¹Dardinger Laboratory for Neuro-oncology and Neurosciences, Comprehensive Cancer Center, and Department of Neurological Surgery, The Ohio State University Medical Center, Columbus, Ohio, USA

Correspondence: E Antonio Chiocca, EA.chiocca@osumc.edu

Traditionally, cancer therapies have been based on the empiric discovery of drugs and/or biological agents that act selectively against tumor cells. Recently, however, a generation of rationally discovered drugs and agents has been developed that shows encouraging benefit in treating cancers. Nevertheless, because the same histologic varieties of cancers in different individuals can respond differently to therapies, molecular correlates of tumor responsiveness are needed to improve the effectiveness of any given treatment.

As an example, the overexpression of members of the family of epidermal growth factor receptors is being investigated as a predictive criterion for selecting patients for treatment with targeted molecular drugs, such as Herceptin (trastuzumab), Erbitux (cetuximab), and Tarceva (erlotinib). The results have indicated that immunohistochemical staining to monitor overexpression is not always predictive of improved response,¹ perhaps at least partly because of the high variability immunohistochemical staining. In fact, analysis by fluorescence in situ hybridization to assess genetic amplification of Her-2-positive tumors revealed discrepancies between the two techniques used for tumor classification. Incorrect classification of a tumor's Her-2 status could lead to underpowered statistics and erroneous results.² Another recent discovery that may compound these difficulties has been the activating mutations identified in the Her-2 kinase domain, which may further contribute toward tumor resistance to tyrosine kinase inhibitors.³

The next logical step in the field of targeted therapeutics is to circumvent the problem of analyzing single-gene alterations by identifying molecular signatures or patterns of gene or protein expression to determine which cancers will prove the best targets for a given therapy.^{4, 5} Consequently, as cancer treatments evolve from "one-size-fits-all" approaches to more "personalized" patient-based strategies, acquisition of tissues from humans with cancers will become more important to discover a tumor's amenability to a proposed therapeutic.

The paper by Yu and colleagues⁶ in this issue seems to provide an initial foray for such arguments in the context of oncolytic virotherapy (OV) using herpes simplex virus type 1 (HSV1) recombinants. OV-mediated therapy is based on selective lysis of tumor cells that spares normal cells.⁷ This multistep process requires an initial entry into and infection of the target cell by the virus, replication of the virus into progeny virions, exit of these progeny virions from the infected cell, and infection of other cells. If therapy is to be efficacious, the tumor cell must (1) possess available receptor(s) to permit viral entry and (2) provide an intracellular environment allowing for robust and efficient replication, and (3) the process of exit and infection by progeny virions must be allowed to occur and cannot be curtailed by host responses, such as interferon- and/or cell-mediated innate responses.⁸ Yu and colleagues provide a significant argument that step (1) is arguably one of the more important events, at least in the context of the particular therapeutic tested and of the tumor cells that were analyzed.

Although study of the natural tropism of a virus to enhance tumor-specific infectivity has been investigated for adenoviruses, the same has not been accomplished for other viruses, such as HSV1. Thus, Yu's group provides a valuable discovery. Expression of the cell surface receptors to which HSV1 normally binds (nectin-1, herpes virus entry mediator (HVEM), and all cellular receptors that bind to HSV1 glycoprotein D (gD), S-phase fraction, and cell doubling time were analyzed in a panel of eight squamous cell carcinoma cell lines and assessed as potential markers for oncolytic HSV (NV1023) sensitivity. Interestingly, Yu's studies revealed that susceptibility to viral infection correlated significantly only with expression of nectin-1. Viral entry in the same panel of cell lines was impeded more significantly by antibodies specific to nectin used to block availability of cell-surface nectin than by blockade of HVEM. It is noteworthy to mention that this is an investigation of a few cell lines derived from one tumor type with one oncolytic virus and that influences determining sensitivity to viral infection may vary depending on tumor type and the kind of OV used.

Cellular infection was enhanced by overexpression of nectin-1, achieved by transient transfection of a plasmid expressing nectin-1. Comparison of the therapeutic efficacy of NV1023 in a rodent model between two cell lines expressing different amounts of nectin-1 revealed a direct correlation between efficacy and expression of nectin-1. However, before nectin-1 expression can be adopted as an important therapeutic predictor for oncolytic HSV-mediated oncolysis, it should be noted that results obtained from comparing different cell lines include more than one variable and hence cannot be considered as conclusive evidence. Nevertheless, this study underscores the importance of investigating tissue specimens for markers predictive of better oncolysis and of using the results of such studies to improve criteria for selecting patients for oncolytic HSV therapies.

Along the same lines, patient selection criteria should also include identification of molecular-genetic identifiers prognostic for patients vulnerable to HSV1-induced side effects. The recent observation of human UNC-93B deficiency in two children with sporadic herpes simplex encephalitis indicates that the presence of monogenic disorders may predispose patients specifically to a single infectious agent without affecting host defense against many other pathogens.⁹ Peripheral blood mononuclear cells from these patients produced reduced amounts of type I and type II interferons in response to HSV1 stimulation but not in response to ten other viruses, thus highlighting the presence of specific susceptibility defects.⁹ The specificity of defects, such as UNC-93B, for a single pathogen that manifests as sporadic events in otherwise healthy individuals makes their identification highly challenging. Whereas studies such as these have several important therapeutic implications, further studies are needed to confirm the validity of these markers as prognostic indicators for oncolytic viral therapy.

Additional promising avenues of research in this field include the design of more potent OV that do not compromise the excellent safety profile demonstrated thus far by these viruses, establishment of treatment strategies with drugs that can synergistically enhance therapy, efficient and specific infection of targeted malignant tissue, provision of high-throughput patient selection criteria to "customize" treatment options, and development of molecular and imaging correlates of responses.^{10, 11, 12} In addition, innate host immune responses to tumor infection with OV have been shown to be responsible for rapid viral clearance that has limited antitumor efficacy.^{8, 13, 14} At later stages in therapy, OV-mediated oncolysis is also thought to set the stage for a systemic adaptive immune surveillance that increases tumor destruction.^{15, 16, 17} Considering the incredible complexity of human cancers, oncolytic viral therapy, and the elicited immune responses, it is highly unlikely that a single influence can be effectively used to predict efficacy. The next generation of clinical trials should determine if effective cancer therapy with HSV1 oncolytic viruses strictly depends on the initial process of infection or whether a variety of patient-, tumor-, and virus-specific features influence the final outcome and effectiveness.