Abstract
The development of cancer and their response to radiation are intricately linked to the tumor microenvironment (TME) in which they reside. Tumor cells, immune cells, and stromal cells interact with each other and are influenced by the microbiome and metabolic state of the host, and these interactions are constantly evolving. Stromal cells not only secrete extracellular matrix and participate in wound contraction, but they also secrete fibroblast growth factors(FGF) molecules, which mediate macrophage differentiation. Tumor-associated macrophages migrate to hypoxic areas and secrete vascular endothelial growth factor (VEGF) to promote angiogenesis. The microbiome and its byproducts alter the metabolic milieu by shifting the balance between glucose utilization and fatty acid oxidation, and these changes subsequently influence the immune response in the TME. Not only does radiation exert cell-autonomous effects on tumor cells, but it influences both the tumor-promoting and tumor-suppressive components in the TME. To gain a deeper understanding of how the TME influences the response to radiation, the American Society for Radiation Oncology and the American Association of Cancer Research organized a scientific workshop on July 26–27, 2018, to discuss how the microbiome, the immune response, the metabolome, and the stroma all shift the balance between radiosensitivity and radioresistance. The proceedings from this workshop are discussed here and highlight recent discoveries in the field, as well as the most important areas for future research.
Historically, the field of radiobiology has emphasized the cell-autonomous effects of radiation on DNA damage and cell death, but recent advances in immunology, stromal biology metabolism, and microbiome research have highlighted the need to understand how radiation is influenced by diverse biologic processes. While radiation serves as a cytotoxic agent that induces DNA damage, it also modifies various components of the tumor microenvironment (TME) and alters the threshold for anti-tumor immune activity. Radiation oncologists must consider the spatial and temporal characteristics of each of these interactions to understand the heterogeneous responses seen after radiotherapy. Ultimately, as radiation is paired with therapies that target other components of the TME, obtaining a more comprehensive understanding of interactions between radiation and the TME will help us improve clinical outcomes for our patients.
Introduction
We are at a crucial moment in the field of radiobiology. The cell-autonomous effects of radiation on DNA damage and cell death are well-established, but it is becoming increasingly clear that we cannot view tumor cells in isolation because they are intricately linked to the tumor microenvironment (TME) in which they reside. Decades of work on clonogenic survival assays has demonstrated that radiation causes DNA damage, which causes irradiated cells to undergo mitotic catastrophe. Because the induction of reactive oxygen species contributes to DNA damage, hypoxia and reoxygenation have become key determinants of the radiation response. Therefore, most radiobiology research on the TME has focused on hypoxia. However, there are also many cell types in the TME which could influence the response to radiation, including stromal cells, endothelial cells, innate immune cells, and adaptive immune cells. Each tumor is defined by differences in the composition of tumor cells and other cellular elements; for example, pancreatic cancer contains a significant stromal component, small-cell lung carcinoma contains mesenchymal, epithelial, neuroendocrine, and neural cells, and lymphoma contains cells derived from the rich lymphatic environment of secondary lymphoid organs. In the clinic, heterogeneity between different histologies accounts for varied clinical responses to stereotactic body radiation (SBRT) using the same radiation dose and fractionation (1).
Although it has been known for awhile that radiation activates the immune system, it is becoming clear that many of the factors that influence responsiveness to immune checkpoint inhibitors (ICI) also influence the response to radiation. Outcomes after ICIs are influenced by the genetic mutational load in the tumor (2), mutations in DNA damage response (DDR) and DNA repair pathways (3), and adaptive resistance, a process in which IFNγ production by activated CD8+ T cells leads to an upregulation of PD-L1 (4). The gut microbiome plays a key role in regulating α-PD1–elicited immune responses, because antibiotic treatment reduces bacterial biodiversity and decreases the effectiveness of immunotherapy (5–7). Metabolic pathways such as glycolysis, amino acid metabolism, and fatty acid β-oxidation influence the response to PD-1 inhibitors (8, 9). Components of the stroma contribute to an immune-excluded environment and mediate epithelial–mesenchymal transition in response to immunotherapy and radiation (10, 11).
To gain a deeper understanding of how these components of the TME influence the response to radiation, the American Society for Radiation Oncology (ASTRO) and the American Association of Cancer Research (AACR) organized a scientific workshop on July 26–27, 2018, which focused on the following key issues: (i) how the microbiome influences radiosensitivity vs. radioresistance, (ii) immune responses to radiation in the TME compared with the periphery, (iii) how metabolism and obesity affect the TME and responses to radiation, and (iv) the role of the stroma in shifting the balance between radiosensitivity and radioresistance.
Influence of the Microbiome on the Response to Radiation
The gut microbiome is rapidly becoming an important clinical biomarker and therapeutic target because microbiome–host cross-talk is important for both metabolism and immune regulation. Dr. Giorgio Trinchieri, Director of the NCI Cancer and Inflammation Program, described the crucial role of the microbiome in setting the threshold for activating the immune response, largely mediated by the IL23–IL22 axis, which promotes maintenance of the intestinal barrier and therefore prevents dysbiosis or dysregulated symbiosis between bacteria and host. This is important in the context of cancer because germ-free mice without microbiota develop larger tumors in the azoxymethane/dextran sulfate sodium colon cancer model (12), and show decreased responsiveness to α-PD-1 therapies (13). Multiple studies have shown that response to immunotherapies are linked to the microbiota in patients. However, differences in geography and diet lead to microbiome variability, which make it challenging to interpret and generalize results from these studies (5–7).
To elucidate mechanisms of microbiome-associated radioresistance, Dr. A. Klopp from MD Anderson Cancer Center illustrated that broad-spectrum antibiotic depletion of the microbiome results in radioresistance of cervical cancer to a single fraction of 6 Gy (14). Dr. A. Facciabene from University of Pennsylvania has shown that vancomycin, but not neomycin or metronidazole, decreased growth of tumors and sensitized the tumors to radiotherapy. Interestingly, vancomycin decreases bacterial diversity by eliminating many subspecies of the Clostridiales family, leading to a decrease in production of bacteria-derived short chain fatty acids (SCFA). Dr. Facciabene's research has shown that some of the absorbed SCFAs (e.g., the C4-SCFA, butyrate) have immunosuppressive effects in the TME, therefore a decrease in SCFA levels can sensitize tumors to radiotherapy by enhancing antigen presentation and T-cell priming. The implication from this work is that inhibitors of SCFA could potentially be delivered in combination with radiation to serve as radiosensitizers. In addition, the microbiome is increasingly appreciated for its role in the processing of metabolites of microbial origin, and many of these byproducts (e.g., tryptophan metabolites) are immunosuppressive molecules. Some of these immunosuppressive pathways have recently been found to modulate the response to radiation in certain forms of cancer (14, 15). Active areas of investigation in this field focus on developing and implementing methods of microbiome manipulation (spore therapy, probiotics, transplant of specific organisms, or fecal transfer etc.) in combination with radiotherapy.
The Immune Response in the TME
Radiation can act as an in situ tumor vaccine by releasing tumor-associated antigens and damage-associated molecular patterns (DAMPs) to induce immunogenic cell death and activate dendritic cells to migrate to the draining lymph nodes and prime antigen-specific CD8+ T cells. While there is a strong interest in studying how localized radiotherapy induces systemic immune responses, effective antitumor immunity is counter balanced by immune tolerance and immune suppression in the TME (16). In preclinical studies using subcutaneous tumor models in the context of CTLA-4 blockade with radiation, hypofractionated radiation regimens (8 Gy × 3 fractions) produce a more robust CD8+ T-cell–mediated abscopal response and improve survival compared with 20 Gy × 1 (17). Hypofractionated radiation can cause double-stranded DNA (dsDNA) to accumulate in the cytoplasm, and this stimulates cyclic GMP/AMP synthase (cGAS)/STimulator of INterferon Genes (STING)/IFNβ signaling, which recruits Batf3+ dendritic cells to activate CD8+ T cells. Dr. C. Vanpouille-Box from Weill Cornell discussed her work showing that radiation doses greater than 12 Gy per fraction induce expression of the DNA exonuclease 3′ repair exonuclease 1 (TREX1; 3′ → 5′), which degrades cytosolic dsDNA, abrogates radiation induction of IFNβ, and decreases systemic antitumor immunity (18). This research suggests that we need to incorporate assays for DNA damage, immunogenic cell death, and innate immune sensing into clinical trials to better understand how radiation leads to immune activation.
Importantly, work by Dr. M. Gough from Earle Chiles Research Institute has shown how antitumor immune responses after combined radiation and ICI depend on preexisting immunity (19). His murine studies have shown that tumor implantation leads to establishment of resident memory CD39+ CD103+ CD8+ T cells in the tumor milieu. Radiation decreases, but does not eliminate, resident memory T cells in the tumor. In addition, his work has shown that treatment with anti-CD40L, depletion of CD8 T cells, or delivery of FTY-720 (which blocks SIP1-mediated exit of lymphocytes from the lymph nodes) prior to tumor implantation prevents tumor control caused by the combination of radiation and immune checkpoint inhibition (19). The timing of the radiation treatment in conjunction with immunotherapy is also likely to be a key factor in determining outcomes, as discussed by Dr. R. Samstein from Memorial Sloan Kettering Cancer Center, based on his analysis of retrospective data.
Dr. D. Schaue from University of California, Los Angeles, described her translational work studying differences in patients' immune response to radiation based on dose, fraction size, site of disease, and volume irradiated. Together with Drs. N. Nickols and A. Kalbasi from UCLA, she has observed an increased myeloid cell infiltrate after hypofractionated radiation to the prostate and an increased T helper cell infiltrate after conventional fractionation for undifferentiated pleomorphic sarcoma. This work emphasizes the need to expand our approach to studying the immune response to radiation beyond the classic paradigm of radiation as “in situ vaccine” discussed above, which focuses on dendritic cells and CD8+ T cells. Finally, the breakout session focused on “hot” (inflamed) versus “cold” (noninflamed) tumors, and emphasized that “cold tumors” can be either “immune-excluded tumors” characterized by altered vasculature and cancer associated fibroblasts, or “immune desert tumors” characterized by anergy, tolerance, or immunologic ignorance (20). Future studies are needed to investigate how radiation influences each of these unique tumor immune microenvironments to enhance the local and systemic antitumor immune response.
How the Metabolome Affects the Response to Radiation
Hypoxia and oxidative stress alter metabolism in the TME, and therefore it is not surprising that metabolic changes in the TME affect the response to radiation. Specifically, metabolic adaptations to hypoxia cause decreased mitochondrial function and increased lactate production. This leads to hypoxia-inducible inhibitory phosphorylation of the pyruvate dehydrogenase E1α subunit, which enhances tumor growth (21). Dr. N. Denko from Ohio State University discussed work suggesting that hypoxia is not a problem of oxygen delivery, but rather is due to excessive oxygen consumption by tumors that cannot be met by supply demands (22). Thus, interventions that specifically target oxygen consumption may sensitize tumors to conventional cancer therapies such as ionizing radiation. In addition, endocrine imbalances associated with obesity affect responsiveness to radiotherapy. Highly glycolytic cervical tumors are associated with alterations in the PI3K/AKT pathway and are resistant to chemoradiation (23), and the glucose inhibitor 2-DG can sensitize these tumors to ionizing radiation, as described by Dr. J. Schwartz from Washington University in Saint Louis (24). In addition, targeting glutathione and thioredoxin alone or in combination with 2-DG enhances the efficacy of radiation in these resistant tumors. Therefore, both redox metabolism and glycolytic flux could be targeted pharmacologically to improve radiation responsiveness.
Fatty acids upregulate PPAR-γ and inhibit p53, which both sensitize cancer cells to radiation. In cervical cancer cells, radiation alone seems to increase uptake of unsaturated fatty acids, in part by regulating the receptor fatty acid uptake receptors such as CD36. As mentioned in the previous section, the antibiotic vancomycin reduces bacterial-derived SCFAs highlighting an intersection between microbiota and metabolism regulation that it is likely to influence tumor growth and the response to therapy through the PPAR-γ pathway. Dr. N. Simone from Thomas Jefferson University presented results from a clinical study showing that patients with diabetes treated with intracranial radiotherapy have reduced overall survival and reduced median intracranial progression-free survival compared with nondiabetic individuals, emphasizing a role of host metabolism in radiotherapy response (25). She discussed results where caloric restriction led to radiosensitization of tumors in preclinical models and also suggested the potential of ketogenic diets, which are low in carbohydrates, to sensitize tumors to radiotherapy. Therefore, dietary host modification of metabolism may improve outcomes after radiotherapy (26, 27). Future studies on the metabolome in radiation biology will help us elucidate specific metabolite signatures associated with favorable responses to radiation, and how dietary modifications cause metabolic changes that improve clinical outcomes after radiation.
Role of the Tumor Stroma in the Response to Radiation
The stroma consists of fibroblasts, which can either form fibrovascular areas throughout a tumor or can form a desmoplastic matrix that separates the tumor core from other cells in the microenvironment. Fibroblasts can secrete collagen and fibronectin, which serve as components of the extracellular matrix, and can also secrete matrix metalloproteinases, which degrade this scaffold. In addition, cancer-associated fibroblasts secrete growth factors such as epidermal growth factor (EGF), hepatocyte growth factor (HGF), FGF, and insulin-like growth factor 1 (28). Neutrophils secrete FGF2 into the extracellular matrix, where FGF2 is sequestered until it is released after proteolysis (29). Dr. R. Muschel from Oxford University has shown that in experimental models of colon cancer, radiation stimulates intratumoral infiltration of macrophages, with simultaneous induction of FGF2 and FGF receptor (FGFR). Dr. R. Muschel's work has revealed that FGF signaling causes macrophages to switch from an M1 phenotype to an M2 phenotype, which mediate resistance to radiation (unpublished). Dr. W.A. Woodward from MD Anderson Cancer Center has studied stromal elements of the TME in inflammatory breast cancer and has illustrated intricate cross-talk between tumor-associated macrophages and mesenchymal stem cells (MSC) (30). M2-polarized macrophages educate MSCs to develop an M2 phenotype themselves, which stimulates invasion of inflammatory breast cancer cells in vitro, and leads to an increase in the aggressiveness of tumors in vivo. Similar to Dr. R. Muschel, Dr. W.A. Woodward has shown that M1 macrophages promote radiosensitivity of inflammatory breast cancer cells, but M2 macrophages induce radioresistance via IL4/IL13-mediated STAT6 phosphorylation (31).
Another element of cross-talk between cancer cells and the stroma involves the often-overlooked role of stromal cells as antigen-presenting cells. Dr. M. Spiotto from the University of Chicago discussed his experimental model of transplanted tumors, in which tumor cells that are poorly antigenic can escape host immune surveillance when they are challenged with stroma (32). Dr. M. Spiotto mentioned that negative signaling pathways induced by the stroma counteract an antitumor immune response, but radiation can increase stromal cells' ability to present antigens and promote antitumor immunity. In addition, Dr. M. Spiotto found that loss of expression of Notch1 increases the expression of ECM proteins and the infiltration of macrophages (33), which echoes Dr. R. Muschel's findings that there is cross-talk between fibroblasts and macrophages. Moreover, Dr. M. Spiotto discussed distinct transcriptomic subtypes identified in lymph node metastases that are absent in primary tumors, and emphasized that this is likely due to the unique lymph node microenvironment. Efforts to understand how the genetic makeups of cancer affects radiation responses were presented by Dr. M. Abazeed of Cleveland Clinic. He showed that patients with non–small cell lung cancer with squamous histologic subtype have an increased risk for local failure, while patients with adenocarcinoma had significantly higher local control rates after SBRT (34). The breakout session emphasized that we need to characterize the tumor stroma with a higher definition utilizing both single-cell–based sequencing, multiplexed imaging, and deconvolution of bulk cell data to identify (i) stromal characteristics that are specific for individual tumors and/or shared across different tumors, (ii) normal stroma versus tumor stroma–specific targets, and (iii) stromal characteristics that characterize response to radiotherapy. In addition, leaders in this field emphasized the importance of developing improved preclinical model systems to study cancer cell–stroma interactions, such as 3D culture systems (organoids and/or bioengineered model systems), syngeneic mouse models and genetically engineered primary tumor models, and patient sample–derived models (patient-derived xenografts and human tumor organoids).
Conclusion
Understanding the contributions from diverse components of the TME will be essential for optimizing radiotherapy (Fig. 1). The microbiome is increasingly appreciated for its role in the processing of metabolites of microbial origin such as bacteria-derived SCFA, and many of these byproducts (e.g., the C4-SCFA, butyrate) have immunosuppressive effects in the TME. The mechanisms by which radiation can induce tissue damage and initiate an immune response are intricately involved with DDR and DNA repair processes. The byproducts of fatty acid metabolism can interact with stromal cells and influence cancer progression (35), but these metabolic processes are only beginning to be understood in the context of the TME. Finally, the stromal cells, which form the extracellular matrix, interact with immune cells, and often influence the threshold for tumor invasion and metastases; therefore, it is not surprisingly that TGF-β and FGF2 are emerging as key mediators of response and toxicity after radiation.
The intersection of radiation and the tumor immune microenvironment. A, Microbiome. Broad-spectrum antibiotics reduce microbiome diversity, which decreases gut microbiota–derived SCFA, leading to increased antigen presentation and T-cell priming, which sensitizes tumors to radiotherapy. B, Immune response. Immunogenic cell death leads to release of DAMPs such as high mobility group box protein 1 (HMGB1), calreticulin, and ATP. Cytosolic DNA activates the cGAS/STING pathway, leading to an increase in IFNβ, which activates Baft3+ dendritic cells that play a key role in cross-presentation and cross-priming of CD8+ T cells. CD39+ CD103+ tissue resident memory T cells also increase sensitivity to ionizing radiation. C, Metabolome. Hypoxia leads to an increase in glucose uptake and glycolysis, which contributes to radioresistance. During hypoxic conditions, mitochondria act as O2 sensors and convey signals to hypoxia-inducible factor (HIF)-1α. Impaired glucose and lipid metabolism leads to an increase in lipid and fatty acid stores, and fatty acid uptake by CD36, can increase radiosensitivity. PPAR-γ inhibition of p53 can also increase sensitivity to ionizing radiation. D, Stroma. FGF2 is sequestered in the extracellular matrix and released upon proteolysis. FGF signaling induces M1 to M2 macrophage conversion. M1 macrophages promote radiosensitivity, while M2-polarized macrophages mediate radiation resistance and promote MSCs to develop an M2 phenotype, which facilitates tumor cell migration and invasion.
The intersection of radiation and the tumor immune microenvironment. A, Microbiome. Broad-spectrum antibiotics reduce microbiome diversity, which decreases gut microbiota–derived SCFA, leading to increased antigen presentation and T-cell priming, which sensitizes tumors to radiotherapy. B, Immune response. Immunogenic cell death leads to release of DAMPs such as high mobility group box protein 1 (HMGB1), calreticulin, and ATP. Cytosolic DNA activates the cGAS/STING pathway, leading to an increase in IFNβ, which activates Baft3+ dendritic cells that play a key role in cross-presentation and cross-priming of CD8+ T cells. CD39+ CD103+ tissue resident memory T cells also increase sensitivity to ionizing radiation. C, Metabolome. Hypoxia leads to an increase in glucose uptake and glycolysis, which contributes to radioresistance. During hypoxic conditions, mitochondria act as O2 sensors and convey signals to hypoxia-inducible factor (HIF)-1α. Impaired glucose and lipid metabolism leads to an increase in lipid and fatty acid stores, and fatty acid uptake by CD36, can increase radiosensitivity. PPAR-γ inhibition of p53 can also increase sensitivity to ionizing radiation. D, Stroma. FGF2 is sequestered in the extracellular matrix and released upon proteolysis. FGF signaling induces M1 to M2 macrophage conversion. M1 macrophages promote radiosensitivity, while M2-polarized macrophages mediate radiation resistance and promote MSCs to develop an M2 phenotype, which facilitates tumor cell migration and invasion.
Radiation oncology must merge with scientific other fields (metabolism, immunology, cancer biology, bioengineering, and microbiology) and emphasize a collaborative, interdisciplinary approach to learn how radiation can most effectively target the TME. There is a critical need to invest in diverse research areas to enhance our understanding of the TME and learn how different components (microbiome, immune response, metabolome and stroma) affect the response to radiation. In order to move the field forward, we must expand our focus beyond the cell autonomous affects of radiation and incorporate discoveries from diverse fields to form the next generation of radiobiology research.
Disclosure of Potential Conflicts of Interest
H.M. McGee is a consultant/advisory board member for AstraZeneca, and reports receiving commercial research support from Adaptive Biotechnologies. D.R. Soto-Pantoja is a consultant/advisory board member for Morphiex Biotherapeutics. No potential conflicts of interest were disclosed by the other authors.
Acknowledgments
The authors would like to acknowledge the ASTRO and AACR scientific leadership who organized the ASTRO–AACR workshop (Judy Keen, PhD; Tyler Beck, PhD; and Michael Powell, PhD) along with W.A. Woodward and A.J. Giaccia for chairing the session. They would also like to acknowledge all speakers who presented their research or moderated sessions or breakout sessions at the workshop, including: Giorgio Trichinieri, MD; Ann Klopp, MD, PhD; Andrea Facciabene, PhD; Phuoc Tran, MD, PhD; Dorthe Schaue, PhD; Robert Samstein, MD, PhD; Claire Vanpouille-Box, PhD; Michael Gough, PhD; Mohamed Abazeed, MD, PhD; A.J. Giaccia, PhD; Julie Schwarz, MD, PhD; Nicolas Denko, MD, PhD; Nicole Simone, MD; W.A. Woodward, MD, PhD; Simon Powell, MD; Catherine Park, MD; Ruth Muschel, Michael Spiotto, MD, PhD; D. Jiang, PhD; D.R. Soto-Pantoja, PhD; A. Nevler, MD; and H.M. McGee, MD, PhD. Due to space limitations, we were unable to include a discussion of all of the research presented. Funding support: P30CA012197, 1K22CA181274 (to D.R. Soto-Pantoja).