Oncolytic viruses: overcoming translational challenges

Abstract

Oncolytic virotherapy (OVT) is a promising approach in which WT or engineered viruses selectively replicate and destroy tumor cells while sparing normal ones. In the last two decades, different oncolytic viruses (OVs) have been modified and tested in a number of preclinical studies, some of which have led to clinical trials in cancer patients. These clinical trials have revealed several critical limitations with regard to viral delivery, spread, resistance, and antiviral immunity. Here, we focus on promising research strategies that have been developed to overcome the aforementioned obstacles. Such strategies include engineering OVs to target a broad spectrum of tumor cells while evading the immune system, developing unique delivery mechanisms, combining other immunotherapeutic agents with OVT, and using clinically translatable mouse tumor models to potentially translate OVT more readily into clinical settings.

Introduction

Oncolytic virotherapy (OVT) is a promising approach in which viruses selectively replicate in and destroy tumor cells while sparing normal ones. The biological amplification of oncolytic viruses (OVs) by viral replication in the tumor cells is one of the major advantages of OVTs over other cancer therapies (1). To increase their utility as anticancer agents, OVs generally are engineered to further increase their antitumor specificity, safety, immunogenicity, and potency (2). OVs have two main mechanisms of action: first, the direct infection of cancer cells and associated endothelial cells (ECs) that results in oncolysis of these cell types in the tumor microenvironment (TME); and second, antitumor immunity elicited by the OV as a consequence of improved antigen cross-priming and recruitment of immune cells into the TME (3, 4).

Clinical trials have extensively demonstrated the tolerability of OVs in patients (5) and in some cases have shown moderate OV-mediated antitumor efficacy (6, 7), such as the recent phase III clinical trials in patients with advanced or metastatic melanoma treated with talimogene laherparepvec (T-VEC) (ref. 8 and Table 1). However, clinical trials with OVs still have not shown robust antitumor efficacy, especially with oncolytic virus monotherapy. In this Review, we provide an overview of the critical limitations of OVs that have hampered their progress in clinics for therapeutic use and summarize innovative research strategies that have been explored to overcome these obstacles.

Table 1. Clinical trials with OVs in last 3 years.

Enhancing the efficacy of OVTs

During the last decade, development of a new generation of therapies based on OVs capable of inducing tumor remissions in preclinical models has been extensively explored (9–11). A perspective on some of the prevalent strategies exploring different avenues to enhance efficacy of OVT is given below.

Enhancing intratumoral viral spread.

Early clinical trials showed that although OVs accessed tumor cells after intratumoral or i.v. administration, viral replication was generally transient and occurred in localized areas of the tumor, resulting in suboptimal antitumor efficacy (12, 13). Subsequent preclinical studies demonstrated that the main sources of physical barriers to OVs were the extracellular matrix (ECM) proteins, polysaccharides, tumor-associated fibroblasts, inflammatory cells, and high interstitial fluid pressure in the tumor mass (14, 15). Hyaluronic acid (HA) and collagen are major components of ECM, and previous preclinical studies have shown that degradation of HA by a proteolytic enzyme, hyaluronidase, reduces interstitial fluid pressure, permitting anticancer agents to reach breast cancer cells (16, 17). Consequently, ICOVIR17, an armed oncolytic adenovirus expressing hyaluronidase PH20, has been shown to degrade the ECM and enhance spread into the solid tumor mass in xenograft mouse models, ultimately improving the outcomes in treated mice (16). We have previously shown that ICOVIR17 degrades the HA in glioblastoma (GBM) tumors, leading to an enhanced distribution of ICOVIR17 within the tumor and a subsequent significant increase in tumor cell death in mouse tumor models of GBM (ref. 18 and Figure 1A). VCN-01, an ICOVIR17 version with improved tumor targeting (19), has shown therapeutic effects in pediatric osteosarcoma (20) and brain tumor mouse models (21) and is currently being tested in two phase I clinical trials in advanced solid tumors (Table 1). In a separate preclinical study, vaccinia virus (VV) GLV-1h255, engineered to express metalloproteinase 9, led to degradation of collagen IV in the tumor, facilitating intratumoral viral dissemination and resulting in tumor regression (22). Degradation of ECM by relaxin-expressing OVs has also shown increased viral distribution and inhibition of tumor growth (23) as well as tumor sensitization to chemo- (24) and radiotherapy (25) in animal tumor models. OVs expressing decorin, an inhibitor of TGF-β, have also been tested in mouse models of lung and bone metastasis (26, 27). Systemic administration of oncolytic adenovirus expressing decorin in an immune-competent mouse model of lung metastasis modulated the antitumor inflammatory and immune responses via activation of CD8⁺ T cells (26).

Figure 1. Strategies to circumvent the obstacles observed in clinical trials using OVs.

(A) Enhancing intratumor viral spread. OVs engineered to express hyaluronidase (HD) are able to break down HA in the ECM, enhancing the ease of intratumor spread of the OV. (B) Sensitize tumor cells to OV therapy. OVs engineered to secrete proapoptotic proteins revert tumor resistance to OV therapies. (C) Optimizing OV delivery. Carrier cells protect OVs from the immune system and increase tumor targeting of OVs. (D) OV-mediated immunotherapy. OV-mediated oncolysis boosts the immune system response against tumor cells, improving overall therapeutic response.

There are contradictory findings with regards to the function of ECM in tumor metastasis. Some studies have shown that ECM promotes tumor metastasis (28), whereas other studies implicate the degradation of HA in cancer progression and metastasis (29), thus raising concerns about the safety of OVs expressing ECM-degrading factors. However, OVs expressing ECM-degrading enzymes have been engineered to express the transgenes in the late phase of viral replication, resulting in a localized degradation of the ECM around OV-infected cells. This strategy minimizes the possibility of exposing uninfected tumor cells to ECM degradation. OVs have also been engineered to express hyper-fusogenic envelope glycoproteins to improve intratumoral viral spread. Preclinical studies have shown that the expression of these proteins in tumor cells induces cell-to-cell fusion, thus allowing the virus to spread without being exposed to the intracellular spaces (30–32).

Strategies to sensitize tumor cells to OVT.

Many studies have shown that advanced tumors have a tremendous capacity to evolve and develop resistance to a wide variety of therapeutic agents (33). OVs attack tumor cells in multiple different ways, and therefore, tumors acquire resistance less frequently compared with other therapies. However, previous preclinical studies have shown that tumor cell lines have variable levels of sensitivity to OV-mediated killing and can acquire resistance to OVs (34, 35). Continuous exposure of tumor cells to reovirus can lead to resistance. This resistance is mediated by increased protein kinase R phosphorylation, which itself contributes to diminished viral replication potential, but also decreases activity of endosomal cathepsin B, which is required for efficient reoviral entry and activation (34). Another study has shown that tumor cells continuously exposed to oncolytic adenovirus can acquire resistance by blocking the lytic phase of the OV (35). To overcome this resistance, several groups have demonstrated that PI3K inhibitors (36), proteasome inhibitors (36), or rapamycin (37) sensitize OV-resistant tumors to virotherapy. Proapoptotic TNF apoptosis–inducing ligand (TRAIL) has been shown to induce apoptosis in a wide range of human cancer cell lines without significant cytotoxicity toward normal cells (38). We have previously shown that oncolytic herpes simplex virus (oHSV) engineered to express a secretable and potent variant of proapoptotic TRAIL (oHSV-TRAIL) was able to target tumor cells resistant to both TRAIL and oHSV by altering cell proliferation pathways and activating caspase-mediated cell death pathways (refs. 39, 40 and Figure 1B). Similar findings were reported with an oncolytic adenovirus-TRAIL combination in a multiple myeloma mouse model (36). Although previous findings had suggested that TRAIL may have potential liver toxicity after systemic administration (41), localized delivery of TRAIL via OVs has been shown to have limited toxicity (39, 40).

Circumventing antiviral immunity.

A substantial proportion of the human population has already been exposed to OVs and thus presents with preexisting humoral and cellular immunity against many of the OVs currently undergoing clinical development, including adenovirus (42), reovirus (43), VV (44), and measles virus (MV) (45). Consequently, OVs administered into the bloodstream are usually neutralized by antibodies, blood cells, complement, and antiviral cytokines (46) and are cleared by phagocytes in the liver and spleen before they reach the tumor mass (47). Rapid OV elimination from circulation following its systemic administration in patients contributes to the promising safety profile of OVT. However, it also results in a reduced antitumor effect. One logical approach to circumventing antiviral immunity has been the coadministration of OVs with immunosuppressive drugs such as cyclophosphamide (48). Previous studies have shown that four daily doses of cyclophosphamide combined with MV or vesicular stomatitis virus (VSV) were able to significantly reduce antiviral antibody titers in mice, thus allowing effective repeated doses of OVs (49). Different viral families and serotypes within the same family can trigger differential immune and inflammatory responses. Several strategies have been developed to circumvent this, including using low-seroprevalent OVs, molecular engineering of chimeric OVs, and switching viral coat proteins. In the case of adenovirus, studies have shown that Ad5/35 (an Ad5 chimeric adenovirus expressing the fiber proteins of Ad35) reduces toxicity and limits the induction of inflammatory cytokines in murine and nonhuman primate animal models (50). In the case of MV, where serotype switching is not an option, the immunodominant epitopes of the viral surface glycoproteins have been modified by mutating key surface residues to reduce viral immunity (51). However, all these strategies have the potential to alter viral tropism (52). Other strategies that reduce viral neutralization are to polymer-coat the virus (53, 54) or use liposome-encapsulated OVs (55), thereby blocking antibody recognition and extending the circulation times of the viruses in mice (53–55). However, these strategies are associated with a decrease in the binding of the virus to its cellular receptors, resulting in reduced tumor cell infection (53–55).

As mentioned previously, the promising safety profile achieved by systemic administration of OVs could be partially due to OV inactivation by preexisting innate and adaptive immunity. Therefore, toxicity studies should be carefully performed on strategies that reduce OV inactivation to determine whether these approaches modify the safety profile of OVT.

Optimizing OVT delivery.

The efficacy of OVT and other cancer therapies depends heavily on the successful delivery of an antitumor agent in the tumor mass. Early clinical trials demonstrated that intratumoral injection was the most efficient and safest way to administer OVs (56, 57). Recent preclinical studies in a Syrian hamster model suggested that innate immunity against the virus promoted the clearance of injected OVs after intratumoral administration, but did not affect antitumor efficacy (58). As a consequence, repeated intratumoral injections of OVs triggered a robust immune response against the virus, resulting in a therapeutic benefit (58). Since several malignancies, particularly metastatic disease, are inaccessible by direct injection, some groups have explored systemic administration of OVs, which would allow OVs to reach even distant metastases via the bloodstream (59). However, in this instance, a very low fraction of virus reaches the tumor in patients, because of viral neutralization in the blood stream; sequestration of the viruses through the fenestrated capillaries of the lung, spleen, and liver; and nonspecific tropism. Analysis of the tumor biopsies from patients treated with an oncolytic VV revealed that those receiving the highest dose (10⁹ PFUs) showed low amounts of virus within the tumor (60). These findings were supported by studies in mouse models, which detected a very low quantity of virus reaching the tumor mass after i.v. administration of 10⁹ PFUs of VSV (ref. 61 and Figure 2A). Preclinical strategies based on mutating the binding site sequence in capsid genes that interact with blood factors and scavenger receptors on macrophages have been shown to partially increase antitumor efficacy by extending viral circulation time and increasing tumor targeting after systemic administration of the virus (62). Other delivery routes such as intracranial (63) and intraarterial (64, 65) have been explored, but with limited success. Although preclinical studies (66) and clinical trials (45) have shown promising results after i.p. administration of oncolytic MV in patients with recurrent ovarian cancer, this delivery route is limited to patients with peritoneal tumors, potentially reducing its efficacy in patients with metastatic cancer.

Figure 2. Optimizing OV delivery with SCs.

(A) Intrinsic immunotherapy of OVs. OVs administered systemically are phagocytosed by immune cells before reaching tumor cells. Furthermore, some OVs infect normal cells via nonspecific tropism, and other OVs are sequestered in the lung and spleen. (B) SCs loaded with OVs migrate to intact tumors. SCs protect OVs from the immune cells and target them to the tumor sites. (C) SCs loaded with OVs migrate to debulked tumors. Tumor debulking releases tumor antigens and causes inflammation in the surrounding area, attracting SCs loaded with OVs to the remaining tumor deposits.

In recent years, cell-based carriers have emerged as the most promising delivery vehicles for OVTs. These carriers protect OVs from neutralizing antibodies, support viral replication and amplification, and specifically target OVs to the tumor mass (ref. 67 and Figure 1C). In general, OV carrier cells can be classified broadly into tumor cells, immune cells, and stem cells (SCs). To determine the most effective cell carrier for the delivery of OVs, it is essential to consider the susceptibility of the carrier cell to the virus and the kinetics of viral replication and release within carrier cell type as well as the kinetics of carrier cell trafficking from the site of injection to the tumor mass (68). Below, we discuss the available evidence supporting the use of each carrier cell type.

Tumor cells.

Tumor cells proliferate readily and are therefore very permissive to virus infection and replication, resulting in a high viral production after infection (69). A growing body of evidence suggests that tumor progression at this stage may be enhanced by circulating cancer cells’ ability to “self-seed,” a process involving cell dissemination into the vascular system away from a primary or metastatic tumor, followed by the cells rehoming to the site of origin (70). Although the molecular mechanism that tumor cells use to target metastatic deposits is not well understood, it seems to be associated with the same receptor molecule repertoire (cell adhesion molecules, chemokine receptors, or integrin ligands) involved in the metastatic process. This cancer cell tropism to metastatic deposits is supported by several preclinical studies (71, 72), which suggest the potential of cancer cells to be used as OV carriers. While the innate tumorigenic potential of cancer cells raises safety concerns, previous studies have demonstrated that irradiated cells can serve as feeder layers for a certain time before dying (73). Using this time window, irradiating tumor cells just prior to OV infection has been shown to reduce their ability to grow without affecting tumor targeting or the production and release of OVs (74, 75). This poses a potential avenue for the safe use of tumor cells as OV carriers.

Immune cells.

Immune cells naturally circulate in a systemic way and specifically migrate to and recognize tumor cells (76). Among the immune cell subsets, T cells and monocytes/macrophages are the most promising carrier cells. Specifically, T cells loaded with VSV (77), reovirus (78), HSV (79), and Newcastle disease virus (80) have been delivered to tumors in mouse models. Previous studies have demonstrated that engineering T cells with chimeric antigen receptors (CARs) increases the delivery of VSV and VV to tumor cells and that CAR expression and function are not affected by the cell’s infection with OV (81). Despite the efficient homing of T cells to tumors, carrier T cells have limited ability to amplify OVs (81), and moreover this clinical application is challenging and expensive. A few studies have shown that the viability of carrier T cells can be significantly improved by attaching VSV to the membrane of T cells, allowing gradual release of oncolytic VSV into the tumor mass (82, 83).

Tumor-associated macrophages (TAMs) are immune cells that localize to hypoxic regions in the tumor mass (84, 85). Administration by i.v. of TAMs loaded with MV into mice that were bearing myeloma tumors resulted in myeloma cell infection and prolonged mouse survival (86). In another study, macrophages loaded with a hypoxia-regulated oncolytic adenovirus showed a synergistic therapeutic effect when combined with chemotherapy and radiotherapy in a metastatic mouse model of prostate cancer (87). Although the use of different immune cells to deliver OVs offers promise, the overall feasibility of employing them as carriers will require extensive study.

SCs.

SC-based therapies are emerging as another promising strategy to treat cancer. Mesenchymal stem/stromal cells (MSCs) in particular have generated immense interest because they can be easily loaded with OVs (88, 89) and home to areas of inflammation and tissue injury in preclinical tumor models (90). In mouse models, MSC-mediated delivery of oncolytic adenovirus to GBM tumors (91) and lung and breast metastatic tumors (92) has demonstrated therapeutic efficacy. We have previously shown that MSCs loaded with oHSV or oncolytic adenovirus can deliver viral progeny to established GBM tumors, reducing tumor growth and increasing mouse survival rates (refs. 18, 93, and Figure 2B). Although a number of studies have demonstrated that MSCs loaded with OVs have better therapeutic efficacy than naked OVs, tumor-homing and biodistribution of MSC-loaded OVs via different routes of administration require more detailed investigation. A few studies have shown that SCs possess immunosuppressive properties (94–96), suggesting that they would not be ideal carriers for OV-mediated immune stimulation. However, recent studies have clarified that OV infection in SCs induces TLR 9 overexpression and activation of the NF-κB pathway, leading to a specific cytokine secretion profile by SCs and generating a pro-inflammatory environment (97).

Approximately 75% of GBM patients undergo tumor debulking (98, 99), and we have shown that delivering human MSCs encapsulated in biodegradable synthetic ECM (sECM) and loaded with oHSV or its proapoptotic variant, oHSV-TRAIL, into the mouse GBM tumor resection cavity significantly increased survival rates (ref. 93 and Figure 2C). Previous studies have shown significant therapeutic efficacy of MSC-mediated ICOVIR17 delivery compared with direct injection in a mouse model of GBM resection (18). In another promising approach, we have recently shown that intracarotid artery–mediated delivery of MSC-oHSV, but not oHSV alone, was able to selectively target metastatic melanoma lesions in the brain (100). Other studies have demonstrated that intracranial administration of immortalized neural SCs (NSCs) loaded with a fiber-modified oncolytic adenovirus, CRAd-S-pk7, results in a significant survival improvement in xenograft models (101, 102). These preclinical studies have led to an ongoing phase 1 clinical trial investigating the therapeutic efficacy of immortalized NSC-CRAd-S-pk7 in patients with GBM tumors (Table 1). However, the ideal NSC carrier cells for clinical use should be autologous to avoid immune rejection. Recent studies have demonstrated that induced NSCs derived from human fibroblasts have tumor-homing capacity in preclinical settings and therefore offer potential use as OV carrier cells (103).

One of the main constraints of using cellular vehicles to deliver OVs is the toxicity of the viral progeny on carrier cells (91, 104, 105), which ultimately reduces the viral delivery and distribution in and around the tumor mass (106). Therefore, increasing the viability of OV-infected SCs and controlling viral replication within the delivery vehicles are critical and have been studied in detail. Previous studies have demonstrated that the DNA synthesis inhibitor, mimosine, temporarily arrested OV replication after NSC loading, allowing OV-loaded NSCs to migrate to a GBM tumor prior to viral-induced NSC lysis in mice (104). Additionally, the ROS inhibitor N-acetylcysteine amide reduced OV-mediated toxicity by preventing ROS-induced apoptosis in carrier cells without reducing viral progeny (107). In another study, EGFP flanked by FLP recombinase sequences was incorporated into the oncolytic adenoviral genome (108). Using this strategy, carrier cells could be engineered to express FLP-recombinase driven by a hypoxia promoter and loaded with proAd-GFP, allowing reactivation of the OV upon reaching hypoxic areas of the tumor site in mice.

Although SCs loaded with OV will ultimately be killed by the lytic cycle of the virus, it is difficult to ensure whether all carrier cells are infected with OVs or whether some cells escape virus-induced death. Additionally, although SCs have been administered to many patients without considerable side effects, the capacity of any carrier cell to acquire oncogenic mutations is a potential safety concern. Preclinical studies in our laboratory have shown that engineering MSC to express HSV-thymidine kinase allowed selective elimination of carrier cells by administration of the prodrug ganciclovir (109). Such studies add a safety parameter and offer the potential for translating SCs loaded with OVs into clinical settings.

Other OV delivery vehicles and strategies.

Tumor microparticles (TMPs) (110) have been used to deliver OVs into tumor cells in immunocompetent mice and have been shown to overcome the nuclear membrane barrier, thus facilitating the entry of the OV into the nucleus and eliminating tumor cells after OV replication (111). For instance, the ultrasound-mediated delivery of microbubble carriers enabled effective delivery of OVs into the targeted cells (112) by increasing the replicating virus at the tumor site (113) as well as improving bioavailability and intratumoral biodistribution of OVs (114).

OV-mediated immunotherapy

Intrinsic immunotherapy of OVs.

Recently, a number of preclinical and clinical studies have shown that OVs are capable of dramatically altering the TME immune landscape, disrupting immune tolerance to cancer cells and leading to improved antitumor activity alone or in combination with assorted immune modulators (refs. 115, 116 and Figure 1D). OV-mediated cell killing is the first in a series of events that culminates in the induction of a robust and long-lasting antitumor adaptive immune response (117). OV infection triggers immunogenic cell death characterized by the expression of damage- or pathogen-associated molecular patterns (DAMPs or PAMPs), which attract and activate DCs and innate immune cells, respectively, in the TME (118). Once at the tumor site, DCs engulf OV-infected cancer cells and capture tumor-associated antigen (TAA) for cross-presentation to naive CD8⁺ T cells, priming them against tumor cells in the lymph nodes (119, 120). These tumor-specific T cells enter the bloodstream to reach the inflamed tumor site, where they exert their cytotoxic effect in the remaining cancer cells displaying TAAs. Compared with other immunotherapies that use specific TAA identification, OVs vaccinate against a patient’s entire TAA repertoire (4).

In mouse tumor models, adenovirus-induced tumor oncolysis elicited specific T cell responses to a panel of putative neopitopes, whereas novel immune checkpoint inhibitor monotherapy failed to trigger such tumor-specific responses (121). In addition, preclinical studies have shown that adaptive antitumor immune responses primed by i.v. injection of reovirus were independent of tumor oncolysis, indicating that viral replication is not critical to inducing OV-mediated immunotherapeutic effects (122). In addition to tumor cells and immune cells, OVs target other subsets of cells present in the TME, such as cancer-associated fibroblasts (CAFs), ECs, and pericytes, thus aiding in the disruption of the TME structure. Previous studies have shown that OVs can infect and replicate in CAFs (123) and ECs (124), resulting in their elimination and subsequently increasing immune infiltration in the TME. Most preclinical studies have concluded that OVs have the potential to convert immunologically inert tumors into highly immune-reactive ones and have the ability to attack tumor malignancies in multiple different ways by targeting different subsets of TME cells and disrupting the tumor landscape (125). Indeed, OVs create an acute localized inflammatory response in the tumor that favors immune cell recruitment and activation and results in a therapeutic antitumor effect. Recent preclinical studies using TOCA 511, a retroviral-replicating vector that encodes a prodrug activator enzyme, cytosine deaminase, resulted in a therapeutic effect mediated by suicide gene therapy as well as antitumor immunity in metastatic colorectal, pancreatic, and GBM tumor mouse models (126–128). These promising results led to several clinical trials in patients with GBM tumors, including the ongoing phase III trials (Table 1).

Combination of OVs with immunomodulators.

Clinical trials using OVs have highlighted that antitumor immunity, which is critical to achieving clinically relevant therapeutic efficacy, is strongly associated with antiviral immunity. This immunity represents a sizeable hurdle for OVT, since it promotes OV clearance prior to achieving therapeutic impact on the tumor mass (129). In recent years, several studies have provided insights into balancing antitumor and antiviral immunity. One of the most promising strategies is the OV-mediated expression of cytokines, such as GM-CSF, which results in the increased recruitment of DCs to the TME, thereby increasing antitumor immunity (44, 60, and 130–133). The most promising clinical trial to date has been performed with T-VEC (oHSV-GM-CSF) (134), an ICP34.5-/ICP47-oHSV that combines immune stimulation and oncolytic lysis. T-VEC was the first OVT approved by the US FDA for the treatment of inoperable melanoma (135, 136) based on the promising results of a phase III clinical trial (8). Clinical trials with T-VEC have demonstrated patient safety (137) and prolonged patient survival compared with GM-CSF alone in unresected stage IIIB–IV cutaneous head and neck melanoma (ref. 138 and Table 1). Recent clinical trials have sought to demonstrate the efficacy of T-VEC in other types of solid tumors, such as breast and pancreatic tumors (Table 1). A similar approach uses JX-594, a genetically engineered VV with a deletion in its growth factor, VGF, and transgene-driven expression of GM-CSF (139). Clinical trials with JX-594 have shown tumor regression in some patients with hepatocellular carcinoma, lung cancer, colorectal cancer, and melanoma in a phase I/II clinical trial (refs. 7, 140–142, and Table 1). Chemokine ligands such as CCL3, -5, -7, -19, and -20 have also been engineered into OVs to stimulate the activity of antigen-presenting cells and enhance tumor infiltration. These strategies have elicited significant antitumor immune responses in mouse models (refs. 143–147 and Figure 3). In other studies, OVs engineered with IL-12 and IL-15 induced proliferation and activated NK and T cells, elicited potent antitumor effects, and prolonged mouse survival (148–151). In a different approach, oncolytic adenoviruses were armed with soluble EGFR-targeting bispecific T cell–engager (BiTE) antibodies. Tumor cells infected with OVs secreted BiTEs that bound specifically to CD3⁺ T cells and EGFP⁺ tumor cells, resulting in an increased persistence and accumulation of tumor-infiltrating T cells in a mouse model of lung cancer (152).

Figure 3. Oncolytic immunotherapy.

OVs engineered to activate immune system release interleukins or chemokines after tumor cell infection, activating the immune response against the tumor. Coadministered checkpoint inhibitors prevent the inhibition of immune response, ensuring that immune cells are constantly activated against the tumor cells. APC, antigen-presenting cell.

Previous studies have also explored strategies to tip the balance toward antitumor immunity by reducing antiviral immunity against the OVT. Specifically, this approach uses two different OVs: adenovirus followed by a therapeutic VV. In a Syrian hamster model, consecutive OV administration evoked immune system responses only against adenovirus, thus allowing the therapeutic VV to induce its antitumor effect. Furthermore, this strategy showed that the administration of two consecutive OVs can also boost the antitumor immune response (153). Another challenging strategy that favors antitumor immunity is based on designing OVs that turn into oncolytic vaccines by expressing highly specific tumor antigens. As such, VSV and VV have been engineered to express human papilloma virus oncogene E7 (VSV-E7) (154), human dopachrome tautomerase (VSV-hDCT) (155), or human oncofetal antigen 5T4 (VV-h5T4) (156). These therapeutic oncolytic vaccines generated antigen-specific CD4⁺ and CD8⁺ T cell responses in mouse tumors expressing the corresponding antigens (154–156). Although these preclinical studies offer promise, specific viral antigens are still the immunodominant epitopes, inducing stronger immune reaction against OVs than the one against the tumor. A comprehensive analysis of the immunodominant epitopes of each OV family would reveal target epitopes that could be specifically mutated to reduce antiviral immunity. Oncolytic vaccines with specific mutations in the immunodominant epitopes of the virus would potentiate antitumor immunity by reducing antiviral immunity, resulting in a more efficient therapeutic approach that might be translated to clinical settings.

Combination of OVs with checkpoint inhibitors.

The most promising strategy that has the potential to revolutionize treatment options is the combination of OVs with immune checkpoint inhibitors (157). Currently, the most widely studied immune checkpoints are cytotoxic T-lymphocyte associated protein-4 (CTLA-4), programmed death 1 (PD-1), and PD ligand 1 (PDL-1). The interaction between CTLA-4 or PD-1 receptors on T lymphocytes and their ligands in tumor cells triggers an inhibitory signal that reduces proliferation of CD8⁺ T cells, resulting in immune tolerance of the tumor (158, 159). To overcome this dampened T cell response, CTLA-4 inhibitor (ipilimumab), PDL-1 inhibitors (avelumab, atezolizumab) or PD-1 inhibitors (lambrolizumab, pembrolizumab, nivolumab) have been tested in clinical trials (160, 161). Studies have shown that preexisting antitumor T cells in the TME predict favorable clinical responses to immune checkpoint inhibitors (157, 162). This evidence has led to the hypothesis that OV-mediated disruption of cancer cell immune tolerance could synergize with the response to checkpoint inhibitors (refs. 10, 163, 164, and Figure 3). In fact, OVs often induce IFN release in the TME, resulting in an upregulation of PDL-1 expression on tumor cells (165). Previous studies have shown that the combination of reovirus and anti–PD-1 increased the ability of NK cells to kill reovirus-infected tumor cells, reduced immunosuppressive Tregs, and increased CD8⁺ T cells. This enhanced the antitumor immune response (166) and induced a robust memory response (10) in mouse tumor models. When combined with other viruses such as VSV or VV, a PDL-1 blockade also enhanced therapeutic outcomes in murine models of acute myeloid leukemia, colon cancer, and ovarian cancer (164, 167). Recent studies have shown that a triple combination of anti–CTLA-4, anti–PD-1, and oHSV–IL-12 resulted in long-term durable cures in most of the mice treated in two syngeneic models of GBM by inducing a profound increase in the ratio of T effector to Tregs (ref. 11 and Figure 3).

The first clinical trial combining T-VEC and anti–CTLA-4 demonstrated tolerable safety and objective responses compared with monotherapies in patients with advanced melanoma (ref. 168 and Table 1). The antitumor effect was observed in noninjected lesions as well as the injected ones, suggesting that the combination treatment induced a systemic effect. In another clinical trial, patients with advanced melanoma were treated intratumorally with coxsackievirus-21 in combination with anti–CTLA-4. The study showed strong evidence of synergistic antitumor effect, enhancing progression-free survival for greater than 6 months in patients who previously had progressed in response to anti–CTLA-4 monotherapies (ref. 169 and Table 1). Recently, a phase Ib clinical trial combining T-VEC with anti–PD-1 has shown objective response rates (62%) and complete response (33%) in patients with metastatic melanoma (ref. 115 and Table 1). Furthermore, a systematic collection of sequential biopsies of injected and noninjected metastases obtained during different time points of treatment regimens showed an increase of CD8⁺ T cells and IFN-γ expression in the majority of injected lesions as well as some noninjected lesions. A subsequent randomized phase III trial has just been completed and will be able to confirm these promising results (ClinicalTrials.gov NCT02263508 and Table 1).

Conclusion and future perspectives

OVs have been associated with a very favorable risk-benefit ratio and therefore offer a promising therapeutic option for cancer. In general, clinical studies performed thus far have demonstrated that OVs have a relatively tolerable toxicity in patients, with clinical trials reporting mild adverse events, few serious adverse events such as neurotoxicity, and minimal mortality(170). Although a number of exciting preclinical and clinical studies have indicated the strong potential of OVs, this strategy needs to be further improved for successful therapeutic efficacy in clinical settings for a broad spectrum of tumor types.

Intratumoral injection remains the most efficient and safest way to administer OVs. With systemic administration, neutralization in the bloodstream, virus sequestration, and nonspecific OV tropism to the tumor all reduce the number of OVs that reach the tumor. As such, different OV engineering strategies that extend OV circulation time after systemic administration should be continuously explored to increase antitumor efficacy. Cell-based carriers such as immune cells, SCs, and tumor cells have been shown to protect OVs from the immune system, support viral replication and amplification, and specifically target the virus to the tumor mass. In determining the most effective cell for OV delivery, it will be essential to consider the carrier cell’s susceptibility to viral infection, replication, and release as well as its tumor-tracking ability. Developing more sophisticated mechanisms to repress viral replication in carrier cells and selectively reactivate OVs once carrier cells reach distant tumor foci will be critical in using carrier cells to deliver OVs to otherwise inaccessible tumors.

As more OVs progress toward clinical trials, having in-depth knowledge of the immune activation profile of each OV type will be crucial. Tumor biopsies and blood samples collected before and after treatment should be evaluated while planning for future clinical trials with viral vectors. Furthermore, patient selection will be an important consideration: immunocompromised patients may not be good candidates because OV-mediated antitumor immunity could be compromised in these patients. Clinical trials with immunotherapeutic OVs must be designed to consider that their antitumor efficacy requires priming and expansion of immune effector CD8⁺ T cells, migration to tumor sites, destruction of cancer cells, and induction of inflammation (76). Therefore, it is essential that clinical trials with immunotherapeutic OVs consider nontraditional end points to assess the benefit of OV treatments (171).

To conclude, OVTs offer tremendous promise for the treatment of cancer. Although patients who are refractory to the current standard of care may well benefit from this novel approach, eagerness to rush through clinical trials might jeopardize their health as well as the integrity of the OV field. Preclinical fervor should be tempered with caution during this precarious phase, and clinical trials should be carefully designed and have rigorous scientific backing.

Version 1. 03/04/2019

Electronic publication

Version 2. 04/01/2019

Print issue publication