Abstract
Oncolytic virotherapy harnesses genetically engineered viruses to selectively infect and lyse tumor cells while stimulating antitumor immunity through pro-inflammatory microenvironment activation. Celyvir represents a pioneering cell-based oncolytic therapy combining mesenchymal stromal cells (MSCs) as carriers of the human oncolytic adenovirus ICOVIR-5. Developed over two decades through collaborative efforts, Celyvir leverages the tumor-homing capacity of MSCs to enhance systemic delivery and therapeutic efficacy of oncolytic virotherapy against solid tumors. Preclinical and clinical studies, including compassionate use programs and phase I trials, have demonstrated Celyvir’s safety, feasibility, and potential efficacy. Mechanistic insights reveal that MSCs with a low pro-inflammatory profile and patients’ baseline immune competence correlate with improved responses. The evolution from autologous (Celyvir) to allogeneic MSC-based delivery (AloCelyvir) aims to overcome manufacturing delays and optimize clinical outcomes. This review summarizes the conceptual foundations, preclinical models, translational milestones, and ongoing clinical trials of Celyvir, illustrating its trajectory from bench to bedside as an improved oncolytic virotherapy modality.
INTRODUCTION
Immunotherapy is a cancer treatment modality based on the use of therapeutic agents that activate or enhance the immune response. Its development has revolutionized the field of oncology and is currently considered one of the most effective therapeutic options for cancer treatment.1,2 Although significant advances have been made in recent years, the concept of boosting the immune system’s own response dates back more than a century. In the 1890s, William Coley observed that tumors in some patients regressed spontaneously following an infection. 1 Coley subsequently administered a mixture of Streptococcus pyogenes and Serratia marcescens bacteria to patients with inoperable tumors, achieving clinical benefits that in some cases led to cancer remission. 3 Since then, research on the immune system’s role in cancer development and treatment has progressed through various strategies aimed at enhancing the patient’s own immune response. Current immunotherapies are generally categorized into five groups:1,4 immunostimulatory cytokines, cancer vaccines, immune checkpoint inhibitors, adoptive cell transfer (such as tumor-infiltrating lymphocytes or Chimeric Antigen Receptor T-cell therapy), and oncolytic virotherapy, which employs oncolytic viruses (OVs) engineered to preferentially infect and lyse tumor cells without damaging normal tissues. 5
Oncolytic virotherapy
Although certain viruses possess natural oncolytic properties, 5 most OVs currently used in immunotherapy are genetically modified to enhance tumor cell selectivity, improve lytic capacity, and promote antitumor immune responses. 5 This dual mechanism represents a major advantage of virotherapy: in addition to direct lysis of tumor cells, viral infection and the resulting cell death create a pro-inflammatory microenvironment that stimulates systemic antitumor immunity.
Viral infection activates several signaling pathways, including those mediated by interferons (IFNs) and receptors such as Toll-like receptors (TLRs). 6 TLRs, located both intracellularly and on the cell surface, detect pathogen molecules common to bacteria and viruses. 6 TLR activation triggers pro-inflammatory pathways, including nuclear factor kappa B (NF-κB), leading to pathogen-associated molecular patterns (PAMPs) production and the initiation of innate antiviral immunity. 6 Lytic viral replication induces direct tumor cell death through various programmed cell death pathways, including apoptosis, autophagy, and pyroptosis. 7 This process releases both new viral particles and intracellular molecules as tumor-associated antigens, facilitating adaptive immune responses. 7 In parallel, cell lysis liberates PAMPs and damage-associated molecular patterns (DAMPs), such as cytokines, IFNs, and interleukins, which recruit and activate antigen-presenting cells, enhancing T-cell-mediated antitumor immunity.6,7 The local release of PAMPs, DAMPs, and IFNs also activates tumor-infiltrating immune cells, helping to overcome the immunosuppressive tumor microenvironment. 8 Moreover, infected cancer cells serve as viral factories, enabling further rounds of infection, replication, and tumor cell lysis. 6
The first genetically engineered OV was reported in 1991: a herpes simplex virus type 1 (HSV-1) mutant known as HSV1716, which lacked the thymidine kinase gene and demonstrated promising results in human glioma xenograft models. 9 Subsequent research expanded OV development to other viral platforms, including vaccinia virus, reovirus, Newcastle disease virus, poliovirus, and adenovirus (Ad), all of which have undergone extensive preclinical and clinical evaluation.5,10 Currently, four OVs have been approved for clinical use: ECHO-7 (Rigvir), 11 H101 (Oncorine), 12 talimogene laherparepvec (T-VEC, Imlygic), 13 and teserpaturev (G47Δ, Delytact). 14 Rigvir, a non-modified picornavirus, was approved in Latvia in 2004 for melanoma treatment. 11 In 2005, H101, a modified Ad, received approval in China for head and neck cancer. 12 However, these agents primarily rely on direct oncolysis, with limited stimulation of antitumor immunity, reducing their standalone efficacy and prompting combination strategies with other therapies. 15 In 2015, the Food and Drug Administration (FDA) and the European Medicines Agency (EMA) approved T-VEC, a modified HSV-1, for unresectable metastatic melanoma. 13 Multiple clinical studies have since confirmed its safety and efficacy in advanced melanoma. For example, one study in 80 patients reported an objective response rate of 57%, including 39% complete responses. 16 Most recently, in 2021, teserpaturev (G47Δ) received approval in Japan for glioblastoma.14,17
Among the broad diversity of viral families, Ads offer several advantages for gene therapy applications, including high-yield production, efficient concentration and purification, broad cell tropism, and a well-established safety profile in clinical settings. 18 Ads are non-enveloped viruses with an icosahedral capsid and a linear double-stranded DNA genome of approximately 36 kb.10,19 They infect a wide range of vertebrates, from fish to rodents and humans, but exhibit strong species specificity. 19 This characteristic is particularly relevant in pre-clinical models of oncolytic virotherapy, as will be discussed later. Human Ads are classified into seven species (A–G) and more than 100 serotypes. 20 Among them, serotypes 2 and 5 from species C, typically associated with mild respiratory or gastrointestinal infections, have been extensively studied and widely employed in gene therapy. 20 In species C Ads, including Ad2 and Ad5, cellular entry is mediated by the coxsackievirus and adenovirus receptor (CAR), which serves as the primary receptor.21,22 High-affinity binding occurs between the distal knob domain of the viral fiber and CAR. 22 Subsequently, cellular integrins (mainly αvβ3 and αvβ5) interact with the RGD motif present in the penton base of the capsid, promoting virus internalization via endocytosis.20,22 Endosomal acidification triggers partial capsid disassembly and the release of viral proteins such as protein VI, which mediates endosomal escape into the cytoplasm. 23 Viral nucleocapsids are then transported along microtubules to the nuclear pore complexes, where viral DNA is imported into the nucleus. 22 Notably, Ad infection is independent of cell cycle phase and typically does not result in genomic integration of viral DNA. 20 This enables infection of both dividing and nondividing cells and reduces the risk of insertional mutagenesis, contributing to a favorable safety profile. 20 Ads are highly immunogenic, with their capsid proteins, DNA, and other viral products acting as potent PAMPs. Recognition by pattern recognition receptors (PRRs) triggers inflammatory responses, including the release of IFN, cytokines, and chemokines. Several TLRs, such as TLR2, TLR4, and TLR9, contribute to innate detection of Ads.24,25 Their activation initiates signaling cascades involving adaptor proteins MyD88 and/or TRIF, ultimately leading to the activation of transcription factors such as NF-κB.
The Ad5 genome encodes four early genes (E1–E4) and five late genes (L1–L5), transcribed before and after viral DNA replication, respectively. 20 Early gene products activate transcription of downstream viral genes, modulate host cell cycle control, and promote viral DNA replication, while late genes primarily encode structural components. E1 is the first viral gene expressed post-infection and serves as the main transcriptional activator of other viral genes. E1A protein interacts with multiple host pathways, and its role as a master regulator has been exploited in engineering conditionally replicative oncolytic Ads (OAds).19,26 One widely used modification targets the retinoblastoma (Rb) pathway, which is frequently deregulated in cancer cells. 27 In quiescent or G1-phase cells, E2F transcription factors are sequestered by hypophosphorylated Rb. Upon Rb phosphorylation, E2F is released and activates transcription. In many tumor cells, Rb loss, mutation, or hyperphosphorylation results in elevated levels of free E2F. Based on this mechanism, a common modification is the Δ24 deletion (removal of 24 base pairs in the conserved region 2 [CR2] of E1A), which prevents E1A from displacing Rb from E2F in normal cells. 26 Consequently, OAds bearing the Δ24 deletion selectively replicate in tumor cells with deregulated Rb pathways, while replication is restricted in normal quiescent cells (Fig. 1). 20
Mechanism of action of the oncolytic adenovirus ICOVIR-5.
ICOVIR-5 (HAd5-DM-E2F-K-Δ24-RGD) is an OAd developed by Prof. Ramón Alemany and colleagues that exploits this approach. 28 In addition to the Δ24 deletion, the endogenous E1 promoter is partially substituted with an E2F-1 promoter, further enhancing tumor selectivity. 28 To address reduced potency observed in certain tumor models, ICOVIR-5 also incorporates a Kozak sequence upstream of the E1 start codon to enhance translational efficiency. Furthermore, ICOVIR-5 includes a classical OAd modification: insertion of an arginine-glycine-aspartic acid (RGD) motif into the H1 loop of the fiber knob to improve infectivity. CAR expression is often lower in tumor cells compared with healthy tissues, whereas integrins are typically overexpressed in many cancers. The addition of the RGD motif allows CAR-independent infection mediated by integrins, enhancing transduction of tumor cells. 29 The antitumor activity of ICOVIR-5 has been demonstrated both in vitro and in vivo in several pre-clinical models, including glioma, neuroblastoma, lung carcinoma, cervical cancer, and melanoma. 28 In clinical settings, ICOVIR-5 as a naked virus was evaluated in a phase 1 trial for patients with metastatic cutaneous melanoma (ClinicalTrials.gov: NCT01864759). 30 Intravenous administration was well-tolerated, although some hepatotoxicity (transaminitis) was observed at higher doses. Despite detection of viral genomes in tumor biopsies, no significant clinical responses were reported in treated patients. 30
Beyond H101 (Oncorine), a total E1B 55 kD and partial E3 genes deleted human type-5 adenovirus, approved in China for head and neck cancers, 12 other OAds with similar design strategies are undergoing clinical evaluation. ONYX-015 (dl1520), a chimeric human group C adenovirus (Ad2 and Ad5) that carries a deletion in the E1B region allowing selective replication in p53-deficient tumor cells, was the first oncolytic virus to enter human clinical trials and has been extensively studied. 31 VCN-01, another Ad5-based OAd designed for selective replication in Rb pathway–deficient tumor cells, showed favorable tolerability in a phase 1 trial, where intravenous administration combined with chemotherapy achieved partial responses in 50% of patients with pancreatic ductal adenocarcinoma. 32 Other candidates, such as Enadenotucirev (formerly ColoAd1), a chimeric group B adenovirus derived from Ad11p and Ad3 33 ; Telomelysin (OBP-301), an attenuated Ad5-based OAd in which viral replication is driven by the human telomerase reverse transcriptase (hTERT) promoter 34 ; CG0070, which combines oncolytic adenoviral replication with GM-CSF expression, and DNX-2401, both designed to selectively replicate in tumor cells with dysregulated Rb signaling, are also in various phases of clinical development.35,36 These examples highlight the growing portfolio of OAds that share molecular features with ICOVIR-5, exploiting tumor-specific defects in cell cycle regulation for selective viral replication.
Limitations of oncolytic virotherapy
Oncolytic virotherapy still faces several limitations that must be addressed to enhance its clinical efficacy. Key challenges include antiviral immunity, potential toxicity, systemic delivery barriers, and limited antitumor efficacy. One of the main obstacles is the host immune response, which often neutralizes the virus before it can exert its therapeutic effect. Following administration—particularly via systemic routes—OVs are recognized as pathogens, triggering antiviral responses that compromise antitumor activity. In this regard, preclinical studies using naked OVs have consistently shown reduced viral replication, accelerated clearance, and diminished antitumor efficacy in immunocompetent murine models compared with immunodeficient models. 37 Viral particles may be cleared by innate immune mechanisms or neutralized by pre-existing antiviral antibodies, which is particularly problematic for OVs derived from common pathogens such as Ads or herpesviruses. 38 To counteract this limitation, most clinical protocols employ multiple dosing regimens, as single-dose administration generally produces suboptimal responses. 39 A critical balance must therefore be achieved between stimulating antitumor immunity and mitigating systemic antiviral immunogenicity—an issue of considerable relevance in the field of oncolytic virotherapy.
Another major limitation is the restricted tumor tropism or homing capacity of OVs. In general, viral particles exhibit limited capacity to selectively migrate to the tumor microenvironment. Intravenous delivery often results in off-target infection or sequestration in nontumor tissues such as the liver, leading to toxicity 40 ; and intratumoral injection is often unfeasible in cases of deep-seated or metastatic disease, as not all tumors are accessible. Although abscopal effects have been reported following intratumoral delivery of OVs, 41 this approach is generally less effective for treating disseminated disease or inaccessible metastases.
While oncolytic virotherapy represents a promising immunotherapeutic strategy for cancer, systemic delivery of naked viral particles still yields only modest clinical outcomes. 42 To overcome these challenges, several strategies are currently being explored, including genetic capsid modifications, incorporation of targeting ligands, or physicochemical surface engineering approaches.43–45 Among these, polyethylene glycol (PEG) conjugation (PEGylation), has emerged as a promising strategy to improve viral pharmacokinetics and immune evasion. PEGylation masks the viral surface, thereby protecting the virus from neutralizing antibodies while retaining infectivity. Early studies demonstrated that PEGylated Ads maintain their ability to infect target cells and are shielded from antibody neutralization, potentially enabling repeated administration. 46 However, the development of innovative delivery approaches and combination strategies remains essential to fully realize the therapeutic potential of OVs.
MSCs as cell-based carriers for OVs
Strategies to optimize oncolytic virotherapy and enhance targeting include the use of nanoparticles, biomaterials, and cell-based carriers.42,47 Several progenitor cells, such as neural or MSCs, 48 immune cells, and even tumor cells, 49 have been explored as delivery vehicles, each offering distinct advantages and limitations. In this “Trojan horse” approach, carrier cells transport OVs intracellularly to the tumor site. Cell-based carriers help the virus to evade immune recognition, as viral particles are shielded from host antiviral responses during peripheral blood circulation. 47 This strategy prolongs viral persistence and enhances delivery to the tumor niche. Notably, it allows the administration of high viral loads without inducing significant toxicity. 50
In addition to immune protection, selecting carrier cells that support active viral replication can amplify viral load en route to the tumor, ultimately increasing the number of virions released in situ. However, replication must be balanced to avoid compromising carrier cell viability and function. Optimal pairing of virus and carrier cell type is essential, taking into account factors such as viral life cycle, carrier cell homing kinetics, and therapeutic context. 50 Using carrier cells with intrinsic tumor-homing capacity can help overcome the limited migratory properties of free viral particles. 7 Tumor homing by carrier cells may be driven by direct tumor tropism, as well as by microenvironmental cues such as hypoxia or angiogenesis, or by the anatomical location of the tumor itself. 51
The intrinsic properties of MSCs and their broad clinical use make them attractive candidates for OV delivery.47,51 First isolated from bone marrow by Friedenstein and colleagues in 1968, 52 MSCs have since been derived from numerous other sources, including adipose tissue, dental pulp, muscle, dermis, umbilical cord, and even menstrual blood.53,54 MSCs are multipotent cells with self-renewal capacity and the ability to differentiate into osteogenic, adipogenic, and chondrogenic lineages, as well as other mesodermal and nonmesodermal cell types. 47 In the absence of a specific marker, human MSCs are defined by their adherence to plastic, expression of CD105, CD73, and CD90 surface markers, and lack of CD45, CD34, CD11b, CD19, and HLA-DR. 47 Their versatility in terms of tissue source, accessibility, availability, and ease of in vitro expansion are among their key advantages for clinical use.
Indeed, MSCs are widely employed in clinical settings. In 2018, the EMA approved the use of allogeneic MSCs (Alofisel) for the treatment of Crohn’s disease. 55 More recently, in 2025, the FDA also approved allogeneic MSCs (Ryoncil) for steroid-refractory acute graft-versus-host disease(SR-aGvHD). 56 Additionally, other MSC-based therapies have been approved outside Europe and the United States, further underscoring their translational potential. These include Cartistem for cartilage defects, 57 Cellgram-AMI for improving cardiac function after acute myocardial infarction, 58 and NeuroNata-R for the treatment of amyotrophic lateral sclerosis, 59 all approved in South Korea; TEMCELL, the first fully approved allogeneic cell therapy in Japan for GvHD, 60 Stemirac, approved in Japan for patients with spinal-cord injury 61 ; and Stempeucel, approved in India for the treatment of critical limb ischemia. 62
MSCs exhibit immunological properties that enhance their clinical versatility. These include low expression of class I HLA molecules and absence of key costimulatory molecules such as class II HLA, CD40, CD80, and CD86. 50 Their low immunogenicity allows for allogeneic MSC transplantation without the need for immunosuppression—a critical feature for immunotherapy applications. 63 However, despite this low immunogenicity, MSCs can elicit humoral and cellular immune responses in vivo and interact with components of the innate immune system, through both direct cell–cell contact and the secretion of soluble factors.63,64
MSCs also exhibit phenotypic plasticity, influenced by their microenvironment and stimulation of TLRs. 64 In particular, stimulation of TLR4 can polarize MSCs toward a pro-inflammatory MSC1 phenotype, characterized by expression of pro-inflammatory mediators, and a reversal of immunosuppressive mechanisms. 65 As suggested by several authors, this inflammatory phenotype should be carefully considered when combining MSCs with OVs to achieve a balanced therapeutic outcome. 63
MSCs possess the ability to home to specific tissues such as sites of injury or tumor niches.47,63 MSC tumor tropism appears to be influenced by tumor microenvironmental factors such as oxygenation status, vascularization, and inflammation. 66 Although chemokines, adhesion molecules, and matrix metalloproteinases are implicated in this process, 63 the precise biological mechanisms remain incompletely understood.
During ex vivo expansion, MSCs are continuously proliferating. OVs can exploit the natural transition of MSCs into S phase—during which Rb and E2F separate—to initiate viral replication. As a result, MSCs can amplify the viral load they carry.67,68 Timing is critical, as viral replication must align with MSC migration to the tumor. Given that MSCs are typically home to tumors within 24–48 h of systemic administration,69,70 viruses with longer replication cycles are preferable, so that the final stages of viral replication and release occur within the tumor microenvironment. 50 In this regard, human Ads, which generally replicate over 72 h, are considered well-suited for delivery by MSCs. 71
Celyvir: Mesenchymal stromal cell-based delivery of OAd
Building on this rationale, the therapeutic strategy known as Celyvir was conceived over two decades ago by Prof. Javier García-Castro and Prof. Manuel Ramírez Orellana (Hospital Infantil Universitario Niño Jesús), in collaboration with Prof. Ramón Alemany (Catalan Institute of Oncology–IDIBELL). This approach employs autologous MSCs as carriers for the human OAd ICOVIR-5, previously described, for the treatment of pediatric solid tumors.72,73
In brief, the production of Celyvir involves the ex vivo expansion of patient-derived MSCs followed by infection with ICOVIR-5, generating the final therapeutic product (Fig. 2). The use of MSCs as a delivery vehicle enables systemic administration of Celyvir. The mechanisms underlying the therapeutic activity of Celyvir are under active investigation and have been a major focus of the evolution of this work. The strategy of using MSCs as carriers for various OVs has also been explored by other groups. For example, MSC-mediated delivery of oncolytic HSV has shown promising results in preclinical models of breast and ovarian cancer, glioblastoma, and brain metastases.63,74,75 Other studies have investigated MSC delivery of oncolytic vaccinia virus, measles virus, myxoma virus, and reovirus.76,77 In many cases, the combination of MSCs with OVs enhances therapeutic efficacy compared with the virus alone.63,77
Mechanism of action of Celyvir therapy. The conceptual infographic illustrates the proposed mechanism of action of Celyvir, in which mesenchymal stromal cells (MSCs) serve as carriers for an oncolytic adenovirus.
Currently, the clinicaltrials.gov database lists six clinical trials investigating the use of MSCs combined with OVs for cancer therapy (Table 1), all in early-phase studies (Phase 1 or Phase 1/2). Four of these trials involve Celyvir therapy using ICOVIR-5, while the others employ the OAd DNX-2401 or a modified measles virus (MV-NIS).
Clinical Trials of MSCs as Cellular Carriers of Oncolytic Viruses
The NCT and/or EudraCT number of each clinical trial, status, type of tumor being treated, the oncolytic virus used, and the trial location are provided. Data collected from the ClinicalTrials.gov repository as of March 2026.
C, completed; NT, not recruiting; R, recruiting.
PRECLINICAL MODELS
Preclinical models remain a critical limiting factor in the clinical development of immunotherapies, often contributing to discrepancies between preclinical and translational outcomes. The pivotal role of the immune system in immunotherapies makes model selection a decisive element, depending on the specific scientific question being addressed. In oncolytic virotherapy, establishing an appropriate preclinical model is particularly challenging, as it must accommodate both efficient viral replication in the tumor and complex interactions between the immune system and the virus. In the case of Celyvir, these challenges are further compounded by the intricate relationship between the cellular and viral components of the therapy. Three key factors must therefore be considered when designing preclinical models for Celyvir: (1) permissiveness to OAd replication, (2) interactions between MSCs and the OAd, and (3) involvement of both innate and adaptive immune responses.
Human Ads, including ICOVIR-5, exhibit species specificity, 19 which restricts their capacity to replicate in murine cells and limits their oncolytic activity in mouse tumor models. Consequently, the antitumor efficacy of OAds is typically evaluated in immunodeficient mouse models that allow for the engraftment of human tumor xenografts. 78 While such models can partially assess viral oncolysis, the absence of a functional immune system precludes the study of virus-induced antiviral and antitumor immune responses. Some animal species do exhibit partial permissiveness to human Ad replication, such as the cotton rat (Sigmodon hispidus) and the Syrian hamster (Mesocricetus auratus).79,80 In this context, Toth and colleagues developed an OAd preclinical model based on the cotton rat. 81 This approach enabled in vivo evaluation of human OAds VRX-007 and Ad.OW94 for the treatment of osteosarcoma, fibrosarcoma, and spindle cell sarcoma, using syngeneic rat tumor lines CCRT, LCRT, and VCRT. 82 The Celyvir strategy was also implemented in this model, utilizing cotton rat MSCs infected with human OAd to investigate tumor homing of the therapeutic cells and activation of adaptive immune responses. 71 Similarly, Thomas and colleagues established an immunocompetent model based on the Syrian hamster, 83 which has since been employed to test various OVs in carcinomas, head and neck cancers, and pancreatic cancer.83–85 These models represent a significant advancement over immunodeficient systems, as they preserve both viral replication and immune competence. However, their use is limited by the need for syngeneic tumor lines from cotton rats or Syrian hamsters and the requirement for specialized breeding and housing facilities.
Semipermissive replicative model
Mice remain the most widely used species in biomedical research, supported by an extensive toolkit of reagents and methodologies, and continue to serve as a primary model for human Ad studies. Despite the inherent limitations, certain murine tumor lines, such as ADS-12 (lung adenocarcinoma), KLN205 (lung carcinoma), CMT64 (lung adenocarcinoma), and KPC-I (pancreatic adenocarcinoma), do permit human Ad infection and semi-productive viral replication.86–88 Our group developed a tumor model based on the CMT64 line, which has been utilized in several studies included in this review. Other investigators have also employed CMT64 in the context of oncolytic virotherapy, testing approaches such as artificial encapsulation of OAds with cancer cell membranes 89 or combining virotherapy with anti-PD-1 immune checkpoint blockade. 90 Among more than 30 CMT64 clones tested, Rincón and colleagues selected clone CMT64-6, which exhibited the highest level of human Ad production, to establish syngeneic tumors in immunocompetent C57BL/6 mice. 71 Although viral replication in this model is only semi-permissive—lower than in human cells or even suboptimal—this system enabled the in vivo research of the antitumor efficacy, tumor homing, and immune responses elicited by human ICOVIR-5, both as a naked virus and in the Celyvir platform. 71
The relative ease of implementing the CMT64-6 model, its versatility for immunological analysis, and the similarity of observed immune responses to those seen in patients have contributed to its adoption by other research groups. It has since been employed in the evaluation of novel OAds such as ICOVIR17K-iRGD, ICO15K-40SAPH20, and ICO15K-E1aPH20.91,92 In the original study, adenoviral particles were detected only in tumors treated with intratumoral ICOVIR-5 injection, but not in those treated systemically with Celyvir. 71 However, subsequent work using immunohistochemistry successfully detected adenoviral presence in tumors following systemic Celyvir administration. 93 Nonetheless, the modest antitumor effect observed in this model (approximately 35% tumor reduction), together with the low levels of adenoviral replication, suggest that this system may underestimate the true oncolytic potential of the therapy.
These challenges highlight a broader limitation in the development of preclinical models for cell-based oncolytic virotherapy approaches such as Celyvir: the complex interplay between carrier cells and OVs. In our particular model, Celyvir was produced using murine MSCs loaded with human ICOVIR-5. This interspecies pairing limits viral amplification during MSC migration to the tumor and may alter cell signaling upon Ad infection, deviating from what is observed in the human clinical setting. Species-specific differences in Ad–host interactions have been well documented, 94 and these evolutionary distinctions likely explain the absence of a viral amplification effect within the carrier cells, resulting in fewer infectious particles reaching the tumor. Combined with restricted viral replication and subsequent rounds of infection, this factor likely contributes to the underestimation of the true oncolytic effect in this model. Moreover, these interspecies differences may also affect how Celyvir therapy interacts with the murine immune system, potentially altering both antiviral and antitumor responses compared with those seen in patients.
Humanized mouse models
Humanized mouse models offer a valuable platform to study immune responses in a setting more closely resembling the clinical scenario. Since immunocompetent mice rapidly reject human cells or tissues, recipient strains are typically immunodeficient, with NSG, NRG, and NOG being the most commonly used for the engraftment of both human immune and tumor cells. 95 Among the available humanized models, we employed a peripheral blood mononuclear cell (PBMC)-based approach to evaluate the Celyvir strategy. 54 This model involves the administration of human PBMCs—comprising a heterogeneous population of T and B lymphocytes, natural killer (NK) cells, monocytes, and dendritic cells—into NSG mice. The immunodeficient background of these strains allows us to use human lung adenocarcinoma cell line xenografts as tumor models. These human tumors support efficient OAd replication, enabling multiple rounds of viral infection and thus a more comprehensive evaluation of the oncolytic effect.
One of the limitations of PBMC-based humanized models is the development of a graft-versus-host disease (GvHD) due to the human-mouse xenogeneic mismatch, which can affect experimental outcomes. 95 This effect restricts the experimental window to a few weeks. 96 Tracking studies of the Celyvir approach conducted by our group and others demonstrated tumor homing within 24–48 h following systemic administration.70,93,97,98 Consequently, PBMCs were administered one day prior to Celyvir treatment in our model, allowing us to examine their influence on both migration to the tumor site and activity within the tumor microenvironment. During the 2-week observation period for tumor growth, we did not detect lethal responses associated with GvHD, although we cannot fully rule out its contribution to certain interactions. In line with findings reported in multiple publications, the humanized model confirmed the pivotal role of the immune system in mediating the therapeutic effect of Celyvir, as in vivo antitumor efficacy was observed in mice that also received PBMCs. 54
Unlike other humanized mouse models that require genetic modifications, the PBMC-based model allows relatively straightforward in vivo manipulation of immune subsets through isolation or depletion strategies. For example, depletion of monocytes and NK cells from administered PBMCs abrogated the antitumor efficacy of Celyvir, highlighting the essential role of these innate immune populations in the observed therapeutic response. However, several limitations of this model—including poor engraftment of myeloid cells, limited antigen-presenting cell function, and reduced survival of NK cells and T lymphocytes 95 —restrict the development and study of adaptive immune responses. 99 This represents a significant constraint, given that adaptive immunity has been shown to be critical for Celyvir efficacy.70,93,97,100
Canine model
In 2018, our group collaborated with the Clinical Veterinary Hospital of Universidad Alfonso X to develop a fully canine-adapted version of the Celyvir strategy for the treatment of spontaneous tumors in canine patients. 101 The first study focused on 27 dogs that had spontaneously developed sarcomas or central nervous system (CNS) tumors, two of the most common pediatric cancer types. 102 A subsequent study, published in 2022, explored the efficacy of the Celyvir approach in 10 dogs diagnosed with high-grade gliomas. 103 The veterinary clinical application of Celyvir was entirely adapted to the canine species, utilizing canine-derived MSCs and the canine OAd ICOCAV17 (Fig. 3). 104 In parallel with our murine preclinical studies,93,97 the MSCs used in this Celyvir strategy were allogeneic, derived from adipose tissue of healthy donor dogs. Consistent with findings from the preclinical models, treatment tolerance was favorable in both veterinary clinical studies, with isolated cases of toxicity appearing to be related either to viral administration or to the advanced disease stage of the patients.101,103 The use of allogeneic MSCs is discussed further in later sections regarding the optimization of Celyvir therapy.
Translational development of Celyvir therapy. The original hypothesis behind Celyvir emerged in 1999. Its study in preclinical models led to its use under compassionate use programs and the launch of a first clinical trial. To optimize and gain a deeper understanding of the therapy, research has continued in parallel using preclinical models and has also been applied in veterinary clinical medicine. These advances have informed the design of new clinical trials. The dashed arrows represent the transfer of knowledge across stages. At the bottom, different versions of Celyvir are schematically depicted, based on the use of various MSCs and OAd combinations.
ICOCAV17 is a conditionally replicative canine Ad derived from canine Ad type 2 (CAV2). The endogenous E1 promoter has been modified by inserting four E2F binding sites, and a 21 base-pair deletion was introduced in the E1A-pRB binding domain (E1aΔ21), analogous to the Δ24 deletion used in human OAds such as ICOVIR-5 and ISC301. 104 Furthermore, ICOCAV17 carries the human PH20 hyaluronidase gene, based on previous evidence that PH20 expression enhances viral distribution and antitumor efficacy in vivo when combined with ICOVIR-5 and other human OAds. 92 Direct intratumoral administration of ICOCAV17 in dogs with various tumor types had previously shown partial responses and disease stabilization. 104 More recently, intratumoral injection of ICOCAV17 in eight canine carcinomas demonstrated 25% partial responses and 75% disease stabilization, improving survival time compared with chemotherapy. 105 Notably, systemic intravenous administration of Celyvir led to clinical responses in 74% of the patients in the first study, including 14.8% complete remissions, with responses even in cases with pulmonary metastases. 101 In the second study, 28.6% of dogs with high-grade gliomas exhibited partial responses, while 42.9% achieved disease stabilization. 103 Although these tumor types differ and direct comparisons are challenging, these data suggest enhanced clinical activity of the Celyvir strategy over naked OAd administration. This improved antitumor efficacy of MSC-delivered OVs versus naked viral administration has been corroborated in our studies and by other groups,54,68,97 even in cases where OAds were administered directly intratumorally. 71
The canine clinical version of Celyvir offers an excellent translational model, as canine cancers share close histological and molecular similarities with their human counterparts, 106 making this model highly valuable for advancing novel cancer therapies. Beyond demonstrating antitumor efficacy, this model has enabled in-depth investigation of tumor-homing capacity, biodistribution, and safety.69,105 The complete adaptation of the Celyvir approach to the canine species provides a comprehensive platform for studying all therapeutic components within an immunocompetent setting that closely mimics the human clinical scenario.
Murine model
The limitations encountered in previous models led us to develop a fully murine, immunocompetent model capable of recapitulating the key features of Celyvir therapy and facilitating its mechanistic study. The inherent species-specificity of human Ad type 5 (Ad5) complicates the in vivo evaluation of ICOVIR-5 in murine MSCs or tumor cells, which is essential for establishing immunocompetent mouse models. To address this, we employed a new Celyvir system using murine MSCs infected with the murine OAd dlE102, a modified type 1 murine Ad (Fig. 3). Analogous to the human ICOVIR-5, dlE102 carries a deletion within the pRb-binding domain, thereby restricting viral replication to cells with free E2F1, such as tumor cells with disrupted pRb pathways.107,108 The selective replication capacity of dlE102 had been previously demonstrated in various murine tumor cell lines, along with attenuated replication in nontransformed cells, at levels comparable to those seen with human OAds in human fibroblasts. 108
Our studies confirmed efficient replication of dlE102 in multiple murine tumor cell lines—including renal cancer, melanoma, and osteosarcoma54,97,100—as well as in murine MSCs. This allows for an amplification effect of the MSC carrier component within the Celyvir strategy, as also reported by Franco-Luzón and colleagues. 109 Such replicative competence in both MSCs and tumor cells likely underlies the robust detection of the murine OAd within tumors following intravenous Celyvir administration, with viral titers comparable to those observed after direct intratumoral injection of the virus. 97 Similarly, Franco-Luzón et al. showed that systemic delivery of dlE102 via Celyvir resulted in greater viral persistence in peripheral blood compared with administration of the virus alone. 109 The enhanced viral replication in both MSCs and tumor cells, together with the superior in vivo antitumor effects observed in this model, surpass those of previous mixed models using human OAd and murine cells, and more closely mirror the results obtained with fully human Celyvir, supporting the relevance of this murine system as a preclinical model.
This approach allowed us to explore Celyvir therapy across different tumor types -including melanoma, renal cancer, and osteosarcoma- as well as in different immunocompetent mouse strains such as C57BL/6 and BALB/c, selected according to the syngeneic tumor model employed. Notably, C57BL/6 mice exhibit a prototypical Th1-skewed immune response, whereas BALB/c mice exhibit a Th2-skewed response.110,111 Nevertheless, Celyvir induced comparable changes in immune cell numbers and activation across the various models studied, irrespective of strain or tumor type.97,100
Furthermore, direct comparisons of the antitumor effects and immune infiltrates among studies demonstrated highly consistent results, with approximately 50% tumor growth reduction observed across all models: renal cancer (51%), melanoma (56%), osteosarcoma (51%), and neuroblastoma.97,100,109 In general, tumor-infiltrating lymphocytes in Celyvir-treated groups exhibited an increased presence of T cells along with a reduced CD4+/CD8+ ratio. Taken together with the various immune alterations observed in both tumor tissues and peripheral blood in these preclinical models, these data suggest a common mechanism of action for Celyvir that appears largely independent of tumor type.
CLINICAL APPLICATION OF CELYVIR
Celyvir exemplifies a case of real translational research, developed over more than two decades through the collaborative work of multiple research groups. Following validation of the original concept in preclinical models, Celyvir has advanced in parallel across preclinical, clinical, and veterinary settings (Fig. 3).
Importantly, the first clinical application of Celyvir preceded the implementation of the European regulatory framework for advanced therapy medicinal products [Regulation (EC) No 1394/2007 of the European Parliament]. In this context, in order to enable clinical translation of the therapy, a clean room facility was specifically designed and constructed at the Hospital Infantil Universitario Niño Jesús in Madrid. This facility allowed the production of Celyvir under appropriate GMP quality standards, paving the way for the first clinical trial. In parallel, while the clean room was being developed and validated, a compassionate use program regulated by the Spanish Medicines Agency (AEMPS) was launched to provide access to Celyvir for pediatric patients with advanced cancer.
The compassionate use program initially focused on pediatric patients with stage IV refractory neuroblastoma, all of whom had received at least two prior lines of chemotherapy.72,73 In these early applications, autologous bone marrow-derived MSCs were used as cellular carriers for the OAd ICOVIR-5. Encouragingly, the first patient treated achieved a complete remission. 72 Among subsequent patients treated under the compassionate use program, outcomes included eight cases of disease progression, one stable disease, three partial responses, and an additional complete response. The complete study suggested that the primary therapeutic mechanism was a secondary immune response triggered by the inflammatory effects of oncolytic viral replication. In addition, systemic administration of Celyvir was well tolerated, with no severe toxicities reported. 73 To date, more than 50 pediatric patients have been treated with autologous Celyvir under compassionate use, with additional cases of partial and complete remissions reported.
Building on this initial clinical experience, the first phase I clinical trial of Celyvir for the treatment of solid tumors was conducted (EudraCT 2008-000364-16), evaluating safety and preliminary efficacy in both pediatric and adult patients. 112 The trial confirmed an excellent safety profile, with no grade 2–5 toxicities observed. Clinical responses were modest, with stable disease reported in 2 of 9 pediatric patients (22%) and in 1 of 7 adult patients (14%). 112
Based on accumulated clinical experience and extensive preclinical research, the Celyvir strategy has since evolved from using autologous MSCs to an allogeneic approach (AloCelyvir). This next-generation version of the therapy, which is discussed in more detail later in this review, leverages the advantages of allogeneic MSCs and is currently being investigated in new clinical trials. Currently, three clinical trials are evaluating AloCelyvir: for extracranial pediatric solid tumors (EudraCT 2019-001154-26) and pediatric brain tumors such as diffuse intrinsic pontine glioma and medulloblastoma (EudraCT 2020-004838-37, NCT04758533) at Hospital Infantil Universitario Niño Jesús in Madrid, and for adult metastatic uveal melanoma (EudraCT 2020–005207-39, NCT05047276) at the Catalan Institute of Oncology— IDIBELL in Barcelona.
MECHANISM OF ACTION OF CELYVIR
A comparative analysis between Celyvir-treated patients who responded to therapy and those who did not revealed the association of several host- and product-related factors with improved clinical outcomes. Baseline immune status appeared to be particularly relevant to therapeutic efficacy. Specifically, responders exhibited higher circulating levels of CD4+ and CD8+ T lymphocytes prior to treatment initiation, suggesting a more competent systemic immune landscape at baseline. 73 In contrast, following Celyvir administration, non-responders displayed a greater increase in peripheral CD8+ T cells, indicative of an antiviral immune response rather than an antitumor effect. 73 Differences were also identified in the phenotypic and molecular profiles of the autologous MSCs used as carriers in Celyvir therapy. 73 MSCs derived from responding patients expressed higher levels of the chemokine receptors CCR1 and CXCR1, both of which have been previously associated with enhanced homing capacity of human MSCs toward gliomas. 113 Furthermore, transcriptomic analyses revealed that MSCs exhibiting a less pro-inflammatory profile after adenoviral infection were associated with improved clinical responses. Specifically, Celyvir products derived from responders’ MSCs showed reduced expression of immune-related genes such as IFN-γ, IL-6, and IL-8, among others. 73 The coordinated efforts of multiple research groups across Spanish institutions have enabled the iterative development and refinement of Celyvir therapy. This translational framework—grounded in the identification of biological correlates of response—has culminated in the design of novel therapeutic strategies and the launch of new clinical trials aimed at optimizing the efficacy of Celyvir.
Whispering in: Immune-silent delivery of Celyvir
The infection of MSCs with the OAd represents the initial step in the preparation of Celyvir therapy and has proven critical in determining its subsequent antitumor efficacy. NF-κB is an inducible transcription factor involved in diverse biological processes such as development, innate immunity, and inflammation. In human MSCs, NF-κB activation occurs in response to a variety of stress signals, cytokines, reactive oxygen species, hypoxia, and bacterial or viral antigens. 114 Accordingly, the activation of NF-κB observed upon ICOVIR-5 infection of human MSCs is consistent with a typical adenoviral response, generally associated with the induction of a pro-inflammatory program involving cytokines such as IL-6, IL-8, TNF-α, CCL5, and IFN-γ. 115
We have demonstrated that Celyvir’s antitumor efficacy is enhanced when using MSCs with a low inflammatory profile.70,73 In this regard, prior analyses of Celyvir preparations used in clinical trials showed that the MSCs derived from responding patients exhibited a reduced pro-inflammatory response. 73 In line with findings from other studies, 116 our data using murine OAd infection in murine cells appears to induce only minimal pro-inflammatory signaling, thereby naturally mimicking the silent infection profile observed in human responders. This observation may help explain the superior antitumor performance of the murine Celyvir model, which utilizes mouse MSCs and a murine OAd, compared with a hybrid model combining mouse MSCs with a human OAd.93,97 These species-specific differences in adenoviral response are likely the result of evolutionary divergence between Ads and their respective hosts.117,118
This concept is further supported by comparative studies using canine and murine OAds—ICOCAV17 and dlE102, respectively—engineered with a similar oncolytic backbone to ICOVIR-5. Despite their structural homology, these Ads elicit distinct patterns of cellular signaling. Notably, canine ICOCAV17 induced a significantly lower pro-inflammatory response than human ICOVIR-5 in both canine and human MSCs, 94 suggesting the involvement of intrinsic mechanisms in canine Ad beyond species origin of cells. One structural difference investigated was the absence of the RGD motif in the penton base of ICOCAV17, a distinguishing feature between canine and human Ads. However, infection with ISC301—a human OAd created by our group lacking the RGD sequence in penton bases—did not attenuate the inflammatory response in MSCs, and the antitumor efficacy of the resulting Celyvir product remained comparable to that of ICOVIR-5. 119 While this finding refutes the RGD at penton base deletion as the underlying mechanism for the reduced inflammation, it does not preclude the broader hypothesis that a less inflammatory MSC infection could enhance Celyvir efficacy. Alternative molecular determinants remain to be explored.
Supporting this ‘silent’ MSC hypothesis, we found that both in our engineered low-inflammatory MSC model (using MSCs knockout for TLR4 or MyD88) and in MSCs derived from responder patients, reduced NF-κB activation and diminished pro-inflammatory cytokine secretion were evident following adenoviral infection.70,119 Importantly, these features were not only induced by viral infection but also appeared to be intrinsic properties of the MSCs employed. A similar innate low-inflammatory profile was observed in canine MSCs used in veterinary applications of Celyvir, which were associated with superior clinical outcomes compared with human MSCs. 94 These findings suggest that the baseline inflammatory polarization state of MSCs—determined by their tissue origin or microenvironment—may critically influence therapeutic efficacy.
Based on these findings, one promising strategy for improving Celyvir is the selection or engineering of an optimal MSC carrier. This approach is further supported by preclinical evidence showing that allogeneic MSCs—derived from unrelated donors—can achieve similar antitumor efficacy and mechanisms of action as autologous MSCs. 93 In addition, several national regulatory agencies have approved the use of MSC-based allogeneic therapies. 120 The low expression of MHC class I and absence of MHC class II on MSCs facilitate their allogeneic use, 63 and allogeneic MSC therapies have already demonstrated favorable safety and toxicity profiles in clinical settings.121,122
In both the compassionate use program and the first clinical trial of Celyvir, the Spanish Medicines Agency required the use of autologous MSCs derived from patient bone marrow.73,112 This process entails approximately 6 weeks of manufacturing from cell isolation to therapy preparation, with dosing based on patient weight. However, 15 of the 34 patients enrolled in the first clinical trial (44%) experienced rapid disease progression before receiving their first dose. 112 The availability of a ready-to-use stock of allogeneic MSCs would significantly streamline the manufacturing timeline and simplify the process. 122 Allogeneic MSCs have also been used in the veterinary application of Celyvir in dogs with spontaneous tumors, with no significant toxicities reported following administration. 101 Biodistribution studies confirmed the capacity of these MSCs to deliver the OV to the tumor site. 123 This modification has already been approved by AEMPS and is being implemented in ongoing clinical trials (EudraCT 2019-001154-26, 2020-004838-37, 2020-005207-39), under the updated designation AloCelyvir.
An essential step in the optimization of Celyvir therapy involves the selection of MSCs with an intrinsically ‘silent’ inflammatory profile. TLRs are pattern-recognition receptors that are commonly present in bacteria and pathogenic viruses. 124 Several TLR family members, including TLR4 and TLR9, have been implicated in the recognition of Ad, as also observed in our earlier work.54,93 TLR4 activation in MSCs has specifically been associated with the induction of a pro-inflammatory phenotype. 65 Consequently, we have investigated TLR-deficient MSCs as a model of inflammation-silent MSCs within the context of Celyvir therapy, that resulted in better antitumor efficacy. 70 In parallel, Prof. Manuel Ramírez Orellana and colleagues at Hospital Infantil Universitario Niño Jesús conducted an in-depth transcriptomic analysis of autologous MSCs derived from patients who either responded or did not respond to Celyvir therapy. This study aimed to identify gene signatures or pathways of potential relevance to therapeutic efficacy. Among the candidate genes validated in a second patient cohort, mitochondrial antiviral-signaling protein (MAVS) emerged as a key factor, displaying differential expression between responders and non-responders. 125 MAVS is a central adaptor in signaling pathways activated by retinoic acid-inducible gene I-like receptors (RLRs), leading to transcription factor activation such as NF-κB.125,126 While RLRs are typically activated by viral RNA,126,127 Ads such as ICOVIR-5—being DNA viruses—nonetheless encode non-coding viral-associated RNAs (VA-RNAs I and II), which may be recognized by RLRs and trigger MAVS-dependent signaling.128,129 These VA-RNAs are immunostimulatory and have been shown to induce IFN-β expression. 129
Despite identifying the involvement of TLR signaling and differential MAVS expressions in MSCs from responding patients, no single receptor appears sufficient to predict or dictate the MSC response to adenoviral infection. TLRs are classically associated with specific molecular ligands—TLR4 with bacterial lipopolysaccharide, TLR5 with bacterial flagellin, and TLR9 with bacterial and viral DNA—, 124 but functional redundancy and cross-activation among TLRs are frequently observed.24,130 Similarly, the RLR pathway, in which MAVS functions, can be activated by stimuli beyond viral RNA, including cellular damage and metabolic stress. 126 Therefore, while candidate genes such as MAVS may inform preliminary screening, we propose that MSC selection for Celyvir therapy should ultimately be based on a functional readout: specifically, the expression or secretion profile of pro-inflammatory mediators.
Indeed, MSCs from different tissue sources exhibit distinct characteristics in terms of growth kinetics, surface marker expression, and cytokine secretion. 131 For example, adipose-derived MSCs, which we employed in murine and canine models, secrete higher levels of IL-6 and IL-8 compared with MSCs derived from skeletal muscle or dermal tissue. 131 In our humanized model, we used menstrual blood-derived human MSCs, which exhibit a relatively low basal pro-inflammatory profile. 54 These cells are not only readily accessible but also demonstrate superior expansion potential compared with bone marrow-derived MSCs, making them attractive candidates for clinical application in Celyvir therapy. 132
At first glance, the finding that MSCs with a low inflammatory profile result in enhanced therapeutic efficacy may appear counterintuitive, particularly within the framework of immunotherapy, which is designed to activate the immune system. However, this paradox underscores one of the fundamental challenges of systemic virotherapy and other immunotherapeutic strategies: while immune activation is desirable within the tumor microenvironment, uncontrolled systemic inflammation may reduce therapeutic efficacy and compromise safety.
For example, cytokine-based immunotherapies such as high-dose IL-2 have been associated with severe systemic toxicity, including capillary leak syndrome and sepsis-like symptoms leading to multi-organ failure. 2 Similarly, CAR-T cell therapies are frequently complicated by immune effector cell-associated neurotoxicity and cytokine release syndrome. 133 Immune checkpoint inhibitors may also trigger systemic inflammatory effects by dysregulating T cell activity. 133 In our case, intravenous administration of the OAd ICOVIR-5 alone has been associated with grade 3 toxicities, including transaminitis. 112 In contrast, intravenous Celyvir administration in pediatric patients was well tolerated, with no severe adverse events and only transient, self-limiting fever as the most common side effect.73,112 Our analysis of peripheral blood from immunocompetent mice and dogs with spontaneous tumors after Celyvir administration showed no significant alterations in hematological and biochemical markers, including stable transaminase levels, effectively ruling out hepatotoxicity typically observed following intratumoral OAd administration.101,134 This favorable safety profile was consistent across both conventional and silent Celyvir treatments. 70
The observed clinical and preclinical benefit of a silent MSC profile appears to be particularly relevant after systemic administration. We propose that minimizing systemic inflammation through the use of silent MSCs enhances the therapy’s ability to evade early immune clearance, allowing more efficient tumor homing and delivery of the oncolytic payload. Ad-induced immune activation includes systemic pro-inflammatory cytokine release, recruitment of cytotoxic immune populations for infected cell clearance, and signaling to uninfected tissues. 135 These effects are largely mediated by circulating IL-6, TNF-α, and IL-1β. 136 In murine models, conventional Celyvir administration resulted in elevated leukocyte, monocyte, neutrophil, and lymphocyte counts in peripheral blood at 48 h, an effect absent in the silent Celyvir group. 70 Of note, innate immune cells, particularly macrophages, act as a first line of defense against adenoviral infection, efficiently clearing virions from circulation. 135
We hypothesize that MSCs with a pro-inflammatory response profile activate innate immune recognition of their viral cargo, leading to premature clearance of the carrier cell. Supporting this notion, our early preclinical studies utilized a Celyvir dose tenfold higher than that employed in clinical settings, yet a reduced dose proved equally or even more effective in tumor control. 93 Given that these findings were not yet available during the initial compassionate-use program, higher doses were used initially; and subsequent clinical experience demonstrated comparable responses with reduced cell doses. As a result, current clinical trials with AloCelyvir employ 0.5 million cells/kg, a marked reduction from the 2 million cells/kg used in the first clinical study. 73 It is likely that excessive dosing of infected MSCs provokes a systemic immune response that offsets the therapeutic benefit, whereas a lower, silent dose preserves therapeutic efficacy while avoiding immune rejection.
Beyond cytokine secretion, pro-inflammatory MSCs may also modulate the expression of specific membrane receptors post-infection. Ads downregulate class I MHC expression on host cells. 137 NK cells, which serve as a bridge between innate and adaptive immunity, 138 rely on inhibitory receptors that detect MHC-I molecules; their absence triggers NK cell-mediated cytotoxicity. 139 Furthermore, IL-2-activated NK cells are capable of lysing both autologous and allogeneic MSCs. 140 Although our data did not reveal differences in MHC-I expression between conventional and silent Celyvir, 70 this could be due to the naturally low basal MHC-I expression in MSCs or species-specific differences in MSC response to human Ad infection. Consequently, the role of MHC-I modulation remains an open question for future investigation in the human clinical setting.
Analysis of immune cell populations in the peripheral blood of treated patients revealed that responders exhibited higher baseline levels of T lymphocytes prior to Celyvir administration. However, during long-term follow-up, this population remained stable in responders, whereas non-responders displayed an increase in T cell counts after therapy initiation. 73 An increase in peripheral CD8+ T lymphocytes has also been observed in patients treated peritumorally with neural progenitor cells infected with the OAd CRAd-S-pk7, although its correlation with clinical outcomes was not assessed. 141 Despite this, responders to Celyvir therapy consistently maintained higher levels of circulating T lymphocytes throughout treatment when compared with nonresponders. Notably, changes in T cell kinetics were less pronounced in the clinical trial setting. 112 This differential response to viral therapy administration could be partially explained by pre-existing adenoviral immunity. In support of this hypothesis, dogs treated with Celyvir—previously vaccinated and seropositive for CAV—exhibited a transient increase in circulating T cells following the first infusion, which remained stable thereafter. 101 In contrast, in our murine model lacking prior adenoviral immunization, systemic administration of Celyvir led to increased peripheral T cell levels and reduced antitumor efficacy compared with “immune-silent” Celyvir, which did not induce such lymphocytic changes. 70 In pediatric populations, seroprevalence data suggest that approximately 50% of children aged five harbor antibodies against adenovirus type 5, with higher rates for Ad1 and Ad2 (70–80%). 142 It is therefore plausible that nonresponder pediatric patients had been previously immunized against Ads, leading to vector neutralization, immunologic targeting of infected MSCs, and diminished clinical benefit. This factor is currently being explored in ongoing clinical trials of allogeneic Celyvir (AloCelyvir).
The therapeutic efficacy of our proposed combination of Celyvir and G-CSF may have been compromised by the associated systemic immune activation. 100 Our rationale was twofold: first, G-CSF–mediated mobilization of immune cells was expected to generate a baseline immune state similar to that observed in Celyvir responders, characterized by high circulating immune cell levels. 73 Second, previous work by collaborators at the Hospital Infantil Universitario Niño Jesús had demonstrated enhanced immune cell recruitment to the tumor, thereby contributing to improved antitumor efficacy. Although increased intratumoral immune infiltration was observed in our model, the antitumor effect of the combination therapy was ultimately comparable to that of Celyvir alone. 100 The timing of G-CSF and Celyvir administration was based on two considerations: the peak of G-CSF–induced immune mobilization occurs approximately 4 days post-initial dose, 143 and Celyvir tumor homing typically peaks 24–48 h after systemic delivery.70,93,97 However, our attempt to synchronize these two events may have inadvertently allowed the systemically mobilized immune cells to target the infused MSCs, thereby reducing therapeutic impact.
Several factors following adenoviral infection and systemic administration of Celyvir may influence the migration of the therapy to the tumor niche. In murine models bearing subcutaneous tumors, Celyvir product was detected at the tumor site at 24–72 h after intraperitoneal administration, with confirmed presence of oncolytic virus.70,93,97 Both the MSCs and the Celyvir product did not only reach the tumor site, but also other off-target organs, primarily the liver, as well as, in the lungs, spleen, and kidney. 70 A comprehensive biodistribution study of AloCelyvir performed in a canine patient with spontaneous lung carcinoma similarly confirmed the presence of migrated Celyvir product within the tumor 48 h post-administration, with further confirmation of in situ release of functional oncolytic virus. 69 Further studies in canine patients with spontaneous tumors treated intravenously with AloCelyvir also identified adenovirus-positive cells in the kidneys, pancreas, enterocytes, metastases, and tumors, including high-grade gliomas.103,123
Although the precise mechanism of MSC tumor homing remains incompletely defined, the process has been associated with the expression of chemokine receptors such as CXCR4, CXCR12, and CCL2. 144 In clinical settings, MSCs derived from responder patients displayed higher expression of CCR1 and CXCR1, receptors previously implicated in MSC migration to gliomas. 113 In vitro migration assays conducted in the absence of immune cells revealed no significant differences between MSCs and Celyvir, indicating that adenoviral infection does not impair the intrinsic tumor-homing capacity of MSCs. Comparable migratory behavior toward tumor cells was also observed when comparing Celyvir with its “immune-silent” variant using MSCs knockout for TLR4 or MyD88. However, in our preclinical studies, these immune-silent versions of Celyvir resulted in greater tumor homing, although analysis of infected MSCs with a silenced profile revealed no significant differences in the expression of surface chemokine receptors compared with standard MSCs. 70 While the potential involvement of unassessed receptors cannot be ruled out, our data suggest that the antiviral immune response following systemic Celyvir administration is a key determinant of the therapy’s migratory efficiency. Supporting this notion, other groups have shown that using cellular carriers to deliver OVs such as vaccinia or myxoma can help overcome systemic immune barriers.76,145
Interestingly, the enhanced tumor migration associated with a less pro-inflammatory MSC profile at the systemic level was accompanied by a more pronounced pro-inflammatory activation within the tumor microenvironment. This was evidenced by changes in tumor cell signaling, immune cell infiltration, and overall therapeutic efficacy. 70 Similar tumor microenvironment changes were shown in canine patients after Celyvir treatments. 101 This apparent paradox is likely mediated by a coordinated series of immunological events. The use of MSCs in Celyvir therapy not only facilitates targeted delivery and systemic shielding of the OAd—minimizing immune activation and antiviral responses—but also amplifies the therapeutic effect by permitting in-transit viral replication. This dual functionality has allowed for the administration of viral loads previously unattainable in patients, with remarkably low toxicity. 73 The MSC’s amplifying capacity may partly account for the improved efficacy observed in the all-murine Celyvir model compared with earlier hybrid models employing murine MSCs with the human adenovirus ICOVIR-5.71,93 Given the species specificity of human Ads, their replication in murine cells is limited, thereby reducing the MSC-mediated amplification effect. 19 Conversely, the fully murine model developed using murine MSCs and the murine adenovirus dlE102 97 supports efficient replication in both carrier and tumor cells. 109 The disparity in viral replication capacity between human and murine Ads within murine hosts affects not only the amplification en route but also replication within tumor cells themselves. This observation suggests that hybrid models may underestimate the true oncolytic potential of Celyvir, whereas the replication-competent murine model provides a more accurate representation of therapeutic efficacy. Indeed, OAd detection levels in tumors following systemic administration of murine Celyvir closely approximated those observed after direct intratumoral injection of dlE102. 97 Again, in canine patients with spontaneous tumors, we were able to detect OAd in the tumor after systemic administration of Celyvir. 101
Altogether, an immune-silent entry of Celyvir enhances the delivery of infected MSCs to the tumor site, where they not only release but amplify the oncolytic viral load. We hypothesize that this localized, high-density viral release—combined with the subsequent infection and lysis of tumor cells—triggers a focused, potent immune response within the tumor, surpassing what would be achieved under conditions of systemic immune activation (Fig. 4).
Shouting back: The immune-explosive response
Decades ago, the initial theoretical foundation of oncolytic virotherapy was centered on its capacity to induce direct tumor cell lysis. However, continued research and clinical application have increasingly highlighted its broader therapeutic potential as an immunomodulatory agent. As articulated by authors such as John C. Bell, the true promise of oncolytic virotherapy lies in its ability to stimulate the host immune system to recognize and eliminate tumor cells. 146 In this context, the antitumor efficacy of Celyvir is partially attributable to the lytic activity of the OV, but more critically to the immune-mediated mechanisms it elicits, an emerging hallmark of oncolytic virotherapy. 146 These mechanisms include direct tumor cell lysis, modulation of the tumor microenvironment, recruitment of tumor-infiltrating lymphocytes, activation of innate immune cells and cytotoxic CD8+ T cells, and remodeling of the tumor vasculature. 7 Several of these processes have been implicated in Celyvir’s therapeutic effects and have been systematically explored throughout the years of development.
Viral oncolysis facilitates the release of DAMPs and tumor-associated antigens or neoantigens from infected and lysed tumor cells, which are subsequently processed by dendritic cells and presented to the adaptive immune system.7,147 Therefore, beyond direct cytotoxicity, efficient viral lysis enhances the exposure of a diverse neoantigen repertoire to both innate and adaptive immune cells, potentially triggering a potent antitumor immune response. 148 The extent of antigen dissemination has been correlated with the therapeutic efficacy of various immunotherapies and may be particularly relevant in tumors characterized by high intratumoral heterogeneity. 148
Nonetheless, the comparable therapeutic outcomes observed across different tumor models suggest that antigen presentation alone may not fully account for the antitumor responses induced by Celyvir. The establishment of a murine Celyvir model enabled systematic evaluation across multiple tumor types, as well as across distinct immunocompetent mouse strains, selected according to the syngeneic tumor line used.97,100 Despite these immunological differences, Celyvir induced similar patterns of immune cell activation and infiltration across all tested models, regardless of strain or tumor type. Moreover, direct comparisons between our studies and those conducted by collaborators consistently demonstrated a ∼50% reduction in tumor growth following Celyvir treatment across diverse tumor types: renal carcinoma, melanoma, osteosarcoma, and neuroblastoma. 109 Overall, treated tumors exhibited increased infiltration by T lymphocytes and NK cells. Clinically, in canine and human patients, Celyvir has shown efficacy in a broad range of pediatric and adult malignancies, including sarcomas, central nervous system tumors, and melanomas.73,103,112 Given this breadth of tumor types and the inherent diversity in their antigenic landscapes, it is unlikely that tumor antigen presentation alone accounts for the consistent antitumor effects observed. This notion is further supported by the general observation that pediatric tumors tend to harbor fewer neoepitopes than their adult counterparts,149,150 although recent whole-genome sequencing studies have challenged this paradigm. 150 Taken together, these findings suggest that while antigen presentation contributes to the immune response, the efficacy of Celyvir likely results from a more complex and multifaceted immunological activation.
Regarding the adaptive immune system, a previous study demonstrated that stimulation with adenoviral antigens of CD8+ T lymphocytes isolated from mice treated intratumorally with ICOVIR-5 induced their activation and consequent IFN-γ production, suggesting the development of an antiviral response by these T cells. However, stimulation of T lymphocytes with tumor cell lysates derived from the same tumor line used in the study elicited a lower response. 71 Thus, activation could arise from recognition of tumor antigens by T cells, but also as a collateral effect of antiviral activation. For instance, stimulation of PRRs by viral PAMPs induces type I IFN production, which exerts antiviral effects by inhibiting viral replication but also possesses direct antitumor properties. 151 The relative contribution of the antiviral versus antitumor immune response at the initiation of immunity remains a frequent subject of debate in oncolytic virotherapy. In our case, the absence of complete responses in murine models, where remissions were not observed, and the modest antitumor effect following adoptive transfer of splenocytes suggest that Celyvir does not elicit a fully developed adaptive antitumor response.70,93
The key involvement of innate immune components, such as monocytes and NK cells, is explicitly demonstrated in a humanized mouse model study, where their depletion abrogates the antitumor effect of Celyvir. 54 The participation of neutrophils and monocytes is supported by their role in recognizing viral PAMPs, triggering signaling cascades that potentiate the innate immune response via multiple receptors. 152 Although tumor immune infiltrate analyses in murine models do not consistently reveal increased infiltration of innate populations such as macrophages, neutrophils, and NK cells, the pro-inflammatory tumor microenvironment appears to polarize these cells toward a pro-inflammatory M1 and N1 antitumor phenotype. The demonstration of reduced tumor growth following depletion of N2 neutrophils in the absence of CD8+ T cells supports the capacity of these innate populations to control tumor growth independently of the adaptive immune system. 153 The role of NK cells in this immune response is complex due to their dual function bridging innate and adaptive immunity. NK cells mount antigen-independent responses to viral infections prior to the development of adaptive immunity, thereby contributing to cancer cell control. They are a significant source of IFN-γ and other cytokines that enhance the immune response. 154 Studies with other OAds have shown that infection activates NK cells through direct virus-immune cell interactions, augmenting the antitumor effect of OAd. 155 Furthermore, NK cell-mediated cytolytic debris can promote cross-presentation of antigens to CD8+ T lymphocytes. 156 Thus, the antitumor immune response may be initiated by NK cell activation against the OAd, where their antiviral activity within the tumor niche influences T cell responses. NK cells, therefore, represent a critical link between innate and adaptive immunity in oncolytic virotherapy.
Celyvir treatment induces increased immune infiltration within tumors, encompassing both innate and adaptive immune populations. This enhanced infiltration is evident not only in cellular density and absolute numbers but also in immune cell localization.71,93,101 Untreated tumors in our models displayed an immune-excluded phenotype, with immune cells confined to the tumor periphery and excluded from the core. Immune cell detection in the tumor infiltrate clearly and visually illustrates the locational shift induced by therapy. 93 These findings recapitulate immune infiltration observed in clinical and veterinary applications of Celyvir,73,101,103 as well as in clinical trials with T-VEC and other OVs.157,158
Multiple mechanisms may underlie the increased central tumor immune infiltrate. Viral replication within the tumor likely promotes immune infiltration as part of the antiviral response. Additionally, tumor cell reduction caused by oncolysis may facilitate immune cell infiltration. The extracellular matrix (ECM) of solid tumors—composed of collagen, fibronectin, elastin, and other components 159 - is altered relative to adjacent normal tissue, restricting T cell mobility and infiltration. 160 The canine version of Celyvir used in veterinary clinics employs the OAd ICOCAV17, which expresses hyaluronidase. 104 This enzyme hydrolyzes hyaluronan, an ECM component, and independently exhibits antitumor properties, having been applied in clinical trials. 161 Overall, Celyvir treatment in canine patients induces remodeling of the tumor microenvironment, characterized by ECM degradation and increased fibrosis, hemorrhage, and necrosis. 101 Similarly, reduced tumor angiogenesis has been observed in murine models treated with Celyvir. 97 As suggested in our publication, the inclusion of hyaluronidase may facilitate tumor disintegration, thereby promoting OAd dissemination and immune infiltration into the tumor core. 101 Consequently, one potential optimization for Celyvir therapy could be the use of OAds that modify the tumor microenvironment, such as the human OAd ICOVIR17, which, in addition to oncolytic modifications similar to ICOVIR-5, also expresses hyaluronidase. 162
As discussed, Celyvir induces an immune response and pro-inflammatory tumor microenvironment; however, the adaptive immune system does not appear to develop a fully effective antitumor response. Combining Celyvir with immunotherapies may enhance intratumoral immune activation and facilitate the development of an adaptive response. Alternative strategies include using MSCs or modified viruses designed to induce localized tumor inflammation. Indeed, we have shown that pre-treatment with the naked OAd ISC301 in a murine model potentiates the antitumor effect of a CAR-T cell. 119 The combination of OVs with CAR-T cells is currently under extensive investigation in preclinical models and a clinical trial (NCT03740256). 163
Celyvir treatment was also associated with decreased PD-1 expression on tumor-infiltrating CD4+ and CD8+ lymphocytes, 97 a finding similarly reported following administration of other oncolytic viruses. 164 However, other studies have described increased PD-1 expression induced by different OAds and myxoma viruses delivered by progenitor cells.76,141 Beyond virus-specific factors, we hypothesize these discrepancies may reflect transient PD-1 expression that declines in the absence of T cell receptor signaling but is sustained during chronic activation. It is important to note that PD-1 expression was measured approximately 30 days after initial Celyvir treatment following multiple doses. Thus, tumor-infiltrating T cells may have been previously activated but subsequently entered a state of exhaustion, manifesting as reduced PD-1 expression. This observation supports the notion that Celyvir does not elicit a fully developed adaptive antitumor immune response and predominantly relies on the innate immune response it induces.
CONCLUSION
In summary, Celyvir represents a novel and multifaceted approach to oncolytic virotherapy, leveraging the unique properties of MSCs to achieve immune-silent delivery while simultaneously eliciting a robust innate immune response within the tumor microenvironment. Future strategies incorporating the selection or even generation of optimized MSC populations, alongside combination with complementary immunotherapies such as immune checkpoint inhibitors, chimeric antigen receptor T cells, or tumor-infiltrating lymphocytes, may overcome current limitations by amplifying adaptive immune activation and achieving durable tumor control (Fig. 4). Developed over more than 25 years from its initial conceptualization to clinical translation, Celyvir has reached a pivotal stage marked by three ongoing clinical trials that implement key optimizations aimed at enhancing therapeutic efficacy. The outcomes of these studies will be critical in defining the next steps for Celyvir and AloCelyvir and its role within the broader landscape of cancer virotherapy, providing valuable insights into the clinical potential of this innovative approach.
Schematic overview of Celyvir’s mechanism of action and potential optimization strategies.
AUTHORS’ CONTRIBUTIONS
A.M.M. and J.G.C.: Conceptualization, literature research, writing, reviewing, and editing. J.G.C.: Funding.
Footnotes
ACKNOWLEDGMENT
The authors gratefully acknowledge the efforts and valuable contributions of numerous researchers from the Instituto de Salud Carlos III, Hospital Infantil Universitario Niño Jesús, Instituto Catalán de Oncología, Universidad Alfonso X, and other institutions who have contributed over the years to the development of Celyvir and AloCelyvir therapies. While we cannot mention everyone by name, every contribution has played an important role in moving this therapy forward. We also appreciate the support of entities and patient organizations such as the Asociación Pablo Ugarte, AFANION, Fundación Oncohematología Infantil (FOI), Cátedra UAM-FOI de Oncohematología Pediátrica y de la Adolescencia, Unidos contra el DIPG, Asociación NEN, Fundación Neuroblastoma, and others. Some of the figures were created with BioRender.com.
AUTHOR DISCLOSURE
The authors declare no conflict of interest.
FUNDING INFORMATION
This work was supported by the