Evolving Status of Clinical Immunotherapy with Oncolytic Adenovirus

Immunotherapy has emerged as a new standard of care for many tumor types, from the adjuvant and neoadjuvant setting to the metastatic stage. In particular, the development of immune checkpoint inhibitors (ICIs) has provided unprecedented clinical benefits in survival and quality of life (1). Despite the success of anti–CTLA-4 and anti–PD-1/PD-L1 antibodies, only a subset of patients will benefit (2). The density, composition, functional state, and distribution of immune cells within the tumor (known as immune contexture) affect the response to distinct immunotherapy strategies, explaining why patients with the same type of cancer differ in the response to the same immunotherapy (3). Accordingly, from an immune perspective, the tumor microenvironment (TME) has been categorized into four major profiles: hot, altered-excluded, altered-immunosuppressed, and cold (1). ICIs and most immunotherapies rely on the presence of immune cells, thereby, cold tumors do not respond well to these therapies (4). These noninflamed tumors show few tumor-infiltrating lymphocytes (TILs), which are mainly caused by poor antigenicity, defects in antigen presentation, and/or immunosuppressive cells in the TME that hamper T-cell priming/activation (5). Failure to recruit dendritic cells (DCs) into the TME prevents T-cell priming in the tumor-draining lymph nodes and, as a result, effector T cells do not infiltrate into the tumor due to the lack of CXCL9 and CXCL10 gradients (6). Moreover, tumors can prevent DC maturation by inactivating DAMP signals or by expressing soluble suppressive factors such IL10, TGFβ, adenosine, and indoleamine 2,3-dioxygenase (IDO), among others (7).

Researchers are testing many approaches to overcome this tumor-induced immune suppression. Several strategies are based on inducing immunogenic cell death (ICD) and subsequent antigen release into the TME, priming the tumor for immune-mediated treatments (8). Although certain types of conventional radiotherapy and chemotherapy can induce ICD, oncolytic viruses (OVs) represent a novel strategy among ICD inducers.

The appeal of OVs relies on their ability to restore the immunologic activity not only by directly infecting and lysing tumor cells, but through the release of cytokines, pathogen-associated molecular patterns (PAMPs), damage-associated molecular patterns (DAMPs), and tumor antigens from virus-infected cells (9). Given the antiviral immune response and physical stroma barriers to intrarumoral spread, OV-mediated infection and lysis may be insufficient to eliminate a large number of tumor cells directly (10), but if the immunosuppressive TME changes, the immune system may eliminate the remaining cancer cells. OVs can also be armed with therapeutic transgenes aiming to kill bystander noninfected cancer cells. Compared to other strategies, OV therapy selectively self-amplify and kill tumor cells, triggering a complex immune stimulation.

Adenovirus (Ad) is a DNA virus that expresses its genes in the nucleus, which means that it can be controlled by replacing viral promoters for tumor-selective promoters. This trait adds a relevant safety layer to oncolytic adenoviruses (OAds), especially when high doses are given systemically. Adenovirus is unique as E1a functions as a master viral switch to regulate the expression of the rest of viral genes, and replacing the E1A promoter by a tumor-selective promoter ensures tumor-selective replication (11). Humans depend on the immune system to detect and eliminate adenovirus infections and, except in immunosuppressed conditions, adenovirus is eliminated efficiently, causing minor or no symptoms.

Adenovirus causes a multimodal cell death process involving necroptosis, pyroptosis, and autophagy (12). OAd-infected cancer cells release new virus progeny, DAMPs and PAMPs (adenovirus proteins and DNA), which will activate nearby DCs and mature them by upregulating costimulatory markers, such as CD80, CD83, and CD86 (9). Mature DCs will process and present tumor-associated, tumor-specific, and viral antigens to T cells in a proinflammatory context, triggering a Th1 profile in the lymph nodes (13). Moreover, adenovirus-infected cells present inflammasome activity due to the released ATP or/and dsDNA sensing in the cytoplasm (Fig. 1; ref. 12). After activation of pattern recognition receptors, DCs produce proinflammatory (e.g., TNFα and IL12) and antiviral (type I IFNs) cytokines, which contribute to tumor antigen cross-presentation and priming of CD8⁺ T cells. Natural killer (NK) cells are also activated and together with CD8⁺ T cells eliminate infected and noninfected tumor cells through IFNγ/granzB and granzB/perforins, respectively. The resultant upregulation of class I MHC-I in noninfected tumor cells due to OAd infection allows them, in turn, to be more exposed to CD8⁺ T cells. A humoral response also occurs: the activation of B cells either by CD4⁺ T cells or in a T-cell–independent manner generates neutralizing antibodies, which target infected cells for NK/type M1 macrophage-mediated cell killing. Simultaneously, distant T cells arrive at the tumor site attracted by the ongoing virus infection, and more macrophages, DCs, NK cells, and neutrophils are recruited into the tumor, reverting the immunosuppressive status from “cold” to “hot” (Fig. 2; ref. 14).

Thus, the antiviral signaling can prime the immune system for a collaterally antitumor response. Preclinical and clinical studies have demonstrated that infected tumor cells induce antitumor responses (15, 16). There is evidence that the immune response is a pivotal for the antitumor activity of OAds independently of virus replication (17, 18). It is challenging to consider that nonreplicating virus may suffice to activate the immune system against the tumor. Many tumors have been injected intratumorally with nonreplicating viruses in the field of cancer gene therapy. In fact, gendicine, an Ad-p53 vector, is approved in China to treat head and neck tumors. However, the experience with gene therapy vectors have not pointed to immune-mediated clinical responses. We believe that virus replication contributes to improve OAd therapy. Virus replication can contribute to eliminate the tumor cells (tumor debulking), to increase the immunogenic cell death, and to release tumor antigens.

The immune-mediated mechanism of action has changed the paradigm in the virotherapy field from a virocentric to an immunocentric point of view. OAds have been designed as immune adjuvants to boost the immune response in the tumor. OAds encoding immunostimulatory molecules or enriched with PAMPs, such as CpG (19) or MyD88 (20), have been combined with different immunotherapies.

Human adenoviruses do not replicate efficiently in mouse tumor models, showing up to a 6-log viral replication deficiency compared with human cancer cells (21). This lack of permissivity represents the main challenge to study the immune-mediated mechanism of action associated to the replication of the virus or oncolysis. Despite this limitation of the preclinical immune competent models, OAds have shown and T-cell responses against tumor antigens and T-cell–mediated antitumor activity (17, 18). OAd monotherapy can elicite T lymphocytes specific against tumor-associated antigens (TAAs) such as mesothelin (22) in mice. Furthermore, taking advantage of the use of adenoviruses as vaccine platforms, OAds coated with TAAs epitopes (23) or expressing them (24) have been tested preclinically as “oncolytic vaccines” with antitumor efficacy and triggered anti-TAA immune activity. The double coating the OAd capsid with both tumor- and pathogen-specific peptides allows to harness the preexistent pathogen-specific CD4⁺ memory T cells in favor of the antitumor response (25).

The current knowledge of tumor immunology has prompted a focus on neoepitope-targeted therapies. OAds treatment increased the tumor neoepitope landscape in mice (16). Moreover, the combination of OAds with DNA vaccines enhanced the immune response against the vaccinated antigens (26). To further illustrate this in this review, we present a concise piece of data that we have obtained with OAds encoding previously published tumor neoepitopes from the mouse models B16-F10 and CMT64.6. OAds with partial E3 deletions (removing 6.7K and 19K) and expressing tumor neoepitopes in a concatenated tandem of minigenes (TMGs; including Ova257 epitope as a control) under a constitutive promoter were generated for each model: ICO15K-d6.7/19K-B16F10TMG and ICO15-d6.7/19K-CMT64.6TMG. When naive mice, without tumors, are immunized strong antiviral responses are detected (i.e., E1b and hexon pool) but the response against the tumor neoepitopes of the TMG reported as immunogenic, such as B16F10-Mut25 or Mut44 (27), is very limited or absent (Fig. 3A). Only the most immunogenic neoepitopes of the TMG according to reported results, Mut17 (27) and Ndufs1 (16, 17), triggered responses (Fig. 3A and B), suggesting that the virus immunodominance masked the putative encoded neoepitope response. This immunodominance has been described for transgenes (28, 29). The antitumor efficacy was tested in the CMT64.6 model. The treatment did not control the growth of established tumors (Fig. 3C and D), despite inducing responses against a cancer neoepitope (Ndufs1) in tumor-bearing animals (Fig. 3E and F). As reported previously, OAd replication does not mediate efficacy in this model (17), and a combination with ICIs (16) or other vaccination strategies (30) is required for efficacy against established tumors. Having mentioned these limitations of a vaccine based on OAds-expressing tumor neoepitopes, it is worth to note that the company EpicentRX is clinically using this vaccination strategy in patients (31), which eventually, if applied to many patients, may clarify the potential of the approach.

After the initial steps of Onyx-015 (32), 40 clinical trials have been completed or are active with OAds (summarized in Table 1; refs. 33–44). Unfortunately, antitumor-specific immunity has been barely investigated. The best results reported in monotherapy have been 6-month complete responders (CRs) in bladder cancer (33) and long-term CRs in glioblastoma (34). Moreover, the kinetics of the regressions in these patients with glioblastoma treated with DNX-2401 occurred several months postadministration after pseudoprogression without detectable viral replication (34), suggesting that the efficacy of the therapy is likely immune mediated. An increase of the tumor immune infiltrates after OAd therapy, reverting the immunosuppression of TME is commonly described previously (34–36). A positive correlation between induced immune response and overall survival was published with ONCOS-102 (36). Although most of the triggered immune cells are probably lymphocytes against OAds, this antiviral response has been postulated to favor antitumor efficacy, as suggested preclinically (45) and in patients (15). Of note, specific T-cell responses against TAAs such as MAGE-A3 (37, 38), MAGE-A1, NY-ESO-1, or mesothelin have been reported after repeated intratumoral injection of ONCOS-102 (37). A major effort should be made to obtain clinical data on antitumor immune responses induced by OVs.

The first clinical trials with OAds tested intratumoral and intravenous administration routes demonstrating significant responses in target lesions after intratumoral treatment (46), whereas limited partial responses in intravnously treated patients (47).

Intratumoral injection ensures delivery at maximum levels independent of tumor-targeting capsid modifications, but it is limited to detectable and injectable lesions. The biodistribution inside the tumor using the intratumoral route is also spatially limited, even by multiple locoregional dosing and image-guided delivery techniques (48), and some injectate may spill over to give considerable systemic overflow (49). Despite efficiency in injected lesions, few responses of distant nontreated metastasis were reported. Intravenous administration potentially reaches multiple micrometastatic and macrometastatic lesions, and it is easily adaptable to different health care centers. The intravenous injection is less appropriate for OVs expressing multiple early genes that can not be selectively controlled, such as herpes virus or vaccinia virus. In contrast, OAds have been administrated intravenously with tolerable toxicity.

Nevertheless, the barriers to successful intravenous therapy are huge: dilution of the virus in the bloodstream that requires higher amounts of viral preparations, neutralization by antiviral antibodies and complement proteins, virus particle sequestration in Kupffer cells and splenic macrophages, limited permeability of tumor vessels, and high interstitial tumor pressure. Despite these barriers, intravenously injected OAds can reach multiple tumor sites (35, 40). Strategies such as cell carriers (41) or capsid shielding (50) might be crucial to improve intravenous delivery of OAds.

Dividing the dose of an agent in multiple administrations within a single cycle is known as dose fractionation. The main value of this approach in the intratumoral administration setting is a more uniform intratumoral distribution of the agent. Closely spaced intratumoral injections deposit virus in different noncontiguous regions of the injected tumor, whereas temporally spaced allow the virus to extravasate into distinct tumor microregions given the changing pattern of microvascular tumor perfusion (51).

On the other hand, splitting the intravenous dose can increase the OAd bioactivity. Preclinically, the predose saturates the Kupffer cells and the postdose is fully available to express transgenes (52) or to reach the tumor (53). In patients, the fractionated dosage slightly increased viral concentration after the last dose (44). Moreover, the innate immune induction also diminished after predosing patients (44). Further evaluations should be done in trials to check how fractionation affects the outcomes in patients.

A therapy cycle is the period of time when the patient is considered under treatment after receiving the therapeutic agent. As mentioned previously, the OAd can be delivered multiple times during a single cycle (fractionated dose), but the number of cycles to repeat the OAd therapy, and also other immunotherapies, is still a controversial issue in the field (54, 55).

Considering that the treatment's efficacy mainly depends on triggering an immune response in the tumor, multiple cycles seem reasonable to heat the tumor as much as possible, as demonstrated preclinically (18). An argument against multiple cycles is that anti-adenovirus neutralizing antibodies (NAbs) impair the treatment efficacy. Hence, repeated cycles seem more appropriate in the intratumoral setting, where the NAbs are not as relevant as systemic administration (18). Different strategies have been published to overcome this neutralization (56), but so far, the only clinical trial with repeated intravenous cycles has been done with enadenotucirev, which reported manageable toxicity and lower activity of the virus throughout cycles (44).

Taking all together, the main trend in clinical trials is to use repeated intratumoral OAd cycles to trigger multiple immune responses in the tumor. Nonetheless, complete responses were achieved with a single-cycle therapy [DNX-2401 (34)] or multiple intravesical administrations [CG0070 (33)]. No clinical studies demonstrated the superiority of repeat dosing versus single. Meanwhile, attention should be paid to avoid continuous therapy without demonstrated benefit for the patient at a considerable cost.

The clinical reality is that few new treatments, including immunotherapies, can replace the current standard-of-care treatment. Thus, new approaches have to be integrated with what is already in clinics. For this reason, and supported by preclinical data, several clinical trials are currently testing the combination of OAds with chemotherapy and immunotherapies (Table 1). Mainstream immunotherapy strategies such as ICIs or stimulators antibodies, immunostimulatory cytokines, TIL therapy, or bispecific T-cell engagers (BiTEs), rely on the presence of immune cells in the tumor. Cold tumors are deemed resistant. The reported recruitment of innate and adaptive immune effectors as a consequence of OV treatment favored the combination with these immunotherapies, led by the combination of T-VEC with ipilimumab (57) and pembrolizumab (58). Regarding OAds, clinical experience showed an increase of CD8⁺ cells in the tumor (34–36, 37), suggesting that OAds can act synergistically with other immunotherapies, as demonstrated in mouse models (59). A relevant aspect of these combinations is the schedule of the different agents.

Chemotherapy-induced cell damage on dividing cells can inhibit the replication of the virus and the immune response. Consequently, timing the combination of chemotherapy with the OAd may be crucial, as demonstrated preclinically with cisplatin: administration of the OAd berore or simultaneously to cisplatin was more efficient than giving cisplatin before the virus (60). Similarly, patients administrated with ONCOS-102 and concomitant cyclophosphamide presented a provocative efficacy (nonrandomized trial) without compromising the induction of antitumor and antiviral T-cell responses (61), suggesting that the virus “immune-kick” is not affected by chemotherapy and both can be synergic (62). However, overlapped adverse effects have to be considered to avoid dose-limiting toxicities.

The effective combination with ICIs or stimulators requires the OAd-triggered innate immune response. Thus, there is a rationale for an injection of OAd followed by these agents, as demonstrated with vaccinia virus (63) or T-VEC (58). However, the disparity in clinical approaches illustrates the need for clarifying data: VCN-01, ONCOS-102, and DNX-2401 are being tested with anti–PD-L1 at different timings (concomitant, 7-day delay, 14-day delay, and other). The results of these trials should provide key data for the proper combination design.

Finally, OAds have been combined with T-cell therapies (64). For example, CAdVEC and TILT-123 are combined with chimeric antigen receptor T cells (CAR T cells) and TILs, respectively (Table 1). The immunogenic nature of OAds could bypass the need for preconditioning and postconditioning regimens for these T-cell therapies, as it is being tested in clinics (NCT04217473). In contrast, the cytokine release syndrome of CAR T cells, characterized by IL6 induction, may add to the IL6 induced by OAds-aggravating toxicities. Thus, the intercalation of OAd and CAR T-cell therapy might be a suitable approach.

The combination of OVs with other approved immunotherapy agents is considered a trans combination. Succesful trans combinations inspired the fusion of the combined agent with the oncolytic virus in a single entity, known as “arming” strategy or cis combination (65). Armed-OAds harbor the gene product or therapeutic transgene in the virus backbone. When the virus effectively infects a cell, the transgene is locally expressed. This avoids systemic toxicity and reduces the cost of producing two GMP products (65).

Nonetheless, when an oncolytic armed virus is used the therapeutic gene will be only expressed while the virus is in the tumor, which is commonly a short time window. If the presence of the combined agent is required over months, cis combination is suboptimal. Similarly, if a different timing of the agents to be combined is needed, this approach might not be suitable. For proteins and RNAs that can be delivered within the virus, the amount of transgene and timing within the viral cycle is crucial for the success of the therapy. In this area, adenovirus stands out from the rest of transgene-carrying viruses (i.e., vesicular stomatitis virus, measles, Newcastle disease virus, herpes, and vaccinia) because it has a highly regulated temporal sequence of viral gene expression starting with the E1A gene, which allows tuning transgene expression. The design of armed-OAd is a key issue for cis combination therapies.

OAds have been modified to encode several different types of transgenes. The function of transgenes can be considered virocentric or immunocentric. Virocentric transgenes would enhance virus cytotoxicity, yields, or spread. For instance, prodrug-converting enzymes or toxins to promote bystander effects. Furthermore, apoptosis stimulates the adenoviral spread at a late stage of virus cycle and OAds have been armed with apoptosis inducers such as p53 (66), TNFα (67), or soluble apoptosis-inducing ligand (TRAIL; ref. 68), among others. OAds modified to digest the extracellular matrix spread more efficiently inside tumors, such as hyaluronidase-armed VCN-01 (69, 70), currently in clinical trials (Table 1).

Immunocentric transgenes aim to enhance immune responses (71). There are two OAds armed with GM-CSF in the clinic: ONCOS-102 (37, 38) and CG0070 (Table 1; ref. 33). Furthermore, OAds expressing combinations of immunostimulatory ligands and cytokines are being tested in phase I clinical trials (Table 1): TILT-123 [expressing TNFα and IL2, (72)] or LOAd703 [CD40L and 41BBL, (73)]. Moreover, armed-OAds evolve in parallel to the immunotherapy landscape to encode anti–CTLA-4 (74), soluble ligands such as OX40L (75), GITRL (76), or a fusion protein sPD1-CD137L (77). An antagonist of crucial molecules for the resistance to immune checkpoint blockade therapies, such as TGFβ, has been expressed from OAds (78).

The generation of BiTEs opened a new opportunity for redirecting the antiviral lymphocytes against TAAs. OAds armed with BiTEs against different tumor targets or cancer-associated fibroblasts have been published with promising results (79–81). The combination of BiTE-expressing OAd with CAR T cells achieved interesting preclinical responses with untransduced and nonspecific CAR T cells in solid tumors (82), which proposes that the combination may overcome different CAR T-cell therapy limitations (antigen loss, transduction percentages, homing to solid tumors). NG-641, currently tested in phase I trial (Table 1) is armed with four transgenes: IFNα to drive dendritic cell priming, CXCL9, and IP-10 to recruit T cells and FAP-BiTE to redirect them.

OAd efficacy depends on the immune system. OAds spread an immunogenic cell death intratumorally that promotes lymphocyte infiltration and overcomes the immunosuppression. Despite the viral immunodominance, this new scenario can reactivate existing or induce the novo antitumor responses. OAds may also be ideal partners of other immunotherapies, scuh as checkpoint inhibitor blockade. The triggered immune response could enhance the homing of other T cells into the tumor, such CAR T cells or ex vivo expanded TILs. The levels of tumor targeting and intratumoral replication to kick off these immune events are unknown, and problably so far are still insufficient when systemic administration ahs been attempted. However, proving that OAds and virotherapy in general elicit immune responses against tumor antigens is a solid basis to continue exploring the potential of this therapeutic strategy.

Farrera-Sal reported personal fees from VCN Biosciences S.L. during the conduct of the study; M. Farrera-Sal also reported personal fees from VCN Biosciences S.L. outside the submitted work. L. Moya-Borrego reported grants from Spanish Ministry of Science and Innovation (MICINN) during the conduct of the study. M. Bazan-Peregrino reported a patent for Use of Viral Vectors in the Treatment of Retinoblastoma issued to VCN Biosciences SL & Hospital Sant Joan De Deu and a patent for “Oncolytic adenoviruses with mutations in immunodominant adenovirus epitopes and their use in cancer treatment” issued to VCN Biosciences SL. M. Bazan-Peregrino is an employee of VCN Biosciences. R. Alemany reported grants from Ministerio de Economia y Competitividad and Generalitat de Catalunya and personal fees from VCN Biosciences during the conduct of the study. No disclosures were reported by the other authors.

R. Alemany is supported by the “Ministerio de Ciencia e Innovacion” BIO2017-89754-C2-1-R and the “Generalitat de Catalunya” 2017SGR449 grants. M. Farrera-Sal received the fellowship 2015 DI 0070 from AGAUR. IDIBELL is a member of Centres de Recerca de Catalunya (CERCA) and is partially funded by this institution. Co-funded by the European Regional Development Fund, a way to Build Europe.