**5. Development of Personalized in Vitro Models for OV Selection**

It can be concluded that the interplay between the OVs and tumor cells is very complex and that defining a single biomarker or set of biomarkers predicting efficacy of a specific OV may not be achievable. An alternative approach is the development of patient-specific assays to screen a set of OVs and identify the optimal OV for a particular patient. Such a predictive assay would need to provide information on the efficacy of OV infection, replication, and oncolysis as well as on the immune response that is mounted by the OVs.

At present, in vitro culture models from patient-derived tumors have become the gold standard in drug development research for GBM. Our group has developed a preclinical screening system based on patient-derived low-passage cell cultures under serum-free conditions for preserving the molecular genetic make-up of the parental tumors [159,160]. Such screening activities have for example led to the identification of viral sensitizers which enhance the oncolytic activity of delta24-RGD in GBM cells [161]. The screening system was also applied to assess the efficiency of infection, replication and cell killing by four different OVs on a panel of primary glioma cell lines, which revealed tremendous intertumoral heterogeneity in viral sensitivities [56] (and unpublished data). Such panels of molecularly characterized cell cultures may also help identify new markers of sensitivity or resistance to tested OVs.

To gain insight in the relationship between oncolytic efficacy and immune stimulation, a co-culture model of glioma cells and (autologous) immune cell populations, could potentially provide useful insight on the immune response that is triggered after treatment with different OV candidates, as well as on the relationship between infectivity, oncolysis and immune activation. For instance, a co-culture of macrophages and delta24-RGD-infected (and permissive) GSCs revealed a shift of the tumor-supportive macrophages M2 to the pro-inflammatory M1 [108]. Such approaches are also being taken for other forms of cancer. A platform has been established for cancers like colorectal and non-small cell lung in which tumor organoids were co-cultured with autologous T cells derived from the peripheral blood of the patient [162]. With such model systems, the T cell-mediated cell killing could be evaluated for individual patients after infection with different OVs [162].

However, the establishment of such primary GSC or organoid-immune cell co-cultures is time-consuming and may not yield a robust OV therapy recommendation within the required timeframe. Furthermore, with serial passaging of primary glioma cells, the diverse clones that characterize the GBM tumor cannot be maintained [59]. The establishment of ex vivo 3D tumor model systems directly from fresh tumor tissue may therefore offer a more attractive approach for performing OV screens on the heterogeneous landscape of GBM as such models still retain architecture and cellular composition of the original tumor, including the presence of immune cell infiltrates. We previously reported that fresh tissue derived organotypic multicellular spheroids (OMS) offer a versatile system for studying OV infection, replication and tissue penetration [163]. Similarly, fresh GBM tumor slices have been employed to assess oncolytic myxoma virus efficacy [164]. Other approaches being developed include the culture of fresh tumor cells in slices, on matrices and in microfluidic systems [165–167]. Culturing fresh tissue also has limitations, since the culturing methods generally favor the tumor cells and not the immune cells [168]. Identification of culture conditions supporting all cell populations would also offer an improvement to these models.

Efforts in the field to generate 3D models from fresh tissue under culture conditions that support and recapitulate the unique immune tumor microenvironment are expected to facilitate investigations into both the dynamics of viral infection and replication in tumor cells as well as the effects thereof on local immune responses [169]. Such systems may in the future offer a tool to screen multiple OVs for a specific patient and select the optimal viral treatment within a clinically-relevant timeframe.
