Effects of Radiation on the Tumor Microenvironment

Abstract

A malignant tumor consists of malignant cells as well as a wide array of normal host tissues including stroma, vasculature, and immune infiltrate. The interaction between cancer and these host tissues is critical as these host tissues play a variety of roles in supporting or resisting disease progression. Radiotherapy (RT) has direct effects on malignant cells, but, also, critically important effects on these other components of the tumor microenvironment (TME). Given the growing role of immune checkpoint inhibitors and other immunotherapy strategies, understanding how RT affects the TME, particularly the immune compartment, is essential to advance RT in this new era of cancer therapy. The interactions between RT and the TME are complex, affecting the innate and adaptive arms of the immune system. RT can induce both proinflammatory effects and immune suppressive effects that can either promote or impede antitumor immunity. It is likely that the initial proinflammatory effects of RT eventually lead to rebound immune-suppression as chronic inflammation sets in. The exact kinetics and nature of how RT changes the TME likely depends on timing, dose, fractionation, site irradiated, and tumor type. With increased understanding of the effects of RT on the TME, in the future it is likely that we will be able to personalize RT by varying the dose, site, and timing of intervention to generate the desired response to partner with immunotherapy strategies.

Introduction

The tumor microenvironment (TME) is defined by a continuous and dynamic interaction between the immune system and cancer cells, which are dependent upon immune evasion for survival.¹ Recently, immune checkpoint inhibitors have been used to shift the balance of the TME from a state of immunosuppression to a state of immune activation, allowing for a sustained and durable tumor response across multiple disease sites. Increasing use of immune checkpoint blockade has also provided the opportunity for combination with radiotherapy (RT) to produce an abscopal effect, in which radiation to 1 site of metastatic disease may produce a regression in a distant, nonirradiated site.² However, the mechanisms that underlie the abscopal effect are not well defined, and there have been only limited reported cases of abscopal effect.³ Nonetheless, recent advances in our understanding of the effect of ionizing radiation on the TME have offered new approaches to unleashing the abscopal effect on a broad basis.⁴ Several ongoing, prospective clinical trials are combining RT and immune checkpoint inhibitors using abscopal response as a surrogate endpoint of efficiency.⁵ Here, we review the complex interactions between RT and the TME, including both the proinflammatory effects and the immune suppressive effects that can either promote or impede antitumor immunity.

The effect of RT on the TME and immune response is highly dependent upon the fractionation of radiation delivered. While several preclinical studies have demonstrated a synergistic response to combined RT and immunotherapy in a variety of tumor types,² there has not been a consensus on the optimal dosing of radiation. Initial studies into dose fractionation in breast and colon cancer models found that multiple fractions were superior to a single ablative dose in combination with anti-CTLA4.⁶ However, other preclinical studies have demonstrated the induction of antitumor T cells with single ablative dosing.⁷ Given conflicting data from preclinical studies, the effect of radiation doses on the immune response were examined. Low-dose radiation at 2 Gy was shown to stimulate nitric oxide synthase by tumor-associated macrophages and create an immunogenic environment.⁸ In contrast, higher doses of radiation were shown to promote tumorigenic macrophages⁹ and cause severe vascular damage, decreasing recruitment of immune cells to the tumor.¹⁰ More recent mechanistic studies highlighted the role of radiation in upregulating 3-prime repair endonuclease 1 (TREX1), which digests cytosolic DNA and reduces radiation-induced immunogenicity.¹¹ Radiation doses above 12-18 Gy were shown to highly induce TREX1, as determined by the size of the single dose rather than total dose. Therefore, fractionation and dosage can significantly alter the immune response to radiation on the TME.

The immune response to radiation occurs in 5 distinct phases.¹² The initial phase is characterized by the release of damage-associated molecular patterns (DAMPs), which activate the NF-kB pathway, leading to a release of proinflammatory cytokines by innate immune cells and initiation of early inflammation.¹³ Radiation also induces tumor cells to release chemokines, creating a positive feedback loop for the recruitment of additional immune cells. In the next phase, innate immune cells participate in antigen presentation to T-cells. This stage is enhanced by granulocyte-macrophage colony-stimulating factor (GM-CSF) secretion following radiation to promote dendritic cell differentiation¹⁴ and by induction of costimulatory molecules on T-cells following radiation.^15,16 Although radiation initially induces an antitumor response, radiation also induces chronic inflammation and rebound immune suppression. In this stage, protumorigenic macrophages are recruited to the tumor in a radiation dose-dependent manner, creating an immunosuppressive TME that supports tumor regrowth and/or resistance. Radiation also induces HIF-a, which induces the expression of PD-L1 in tumor cells and tumor associate macrophages, which results in immune suppression.¹⁷ As such, radiation appears to have a temporal effect on the immune response of the TME in which there appears to be a window of anti-tumor response.

In this article we will review the impact of RT on the TME with an emphasis on how RT-induced changes in the microenvironment shape the antitumor efficacy of RT. We will focus heavily on aspects such as non-T cell lymphoid cells and innate cells which have been less comprehensively addressed in other reviews. While any discussion of RT effects on the immune compartment of the TME must necessarily describe effects on tumor vasculature and tumor stroma, we will not separately discuss the tumor vasculature and endothelium but will provide a discussion on the tumor stroma and fibroblasts. We additionally refer the reader to a thorough review by Harrington et al¹⁸ for a more detailed examination of these subjects.

Lymphocytes

There is a growing body of data examining the effects of RT on T cells in the TME. These findings have been expertly summarized in a number of recent reviews on the topic such as those from Sharabi et al¹⁹ and DeMaria/Formenti et al.^4,20 We will provide a brief review here. The effects of RT on T cells in the TME can be broadly divided into 2 overlapping categories: the effects on existing T cell responses and the generation of new T cell responses.

With regards to the augmentation of existing T cell responses, numerous effects have been described. Radiation can sensitize tumor cells to T cell cytotoxicity by increasing expression of major histocompatibility complex (MHC) I²¹ and Fas.²² Radiation can induce inflammatory cytokines, such as interleukin-1B, tumor necrosis factor-alpha, and interleukin-6, which support the function, expansion, and differentiation of antigen-experienced T cells.^23-25 It should be noted that most of the existing cytokine induction data is from analysis of irradiated normal tissues and not the TME. Finally, radiation can induce the homing and infiltration of T cells into the TME.

Two studies from Hammerling et al published roughly a decade apart elegantly demonstrate that RT can normalize tumor vasculature facilitating T cell homing and infiltration into the TME.^8,26 Other studies confirm that radiation induces T cell homing and infiltration into the TME^27,28 and additional contributing mechanisms, such as radiation induced chemokines²⁹ and vascular adhesion molecules,³⁰ have been described. This induction of tumor infiltrating lymphocytes (TILs) is one of the most commonly discussed immune modulatory effects of RT.

The idea of transforming a “cold” tumor into a “T cell inflamed” tumor is a common rationale for combining RT with immunotherapy. Despite the preclinical data, the clinical data validating this notion are relatively sparse. There are some data that suggest an increase in T-cell proliferation (Ki-67+ T cells) post-RT without demonstrating an overall increase in TILs. An examination of resected oral squamous cell carcinomas revealed that increased proliferation of TILs correlated with preoperative RT.³¹ A recent study of stereotactic ablative body radiotherapy combined with cytoreductive nephrectomy in patients with metastatic renal cell carcinoma likewise demonstrated an increase in proliferating CD8+ T cells post-RT.³² Several studies have demonstrated increased TILs after neoadjuvant chemoradiotherapy in esophageal,³³ cervical³⁴ and rectal cancer.^35-37 It is unclear if the increased TILs are a product of RT, chemotherapy, or the combination of the 2. Another study performed in cervical cancer patients reported decreased TILs after RT; it appears these biopsies may have occurred during and immediately after radiation not allowing sufficient time for infiltration by new lymphocytes.³⁸ To address this gap in the literature, we provide here previously unpublished data (courtesy of: Monjazeb AM, Canter RJ, Murphy WM, Schalper KA) examining TILs in a cohort of 29 patients with soft tissue sarcomas treated with standard neo-adjuvant RT (Fig. 1). Comparing the diagnostic core biopsy with the post-RT resection specimen (4-12 weeks post-RT) we see significant increases in CD4+, CD8+, and CD20+ TILs post-RT (Fig. 1).

Figure 1.

Tumor infiltrating lymphocytes at baseline and after neoadjuvant radiotherapy in clinical sarcoma specimens. (A) An example of multiplexed immunofluorescence staining for tumor infiltrating CD8+ or CD4+ T cells and CD20+ B cells. (B) Quantification of tumor infiltrating lymphocyte staining before (pre) and after (post) neoadjuvant radiotherapy. Paired baseline biopsy preradiotherapy and tumor resection postneoadjuvant radiotherapy samples from 30 patients treated at UC Davis were used to generate a tissue microarray. Samples were stained with DAPI, anti-CD4, anti-CD8, anti-CD20, and anti-Cytokeratin at the Schalper laboratory.

RT also induces well described changes in the TME that can contribute to the activation, function, and localization of pre-existing T cell responses, but that also have the capacity to prime de novo antitumor T cell responses. RT treated tumor cells undergo immunogenic cell death due to radiation induced expression of danger-associated-molecular-patterns (DAMPs). The release of HMGB1 by irradiated cells can activate Toll-like receptor 4 and induce antigen uptake and cross-presentation by dendritic cells.³⁹ Likewise, translocation of calreticulin to the cell surface of irradiated cells can increase phagocytosis of these cells.⁴⁰ Finally, DNA from irradiated tumor cells has been shown to be a critical signal, via the c-Gas-STING pathway, for type I interferon production by TME dendritic cells.⁴¹ Type I interferon signaling is one of the major immune modulatory signals induced by RT and is central to radiation induced TILs and antitumor effects.⁴² In fact, it has been suggested that suppression of type I interferon signaling may mediate resistance to checkpoint inhibitors and that mechanism of resistance can be reversed by RT.⁴³ The immune modulatory effects of RT on T-cells in the TME is an active field of study.

Relatively little attention has been given to the effects of RT on other lymphoid cell types. Natural Killer (NK) cells are innate lymphocytes first recognized for their ability to kill cancer cells without prior sensitization or MHC restriction.⁴⁴ While traditionally considered to play a role in hematologic malignancies growing evidence suggests a role for NK cells in solid tumors as well.⁴⁵ NK cell activity is mediated by cell surface markers such NKG2D (activating) and killer cell immunoglobulin like receptors (KIRs; activating and inhibitory). The cognate ligands for these receptors are upregulated on cancer cells, virally infected cells, and, in general, as part of the cellular stress response. NK cells also play a critical role in antibody dependent cell cytotoxicity via the CD16 receptor which binds Fc on antibody coated cells. As with other lymphocytes, NK cells may be radiosensitive, and RT has been noted to affect the viability and activity of circulating NK cells.

In vitro work suggests that NK cells are radiosensitive and respond like acutely responding tissues.⁴⁶ Clinical data from our lab demonstrated a decrease in circulating total and cytotoxic (CD56^dimCD16⁺) NK cells after ablative doses of RT.⁴⁷ However, some data suggests that NK cells may be more radioresistant than other lymphoid populations.⁴⁸ In fact, it has been reported that mature NK cells are relatively radioresistant but their precursors are radiosensitive.⁴⁹ In vitro data suggests that RT, at doses >30 Gy, affects the cytotoxic function of NK cells before death or apoptosis is observed.^50-52

Patients in 1 clinical study had no difference in the number of circulating CD56+ cells after fractionated RT but did demonstrate a modest decrease in NK cell activity.⁵³ Other clinical studies corroborate this finding,⁵⁴ including studies in breast cancer patients after adjuvant chemo-radiotherapy⁵⁵ and in endometrial cancer patients after RT.⁵⁶ These effects might be dose dependent since data suggests that low-dose RT can increase the activity of NK cells⁵⁷ and clinical data from cervical cancer patients demonstrated an increase in the cytotoxic activity of circulating NK cells after RT of the primary tumor, suggesting systemic activation of NK cells.⁵⁸ Interestingly, in our work described above examining circulating NK cells after ablative RT, despite the decrease in circulating NK cells, we observed a robust increase in TIM3+ NK cells⁴⁷ which have been reported to be the most functional NK cells.⁵⁹

RT also has robust effects on the interaction between NK cells and tumor cells in the TME. A number of early studies demonstrated that irradiation targeted cancer cells for NK cell mediated cytotoxicity. Irradiation has been reported to increase the NK cell killing of 3LL Lewis lung carcinoma and MCA105,⁶⁰ as well as K562 cells.⁶¹ Likewise, coculture of human cancer cells with human primary NK cells or the NK-92 cell line yielded no cytotoxicity, but NK cell mediated cytotoxicity was observed when tumor cells were irradiated.⁶² The authors suggest that radiation induced release of Smac sensitizes cancer cells to granzyme b mediated killing by NK cells. Another study demonstrated the ability of NK cells to eradicate murine 4T1 breast cancer.⁶³ The addition of low-dose chemoirradiation increased plasma levels of NK cell-activating cytokines. NK cell activity, and NK cell mediated elimination of tumors. Thus, RT may not only increase the cytotoxic ability of NK cells but may facilitate their translocation into the TME. Another report also supports this notion, demonstrating that RT induces NK cell migration into tumors in a manner dependent on the CXCL16 chemokine which binds CXCR6 expressed on activated NK cells.⁶⁴

With regards to the increased NK cell cytotoxic activity observed after RT this is most likely due to the upregulation of cellular stress markers on irradiated tumor cells sensitizing them to NK cell recognition and killing. A seminal study by Raulet et al demonstrated that genotoxic stress can upregulate NK cell ligands on the cell surface of nonmalignant cells in an ataxia-telangiectasia mutated/ataxia-telangiectasia and Rad3 related-dependent manner.⁶⁵ Studies from our lab have demonstrated that RT upregulates the expression of stress induced NK cell ligands such as MICA/B and Fas on human and murine cancer cells as well as on primary resected human tumors treated with neoadjuvant RT.⁶⁶ Other studies corroborate the increase of NK cell ligands after RT.⁶⁷ The upregulation of these ligands on radioresistant stem-like cells surviving after RT may be critical for NK targeting of this surviving fraction and tumor control after RT.^66,68

Furthermore, we find that adoptive transfer of NK cells, while having no effect as a single modality, drastically increases the efficacy of RT.⁶⁶ A recent canine study from our group demonstrated that RT increased NK cell tumor cytotoxicity in a dose dependent manner in vitro.⁶⁹ Using a patient derived xenograft model with spontaneous canine sarcomas we observed that, after focal tumor irradiation, adoptively transferred NK cells homed to tumors and induced significant antitumor effects. An ensuing pilot clinical trial of NK cell adoptive transfer and focal RT in companion canines with spontaneous osteosarcomas provided preliminary data on the feasibility and efficacy of this approach.

B cells play a critical role both in humoral immunity and antigen presentation and further examination of their role in RT induced immune modulation is warranted. RT has been demonstrated to induce antitumor humoral immunity in prostate cancer patients.⁷⁰ Two preclinical studies by Sharabi et al and Guha et al have also pointed to B cell activation and humoral immunity as important mediators of the antitumor effects of RT combined with immunotherapy.^71,72 Data examining the effects of RT on TME B cells is extremely sparse. Our clinical data above (Fig. 1), however, demonstrates a significant increase in TME B cells post-RT. Clearly more studies are needed addressing the effects of RT on B cell infiltration and function in the TME.

Rebound Immune Suppression

As with of the other body systems, the immune system functions to maintain homeostasis. Too much inflammation can lead to autoimmunity and inflammatory disease whereas immune suppression can lead to overwhelming infection and malignancy. Thus, it stands to reason that every perturbation made will be met with an opposite reaction to maintain balance and homeostasis. We have coined this process “rebound immune suppression”.⁷³ Indeed, the emerging biology of the PD-1/PD-(L)1 axis demonstrates how finely tuned the immune system is and how delicate this balance is. PD-1, although widely considered a marker of exhaustion, is an early activation marker^74,75 rapidly upregulated after T cell receptor engagement presumably as a means to temper T cell activity and maintain peripheral tolerance and only mediates exhaustion when engaged by its ligands. Likewise, the inflammatory signals induced by RT often trigger counterregulatory immune suppressive mechanisms which limit the proinflammatory effects of RT. For example, as outlined above, RT induces expression of potent proinflammatory type I and II interferon cytokines. These cytokines, in addition to their above described inflammatory effects, can also upregulate expression of PD-L1.⁷⁶ Indeed, it has been reported that radiation induces upregulation of PD-L1 in the TME which can mediated resistance to RT.^77,78 These findings form the basis for the many trials combining RT with PD-(L)1 checkpoint inhibition.

Indolamine 2,3-dioxygenase (IDO) catalyzes the breakdown of tryptophan to kynurenine and other downstream catabolites and has been implicated as a master orchestrator of immune suppression in the TME. By altering the metabolic landscape of the TME through starvation of tryptophan and induction of immune suppressive catabolites IDO has been implicated in T cell suppression, T reg induction, and induction of immune suppressive myeloid cells (myeloid-derived suppressor cells (MDSCs)) and tumor-associated macrophages (TAMs). Studies from our lab and corroborated by Welsh et al indicate that the inflammatory effects of RT can induce IDO upregulation.^73,79 The above studies demonstrate post-RT that IDO expression is associated with increased TAMs or increased MDSCs and that IDO inhibition improves the efficacy of RT and results in reduction of TME immune suppressive elements including Transforming growth factor beta (TGFβ), TAMs, MDSCs, and Tregs. As described below, this same process of rebound immune suppression can be seen with many innate cell types in the TME after RT, which at first may have an inflammatory function but later switch or are replaced by cells with suppressive function.

Overall, our understanding of this process of rebound immune suppression needs to increase and it suggests that there may be a window period post-RT where proinflammatory effects dominate presenting an opportunity for immune modulation but after which increasing immune suppression limits the immune response and in-situ vaccination effect. The clinical data from the PACIFIC trial supports this notion as patients who started checkpoint inhibition within 14 days of completing RT appeared to have much better outcomes than patients who started later.⁸⁰

Innate Immunity

The innate immune system consists of cells that do not express antigen-specific receptors (T cell and B cell receptor) and often serve as the regulators of an immune response controlling the initiation and elimination of an inflammatory response. The innate system is broadly composed of macrophages, NK cells, and DCs. As early detector of cell damage and initiators of inflammation, activation of the innate immune system likely serves as one of the main mechanisms driving the extraordinary efficacy of RT. Evidence of the importance of innate immunity in the response to RT come from studies that demonstrate reduced efficacy for RT in preclinical models of cancer which are deficient in innate immune cells including NK cells,⁸¹ macrophages,^82,83 and DCs.⁸⁴ These findings are further supported by numerous observations from patients; 1 study in hepatocellular carcinoma, for example, showed that increased numbers of circulating myeloid cells following RT correlated with poorer responses.⁸⁵ Thus, given that innate immunity has such an important role in determining the response to RT, multiple groups have explored the mechanisms by which RT interacts with the innate immune system. We discuss the findings from these studies below in the context of the different functions of the innate immune system: initiation of inflammation, activation of the adaptive immune response, and resolution of an immune response.

When cells of the innate immune system detect that there is a problem, for example an infection or tissue damage, they activate a program of inflammation that leads to activation of the adaptive response (T and B cells) that requires maturation of dendritic cells or macrophages into antigen-presenting cells and appropriate expression of MHC molecules and co-stimulatory signals. Interestingly, RT has been shown to upregulate MHC class I and stimulate presentation of unique antigens,^72,86,87 as well as costimulatory molecules^87,88 by dendritic cells.

The importance of dendritic cell (DC) in mediating the efficacy of RT was shown by Dewan et al, where fractionated RT along with anti-CTLA4 had significant abscopal effects in part through the generation of increased numbers of Batf3 DCs.⁶ Batf3 dependent DC cells are an important subset of dendritic cells with their ability to efficiently cross-present antigens and regulate tumor growth by enhancing CD8+ T cell migration to the TME and fostering effective T cell response.^89,90 Abscopal effects were abolished in the Batf3−/− mice consistent with other observations demonstrating the critical role of Batf3 DC in regulating RT-induced antitumor immune responses.^89,91,92 In addition to its effects on DC, RT further contributes to the adaptive immune response by encouraging innate immune cells to establish an inflammatory milieu in irradiated tissue in part through stimulating the release of complement and proinflammatory cytokines and chemokines by innate immune cells.^93,94

Several groups have shown that they can improve the response to RT in murine models and early human trials by increasing the growth and differentiation of dendritic cells. One way to increase the number of DC is the cytokine GM-CSF which has been shown to be a crucial pathway for the growth, maturation, and migration of DC.^95,96 Several human trials of GM-CSF in melanoma and breast cancer have demonstrated the efficacy of GM-CSF administration alone with improved survival compared to historical controls^97,98 and an increase in circulating DC.⁹⁹ Based on these successful early studies, trials of GM-CSF and RT were initiated. In 1 trial of metastatic patients of various histologies, exogenous administration of GM-CSF with a course of fractionated RT (35 Gy in 10 fractions) found evidence of an abscopal, and hence systemic, anti-tumor immune response in 27% of the patients.⁹⁶

Another cytokine for DC-specific growth similar to GM-CSF that has been shown to enhance the response to RT is the Fms-like tyrosine kinase 3 ligand (FLT3L).^100-102 FLT3L binds and activates FLT3 on hematopoetic progenitors and serves a critical role in steady-state maintenance of DC¹⁰³ and increased levels of FLT3L during inflammation mobilizes DC.¹⁰⁴ Two studies using preclinical models of nonsmall cell lung cancer demonstrated reduced tumor growth, metastases, and improved survival with administration of RT and FLT3L in a T-cell dependent manner.^101,102 Preclinical data in a murine model of hepatocellular carcinoma has also shown that the efficacy of RT can be enhanced by augmenting DC function through the use of exogenous IL-12 to help DCs better generate cytotoxic T cells.¹⁰⁵

With the recent recognition of the need to alleviate the intrinsic tumor immunosuppression to allow antitumor immunity to progress, much activity has been devoted to targeting the pathways and cells that mediate immunosuppression. Interestingly, many of the cellular targets are innate immune cells such as macrophages. Since RT generates both an antitumor immune response and the corresponding suppressive immune control mechanisms, combinations of RT with agents that target intratumoral immune suppression are thought to allow for an enhanced antitumor immune response following RT. Preclinical models strongly support this notion and clinical data are just emerging that suggests that this strategy may be efficacious in the clinical setting.

One of the most successful regimens targeting intratumoral immunosuppression has been targeting immune suppression with checkpoint inhibitors, which are agents that target the PD-1/PD-L1 and CTLA-4 pathways. Innate immune cells are one of the key sources of signal for the PD-1/PD-L1 pathway with dendritic cells and macrophage serving as one of the primary, nontumor sources of PD-L1 in the TME. Thus, the underlying mechanism of checkpoint blockade likely involves disrupting the effects of innate immune cells on immune response in tumors. To date, a large amount of data has demonstrated the efficacy of checkpoint inhibitors in the preclinical and clinical setting in combination with RT. As several excellent recent reviews have examined the role of combining checkpoint blockade with RT in detail, we will not discuss combinations with checkpoint blockade further here though it should be recognized that including one of these agents as a part of any immune-directed therapeutic regimen will be an important consideration for the foreseeable future.^2,106 Likewise, there is a rising interest in immunotherapies that directly activate innate immunity. The rationale and potential of combining innate immune system agonists with RT has also been recently reviewed.¹⁰⁷

Beyond checkpoint blockade, macrophages serve as the main source of immunosuppression within the TME following RT. As evidence of the importance of macrophages, various studies have revealed a strong negative correlation between the presence of macrophages and survival in various solid tumors including breast, colon, bladder, and lung cancer.^108-110 As described above, macrophages are often associated with resistance to RT and chemotherapy by providing both prosurvival signals and tissue repair functions that protect and/or repair the damage done by these therapies. Various studies have shown that macrophages, the most abundant cells of the TME, are altered by RT to support tumor growth after being damaged and sensing damage resulting from irradiation. For example, Leblond et al found an increase in density of pro-tumor M2 macrophages in the TME post-RT in glioblastoma.¹¹¹ Kioi et al showed that the RT-recruited macrophages help rectify the damage done by RT by promoting vasculogenesis.¹¹²

Given, the protumor role of macrophages following RT, multiple groups have shown that blocking macrophage recruitment via targeting CD11b,⁸³ CCL2¹¹³, or CSF-1R,^82,114,115 enhances the efficacy of RT in preclinical murine models. In a squamous cell carcinoma model Ahn et al found that administration of a CD11b antibody enhanced the efficacy of RT by blocking myeloid cell recruitment to the tumor site after RT leading to delayed regrowth in part through impaired angiogenesis.⁸³ Other studies have revealed that inhibition of macrophages following RT increases both the antitumor immune response⁸² and prevents protumor repair mechanisms such as angiogenesis and matrix remodeling.^83,112 These studies all demonstrate that targeting macrophages can synergize with RT, however, given the potentially positive role of macrophages in producing cytotoxic antitumor immune responses, other groups have sought to preserve the proinflammatory activation capacity of macrophages while preventing their suppressive differentiation to even further synergize with RT. Interestingly, agents targeting tumor-associated macrophages such as the CCL2 inhibitor carlumab have had limited effect as single agents¹¹⁶ and in fact may only have efficacy when combined with other agents such as RT that perturb the tumor immune microenvironment.^82,117

In order to preserve the macrophage capacity to activate antitumor immunity while preventing their differentiation into protumor, immunosuppressive phenotypes, several groups including our own have examined the potential of targeting the pathways that lead to protumor phenotypes in macrophages including IL-4,⁸² arginase 1,¹¹⁸ TGF-β,¹¹⁹ and Tyro3/Axl/Mer tyrosine kinases^120,121 in combination with RT. Targeting macrophage differentiation led to improved antitumor immunity, particularly cytotoxic CD8+ T cells, resulting in dramatically enhanced responses to RT. Though each of these strategies targets a distinct pathway found in myeloid-macrophages, they result in reduction but likely not elimination of immunosuppressive differentiation suggesting that even modest reductions in tumor-associated immunosuppression can have profound effects on therapeutic responsiveness to RT.

Transforming Growth Factor Beta

TGFβ is a multipotent cytokine with both tumor suppressive and tumor promoting properties. In preinvasive disease, such as carcinoma-in-situ, TGFβ acts as a tumor suppressor primarily through its growth inhibitory functions. However, once a tumor becomes invasive, TGFβ is tumor promoting via roles in epithelial to mesenchymal transition,¹²² angiogenesis,¹²³ tumor cell motility and metastasis,¹²⁴ cancer-associated fibroblast (CAF) proliferation,¹²⁵ and immunosuppression.¹²⁶

TGFβ is produced and released into the TME in its latent form, and can be activated through multiple mechanisms (reviewed in¹²⁷), including radiation. ProTGFβ is synthesized as a homodimer consisting of the latent-active peptide (LAP) and the active cytokine TGFβ. The LAP is cleaved from the active cytokine in the Golgi by furin-type enzymes, however the homodimeric LAP forms a cage around the dimeric TGFβ preventing its association with cellular receptors. Disulfide bonds form between LAP and the latent-TGFβ-binding protein to form the large latency complex (LLC). The LLC is anchored in the extracellular matrix through covalent binding of the N-terminal region of latent-TGFβ-binding protein to matrix proteins by transglutaminase.

Release of active TGFβ from the LLC can occur through proteolytic cleavage by matrix metalloproteases (ie MMP-2 and MMP-9),¹²⁷ disruption of the LAP-TGFβ interaction by thrombospondin-1,¹²⁷ or physical force-dependent activation via unfastening of the “straightjacket” domain by integrin binding and stretching of LAP resulting in conformation change releasing bound TGFβ.¹²⁸ Radiation-mediated generation of reactive oxygen species can also activate TGFβ by modifying LAP causing disruption of its interaction with, and therefore activation of, TGFβ.¹²⁹ Elevated levels of TGFβ can be detected within 1 hour of radiation in vivo, related to activation of the LLC already deposited in the extracellular matrix.¹²⁹ Additionally, increased transcription of pro-TGFβ is induced 3-7 days following radiation, particularly within macrophages and neutrophils (pending publication), consistent with the timing of increased TGFβ during wound healing.^130,131

Once activated, dimeric TGFβ binds the heteromeric receptor consisting of 2 copies of the type I receptor (TGFβRI) and 2 copies of the type II receptor (TGFβRII). Binding of TGFβ to its receptor leads to phosphorylation of the serine/threonine kinase, TGFβRI, leading to phosphorylation of the intracellular signaling mediators, Smad2 and Smad3, which are then capable of binding the common Smad, Smad4, leading to nuclear translocation of the Smad complex. Once in the nucleus, Smad3 and Smad4 bind DNA at Smad-binding elements, termed CAGA boxes, and regulate transcription of TGFβ target genes.

TGFβ contributes to an immunosuppressive TME through effects on all immune subsets. TGFβ skews innate immune populations towards phagocytic anti-inflammatory macrophages,¹³² while inhibiting dendritic cell activation, maturation, and migration,^133,134 thereby hampering effective tumor antigen presentation. Furthermore, TGFβ is known to suppress T cell effector function, in part, through Smad-mediated downregulation of the target genes granzyme, perforin, and interferon.¹²⁶ TGFβ promotes regulatory T cell differentiation, further suppressing effector T cells. TGFβ has also been shown to inhibit central memory T cell differentiation by a noncanonical SMAD and mammalian target of rapamycin independent mechanism.¹³⁵ These finding are not universal as other studies have suggested that TGFβ is critical for the differentiation and maintenance of memory CD8+ T cells.¹³⁶ TGFβ can have differing effects on different T cell subpopulations and these nuances should be noted, as depending on the context TGFβ may play a critical role in support of T cells and is not purely immunosuppressive. For example TGFβ has also been demonstrated to play a central role in development, migration, and retention of tissue resident memory T cells in the gut.^137,138

TGFβ promotes stromal fibrosis further contributing to immune escape. TGFβ is a critical mediator of wound healing, enhancing fibroblast migration to the site of injury, resulting in deposition of collagen, inhibition of MMPs, and aiding the transition from the inflammatory phase to the proliferative phase.¹³⁰ Radiation-related adverse events such as pneumonitis and fibrosis may be viewed as impaired wound healing responses, and are associated with elevated levels of serum TGFβ^139,140 and polymorphisms in TGFβ.¹⁴¹ Recent data links TGFβ-mediated fibrosis and immunosuppression. Subsets of patients with metastatic cancer, including urothelial and colorectal cancers, who failed to respond to checkpoint blockade and exhibited a T cell excluded phenotype, harbored an elevated stromal/fibroblast TGFβ gene expression signature.^142-144 Together, these data suggest that TGFβ promotes a suppressive tumor stroma which may exclude T cell infiltration into tumors rendering them resistant to T cell-directed immunotherapy. We and others have previously shown that blockade of TGFβ signaling improves response to radiation dependent upon CD8 T cells, and synergizes with immune checkpoint blockade.^145-149 In addition to the stromal exclusion of T cells, TGFβ may also overpower the antitumor effects of infiltrating immune cells as observed in certain subtypes of breast cancer.¹⁵⁰ Based on these data, we conclude that TGFβ contributes to immunosuppression through promotion of fibrosis and collagen deposition leading to exclusion of T cells; T cells that are able to infiltrate the tumor are rendered less effective by TGFβ mediated suppression of T cell cytotoxicity. Combination of radiation, TGFβ inhibition, and immune checkpoint blockade may be more effective than radiation and checkpoint blockade alone.

Cancer Associated Fibroblasts and Fibrosis

Radiation efficacy can be limited by alterations in the tumor stroma. For example, pancreatic ductal adenocarcinoma (PDAC) is an aggressive malignancy with a poor prognosis characterized by a fibrotic stroma and poor immune infiltrate. PDAC is relatively radioresistant with poor drug penetrance and elevated levels of hypoxia limiting the efficacy of chemoradiotherapy.¹⁴⁹ An additional benefit of radiation is its ability to expose tumor antigens and create a focal inflammatory response,^26,86,151 with efficacy dependent on CD8 T cells,^149,152,153 which is also limited in PDAC. CAFs may be the link between these phenomena in PDAC as well as in other cancers. CAFs are a chief source of extracellular matrix fibrotic components, such as collagen, hyaluronic acid, and fibronectin, which result in impaired drug penetration, poor immune cell infiltration, and reduced vascular patency.¹⁵⁴ Furthermore, CAFs can secrete cyotkines /chemokines and participate in direct cell to cell interactions that govern the functional fate of innate and adaptive immune cells in the TME.¹⁵⁵ CAFs may also play a metabolic role in the TME. Recent studies demonstrate that CAFs may metabolically support malignant cells by secretion of alanine¹⁵⁶ but this could also have some positive immune consequences as T cell function has also been shown to rely on extracellular alanine.¹⁵⁷

Recent data suggests 4 CAF subtypes in breast cancer, with the immunosuppressive CAF-S1 subtype expressing fibroblast activation protein (FAP).¹⁵⁸ PDACs express high levels of FAP compared to normal pancreas.^159,160 Given the dependence of high-dose radiation on CD8 T cells, combination radiation with immunotherapy has been attempted to enhance PDAC tumor clearance, but with little success, in part attributed to impaired ability of immune cells to penetrate the fibrotic stroma and interact with tumor cells.^{118,142,143,149,161} As mentioned above, fibroblast derived TGFβ sequesters immune cells outside of tumors leading to resistance to immune checkpoint blockade.^142,143 Mouse models targeting CAFs resulted in improved drug penetrance and CD8 T cell infiltration.¹⁶² However, tumor infiltrating T cells have impaired efficacy due in part to upregulation of immune checkpoint ligand expression on CAFs and other stromal cells.^163,164 CAFs polarize the tumor immune cells to an immunosuppressive phenotype characterized by M2 macrophages expressing Arginase and regulatory T cells, achieved in part via expression of IL-6, IL-10, CXCL12, and TGFβ.^158,165-167 Depletion of CAFs by targeting FAP, resulted in improved antitumor immunity characterized by higher levels of interferon-gamma and tumor necrosis factor-alpha.¹⁶⁸ Additionally, in a melanoma tumor model, administration of a vaccine targeting tumor antigen and FAP resulted in tumor clearance as a result of antigen spreading.¹⁶⁹ However, combination of radiation, aPD1, and targeting of FAP using a highly selective orally bioavailable FAP small molecule inhibitor, or a vaccine strategy to deplete FAP expressing cells, failed to improve survival in murine models of PDAC, despite increasing CD8 T cell infiltration and polyfunctionality.¹⁶⁰ These data suggest additional suppressive pathways or key mediators may be contributing to resistance to radiation and immunotherapy in fibrotic tumors.

Targeting upstream mediators of fibrosis has shown efficacy in combination with immuno- and chemotherapy. Stromal stiffness contributes to tumor progression though mechanical forces signaling through integrins and adhesion kinases resulting in enhanced invasion and growth via PI3K activity.¹⁷⁰ One such adhesion kinase, focal adhesion kinase, is elevated in PDAC regulating the fibrotic and immunosuppressive microenvironment. Combination focal adhesion kinase inhibition with gemcitabine and αPD1 led to significantly improved survival in murine PDAC.¹⁷¹ In addition, immunosuppressive subtypes of B cells have been shown to be drivers of fibrosis, and targeting B cells by genetic ablation of IgA+ cells or using BTK inhibitors can attenuate carcinogenesis and induce CD8 T cell-mediated tumor regression,¹⁷² as well as improve the efficacy of chemotherapy.¹⁷³ Interestingly, reprograming CAFs to a less immunosuppressive phenotype shows promise for improving T cell function and response to checkpoint blockade using tumor-microenvironment activated angiotensin receptor blockers.¹⁷⁴

Radiation fibrosis is a well-characterized deleterious treatment side effect. Radiation mediated fibrosis occurs via a feed-forward loop resulting in enhanced TGFβ production. Radiation increases lactate dehydrogenase-A activity increasing lactate levels, acidifying the extracellular compartment and activating latent TGFβ.¹⁷⁵ This leads to myofibroblast differentiation and excess deposition of extracellular matrix proteins.¹⁷⁵ Therefore, CAF differentiation and fibrosis can be driven by radiation, but also contribute to radiation resistance and immunosuppression. Further investigation into preventing fibrosis, either at baseline or following radiation, may result in improved responses to immunotherapy.

Conclusions

Given the expanding scope of immunotherapy strategies in the treatment of cancer, improved characterization of the effects of RT on the TME is needed. The inflammatory and immune modulatory effects of RT have been used to rationalize many clinical trials combining RT and immunotherapy, but, in many cases, without robust empirical data to support these approaches. The equivocal outcomes, to date, of clinical trials combining RT and immunotherapy suggest that more learning is required to optimally combine these modalities. The data reviewed herein clearly demonstrate that radiation can induce profound changes in the TME but the exact nature of these changes requires further investigation.

Radiation can induce inflammatory changes and increase the number and functionality of T cells, NK cells, and antigen-presenting cells through numerous mechanisms including interferon and toll-like receptor signaling in immune cells and inducing immunogenic cell death or upregulation of cellular stress markers and MHC on tumor cells. However, radiation can also induce suppressive immune changes in the TME such as induction of TGFβ, IDO, and PD-L1 with resultant increase of suppressive cells within the TME such as Tregs and TAMs. Overall, it is likely that acutely radiation induces inflammation which can lead to antitumor effects but that over time, in response to the sustained/chronic inflammation in the TME, counterregulatory immune suppressive mechanisms are activated in a process of rebound immune suppression. The time course and extent of each phase of this process are likely to vary based on dose, fractionation, tumor type, and site irradiated. With increased understanding of the effects of RT on the TME, in the future it is likely that we will be able to personalize RT by varying the dose, site, and timing of intervention to generate the desired response to partner with immunotherapy strategies.