Oncolytic virus (OV) therapy is an approach to treating cancer in which patients are infused with live viruses, either naturally selected or genetically engineered to infect and kill cancer cells, while leaving normal cells intact [1], [2]. When conceived of over a quarter century ago, direct oncolysis was intended to be the primary antitumor mechanism of this therapeutic strategy. As such, improving viral oncoselectivity, immune system evasion and cancer cell killing were the main drivers of OV research and development [3]. However, as the field matured, data from immunocompetent animal models and human clinical trials demonstrated that OVs work predominantly as an immunotherapeutic, by stimulating potent antitumour immunity. In particular, the response to OV therapy is now known to be heavily reliant on cytotoxic CD8+ T cells [1], [4], suggesting that OVs impinge upon the cancer immunity cycle (CIC) [5] for their activity. Although the molecular mechanisms underpinning this biology are not completely understood, it is commonly suggested that OVs engage antitumour immunity by eliciting immunogenic cell death (ICD) in tumours [6], [7], [8]. In this mechanism, oncoselectivity is still theorized to be the foundational principle of OV therapy, where it is proposed that infection and lysis of cancer cells causes the release of tumour antigens and danger-associated molecular patterns (DAMP) into the tumour microenvironment (TME). Recognition of DAMPs by innate immune cells is in turn thought to induce the expression of pro-inflammatory cytokines, creating an immunosupportive milieu for the uptake of tumour antigens by antigen presenting cells (APC) such as dendritic cells (DC) within the TME. These cells are then proposed to migrate to the tumour draining lymph node (TdLN), where they activate anti-tumour CD8+ T cells that traffic back to the tumour and mediate potent tumour cell killing [7], [8].

Much like for the original theory that direct oncolysis is the primary driver of tumour regression after OV therapy, there is unfortunately relatively little evidence in immunocompetent animal models and human patients supporting ICD as the mechanism of action for OV therapy. Further, this model fails to account for several emerging concepts in the field. The first is that OVs have been shown to elicit cancer regression against tumours that are not robustly infected by virus. Indeed, studies have demonstrated that some tumour cell lines non-permissive to viral infection in culture can still respond to OV therapies when implanted into mice, including to Newcastle Disease virus (NDV) [9], [10], reovirus [11] and adenovirus delivery [12]. Similarly, several tumours naturally resistant to virus infection in vivo have been shown to still respond to OVs such as Herpes Simplex Virus (HSV) [13], reovirus [11] and Vesicular stomatitis virus (VSV) [14], [15], [16]. Moreover, some tumours rendered non-permissive to productive OV infection, either through pre-immunization of mice with virus [17], the use of single cycle replication incompetent virus [18], or genetic engineering of tumour cell lines to disrupt virus entry [19], [20], can still undergo regression with OV treatment. For example, we recently found that intravenous (i.v.) administration of VSV to mice bearing tumours genetically ablated for viral entry receptors still elicited tumour regression and durable cures [20]. This observation is consistent with data emerging from human clinical trials, where oncolytic VSV treatment shows little evidence of tumour infection despite some patients responding [21]. Importantly, in our study, treatment response in murine models was dependent on cross-presenting classical DCs (cDC1) and CD8+ T cells [20], demonstrating that OVs can produce antitumour immunity via an ICD-independent mechanism.

The second concept not well explained by the ICD model is that, despite being designed to selectively replicate in cancer cells, many OVs still infect non-malignant cells, particularly after systemic delivery. Within the TME, OVs such as Vaccinia virus (VACV), VSV, HSV, NDV and adenovirus have been shown to infect endothelial cells (EC), fibroblasts and pericytes [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32]. Importantly, many of these infections are not simply incidental findings noted upon imaging of the tumour bed. Rather, they have been found to contribute to OV-induced anticancer activity. For example, EC infections destabilize tumour neovasculature, resulting in tumour cell death [22], [23], [26], [30], [33], [34], [35]. Outside of the TME, reported reservoirs of OV infection include the liver, spleen and lymph nodes (LN), amongst others. These, too, have been shown to impact OV activity. For instance, our group found that secondary lymphoid organ (SLO) infections play a key role in eliciting antitumour activity, by acting as potent immune system adjuvants for established anticancer immune responses [20]. In contrast, infections in the liver, which have been most extensively described for adenovirus but are still reported to occur with other OVs, are suggested to be deleterious to OV activity, by sequestering virus away from the tumour [37], [38].

In this review, we aim to summarize current knowledge on the role of non-cancer cell infections in OV-induced antitumour activity, with particular emphasis on the influence of these infections on tumour-specific immunity. We will also focus primarily on infections of non-malignant cells within the TME and SLOs, although will include a brief commentary on the role of liver and peripheral blood mononuclear cell (PBMC) infections. Finally, we will discuss how a better understanding of non-cancer cell infections can be leveraged to produce next generation OV platforms with enhanced antitumour activity.