**5. The Future Directions of Organoid Models in Radiation Biology**

Despite a growing role for organoids in radiobiology (as well as other fields of biology) and continuous advances of the models to faithfully simulate the tissue of origin, organoids still have limitations. However, these drawbacks may represent opportunities. Opportunities for researchers to optimise and improve current organoid systems, and opportunities to complement their research with other techniques, such as clinical imaging techniques for enhancing patient treatment response predictions.

While organoids consist of heterogeneous cell types and are cultured in three dimensions, they still lack important microenvironmental cues, such as sympathetic and parasympathetic innervation and immune cells (such as macrophages and cytokines). These are crucial factors in both development and regeneration of tissue. There is growing evidence for the role of parasympathetic innervation in salivary gland development [118] and regeneration [119], including following radiation-induced damage [120]. Finding means to accurately mimic autonomic innervation in organoids may be important to fully utilising them as models for regeneration in tissues with a similar architecture to the salivary glands. Similarly, the lack of stroma and immune cells in organoids in response to both injury and treatment are important factors which still need addressing especially considering the rising number of applications of immunotherapy. In the aforementioned study of Ooft et al. [92], while the patient-derived organoids were predictive of patient response to irinotecan-based treatments, they were not predictive of 5-FU–oxaliplatin combination therapy, which the authors suggest may, at least in part, be done to the lack of crucial stroma and immune system interactions. Recent advances have been made to overcome these issues, with Neal et al. [121] successfully developing patient-derived organoids with the T-cell spectra of the original tumours capable of modelling the immune checkpoint blockade. Alternatively, co-culturing of organoids with immune cells will offer a theoretically more realistic tissue response. Indeed, in co-culture experiments with macrophages and mammary organoids, macrophages were shown to migrate to organoids with an increased migration rate towards irradiated organoids [122].

Furthermore, stroma also plays an important role in radiation responses, of both normal and tumour tissue. In organoid cultures derived from whole tissue biopsies (without stem cell selection) stromal cells and effects can be found within the culture system [123]. However, in organoid cultures from selected stem cells stroma is absent, and therefore stromal co-culturing is necessary to recapitulate the effects of the tissue's stroma. In prostate organoids, an increased viability and maintained branching was induced upon co-culture with prostate stroma [124]. Furthermore, the generation of organoids derived from prostate cancer was also improved upon stromal co-culture. These effects were suggested to be primarily due to direct contact with stromal cells and the expression of factors, such as TGF-β, by the stromal cells [124]. Besides the advance that this model represents in development and disease studies, the co-culture of organoids with tissue-specific stromal cells could have important implications for treatment responses, due to the important role of stromal cells [125] and the effects of signalling factors, such as TGF-β [126], in tissue responses.

Radiation-induced bystander effects have been suggested to act both proximally [127,128] and distally [129,130] to the site of irradiation; however, organoids derived from a single tissue currently do not recapitulate such interactions. Various anti-cancer therapies, including radiation, are known to induce senescence and an induction of a senescent associated secretory phenotype [131,132] which, it has been suggested, can in turn contribute to therapy-induced normal tissue side effects [133]. Studies using cultured media from irradiated cells has long been shown to induce paracrine bystander effects in non-irradiated cells [134] and such techniques may be insightful into the effects of secreted SASP proteins on untreated cells or organoids. Indeed, our group recently demonstrated that cultured media from irradiation-induced senescent organoids inhibits organoid forming efficiency in freshly passaged salivary gland-derived organoids [135]; however, these models still lack a true interaction between treated and non-treated organoids and the potential paracrine effects of other tissues in their vicinity in vivo.

Furthermore, both organ–organ, tumour–organ and vasculature interactions are generally absent in organoid cultures. Some of the glioblastoma organoid studies mentioned above elegantly show that cancer cells and healthy cells can be cultured together as organoids allowing for the study of tumour invasion [103,115,116]. Moreover, these models may be useful in revealing new therapeutic targets for tumour radiosensitisation or normal tissue radioprotection. Implementing organoid models alongside newly-established microfluidic devices which allow for the study of metabolic gradients [136] in radiation studies has the potential to reveal valuable insights of how such as signalling gradients can influence both irradiated and non-irradiated cells in perhaps a physiological relevant setting than organoids alone. Indeed, gut-on-a-chip models have recently been utilised in studies of radiation-induced intestinal injury and faithfully mimicked epithelial cell loss due to reactive oxygen production as seen in vivo [137] and may represent an excellent model for complementary studies to the abovementioned "mini-gut" organoid models in radiation studies. Organ-on-a-chip devices have been established for various other tissues, including lung [138], kidney [139] and liver [140]. The capacity of these platforms to mimic functional mechanics, such as breathing movements in

lung-on-a-chip [138], could potentially offer more physiologically relevant models to complement and add a translational element to findings from organoid radiation studies. Furthermore, multiple chamber "on-a-chip" devices [141] could overcome the limitation of organoids of studying organs in isolation, in which each chamber potentially could contain cells from different tissues, vessels, stroma or nerves.

Radiation-induced endothelial cell loss and vascular damage are known to be major contributors to the response of both normal tissue and tumours [142,143]. Vasculopathy significantly increases the chances of ischemic stroke following radiation treatment [144], while preclinical models have been used to demonstrate that vascular remodelling is a major contributor to radiation-induced lung toxicity [145]. The vasculature of a tissue is essential for nutrient availability and regeneration following damage, as well as effective engraftment after tissue transplantation [146]. Furthermore, the response of tumour vasculature, particularly vasculogenesis, has also been shown to play a key role in tumour recurrence following radiation treatment [147]. While radiation can initially control the tumour, a reduced flow through tumour blood vessels and increased hypoxia can induce the hypoxia inducible factor-1 pathway. This in turn can activate pathways to re-promote vasculature and can subsequently cause tumour regrowth [143]. Therefore, it is important that in vitro models, particularly tumour models, can recapitulate such vasculature features. Recently, many techniques have been established to engineer vascularisation of organoids, including bioprinting, implantation into highly vascularised tissue and growing organoids in the vicinity of endothelial cell monolayers [148]. Vascularised organoids, such as recently-established vascularised cortical organoids [149,150] and tumour organoids [151], offer new opportunities to study disease pathology but also to study the impact vascularisation can have on treatment (including radiation) responses.

Finally, although there are many different protocols and technical considerations for the isolation and propagation of organoids, they are often arduous and time consuming. In order to have enough cells or organoids to test still often requires weeks to months of culturing. This is of particular importance for the development of organoids as a model for predicting patient responses in proposed precision therapies, where it is frequently necessary to treat patients as soon as possible. Protocols are being established to reduce culture time of organoids while maintaining fidelity of the systems of various different tissue origins (such as the aforementioned glioblastoma organoid model [117]); however, it is important that organoid models are further optimised for rapid and accurate screening of responses before implementation in a personalised medicine.

Despite their limitations, the future of organoid models in the field of radiobiology remains bright. As highlighted, many valuable studies are already overcoming the shortcomings of organoids, and as our knowledge and availability of organoid models grow, so too will their place in radiobiology. New organoid models can potentially shed some much-needed light on tissues which are perhaps less studied or highly limiting to the clinical application of radiation treatment, such as the liver. Moreover, while it could be questioned if a response prediction accuracy of approximately 80–85% is good enough, this will surely only improve as the models themselves are further optimised. Combining clinical patient imaging techniques currently used to predict patient responses, such as PET/CT, with the in vitro predictions from organoids may in the future bring around more accurate means to forecast treatment outcomes. Organoids could also potentially be used in discovery and validation of radiation biomarker and in radiomics. Understanding the mechanisms behind tissue regeneration are key to mitigating radiation-induced side effects, whether it is by stem cell therapy or through druggable targets to protect against damage, and organoids have already proven themselves as excellent models for such studies.

**Author Contributions:** Writing—original draft preparation, P.W.N.; writing—editing and reviewing, P.W.N. and R.P.C.; All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was funded by the Dutch Cancer Society (KWF-grant number 10417).

**Acknowledgments:** While we have tried to give a broad overview of the state of organoids in radiation biology, there are many other important works in this area of the field that unfortunately we could not include due to manuscript constraints.

**Conflicts of Interest:** The authors declare no conflict of interest.
