Clinical advances in cancer research are often slow to materialize, in part because the efficacy of a treatment has to be balanced against its potential toxicity to normal tissues. Infection of tumours with oncolytic (cancer-killing) viruses has been explored as a new type of treatment that is not cross-resistant with approved cancer therapies and, being target-specific, may have fewer toxic side effects. On page 99 of this issue, Breitbach et al.1 describe a phase I clinical trial in which an intravenously delivered oncolytic poxvirus was capable of replicating selectively in metastasized tumours. This is a milestone in the development of an effective oncolytic agent for systemic administration.

Oncolytic viruses became a focus of attention for cancer therapy following observations that natural viral infection or vaccination can lead to spontaneous regression of malignancies2. Unhindered by interferon-mediated antiviral defence, which is compromised in many tumours3, these viruses specifically attack cancer cells by gaining entry through receptors that are overexpressed in these cells and/or by exploiting molecular pathways associated with malignant transformation for their replication4,5. As the virus starts to replicate at the tumour site, its destructive effect increases. Strategies are being devised to make this process even more efficient by deploying genetically engineered oncolytic viruses that carry therapeutic or immunomodulatory transgenes.

In advanced cancer, systemic dissemination of solid tumours is linked with a poor prognosis. Before oncolytic viruses can be used to treat such metastases, they must be able to reach and replicate in metastatic sites following intravenous administration. But there are obstacles to be overcome, including the antiviral immune response, and the uptake and destruction of the virus by the endothelial reticulum system in the liver and spleen.

Breitbach et al.1 take up the challenge using a genetically engineered oncolytic poxvirus known as JX-594. This is a smallpox-vaccine derivative of Wyeth-strain vaccinia virus carrying an inactivated thymidine kinase gene to increase tumour specificity, and expressing two transgenes: one encoding human granulocyte–macrophage colony-stimulating factor (GM-CSF) to stimulate anti-tumour immunity and the other β-galactosidase, a surrogate marker for detecting viral gene expression.

The authors tracked the virus in 23 cancer patients, all with advanced solid tumours that were resistant to other treatments. Patients were each given one dose of JX-594 at one of six different dosage levels by intravenous injection; these were all well tolerated. The maximum feasible dose was 3 × 107 plaque-forming units (PFU) per kilogram of body weight (corresponding to a total dose of about 2 × 109 PFU). This dosage is in line with doses of other oncolytic viruses that can safely be given intravenously, including adenovirus, reovirus, paramyxovirus (Newcastle disease virus and measles) and Seneca Valley virus.

Breitbach et al. demonstrated such dose-dependent delivery of the virus (at 8–10 days after intravenous administration) to metastatic tumour deposits from a variety of tumour types, including leiomyosarcoma, mesothelioma, and lung, ovarian and colorectal cancers. In eight patients who had received 109 PFU or more per dose, delivery and replication were confirmed by quantitative polymerase chain reaction in five patients and by immunohistochemistry using a polyclonal anti-vaccinia antibody in six patients: granular cytoplasmic staining evident in tumour tissue was indicative of replicating virus (viral factories; Fig. 1).

Figure 1: Common oncogenic mutations in cancer cells encourage replication of the genetically engineered oncolytic JX-594 virus1.
figure 1

The virus takes advantage of a cancer cell's uncontrolled epidermal growth factor receptor (EGFR)–RAS signalling pathway. To replicate, this thymidine kinase (TK)-deficient virus relies on expression of TK by cancer cells. The newly assembled viruses then leave the cell to infect other tumour cells. These viruses also secrete GM-CSF, a factor that stimulates anti-tumour immunity. In normal cells, however, viral replication is blocked because this virus cannot efficiently exploit the cell's replication machinery.

Although JX-594 administration seemed to result in disease control in a dose-dependent way, with patients treated with the higher doses benefitting the most, viral infection and replication in metastatic deposits did not consistently affect clinical outcome. Some patients experienced clinical benefit — defined as disease stabilization for more than ten weeks — even when there was no evidence of viral replication in their tumour biopsies. By contrast, two out of six patients who were JX-594-positive by immunohistochemistry had progressive disease at first evaluation, even though replicating virus was detected in their metastatic tumours.

The explanation for these discrepancies may be down to several factors. For example, patients were allowed only one viral dose and treatment cycle: as with other cancer therapies, it is unlikely that a single round of treatment would be enough to stop tumour growth. Sampling variability in patients, whether positive or negative for JX-594, may also have confounded the results. Reassuringly, the normal tissue of patients in whom replication was detected was negative for replication by immunohistochemistry.

The limitations notwithstanding, these results convincingly demonstrate successful dose-dependent delivery and replication of an oncolytic virus in metastatic disease sites, following intravenous administration in patients with primary solid tumours. Although oncolytic viral replication in metastatic disease sites after systemic administration has been reported before, those studies are undermined by detectable replication only in isolated patients or by methodology unable to distinguish properly between input and progeny virus. Promising preclinical data, however, point to several strategies for enhancing systemic delivery of oncolytic viruses, including the use of cell carriers, cationic liposomes and polymers.

Large randomized trials to test oncolytic viruses in cancer treatment are ongoing or soon to be activated. These will investigate the potential synergistic cytotoxicity between oncolytic viruses and more conventional therapeutic approaches such as chemotherapy, small-molecule cell-cycle inhibitors, radiation therapy and anti-angiogenesis agents6,7,8,9. In addition, they will exploit induction of a systemic antitumour immune response in association with oncolytic tumour-cell death and expression of immunomodulatory transgenes10.

Examples of such trials include the soon-to-be-completed phase III trial of an attenuated strain of herpes simplex virus-1 that encodes GM-CSF in patients with metastatic melanoma; the recently activated phase III trial testing addition of reovirus to paclitaxel/carboplatin chemotherapy in patients with recurrent head and neck cancer; and a randomized phase II trial comparing JX-594 with the best supportive care in patients with hepatocellular carcinoma for whom treatment with the drug sorafenib has failed.

In contrast to Asian countries, no virotherapy agent has so far been approved in the United States or Europe. The outcome of these trials may change this, generating additional valuable clinical tools for oncologists.