Current development in adenoviral vectors for cancer immunotherapy

Abstract

Adenoviruses are well characterized and thus easily modified to generate oncolytic vectors that directly lyse tumor cells and can be “armed” with transgenes to promote lysis, antigen presentation, and immunostimulation. Oncolytic adenoviruses (OAds) are safe, versatile, and potent immunostimulants in patients. Since transgene expression is restricted to the tumor, adenoviral transgenes overcome the toxicities and short half-life of systemically administered cytokines, immune checkpoint blockade molecules, and bispecific T cell engagers. While OAds expressing immunostimulatory molecules (“armed” OAds) have demonstrated anti-tumor potential in preclinical solid tumor models, the efficacy has not translated into significant clinical outcomes as a monotherapy. However, OAds synergize with established standards of care and novel immunotherapeutic agents, providing a multifaceted means to address complexities associated with solid tumors. Critically, armed OAds revitalize endogenous and adoptively transferred immune cells while simultaneously enhancing their anti-tumor function. To properly evaluate these novel vectors and reduce the gap in the cycle between bench-to-bedside and back, improving model systems must be a priority. The future of OAds will involve a multidimensional approach that provides immunostimulatory molecules, immune checkpoint blockade, and/or immune engagers in concert with endogenous and exogenous immune cells to initiate durable and comprehensive anti-tumor responses.

Graphical Abstract

Oncolytic adenoviruses (OAds) emerged as a promising anti-tumor and immunomodulatory platform to address solid tumor complexities. This review discusses developing OAd-based strategies, including those being clinically tested as single agents and in combination with other therapeutics. Also discussed are aspects of OAd research that must be considered to improve clinical outcomes.

Introduction

In 2020, more than 19 million people were diagnosed with cancer, and nearly 10 million died from cancer-related causes globally.¹ To combat this, novel therapeutics, such as gene and cellular therapies, are being pursued aggressively. Cancer is the most frequent disease targeted in gene therapy clinical trials, with adenoviral (Ad) vectors being one of the most used vectors worldwide.² In recent years, preclinical and clinical results have demonstrated that these Ad-based gene therapies have potential as effective cancer immunotherapeutic agents.

Ads are composed of a linear double-stranded DNA (dsDNA) genome packaged in a non-enveloped icosahedral capsid approximately 90 nm in diameter. Ads possess several features, which make them an ideal vehicle for gene transfer and as oncolytic vectors. First, Ads infect both dividing and non-dividing cells, with a broad cellular tropism. Ad capsids can also be genetically modified to alter the tropism specificity, allowing for a wider or more restrictive set of cellular targets. Second, Ad replication machinery is well studied and is frequently altered to limit replication to specific cells or tissues. Moreover, genome modifications allow for the expression of a variety of transgenes, with later-generation Ad vectors able to accommodate large or complex transgene expression cassettes. Third, Ads can be grown to high titers (up to 1 × 10¹³ viral particles/mL) and produced under clinical good manufacturing practice (GMP) standards. Finally, Ads naturally induce a profound immune response, but clinical symptoms are generally mild, and the infection is quickly cleared in immunocompetent individuals.³

Based on these beneficial characteristics, Ad vectors have been under investigation for decades as a tool for a variety of purposes, from correcting genetic disorders, to vaccines, and as cancer therapeutics. Inducing significant immune responses is a hallmark of the expansive utility of Ads for cancer gene therapy agents. As gene therapy vectors for genetic disorders, an immune response to Ads should be avoided as elimination of transduced target cells will result in the loss of therapeutic transgene expression. As such, less immunogenic Ad vectors have been developed as described below. Ads are excellent vaccine platforms as they stimulate an immune response to function as an adjuvant to the vaccine target antigen (e.g., SARS-CoV-2).^4,5 However, in the context of cancer, where the development of anti-tumor immunity is key, Ads fall short when used as unmodified single agents because the host immune response is skewed toward an antiviral response, failing to generate an efficacious anti-tumor immune response.

Treatments for solid tumors must incorporate multipronged strategies to address the daunting complexity of the tumors themselves, peritumoral extracellular factors, tumor-associated cellular support structures, and significant immunosuppression. Fortunately, Ads provide an opportunity to introduce solutions to address multiple facets of cancer immunotherapy simultaneously and synergize with other potent therapeutic agents as part of combinatorial treatment approaches. In this review, we discuss promising developments in Ad vectors for cancer treatment, and how those modifications induce, supplement, and enhance immune responses (both endogenous and exogenous) against solid tumors, and how these vectors are used as part of a multifactorial approach for cancer therapy.

Adenoviral vectors

Ads are well characterized and extensively sequenced, thus it is easy to introduce genetic modifications using standard molecular techniques (Figure 1). Oncolytic Ads (OAds) were developed to retain their replication capacity, but vector replication is restricted to malignant cells (Table 1). These vectors directly lyse cancerous cells, induce immunogenic cell death (ICD), and stimulate systemic immune responses leading to abscopal effects.6, 7, 8, 9 Details of adenoviral vectorology have been extensively reviewed elsewhere.^3,10

Figure 1.

Genomic structure of adenoviral vectors

Wild-type adenoviral genomes are composed of inverted terminal repeats (ITRs) positioned at the termini denoted in gray; early genes 1A (E1A), E1B, E2, E3, and E4 represented in orange; late genes 1–5 (L1–5) in red; and a packaging signal (Ψ) that is responsible for packaging genomes into virion capsids represented in white. The transgene capacity is represented in green. The 24-bp deletion present in some oncolytic adenovirus virus (OAd) constructs is represented by a light-blue triangle, and the dark-blue triangle shows an example of a cancer-specific promoter insert. Striped sections represent additional deletion sites present in some constructs. The typical transgene capacity for OAds is 2–3 kb, while first-generation and second-generation Ads can accommodate up to 8 and 12 kb. Helper-dependent (HDAd) vectors have the largest transgene capacity, able to encode approximately 34 kb.

Table 1.

Classes of adenoviral vectors

Adenovirus vector	Feature
Oncolytic adenovirus (OAd)	•malleable cellular tropism •highly immunogenic •modified for preferential replication in malignant cells •restricted transgene capacity (2–3 kb)
First-generation adenovirus (FGAd)	•E1 and/or E3 deletion (replication deficient) •transgene capacity (up to 8 kb)
Second-generation adenovirus (SGAd)	•E2 and/or E4 deletion (in addition to E1/E3) •reduced immunogenicity •greater transgene capacity (12 kb)
Helper-dependent adenovirus (HDAd)	•devoid of all viral genes •least immunogenic •highest transgene capacity (34 kb)

Since Ads can be grown to high titers and produced under clinical GMP standards, non-replicating Ads have been developed as vaccines and gene therapy vectors for genetic disorders. The early Ad vectors, so-called first-generation vectors (FGAds), were developed to be replication deficient by deleting E1 genes, as these genes can be provided in trans using cell lines such as HEK293.^11,12 Some first-generation vectors have additional deletions in the E3 region allowing the insertion of larger transgenes. While these vectors are replication deficient, leaky expression of the late Ad genes (i.e., Ad capsid proteins) induce immunogenic responses against Ad,¹³ and thus their transgene expression is short lived.^14,15 To increase the transgene capacity and decrease the immunogenicity of Ad vectors, further genomic deletions were made to establish second-generation Ad vectors. Although the transgene capacity was increased, the expression of these transgenes and the production yield of these vectors are reduced.¹⁶

The latest generation Ads are helper-dependent adenovirus (HDAd), or gutless, vectors devoid of all viral coding sequences, retaining only the packaging sequence and viral cis-acting elements needed for vector genome replication and packaging. These HDAds have a transgene capacity of ∼34 kb and can confer long-term transgene expression (>10 years in non-human primate studies);¹⁷ however, production is more complex as these vectors require a helper virus to provide all viral components in trans.^18,19

Adenoviral vectors for cancer gene therapy

Humans have developed an intricate immune system that senses and responds to Ad infection at multiple levels. This response is divided into two general categories: the innate immune system and the adaptive immune system. The innate immune system functions to sense the initial infection, generate signals to warn neighboring cells, and promote an adaptive immune response. To sense the presence of an infection, innate pattern recognition receptors (PRRs) recognize Ad pathogen-associated molecular patterns (PAMPs), such as viral capsid components and viral DNA. A detailed explanation of these interactions was recently reviewed in depth by our group.³

These innate immune responses educate host adaptive immune responses against Ads, and adaptive immune cells, such as B and T cells, become activated upon binding PAMPs presented by major histocompatibility complexes (MHCs) and then initiate their respective effector responses, such as the generation of neutralizing antibodies or cytotoxic lymphocytes.

The culmination of PRR activation, cytokine and chemokine production, and the expression of type I and type II interferons (IFNs), is a highly immunogenic microenvironment whereby the adaptive immune system can identify and eliminate virally infected cells.

In addition to PAMPs, the lysis of tumor cells by OAds releases damage-associated molecular patterns (DAMPs) which activate PRRs and stimulate an adaptive immune response through ICD.²⁰ ICD induction via viral replication results in the recruitment of professional antigen-presenting cells (APCs) leading to the development of tumor-specific T cells and, subsequently, anti-tumor immunological memory.^20,21 Generating this type of cellular response is the ultimate goal of Ad gene therapies for cancer. However, Ad-dependent oncolysis strategies are limited in their ability to produce complete anti-tumor immune responses.

Tumors are often characterized by a broadly immunosuppressive tumor microenvironment (TME). Meaning, non-malignant cells in and around the tumor, extracellular components, and the tumor cells themselves prevent or attenuate the activation of anti-tumor immune system components, thereby allowing the tumor to escape immune-mediated killing and continue proliferating. Ad-based gene therapies present an opportunity to stimulate anti-cancer immune responses by potentiating the initial antiviral response and shifting the target to tumor cells or components of the TME.

Generation of oncolytic adenoviral vectors

Capsid modifications

The most commonly used serotype 5-based Ads utilize the Coxsackievirus and Ad receptor (CAR) for entry into the host cell. Although there is some debate on the role that CAR plays in tumorigenesis, some studies have suggested that CAR expression is related to tumor progression.²² For example, the expression of CAR in prostate, lung, and brain cancers is greater than that of healthy tissues, which establishes a natural selectivity for Ad5-based therapies.²² However, in colorectal and gastric cancer lines, CAR expression is reduced as the tumor develops, which reduces the efficacy of Ads that rely on CAR for entry. Thus, some Ad constructs increase infection of malignant cells via “fiber switching” from serotype 5 to non-group C serotype Ad fibers; allowing for improved infectivity in a broad range of cancer types.²³ One example of fiber switching utilizes the serotype 3 knob domain allowing for desmoglein-2 targeting, which is often expressed on cancer cells of epithelial origin.^24,25 Another fiber-switching approach utilizes serotype 35 fibers, which exploit the CD46 receptor for cellular entry, as many cancers express high levels of CD46.²⁶ This modification also makes it possible to infect and modify T cells,²⁷ dendritic cells,²⁸ and other hematopoietic cells.²⁹ A third key fiber modification is the incorporation of an Arg-Gly-Asp (RGD) sequence into the Ad fiber. This RGD alteration enhances αv integrin binding resulting in improved Ad infectivity and proliferation compared with its non-RGD predecessor.³⁰ These fiber modifications allow for a CAR-independent entry, circumventing the issue of variable CAR expression across cancer types.

Replication specificity

While enhancing cancer cell infection is an important aspect of OAd effectiveness, a cancer cell-specific entry mechanism is yet to be developed. Instead, cancer-specific replication can be achieved by manipulating the Ad genome. The earliest iterations of OAd vectors were novel because their replication was limited to aberrantly dividing cells. The next iteration of that concept is to capitalize on cancer-specific gene programs.

The early 1A (E1A) and E1B genes are responsible for viral replication by initiating the S phase in the host cell while simultaneously preventing apoptosis by attenuating the function of cell-cycle checkpoint proteins, such as p53 and retinoblastoma (Rb) protein. This manipulation by the virus bypasses the G₁ cell-cycle checkpoint that functions to limit excessive cellular growth and replication. In this way, Ads can piggyback off host cell DNA replication machinery and further their own proliferation.

Ad replication in non-malignant cells is initially dependent on the sequestration of host Rb protein by E1A. Rb is responsible for regulating the cell cycle by directly repressing E2F transcription factors which promote cell-cycle progression. Thus, a deletion in the Rb interaction domain (24 bp) in E1A establishes a viral replication cycle that is relegated to malignant cells that lack a functional Rb protein.³¹

Another popular strategy to promote tumor specificity is to restrict E1A expression with a cancer-specific promoter that is associated with the aberrant transcriptional profile of that tumor type. Several promoters have been explored for tumor-specific OAd replication, such as Cox2 in the context of gastric cancer,³² and telomerase reverse transcriptase (hTERT) for prostate cancer³³ and ovarian cancer,³⁴ as well as osteo- and soft tissue sarcomas.^35,36

Modifications to enhance Ad immunogenicity

Once tumor-specific replication occurs, the next phase of OAd activity is the induction of immune responses against cancer cells through OAd PAMPs or other antigens displaying via MHCs with subsequent upregulation of IFN-stimulated genes, for effector cells to “recognize” and carry out their functions (Figure 2A).

Figure 2.

Current OAd mechanisms of immunostimulation

(A) Oncolysis is the primary method of action of OAds. When the tumor cell is lysed, via apoptosis or immunogenic cell death, cellular contents including tumor-associated antigens (TAAs) and viral particles are released in the intracellular space, which are then taken up by antigen-presenting cells (APCs). APCs then present these antigens (DAMPs, PAMPs, and TAAs) to effector cells, which are then directed to act against the remaining tumor cells. (B) OAds that encode cytokine transgenes enhance effector cell function, promote an immunogenic tumor microenvironment, and improve effector cell persistence. (C) OAds encoding bispecific T cell engager (BiTE) molecules enhance tumor targeting by endogenous and adoptively transferred T cells when tumor cells lack the specific antigen required to activate T cell function via the T cell receptor (TCR). (D) Immune checkpoint blockade antibodies generated by armed OAds counteract T cell hypofunction and exhaustion (gray T cell) by sterically hindering the binding of immune checkpoint receptors and their ligands such as PD-1/PD-L1.

MHCs are membrane-bound protein structures present on all cell types, which function to present peptides to surveilling immune cells. These peptides are generally non-immunogenic or “self” antigens; however, when cells become infected with a pathogen or generate mutated proteins, such as tumor-associated antigens (TAAs), the MHC is then loaded with these peptides which initiate immunostimulatory signals upon binding with immune cell components, such as the T cell receptor. Naturally, many viruses including Ads have evolved mechanisms that prevent antigen presentation via MHC to prolong proliferation. In the case of Ads, the E3/19K protein binds MHC class I molecules in the endoplasmic reticulum, preventing their transport to the plasma membrane, allowing the infection to go undetected.³⁷ Since cancer cells often suppress MHC to prevent the presentation of TAAs,³⁸ deleting the E3/19K gene from OAds can increase MHC class I cell surface expression and enhance tumor immunogenicity.³⁹ However, this could cause an issue whereby immune cells are directed toward the Ad components presented by MHC, rather than TAAs, which may limit the OAd lytic efficacy. One potential approach to overcome that issue is cloaking the Ad capsid with TAAs. Cancer vaccine platforms, such as ExtraCRAd (Ad5, Δ24, CpG), an OAd artificially wrapped in a TAA-coated lipid bilayer, and PeptiCRAd (Ad5/3, Δ24, ΔE3-CR1-α, gp19K, and 14.7K), an OAd conjugated with tumor-derived peptides, have shown improved tumor cell infection and oncolysis compared with naked virus, and improved the number of tumor-specific T cells infiltrated into the TME.^40,41 The TAA-specific immune response can be further enhanced by combining OAd platforms like these with other cancer vaccines, or by encoding the TAA sequence as an OAd transgene.^42,43

The key to an effective Ad vector is striking a balance between vector dissemination and immune activation. Ad infection is quite common, with most people having developed adaptive immunity to the most prevalent group C Ads, such as Ad5.⁴⁴ Therefore, in developing OAds, it is critical to consider the antiviral immune responses to best harness the adaptive immune system for targeting tumors. Fortunately, there are numerous clinical trials underway that are contributing to our understanding of how to maximize the anti-tumor efficiency of OAds.

Clinical trials with unarmed OAds

In most ongoing OAd clinical trials, the vectors used are based on Ad5. Because of this, most Ad vectors are administered intratumorally as this prevents inactivation by neutralizing antibodies and/or the complement system. Furthermore, direct tumor injection ensures that a high proportion of the OAd dose is available to transduce the target cancer cells with limited toxicity to normal tissue (e.g., liver).

The only OAd to be approved for clinical use to date is the E1B-deleted Oncorine (H101), which was approved by the Chinese FDA in 2005 for the treatment of head and/or esophageal squamous cell carcinoma in combination with chemotherapy. Intratumoral H101 combined with chemotherapy resulted in a 78.8% response rate, a dramatic improvement from the 39.6% response rate with chemotherapy alone.⁴⁵ H101 is currently being investigated in a phase II study for malignant pleural ascites with intraperitoneal administration (NCT04771676) and has shown promise when combined with transarterial chemoembolization for treating advanced hepatocellular carcinoma for which a phase III trial is underway in China (NCT03780049). A similar OAd developed in the US, ONYX-015, did not demonstrate significant benefit in two pancreatic carcinoma patients and is no longer being clinically evaluated.

A promising Δ24-RGD OAd (RGD-modified fiber, Rb binding domain of E1A deleted), DNX-2401, has demonstrated safety and anti-tumor efficacy in a recent phase I clinical trial for gliomas. In this study, a single intratumoral injection of DNX-2401 resulted in reduced tumor mass in most patients, including three patients achieving complete responses. Importantly, this study also shows the ability of OAds to stimulate a robust immune response, with an increase in macrophage and T cell infiltration to resected tumors in response to DNX-2401 treatment with concomitant expression of markers of ICD, such as increased HMGBI, HSPs, and DAMPs. DNX-2401 is currently in clinical trials for patients with high-grade gliomas (NCT03896568) and pediatric diffuse intrinsic pontine glioma (NCT03178032).

Telomelysin (OBP-301), an Ad5-based OAd, contains the human telomerase-specific promoter driving Ad E1 gene expression to confer cancer cell selectivity.⁴⁶ In a recent report from a phase I clinical trial of esophageal cancer patients, intratumoral Telomelysin combined with radiotherapy resulted in an impressive 91.7% objective response rate and 8/13 patients achieving local complete responses.⁴⁷ As with other OAds, Telomelysin treatment induces T cell infiltration into tumors. Clinical trials of Telomelysin in combination with chemoradiation (NCT04391049), phase II studies combining Telomelysin with immune checkpoint blockade (ICB) (NCT03921021), and radiation with ICB (NCT04685499) in esophageal and head and neck squamous cell carcinoma (HNSCC) are currently ongoing.

Finally, an OAd based on group B Ads 11 and 3, Enadenotucirev, was generated by directed evolution for its oncolytic activity as well as its potential as a systemically administered OAd. In some cases, intratumoral administration is difficult due to the location or mass of the tumor but also has the disadvantage of restricting the OAd lytic effect to a single treated tumor site. In a phase I study, systemic administration of Enadenotucirev led to stable disease in 5/61 patients with advanced solid tumors (NCT02028442). Repeated dosing was deemed safe because the number of adverse events decreased after subsequent administrations, but most patients developed neutralizing antibodies. Consequently, the development of this humoral immune response decreased the oncolytic activity in successive treatment cycles.⁴⁸ Nevertheless, Enadenotucirev is currently being evaluated in clinical trials, one in combination with chemoradiation for the treatment of advanced rectal cancer (NCT03916510) and another combined with ICB for advanced solid tumors (NCT02636036).

Immunomodulation “armed” Ad gene therapy in clinical trials

The results from studies using unarmed Ads highlight the potential for these vectors to address some of the challenges induced by the immunosuppressive TME as well as the tumor itself. Specifically, the absence of anti-tumor immune cells and/or the prevention of their anti-tumor function. Non-immunogenic or “cold” tumors can either: lack immune cell infiltrates, which are relegated to the tumor mass periphery; or if immune cells are present within the tumor, they are inactive due to inhibitory signals and therefore non-functional. Introducing an OAd instigates an immune response in the proximity of the tumor, turning an immunologically cold tumor into an immunogenic “warm” or “hot” tumor by increasing immune cell trafficking to the tumor and immune cell activation. Increasing the intratumoral immune cell population (APCs, cytotoxic lymphocytes, etc.),³⁷ and promoting anti-tumor phenotypes is a defining feature of OAds. However, the amount of immune stimulation generated by infection and replication of OAds discussed thus far is insufficient to eliminate most tumors. Therefore, OAds have been further modified and armed with transgenes to enhance the anti-tumor efficacy.

Multiple strategies have been explored to improve solid tumor immunogenicity. Initial OAd vectors aimed to capitalize on the expression of co-stimulatory molecules to be presented at the surface of infected tumor cells. For example, the CD40-CD40L interaction functions to promote APC activation, specifically dendritic cell (DC) maturation, and induces T helper type 1 immune responses, which primes the activation of effector lymphocytes, such as cytotoxic T and NK cells.^49,50 Hence, CD40L has been incorporated into a replication-defective Ad vector, FGAdCD40L, which was able to control tumor growth in immunocompetent murine models⁵¹ and determined to be a safe treatment option for patients with solid cancers, alone or in combination with cyclophosphamide.^52,53 To bolster the effect of CD40L, the replication-competent vector LOAd703 was developed to express two co-stimulatory molecules: first, CD40L featuring a trimerized membrane-bound isoleucine zipper motif (TMZ-CD40L), and second, 4-1BBL. The 4-1BB/4-BBL interaction between T cells and APCs provides a co-stimulatory signal that leads to T cell activation, proliferation, and the production and excretion of cytokines, such as interleukin-2 (IL-2) and IFN-γ.⁴⁸ LOAd703 is currently being tested in clinical trials for safety and efficacy in several solid tumor types (NCT03555149, NCT03225989, NCT02705196, and NCT04123470). In an ongoing phase I/II trial studying pancreatic cancer, LOAd703 in combination with standard of care treatment was found to reduce the number of circulating myeloid-derived suppressor cells and/or increase the number of effector memory T cells in a majority of trial participants (NCT02705196). Furthermore, Musher et al. observed elevated tumor antigen-specific T cells in 10/13 subjects.⁵⁴

The incorporation of transgenes into Ad genomes was a crucial step in OAd development. Critically, OAds armed with transgenes encoding for immunostimulatory molecules increases the potency of immune responses locally at the tumor site, which addresses the toxicity associated with systemic administration of the same molecules. Thus, transgenes provide a mechanism to revitalize immune cells present within the tumor mass and/or to increase immune cell trafficking to the tumor (Figure 2B).

The first oncolytic product approved by the US FDA and the European Union, talimogene laherparepvec (T-VEC), is an oncolytic herpes simplex virus (oHSV), which encodes granulocyte-macrophage colony-stimulating factor (GM-CSF). This was a significant step in the feasibility of utilizing armed OVs clinically. One advantage of oHSV as a vector is that it has a larger modification capacity providing the ability to express multiple transgenes within an oncolytic vector.⁵⁵ Typical OAds provide approximately 2 kb of capacity; however, we have developed our armed CAd system to retain oncolytic activity and provide multiple transgenes within our HDAd vector, as we discuss below.

Similar to T-VEC, the serotype 5 OAd construct CG0070 expresses GM-CSF and utilizes a human E2F-1 promoter insert for Rb-deficient tumor-specific replication.⁵⁶ This cytokine induces the maturation of professional APCs, macrophages, and DCs, which can then initiate cell-mediated immune responses via T cell activation. CG0070 is undergoing multiple late-stage clinical trials as a treatment for non-muscle invasive bladder cancer (NCT04452591, NCT04610671, and NCT04387461). An OAd5/3-Δ24-expressing GM-CSF, CGTG-102, demonstrated early evidence in a small number of patients of tumor antigen (Survivin, CEA + NY-ESO, c-myc + SSX2, MAGE-3, WT-1)-specific T cell generation after OAd treatment.⁵⁷ An increase in tumor antigen-specific T cells was also seen after an additional modification was included in this virus (OAd5/3-E2F-Δ24-GMCSF), CGTG-602.⁵⁸ Treatment with CGTG-102, now known as ONCOS-102, also elevated circulating CD8+ T cell populations in additional phase I/II clinical trials and is presently undergoing evaluation as a treatment method for mesothelioma and multiple peritoneal malignancies in combination with ICB or chemotherapeutic drugs (NCT02879669 and NCT02963831).^59,60

Among all of the OAd vectors encoding various cytokines, including but not limited to IL-2 and tumor necrosis factor α (TNF-α) (NCT04217473 and NCT04695327),61, 62, 63 IL-18,⁶⁴ and IL-24,⁶⁵ IL-12-encoding vectors have emerged as some of the most promising across multiple model systems and clinical trials.^66,67 IL-12 is a cytokine produced by APCs to stimulate NK and T cells, thus IL-12 armed OAds are potent vectors for promoting inflammatory responses in the TME. As such, OAds expressing IL-12 are currently being explored across several tumor types, including pancreatic^7,68 and prostate cancer (NCT02555397),⁶⁹ glioblastoma (NCT02026271),⁷⁰ and other solid tumors (NCT03740256).

A multifaceted OAd called NG-641 encodes chemokines CXCL9 and CXCL10 to recruit T cells, and two factors to promote T cell activation: IFN-α and a fibroblast activation protein (FAP)-targeting bispecific T cell activator (FAP-TAc). This engager molecule is purported to reduce the tumor-associated fibroblasts in tumor stroma.⁷¹ NG-641 is currently being evaluated in a phase I clinical trial for advanced solid tumors (NCT04053283).

The results from studying preclinical models and the initial evidence demonstrated in clinical trials underline the importance in using immunomodulatory molecules to improve OAd anti-tumor efficacy. Armed OAds can both initiate and perpetuate an immune response. These findings also introduce another feature added to the OAd arsenal; the ability to supply supplemental targeting proteins that allow T cells the potential to bind and be activated by more than a single antigen, which helps facilitate T cells to switch their target from OAds to cancer.

Engager armed OAd

Bispecific T cell engagers (BiTEs) are a class of immunotherapeutic molecules characterized by one single-chain variable fragment (scFv), commonly CD3, connected to a second scFv that binds a tumor membrane antigen. BiTEs function to encourage anti-tumor T cell activity in an MHC-independent manner, thus allowing for activated T cell populations within the TME to function despite their endogenous T cell receptor specificities. BiTEs have shown to be effective for some patient populations in the context of liquid malignancies, such as FDA-approved Blinatumomab, a CD3xCD19 BiTE for B cell malignancies.⁷² Unfortunately, recombinant BiTEs have a restrictive half-life while in circulation, which necessitates continuous infusions and is subsequently associated with severe toxicities for targeting solid tumors.^73,74 Yet, BiTE-armed OAds and other oncolytic viruses present an opportunity for reduced systemic toxicity, and enhanced lymphocyte function associated with localized expression and persistence (Figure 2C).

Enadenotucirev is among the first OAds to express a BiTE molecule. The first iteration, EnAd-EpCAM-BiTE, was designed to bind an epithelial cell adhesion molecule (EpCAM), which is frequently overexpressed on tumor cells.⁷⁵ Soon after, a sister BiTE-containing OAd called EnAd-FAP-BiTE, was shown to target FAP present on cancer-associated fibroblasts, a major component of the TME that reduces the efficacy of cellular therapies.⁷⁶ Both constructs effectively improved the function of endogenous T cells in the presence of tumors in preclinical models and laid the groundwork for future clinical study. Similarly, ICO15K-FBiTE, an OAd armed with a FAP-BiTE, produced comparable results.⁷¹ These studies demonstrate that targeting TME components as well as malignant cells is not only possible but important for treating complex solid tumors and implies the necessity for additional T cell activation with increased persistence.

Combination strategies

Immune checkpoint inhibitors

As we have discussed, OAds can boost immune cell infiltration and functionality, and improve tumor cell targeting. However, as monotherapies, the anti-tumor effects of OAds, including armed OAds, are insufficient to produce a complete and durable immune response. The potent immunostimulatory activity of OAds to recruit and activate T cells suggests a natural pairing for ICB strategies that allow for these effector cells to overcome the inhibitory molecules expressed by the tumor. The encouraging potential for this multifactorial approach is evident by the plethora of combinatorial clinical trials described above.

A major aspect of the TME is the expression of immune checkpoint molecules, which normally function to prevent aberrant immune cell killing. Immune checkpoints can occur through several molecule interactions: cytotoxic T lymphocyte-associated protein 4/CD80/86, which prevents T cell co-stimulation signaling; programmed cell death protein 1 (PD-1)/programmed cell death-ligand 1 (PD-L1), which attenuates cytotoxic T cell activity and promotes an immunosuppressive environment; T cell immunoglobulin and mucin domain 3/galectin 9 leads to immune tolerance; and lymphocyte activation gene-3/MHC class II binding alters T cell activity directly and indirectly.^77,78 Each of these inhibitory co-stimulatory molecules attenuate T cell function upon ligand binding.⁷⁹ Thus, ICBs that block this interaction have rapidly developed into an ever-expanding cancer immunotherapy field. Although these antibodies show clinical benefit by reversing T cell hypofunction, as evident by the sheer volume of regulatory approvals, these antibodies have limited therapeutic efficacy as monotherapies in many individuals.⁸⁰ These clinical results suggest that patients may lack tumor-specific T cells and/or limited infiltration of these T cells at tumor sites. However, as discussed above, OAds, including armed OAds, accelerate the development of tumor-specific T cells. ICB in combination with Ad vectors provides a method for direct tumor targeting and a multipronged strategy for improving T cell function against cancer cells (Figure 2D).

In a phase II study of glioblastoma and gliosarcoma, DNX-2401 (Δ24-RGD OAd) is being used in combination with Pembrolizumab (αPD-1) (NCT02798406). Preliminary results indicate at least some partial response and 9-month survival for the first seven patients enrolled.⁸¹ Similarly, ONCOS-102 (Ad 5/3 Δ24 GM-CSF) administered before treatment with Pembrolizumab is also under evaluation in patients with advanced or unresectable melanoma in a phase I trial (NCT03003676). A different PD-1 inhibitor, Nivolumab, is being tested in combination with Enadenotucirev in a phase I trial for epithelial tumors (NCT02636036). A recent interim report of 31 patients with metastatic colorectal cancer highlighted the safety profile of the Enadenotucirev/Nivolumab combination and remarked on the induction of CD8 T cell infiltration and upregulation of markers of T cell activation in 6/8 paired tumor samples.⁸²

Several other Ad vectors in conjunction with Pembrolizumab are also under evaluation for a variety of solid tumors, including: a replication-deficient Ad expressing a HSV-tk suicide gene for metastatic triple-negative breast cancer and non-small cell lung carcinoma (NCT03004183); the FAP-TAc, chemokines, and IFNα-armed OAd NG-641 for HNSCC (NCT04830592); and the GM-CSF-armed OAd CG0070 for bladder cancer (NCT04387461).⁸³

However, since systemic administration of ICB is frequently associated with adverse events and toxicities,⁸⁴ the next phase in Ad vector innovation involves the inclusion of ICB transgenes packaged into the vector genome. OAds armed with αPD-L1 transgenes, such as CAdTrio or Ad-Cab, exemplify a growing number of vectors currently in preclinical and clinical testing. Ad-Cab, for instance, encodes for a hybrid Fc-PDL1 antibody, which activates antibody-dependent cell cytotoxicity from multiple effector cells.⁸⁵

OAds with chimeric antigen receptor T cells

OAds armed with ICB have shown that the endogenous immune system benefits from this multipronged immunomodulation.⁷ Unfortunately, in the context of cancer, various factors can affect endogenous immune cells and reduce their prevalence and/or function. For example, over the course of chemotherapy treatments, immune cells suffer collateral damage. Adoptive cellular therapies present a solution by supplying a patient with a functional immune component that was activated and possibly modified exogenously before reintroduction. A well-studied subset of adoptive cellular therapies involves the addition of a chimeric antigen receptor to autologous T cells (CAR-T cells). CAR-T cells armed against various antigens have repeatedly shown clinical efficacy in treating hematological malignancies as monotherapies. However, this success does not translate well when treating solid tumors.⁸⁶ Since OAds generate efficient targeted activation of immune effector cells, the synergistic potential provided by OAds has been demonstrated with CAR-T cells.

We have reported on the efficacy of a locally produced PD-L1 mini-antibody after intratumoral administration of our combination adeno-immunotherapy platform (CAd), CAdPDL1, and the enhancement of clinically tested HER2-specific CAR-T cell (HER2.CAR-T cell) anti-tumor function in prostate and squamous cell carcinoma xenograft models. This local blockade of PD-1/PD-L1 interaction improved CAR-T cell anti-tumor activity and showed improved safety compared with the combination of HER2.CAR-T cells and systemically administered anti-PD-L1 IgG.⁸

Another armed OAd expressing RANTES and IL-15,⁸⁷ delivered in conjunction with CAR-T cells directed at the tumor antigen GD2 (GD2.CAR-T cells) improved CAR-T cell tumor trafficking in a neuroblastoma xenograft model. As expected, intratumoral OAd-IL15/RANTES injection boosted GD2.CAR-T cell infiltration and persistence at the tumor site, significantly enhancing animal survival.⁸⁸

A third combined strategy brings together OAds expressing TNF-α and IL-2 (OAd-TNFa-IL2, or TILT-123) with a CAR-T cell directed to mesothelin (meso.CAR-T cell) in a preclinical pancreatic ductal adenocarcinoma (PDAC) model.⁶² When these two therapeutics were combined in a xenograft mouse model, a single intratumoral dose of OAd-TNFa-IL2 followed by systemic meso.CAR-T cell administration significantly improved median survival compared with meso.CAR-T cells alone. This finding emphasized how TNF-α and IL-2 produced at the tumor site improved T cell infiltration and proliferation, leading to improved T cell anti-tumor function, which slowed tumor growth and thus improved survival. This local OAd/systemic CAR-T cell combination prevented the formation of lung metastases, demonstrating the importance of combinatorial treatments and their synergistic potential.⁶²

Since CAR-T cell persistence at the tumor site is one of the major hurdles for solid tumor treatment, our group also brought together ICB and local proinflammatory cytokine production by introducing an IL-12 transgene into CAdPDL1 to create CAd12_PDL1. We observed that CAd12_PDL1 improved the anti-tumor effects of HER2.CAR-T cells. HER2.CAR-T cells in combination with CAd12_PDL1 were able to control the growth of both primary and metastasized HNSCC tumors in an orthotopic xenograft model.⁸⁹ The addition of IL-12 to CAdPDL1 maintained HER2.CAR expression of adoptively transferred T cells, allowing for continued HER2.CAR-T cell function in vivo.

Although the addition of cytokines into OAd vectors improved adoptively transferred CAR-T cell persistence, residual tumor could remain due to the heterogeneous nature of tumors, whereby the CAR-T cell target protein is not ubiquitously expressed by the tumor mass. Meaning, even though the T cells are activated and directed toward a tumor antigen, if a subset of tumor cells do not express the target antigen, then the CAR-T cell cannot recognize and kill those tumor cells. Therefore, to expand the number of possible CAR-T cell targets, the OAd-BiTE principle went a step further, as exemplified in a combined EGFR-targeting OAd-BiTE with CAR-T cells targeting folate receptor alpha (FR-α). These FR-α.CAR-T cells had previously been shown to be safe in patients treated for metastatic ovarian cancer; however, efficacy was limited by the lack of FR-α.CAR-T cell longevity.⁹⁰ The group observed increases in FR-α.CAR-T cell killing, proliferation, and IFN-γ production when combined with OAd-EGFR.BiTE in vitro. In vivo, this combination slowed tumor growth in colorectal carcinoma and PDAC xenograft models where tumor cells expressed high levels of EGFR but variable levels of FR-α.⁹¹

Based on work done with the CD44 variant 6 (CD44v6) antibody (bivatuzumab) to safely treat patients with HNSCC, and the preclinical studies of CD44v6-targeted CAR-T cells (and currently a clinical study as well [NCT04427449]), our group designed a CD44v6 BiTE-armed Ad, which also expressed the IL-12 and anti-PD-L1 transgenes. This construct, as part of our CAd system, enabled HER2.CAR-T cells to eliminate multiple CD44v6+ cancer cell lines and durable control of HER2-positive and HER2-negative HNSCC orthotopic tumors in vivo.⁹²

While most of these studies were performed in immunodeficient models, and thus it is difficult to extrapolate these results into a dynamic immunocompetent setting, it appears that the superior strategy will involve cooperative OAd and CAR-T cells to initiate and perpetuate immunogenicity in the highly immunosuppressive TME. To that point, we have recently initiated a “first-in-human” clinical trial to study the combination treatment of CAdTrio (expressing IL-12, αPD-L1, and an HSV-tk “safety switch”) and CAR-T cells at Baylor College of Medicine (NCT03740256). The data from this clinical trial will direct us in further optimizing a potent OAd to be used in combination with CAR-T cells.

Future directions for combinatorial therapies with OAds

The recent data from oncolytic viruses, including OAds, have emphasized the necessity of a multidimensional strategy to confront the plethora of challenges presented by solid tumors. Improving tumor immunogenicity to the extent that both endogenous and, potentially, adoptively transferred lymphocytes can eliminate tumors and establish durable immune responses will require additional modifications for significant systemic responses. Targeting aspects of the TME is also a crucial consideration to be made because stromal components, such as CAFs, play a role in determining the efficacy of cancer immunotherapies.

In addition, most Ad-based vectors rely on intratumoral administration for delivery because systemic administration poses several barriers to treatment, including liver sequestration and pre-existing neutralizing antibodies, which would prevent functional vectors from reaching the tumor sites. Ad vectors derived from rare or zoonotic serotypes might be effective upon initial systemic injection; however, subsequent injections have diminishing returns.⁹³ Hence, strategies are being explored to overcome these roadblocks, which involve further vector manipulations.^40,94, 95, 96

One novel method for systemic delivery involves “cloaking” Ad vectors inside cell carrier systems ex vivo, which can then be administered intravenously. These cellular Trojan horses protect Ad vectors as they are trafficked to the tumor, where the virus is then expressed or released in the proximity of the tumor mass.97, 98, 99, 100

These efficacy and delivery hurdles aside, one of the most pertinent barriers to improving OAd cancer immunotherapy strategies is the limitations of our model systems. In vitro systems, including 3D model systems, have the capacity to recreate certain tumor conditions, which allows for a more accurate assessment of vector spread and infectivity. Still, these models have generally simple cellular populations and lack an honest representation of the TME. Furthermore, in vitro systems cannot address questions of systemic immune responses, which are critical considerations for assessing clinical efficacy and toxicity as discussed earlier.

Murine models are thus the standard for OAd evaluation; however, human-derived Ads show limited infectivity and replication capacity in murine cancer cells. Thus, functional OAd studies are relegated to immunodeficient models, where once again the interaction of the immune microenvironment is difficult to recreate, and we can no longer make any inferences on how endogenous immune cells would react to OAd intervention. Our group and others have thus made strides toward developing humanized mouse models to overcome these challenges and generate a preclinical model which will better predict the overall impact of local OAd treatment to both local (TME) and systemic host immune responses. Using humanized mice, we have recently described the ability of clinically testing CAdTrio (oncolysis plus IL-12, αPD-L1 and an HSV-tk safety switch expression) to stimulate both local and systemic host immune responses. However, CAdTrio alone showed limited metastatic tumor growth control, and there was no anti-tumor effect with HER2.CAR-T cell alone in humanized mice harboring multiple pancreatic tumors. Combination of CAdTrio and HER2.CAR-T cells repolarized the TME and induced curable responses of both primary and metastatic tumors in this advanced solid tumor humanized mouse model.⁷

The eventual culmination of these efforts will allow for shorter lag times from the bench-to-bedside, and the equally important return trip from the bedside back to bench, such that we can further improve our OAd therapeutic platforms.