important choices—what therapeutic gene should be introduced and what vector should carry it to the target? We already have a range of genetic prodrug activation therapy (GPAT) strategies available, based on introducing tumour specific “suicide” enzymes that can generate toxic metabolites from otherwise benign prodrugs. Common examples include the herpes simplex virus thymidine kinase (HSVtk), which confers sensitivity to the antiviral drug ganciclovir, or theEscherichia coli derived cytosine deaminase that converts 5-fluorocytosine into the antineoplastic 5-fluorouracil (5-FU).2 Several other enzyme based systems are in development with the intention of either increasing tumour sensitivity to drug therapy or enhancing resistance of normal tissues to the side effects of anticancer agents.3

Each of the GPAT systems is capable of killing malignant cells effectively but only if they can be introduced into the appropriate target and expressed at a satisfactory level. The vectors currently available to us are unable to guarantee that every neoplastic cell will receive and express the transferred gene. As a result, anticancer gene therapies require augmentation of cell killing, either by approaches that cause an immune response against the tumour or by generating a “bystander effect”. This latter term refers to the well recognised phenomenon of GPAT killing many more cells in a tumour mass than are actually expressing the therapeutic construct.4 In the case of 5-FU generation, the cytotoxic agent itself is able to diffuse into adjacent cells. With the HSVtk approach, the effect is mediated by intercellular gap junction communications.5

The other method of enhancing cell killing with GPAT is to recruit the power of the immune system.6 Immunotherapeutic strategies have, in their own right, made up the majority of anticancer gene therapy protocols and there is now further evidence that immune mediated approaches can be particularly effective in combination with GPAT.7 In particular, aggressive cytotoxic gene therapy can lead to enhanced systemic antitumour immunity if the mode of non-apoptotic cell death is rendered as immunologically “dangerous” as possible.8

In this issue of Gut, Lechanteuret al describe the effects of combining GPAT with the transfer of an immunostimulatory cytokine gene in a rat model of peritoneal carcinomatosis (see page 343).9 Tumour deposits were created by injecting 10⁶ cells of the DHD/K12 colorectal lineage into the peritoneum. After seven days of growth, the animals received weekly intraperitoneal injections of retrovirus producing cells expressing the HSVtk enzyme and either granulocyte macrophage-colony stimulating factor (GM-CSF) or interleukin 12 (IL-12) for a total of three doses. Ganciclovir was administered in a series of three day courses, beginning on day 10. The authors report that while HSVtk GPAT alone slightly increased the mean survival time from 63 to 72 days, addition of the cytokine genes significantly improved their results, with 60% of the GM-CSF group and 40% of the IL-12 cohort being tumour free after almost 16 months of follow up. Their successful response is attributed to a combination of prodrug activation, the bystander phenomenon, and the adjuvant effect of proinflammatory cytokines on the immune system.

These results are clearly encouraging although exactly how easy they will be to replicate in disseminated human cancer is another matter. Relatively small, one week old tumours are likely to pose less of a physical barrier to the penetration of viral vectors than large tumour masses. In the particular case of adenoviral gene therapy for intraperitoneal malignancy, immunohistochemistry and laser scanning cytometry have shown a depth of penetration of between 1 and 10 cells from the cell surface.10

Creating the conditions in which a high enough number of cancer cells express the therapeutic constructs and invoke bystander effects or immunological responses may require vectors with greater tumour selectivity. To this end, there has been renewed interest in the use of preferentially replicating bacteria that can colonise tumour deposits at concentrations orders of magnitude higher than in normal tissues.11 Such bacteria have a number of potential advantages over viruses. They are able to survive and multiply in the extracellular (as well as intracellular) microenvironment, are actively motile within tumour masses, and their relatively large size facilitates the carriage of multiple therapeutic constructs such as GPAT or immunostimulatory elements.12

Gene therapists can now arm themselves with a whole range of different cytotoxic strategies. In much the same way as “conventional” anticancer chemotherapy utilises multiple drugs acting on different targets, it seems the way forward will lie in finding the ideal combinations of vectors and therapeutic genes.13