Viruses for Tumor Therapy

Abstract

Oncolytic virotherapy exploits live viruses with selective tropism for cancerous cells and tissues to treat cancer. As discussed here, the field has progressed considerably as a result of both the successes and failures of previous and on-going clinical trials for various cancers. These studies indicate that oncolytic viruses are remarkably safe and more efficacious when virus replication stimulates sustained antitumor immune responses. In the future, virotherapy should be combined with immunomodulatory reagents that target immune tolerance to established cancers.

Introduction

The field of oncolytic virotherapy began as an observational science more than a century ago when it was noted that cancer regressions sometimes occurred spontaneously in patients following certain viral infections, (Kelly and Russell, 2007). These early anecdotes spawned a small number of clinical studies beginning in the 1940s using unmodified, and sometimes dangerous, test viruses. Although there were often glimmers of activity in these studies, the field of anticancer viral therapy languished for several decades, in part because of the early success of chemo and radiation therapies but also due to our limited understanding of the biology of these complex biological agents. The recombinant DNA revolution of the last 30 years has now provided the tools necessary to better understand, at the molecular level, how viruses attack and usurp host cell machinery. These advances coupled with those in the field of cancer biology have reignited interest in the use of replicating viruses as cancer therapeutics. Over the last two decades in particular, a variety of DNA and RNA viruses shown or engineered to be selective for cancer cells have transitioned from preclinical studies into early phase clinical testing and more recently into randomized clinical trials. The current status of ongoing clinical trials, and the candidate oncolytic viruses that are in various stages of development, have been summarized in great detail within many recent reviews, and the reader is encouraged to consult these for details regarding specific viruses and cancers under active clinical investigation today (for example, see Bourke et al., 2011; Donnelly et al., 2012, 2013; Eager and Nemunaitis, 2011; Russell et al., 2012; Patel and Kratzke, 2013; Sze et al., 2013; Vacchelli et al., 2013; Miest and Cattaneo, 2014). Our intent is to summarize some of the general trends currently emerging based on the clinical experiences to date, and to comment on the future prospects for oncolytic virotherapy assuming a more prominent role as a licensed modality that would join the current standard therapeutic trio of clinical oncology practice: i.e. surgery, chemotherapy and radiotherapy.

Oncolytic Viruses Target Cells with Malignantly Altered Signaling Pathways

The startling recent advances in the sequencing of cancer patient genomes continue to re-enforce the notion that cancer is a complex, heterogeneous disease that defies treatment with agents that target only a single genetic mutation. As Bert Vogelstein noted in 2008, to make significant advances in cancer therapy, “the focus should shift from hunting for individual genes that cause certain cancers, to disrupting broader biological pathways that support cancer growth” (Hayden, 2008). We argue that oncolytic viruses are indeed just such agents, because they thrive in tumor cells where pathways are malignantly activated or disrupted and can exploit the deregulated metabolic processes that characterize cancerous transformation. However, because different oncolytic viruses likely benefit or even require specific alterations in host cell pathways, it has been difficult to identify individual molecular markers that predict specific antitumor efficacies for each oncolytic virus. While this remains an active area of research, the explanations for the precise tumor specificities of specific viruses vary widely from one virus/cell scenario to the next (Russell et al., 2012). Despite this, several overarching themes have become apparent to rationalize selective virus targeting of cancer cells. There is little doubt that the unbridled metabolism of tumor cells provides a selective niche for many viruses that benefit from dysregulated cell growth in general. Additionally, most, if not all, cancer cells during the transformation process undergo alterations that sacrifice elements of their potent cellular, innate antiviral response pathways. Thus, cancer cells in general become collectively susceptible to many more viruses than their parental nontransformed cellular counterparts and are often less responsive to the induction of the antiviral state by self-protective cytokines such as the type I and II interferons (IFNs) or tumor necrosis factor (TNF). There is a complex array of cellular defenses that have developed over evolution to combat virus infections, and, not surprisingly, each successful virus family has developed different strategies to overcome these collective antiviral responses, at least in their specific evolutionary hosts. Given the diversity of genetic alterations documented in cancer cells, it remains unlikely that a single “magic bullet” virus will ever be identified that would treat all cancers equally. Instead, viruses with differing cellular attack mechanisms will have to be matched with pathway-specific cancer cell defects.

Delivery Issues: In vivo, In situ, Ex vivo, or FedEx?

Oncolytic viruses are quite unique as cancer therapeutics, since they are capable of productive replication within the tumor bed and have the potential to “self-amplify, ” thus spreading within and between tumors. This property allows them to be administered in multiple different ways to the patient, including systemic infusion, intratumoral injections, and/or combinations thereof. For localized cancers, for instance, those contained within the skull, it may be appropriate to use strategies like convection-enhanced delivery, which uses a surgically implanted catheter to focus the virus payload in the local vicinity of the brain tumors. Similarly, in diseases like hepatocellular carcinoma, where the standard of care includes locoregional therapy through direct injections or radio frequency ablation, intratumoral virus administration is easily implemented. For patients with metastatic disease, it seems reasonable to propose that intravascular infusion would be the preferred route, as it potentially provides access to all vascularized tumor sites within the body. However, if virus-induced acquired antitumor immunity is in fact the end goal, the blanket infection of every available cancer cell in situ may not be necessary. For example, the recent Amgen melanoma trial (see Table 1) is a clear example where the effective treatment of metastatic disease was accomplished through direct peripheral tumor injections. In this case, the induction of systemic antitumor immunity in therapy responders was accomplished by the direct infection of only a limited subset of tumor lesions manifested by the patient.

Table 1.

Clinical Trials of Oncolytic Viruses

Virus Family	Examples	Genetic Modifications	Target Cancers	Clinical Trial Sponsor
Adenovirus	Oncorine (H101)	Ad-E1b−	Liver, lung, head/neck, pancreas	Shanghai Sunway (approved in China)
	CGTG-102	Ad-GMCSF+	Solid tumors	Oncos
	DNX-2401	Ad-d24RGD	Brain	DNAtrix/Erasmus Medical Center
	ICOVIR-5	Ad-DM-E2F-K-d24RGD	Melanoma	Institut Catala d’Oncologia
	CG0070	Ad-GMCSF+	Bladder	Cold Genesys
	Colo Ad1	Ad3:Ad11p hybrid	Metastatic solid tumors	PsiOxus
Herpesvirus	T-VEC	HSV1-ICP34.5−/4−GMCSF+	Melanoma	Amgen/BioVex (completed Phase III)
	Seprehvir	HSV1716-ICP34.5−	Lung, various solid tumors	Virttu Biologics and Children’s Hospital
	G207	HSV1-ICP34.5−/6−	Brain	MediGene
	HF10	HSV-HF strain	Head/neck, skin, breast, melanoma	Takara Bio
Poxvirus	Pexa-Vec/JX594	Vaccinia Wyeth TK−/GMCSF+	Liver, colorectal, head/neck, others	Jennerex (multiple trials)
	GL-ONC1	Vaccinia Lister-GFP+ F14.5L−/TK−/A56R−	Peritoneal cavity, head/neck, others	GeneLux (multiple trials)
Paramyxovirus	Measles virus	MV-NIS+, MV-CEA+	Ovarian, peritoneal, myeloma, others	Mayo Clinic/NCI (multiple trials)
	Newcastle disease virus	Natural isolate (HUJ)	Glioblastoma multiforme, neuroblastoma, sarcomas	Hadassah Medical Organization
Reovirus	Reolysin	Reovirus-serotype 3	Diverse cancers	Oncolytics Biotech Inc (multiple trials)
Rhabdovirus	Vesicular somatitis virus	VSV-IFNb+	Hepatocellular carcinoma	Mayo Clinic
	Maraba virus	MAGE A3+/Matrix glycoprotein mutations	Lung, colon, melanoma	NCIC Clinical Trials Group
Picornavirus	CAVATAK	Coxsackievirus-A21	Melanoma	Viralytics (multiple trials)
	PVS-RIPO	Polio:Rhino virus chimera	Glioblastoma	Duke University
	Seneca Valley V.	Natural isolate (NTX-010)	Neuroendocrine tumors	Children’s Oncology Group
Parvovirus	H-1 PV	Natural isolate	Glioblastoma	Heidelberg University Hospital

In contrast to treating cancer in situ, one particular subset of virotherapy is to target and eliminate potential cancer cells that can contaminate self-transplant tissues or cells ex vivo, prior to tissue engraftment into the patient. This ex vivo purging strategy offers the prospects of delivering oncolytic virus to all of the potential cancer cells within a given patient transplant sample (for example, in order to eliminate even low levels of cancer stem cells that might reside within an autologous hematopoietic stem cell transplant specimen) (Rahman et al., 2010).

Another virus delivery strategy that has received a great deal of attention but has not yet been exploited in oncolytic virotherapy clinical trials involves using patient cells as virus carriers. This delivery mechanism uses cells, particularly immune cells that exhibit a natural predilection to migrate to tumor sites within the body, as cellular carriers that can be infected with therapeutic virus ex vivo, then infused back into the patient with the hope that they will deliver live oncolytic virus to distant metastatic cancer sites within the patient (Willmon et al., 2009). This has been called the “Trojan Horse” strategy (or, less historically, FedEx delivery of virus to the cancer tissue zip code) and has proven effective in a variety of animal models. One of the issues remaining prior to clinical development of this strategy has been how to best identify the most effective carrier cell for any individual cancer patient. Many cell candidates are under testing as potential carriers for oncolytic viruses in preclinical models (for example, tumor-infiltrating lymphocytes or mesenchymal stem cells) (Willmon et al., 2009). On the clinical front, a virus: cell carrier platform was recently developed that couples autologous adipose-derived mesenchymal stem cells with an oncolytic measles virus (Mader et al., 2013), and this antitumor strategy received US FDA approval to move into clinical testing in ovarian cancer patients early in 2014.

Is Oncolytic Virotherapy Really Anticancer Immunotherapy in Disguise?

The definition of what precisely constitutes “oncolytic virotherapy” has become increasingly more blurred as it becomes clear that any oncolytic virus can be viewed in different perspectives. For example, oncolytic viruses can be considered as delivery vehicles for death signals specific for cancer cells, as gene therapy vectors for expression of therapeutic anticancer genes, or as oncotropic agents with immunostimulatory properties that upregulate antitumor immunity. Perhaps the single biggest change in our thinking about how to develop oncolytic virotherapy is the recognition that it is no longer regarded as critical for the therapeutic virus to directly infect and kill every last cancer cell in the patient. Instead, the most successful virotherapy results in the virus-induced triggering of more effective antitumor immune responses and/or the diminishment of immune suppression that shields tumor cells (sometimes called breaking immune tolerance). Induced antitumor immune responses that can follow active virus replication within tumor beds have emerged as key determinants of successful anticancer virotherapy (Tong et al., 2012; Bartlett et al., 2013). Indeed, some patients undergoing oncolytic virotherapy seem to spontaneously undergo a dramatic and therapeutic transition from induced antiviral responses against the oncolytic vector and then progress onto more effective acquired antitumor immunity. Those patients who have developed enhanced antitumor immune responses following oncolytic virotherapy can be called “elite responders” and have the highest chances to go on to durable disease regressions. Currently, the biggest challenge to the oncolytic virotherapy field is to now carefully study virus-treated patients who become elite responders well after the input virus has been eliminated from their systems. It will be critical to follow the detailed phenotypic characteristics of this patient subset, both before and after their virotherapy regimen, to assess if there are any trackable immune or genetic parameters that can be correlated with their elite responder status. Criteria that assess immune responses, such as HLA-profiles, cytokine expression levels, immune cell activation stages, changes in antitumor antibodies, suppressor cell activation levels, etc., all need to be evaluated to identify measureable parameters that can be used to preidentify elite responders a priori and characterize those patient markers that specifically correlate with tumor regressions.

It is reasonable to posit that elite responders somehow engage their cellular immune systems to respond more effectively to tumor cell epitopes after virotherapy treatment. Assuming that selective virus replication within at least some of the tumor beds within a patient is a prerequisite for this response, there are several possible mechanistic explanations. For example, the cellular cytopathologies caused by productive virus replication in target cancerous tissues could result in improved, or altered, antigen presentation pathways that prime reactive lymphocytes against tumor epitopes. Alternatively, the support for cancer cells provided by resident stromal cells or inflammatory myeloid cells might become compromised. These important support cells can include regulatory immune cells that protect the cancer microenvironment from effective antitumor immunity, inflammatory cells and fibroblastic cells that produce supportive cytokines and growth factors, as well as newly acquired vascular beds that feed the tumors. Thus, disruption of these support cells might allow more leeway for effective acquired immune response pathways to develop against cancer cells. Finally, virotherapy in the elite responder patients might be a trigger to break immune tolerance, either by compromising the functioning of resident immune suppressor cells that infiltrate the tumors (such as regulatory T lymphocytes) or by causing a more global tissue support imbalance that destabilizes local tumor-educated but anergic T cells or myeloid suppressor cells. Perhaps the most intriguing possibility to tip the balance further in this therapeutic direction lies in the prospects that certain chemotherapeutic drugs, or immunomodulatory reagents, have been used to achieve an elite responder status in more patients (Ottolino-Perry et al., 2010; Melcher et al., 2011; Tong et al., 2012; Wennier et al., 2012). With more complete information about the immune pathways that are specifically triggered by viral infection in tumor tissues of elite responders, it should be possible to devise specific therapeutic strategies to more effectively skew patient immune responses to more effective anticancer modalities.

More detailed studies of elite responders are critical to the field, because current animal models do not reflect the accurate range of immunological backgrounds encountered in true human cancers. The two major classes of animal cancer models used to evaluate candidate oncolytic viruses prior to clinical trial are xenograft models of human cancer in immunodeficient mice and variously derived syngeneic/genetic cancers within immunocompetent animal hosts. The advantage of the former class of models is that the replication properties of the test virus can be studied within bona fide human cancer cells, but the drawback is that the cancer graft can only be maintained under immunodeficiency conditions in the recipient test host. Also, the coterie of support stromal and inflammatory cells that are commonly found in true human tumors are frequently absent in the xenograft models. The advantage of the second class of models is that the host animal immune system can be intact, and even accurately mimic some of the immune tolerance aspects found in human cancers, but the cancer cells themselves are not human and may exhibit very different tropism properties with respect to the test viruses. Furthermore, the support interactions with stromal or inflammatory cells within tumor beds can be quite different from those found in the human disease. Thus, although the human oncolytic virus trials to date have been well-crafted for safety, the efficacy profiles have been essentially unpredicted by the animal models used to validate the choices of cancer targets and delivery modalities used for the trial.

Devising Combination Strategies to Improve Virotherapy Efficacy

One of the more exciting developments in current investigative oncology lies in the development of drugs and therapeutics that target the endogenous suppressor leukocytes thought to dampen effective immune responses to cancer (Vanneman and Dranoff, 2012; Gao et al., 2013; Ribas and Wolchok, 2013). Reagents such as monoclonal antibodies targeting the immune inhibitory molecules PD1, PD1L, or CTLA4 are under development as immune adjuvants to selectively assist in breaking tolerance to tumor antigens within cancer patients, often with the prospects of boosting the efficacy of cancer vaccines, but such reagents should also be particularly effective as combinatorial therapeutics with oncolytic viruses (Ottolino-Perry et al., 2010; Melcher et al., 2011; Tong et al., 2012). Although one potential concern is that inhibition of immune suppressor pathways might exacerbate viral infections (Frebel and Oxenius, 2013), the oncolytic viruses tested to date have proven to be remarkably safe even in the most severely immunosuppressed patients or immunodeficient animals. However, the potential for runaway virus infections in a specific subset of patients receiving multiple modalities of immunotherapy in addition to virotherapy cannot be eliminated entirely, and clinical vigilance will need to be continually exercised.

In some respects, immunological augmentation of oncolytic virotherapy is already being tested, in that several oncolytic viruses in clinical trial have been engineered to express proimmune cytokines, such as GM-CSF or IFN-β (see Table 1) (Else-dawy and Russell, 2013). Also, a number of chemotherapeutic drugs in use, such as the nucleoside analog gemcitabine, are already known to reduce the activity of endogenous immune suppressor cells and are potentially synergistic with a variety of oncolytic viruses (Ottolino-Perry et al., 2010; Tong et al., 2012; Wennier et al., 2012). It remains to be seen whether the most attractive clinical course will be to attempt to compromise immune tolerance from the beginning of the virotherapy or whether it improves efficacy to first promote transient immunosuppression to allow for a longer time window for virus replication and dissemination, followed by suppressor immunomodulation with targeted drug therapies. Although animal models can provide clues as to the best strategy, there is simply no substitute for rational approaches to clinical trials.

Lessons Learned (and Learning!) from Oncolytic Virotherapy Clinical Trials

In the laboratory with particular mouse models, oncolytic viruses can be exceedingly effective, with a single dose leading to complete and long-lasting cures (Naik et al., 2012). However, even in well-controlled settings with clonal tumor cell lines, there are numerous mouse models that are refractory to monotherapy. These “preclinical failures” have been incredibly informative and as described above have advanced concepts in the field, including a better understanding of the importance of the tumor microenvironment, innate/adaptive immune responses, and the challenges associated with tumor heterogeneity. Not surprisingly, the path to clinical approval of oncolytic viral therapeutics has not been a straight line, and, as of 2013, no oncolytic viruses have yet been approved by regulatory agencies in the US or EU. There are over a dozen candidate viruses undergoing in excess of 30 approved clinical trials around the world, and the likelihood continues to grow that at least some of these viruses will receive approval in the US/EU markets at some point in the future. Table 1 illustrates a representative spectrum of oncolytic virus trials currently in progress in 2013, as registered by the National Cancer Institute at clinicaltrials.gov, or that are about to be launched.

Results from clinical studies are now supporting several key concepts first uncovered in mouse tumor models: (1) oncolytic viruses can be delivered systemically to sites of metastatic disease by intravenous infusion, (2) physical barriers that limit intravenous virus delivery can be overcome by dose escalation, (3) tumor cells are not the only therapeutic targets of many oncolytic viruses (for example, the tumor vasculature), and (4) immune cells can be both dose-limiting barriers and protherapeutic carriers of oncolytic viruses (Breitbach et al., 2011,2013; Adair et al., 2012).

As with our mouse experiments, it is important to recognize and learn from the “clinical failures.” For example, the low response rate in early trials that used the reovirus-based agent Reolysin as a monotherapy led to this virus being tested in combination with chemotherapy in ongoing randomized trials. The Jennerex Biotherapeutics virus, Pexa-Vec (JX-594), which consists of an attenuated vaccinia virus carrying the gene encoding GM-CSF, recently failed to show a survival benefit over best supportive care in a randomized Phase 2 study in second line hepatocellular carcinoma (HCC) patients. As we now better recognize the critical importance of virus-initiated antitumor immune responses for long-term survival, it becomes even more critical to design virotherapy trials to include patients whose projected median survival will provide ample time to mount an effective therapeutic immune response to their cancers. Indeed, in a small Phase 2 trial in front line patients with HCC, a survival benefit was observed following Pexa Vec therapy (Heo et al., 2013), suggesting that, as a monotherapy, this virus should best be tested in patients earlier in their disease progression, when they possess more robust immune response capabilities.

Perhaps the single most remarkable observation made to date is that, after more than two decades of trials and literally thousands of patients, there have been no oncolytic-virotherapy-attributed deaths, and the frequency of clinically serious adverse events has remained remarkably low, especially when compared to the standard regimens of chemotherapy, surgery, and radiation. This is an even more impressive statistic if one takes into account that most of the cancer patients enrolled in virotherapy trials have already failed standard-of-care therapies and have entered the trial only after their cancers have been sculpted into increasingly treatment-refractory diseases by virtue of their previous intervention failures. Indeed, for those trials where both treated/failed and previously untreated patients have been enrolled (such as the recent Amgen trials for metastatic melanoma), it will be highly instructive to compare detailed response level and survival benefit between these two patient populations.

Conclusions and Prospects

Despite the impressive progress to date, key challenges remain before oncolytic viruses will become the fourth pillar of approved clinical oncology. It is now clear that oncolytic viruses cannot be viewed as stand-alone therapies for any cancer. Instead, they should be regarded as potentially powerful adjuncts to the standard-of-care armamentarium and will likely function best in combination with immunotherapies that improve acquired immune responses against cancers that have already been selected within the individual patient for functional tolerance. The prospect is remote that any single oncolytic virus treatment will infect and kill every last cancer cell within any given patient, including potential cancer stem cells that may remain quiescent and refractory within tissue niches that are inaccessible to virus. It is also clear that cancer tissues are heterogeneous and always include complex mixtures of transformed cancer cells that are variably permissive to the oncolytic virus and which are supported by a complex colony of nontransformed support cells that may or may not be susceptible to the virus. Also, malignant cells can evolve resistance to specific viruses over time, much like they can for chemotherapeutic drugs, monoclonal antibodies, or antitumor cytokines such as TNF or TRAIL, and so there is a limited window of opportunity for any one unique virotherapeutic to be effective. Serial dosing with antigenically distinct oncolytic viruses might offer one alternative strategy to outpace acquired immune compromise of therapeutic virotherapy. Delivery issues remain as well, but the most critical challenge is to identify which virus and delivery method best optimizes the engagement of patient immune cells that mediate the transition to acquired effector immunity against tumor antigens. This transition appears to occur spontaneously in elite responder patients but likely can be specifically triggered with the appropriate cotherapies in other cancer patients as well. But, in our opinion, perhaps the single most important development will be the exploitation of virotherapy much earlier in the disease process, as a first responder component of the front line therapeutic regimens. In this scenario, the prospects for long-term durable responses look to be the highest.