In situ vaccination by radiotherapy to improve responses to anti-CTLA-4 treatment

Abstract

Targeting immune checkpoint receptors has emerged as an effective strategy to induce immune-mediated cancer regression in the subset of patients who have significant pre-existing anti-tumor immunity. For the remainder, effective anti tumor responses may require vaccination. Radiotherapy, traditionally used to achieve local tumor control, has acquired a new role, that of a partner for immunotherapy. Ionizing radiation has pro-inflammatory effects that facilitate tumor rejection. Radiation alters the tumor to enhance the concentration of effector T cells via induction of chemokines, cytokines and adhesion molecules. In parallel, radiation can induce an immunogenic death of cancer cells, promoting cross-presentation of tumor-derived antigens by dendritic cells to T cells. Newly generated anti-tumor immune responses have been demonstrated post-radiation in both murine models and occasional patients, supporting the hypothesis that the irradiated tumor can become an in situ vaccine. It is in this role, that radiation can be applied to induce anti-tumor T cells in lymphocyte-poor tumors, and possibly benefit patients who would otherwise fail to respond to immune checkpoint inhibitors. This review summarizes preclinical and clinical data demonstrating that radiation acts in concert with antibodies targeting the immune checkpoint cytotoxic T-lymphocyte antigen-4 (CTLA-4), to induce therapeutically effective anti-tumor T cell responses in tumors otherwise non responsive to anti-CTLA-4 therapy.

Introduction

From the inception of carcinogenesis, the immune system detects and eliminates nascent tumors in a process described as cancer immunosurveillance. Stress-induced ligands and altered antigenicity render transformed cells susceptible to natural killers (NK) cells, γδ and conventional α/β T cells. Tissue disruption and unscheduled cell death that occur during tumor progression to invasion generate dangers signals in the form of damage-associated molecular pattern (DAMP) molecules that alert the immune system of a potential threat, activating both innate and adaptive immunity [1]. However, occasionally elimination of cancer cells is incomplete and cancer cells that have acquired the ability to evade immune control emerge, as a result of the selective pressure of the immune system. Thus, cancers rise to clinical detection after a long and complex crosstalk with the immune system, while a dominant immune suppressive tumor micro-environment has also emerged. The latter is enriched in cells with regulatory and immunosuppressive function that secrete cytokines such as transforming growth factor-β (TGFβ) and IL-10, which counteract immune-mediated rejection [2]. Noticeably, in some patients robust anti-tumor T cell responses are detectable at clinical diagnosis and their presence in the tumor specimen has been associated with a better prognosis [3, 4]. Patients who retain such anti-tumor immunity appear to benefit the most from immunotherapy, even at advanced stages of the disease [5]. For example, responses to immune checkpoint inhibitors rely on the patient's pre-existing anti-tumor T cells [6, 7]. Unfortunately, only a small fraction of cancer patients retains sufficient anti-tumor immune responses. Among solid tumors patients, melanoma carriers are most likely to respond to immune checkpoint inhibitors targeting CTLA-4 or programmed cell death-1 (PD-1) [8, 9], possibly because of their high mutational load [10].

Because responses to anti-CTLA-4 often are durable [11, 12], identifying combination treatments that can convert patients unresponsive to CTLA-4 inhibition into responders is an active area of investigation. Potential candidates include other immunotherapies, standard chemotherapy, targeted agents [13-15], and radiotherapy has earned a prominent place, due to substantial pre-clinical data [16-20] and rapidly accumulating clinical observations [21-23] that it can induce therapeutically effective anti-tumor immunity when combined with CTLA-4 blockade. Several clinical trials are currently ongoing to test radiotherapy in combination with the FDA-approved anti-CTLA-4 antibody ipilimumab (Yervoy®, Bristol Meyers-Squibb, New York, New York) (Table 1).

Table 1.

Ongoing clinical trials testing the combination of CTLA-4 blockade and radiation therapy (RT).

Identifier	Condition	Intervention	Phase
NCT01689974	Metastatic melanoma	Ipilimumab Radiation Therapy and Ipilimumab	Phase 2
NCT02254772	B-cell Lymphomas, multiple types	Ipilimumab TLR9 agonist SD-101 Radiation therapy	Phase 1 Phase 2
NCT01703507	Recurrent Melanoma Stage IV Melanoma Tumors Metastatic to Brain	Ipilimumab Whole-Brain Radiation Therapy (WBRT) Stereotactic Radiosurgery (SRS)	Phase 1
NCT02239900	Liver Cancer Lung Cancer	Ipilimumab Stereotactic Body Radiation Therapy (SBRT)	Phase 1 Phase 2
NCT01970527	Recurrent Melanoma Stage IV Melanoma	SBRT Ipilimumab	Phase 2
NCT01557114	Malignant Melanoma	Ipilimumab Radiotherapy	Phase 1
NCT02221739	Non-small Cell Lung Cancer	Ipilimumab Intensity-modulated radiation therapy (IMRT)	Phase 2
NCT01449279	Melanoma	Ipilimumab Radiation Therapy	PILOT
NCT01565837	Melanoma	Ipilimumab Stereotactic Ablative Radiosurgery (SART)	Phase 2
NCT02107755	Liver Metastases Lung Metastases Recurrent Melanoma Stage IV Melanoma Tumors Metastatic to Brain	Iipilimumab Stereotactic radiosurgery	Phase 2
NCT01996202	High Risk Melanoma	Ipilumimab Radiation	Phase 1
NCT02097732	Metastatic Melanoma Brain Metastases	Ipilimumab Stereotactic Radiosurgery (SRS)	Phase 2
NCT01950195	Newly Diagnosed Melanoma Metastases in the Brain and Spine	SRS Ipilimumab	Phase 1
NCT01711515	Cervical Carcinoma	Cisplatin External beam radiation therapy Brachytherapy Ipilimumab	Phase 1
NCT01935921	Stage III and IV Head and Neck Cancer	Cetuximab IMRT Ipilimumab	Phase 1
NCT02115139	Melanoma Brain Metastases	Ipilimumab WBRT	Phase 2
NCT01860430	Head and Neck cancer	Cetuximab/IMRT Plus Ipilimumab	Phase 1
NCT01730157	Ocular Melanoma Extraocular Extension Melanoma Metastatic ocular melanoma	Ipilimumab yttrium Y 90 glass microspheres	Phase 0
NCT02406183	Metastatic Melanoma	Ipilimumab SBRT	Phase 1

Here we review the available data that has informed the rationale for exploiting the synergy of radiation and CTLA-4 blockade.

1. Radiation-induced in situ tumor vaccination

Over the past decade, an improved understanding of the effects of local radiation on tumor-host interactions has led to the recognition that radiotherapy may have a novel role as an inducer of acute inflammation and immunogenic cell death, capable to convert a tumor into an in situ vaccine [24-26]. Pioneering work implicating T cells in determining the response to radiation was published several decades ago [27]. More recently, the demonstration that T cells mediate the abscopal effect (out-of-field responses) of radiation in a pre-clinical tumor model [28] has provided a putative mechanism for the intriguing clinical observation that rare patients with disseminated cancer experienced systemic tumor regression after irradiation of a single tumor site [29-32].

1.1. Radiation induces an immunogenic death of cancer cells and priming of tumor-specific T cells

Multiple mechanisms that contribute to radiation-induced anti-tumor immunity are emerging and the signals generated by irradiated dying tumor cells are being elucidated.

Priming of anti-tumor immune responses by cytotoxic treatments has been shown to require the presence of an immunogenic cell death (ICD) [33]. ICD relies on the orchestration of specific molecular signals that stimulate cross-presentation of tumor cell antigens by dendritic cells (DCs) to T cells [34]. Mediated by endoplasmic reticulum (ER) stress and autophagy, ICD is characterized by cell surface translocation of calreticulin (CRT), and extracellular release of high-mobility group protein B1 (HMGB1) and ATP [35-37]. CRT is an ER-associated chaperone protein that when expressed on the surface of dying cells promotes their phagocytosis acting as an “eat-me” signal for DCs [38]. HMGB1 is a nuclear protein found in almost all mammalian cells and is secreted by a variety of immune cells to induce pro-inflammatory effects when bound to pattern-recognition receptors (PRRs) such as the toll-like receptor-4 (TLR4) [39]. During cell death extracellular release of HMGB1 delivers both chemotactic and maturation signals to DCs, promoting phagocytosis of dying tumor cell and DC migration to lymph nodes (LN) to cross-present antigens and prime T cells [40, 41]. ATP released from dying cells binds to P2X7 purigenic receptor on DCs and activates the inflammasome, leading to secretion of interleukin (IL)-1β [37]. While in vivo the relative contribution of each of these factors to priming of anti-tumor T cells remains undefined, radiation has been shown in vitro to generate all three ICD molecular signals [42-44].

Recently, additional mechanisms that play a critical role in radiation-induced anti-tumor T cell priming have been described. Gupta et al., showed that a single 10 Gy radiation dose induced the activation of intratumoral DCs, measured as up-regulation of co-stimulatory molecules CD86 and CD70 [45]. Activation of intratumoral DCs was shown by Burnette et al. to first require their production of type I interferon (IFN-I) and to be critical for radiation-induced T cell cross-priming [46]. Moreover, Deng et al. demonstrated that tumor-derived DNA induced IFN-I production by DCs via the stimulator of IFN genes (STING) pathway [47].

It is important to notice that the experimental settings demonstrating priming of T cells by radiation consist of relatively immunogenic mouse tumors and/or used strong model antigens such as ovalbumin (OVA) as reporter antigens [48, 49]. In the setting of poorly immunogenic murine tumors, that better model the clinical reality of cancer, radiation by itself is generally insufficient to prime T cell responses [50, 51].

One explanation for the suboptimal vaccination induced by radiation it the possible concomitant activation of immunosuppressive signals. Our recent data support this hypothesis. In two poorly immunogenic murine carcinomas blockade of TGFβ, which is activated from its latent form by reactive oxygen species (ROS) generated by radiation [52], was required for priming of CD8 T cells specific for three endogenous antigens: The anti-apoptotic protein survivin, the transcription factor twist-1 and, the gp70 envelope glycoprotein of an endogenous retrovirus [53].

Another mechanism of radiation-induced immunosuppression is mediated by the conversion of ATP by CD39 and CD73 into adenosine, which suppresses anti-tumor T cell activation, survival and effector function via the A2a adenosine receptor (A2AR). Adenosine also negatively modulates differentiation and function of DCs and natural killer (NK) cells [54, 55]. Thus, at least in some tumors adenosinergic pathways may limit the ability of radiation to induce effective anti-tumor immunity.

Other mechanisms that have been suggested to hamper radiation-induced immunization are a relative increase in regulatory T cells (Tregs) and in immunosuppressive myeloid cells post-radiation [56, 57].

Information about the breath and specificity of the anti-tumor T cell responses primed by radiation is very limited. The antigenic repertoire of a given tumor classically includes self-antigens, like overexpressed differentiation antigens and Cancer-Testis (CT) antigens [58]. Although T cells can recognize them, some degree of T cell tolerance usually exists as the result of elimination of T cells with high affinity T cell receptor (TCR) during T cell ontogeny. In experimental models, radiation fails to induce T cell responses to these relatively weak antigens [53]. However, increase in T cell responses to self-antigens such as survivin could be detected in occasional patients undergoing radiotherapy [51]. Interestingly, it has been shown that radiation induces the expression of novel epitopes derived from protein that are transcribed in response to radiation-induced damage of cancer cells [59]. Nevertheless it is unclear whether T cells specific for radiation-induced antigens are elicited that contribute to the rejection of the irradiated tumor. Finally, a class of tumor antigens that can elicit strong T cell responses are mutated neo-antigens [60]. Recent evidence suggests that such mutated neo-antigens are the targets of T cells during successful tumor rejection and our group is actively investigating whether radiation can induce such responses.

1.2. Radiation promotes a pro-immunogenic tumor microenvironment

In addition to the signals released by dying cells after exposure to ionizing radiation, signals released by tumor and stromal cells that survive radiation damage modulate the tumor microenvironment [61, 62]. Effects of radiation that facilitate the effector phase of anti-tumor immune responses include upregulation of specific chemokines and cell surface receptors, as well as vascular changes.

Cancer cells are the source of some of the chemokines induced by radiation. For example, in vitro, mouse and human breast cancer cells, and mouse prostate, colon carcinoma and fibrosarcoma cells increased levels of CXC chemokine ligand (CXCL) 16 upon radiation exposure to doses in the range of 2 to 12 Gy [19, 63]. In vivo, CXCL16 was critical for efficient recruitment of effector CD8 T cells to irradiated mouse 4T1 breast tumors [19]. In the B16-OVA mouse melanoma model, after radiation the cancer cells produced CXCL9 and CXCL10 in response to IFN-γ produced by infiltrating hematopoietic cells [64].

Radiation has also been shown to promote normalization of aberrant vasculature and activation of endothelial cells with upregulation of vascular adhesion molecules, leading to improved tumor infiltration by T cells [48, 65]. Interestingly, in a mouse model of pancreatic cancer vascular normalization was achieved with a single 2 Gy dose of radiation, improving tumor rejection by adoptively transferred T cells [65].

Finally, a number of phenotypic changes defined by some investigators as “immunogenic modulation” have been described to occur in neoplastic cells surviving radiation exposure [62]. For instance, post-radiation upregulation of Major Histocompatibility Complex (MHC) class I molecules and Fas death receptor on tumor cells promoted tumor rejection by CD8+ T cells in vivo [59, 66, 67]. Moreover, RT induces expression of natural-killer group 2, member D (NKG2D)-ligands, powerful stimulators of both NK and CD8+ T cells [68].

1.3. Role of radiation dose and fractionation

Many of the pro-immunogenic effects of radiation on the cancer cells described above can be induced, at least in vitro, by radiation doses varying from about 2 Gy to as much as 30 Gy or more, but the optimal radiation regimen to induce clinical anti-tumor immunity remains to be defined. Only a few studies have compared different radiation doses and fractionation, or conducted the same experiments in tumor cells with different intrinsic radiosensitivity, [69]. In vitro, a dose-dependent increase of the ICD signals and of some chemokines and surface molecules were reported [19, 43, 59]. In vivo, the nature of the pre-existing tumor microenvironment at the time of radiation and the response of normal stromal cells present within the field of radiation are important determinants of the development of effective anti-tumor immunity. For instance, Klug et al overcame the immunosuppressive effect of hypoxia by a low dose single dose radiation (2 -5 Gy) that resulted in re-programming of tumor-infiltrating macrophages [65]. On the other hand, priming of OVA-specific T cells was shown to be more pronounced when B16-OVA tumors were irradiated with 15 Gy given as single dose rather than as 5 fractions of 3 Gy each [48]. In another study using the same tumor model, Schaue et al. reported that while a single dose of 15 Gy primed OVA-specific T cells, it also led to a relative increase in Tregs, and showed that the best ratio of anti-tumor T cells to Tregs was achieved when radiation was given in two fractions of 7.5 Gy [70]. These examples highlight the complex interaction between the irradiated tumor and the host immune system, and suggest that studies to optimize the radiation regimen need to take into consideration multiple parameters, likely to also be tumor-type specific. Moreover, combinations of radiation with different immunotherapy strategies may require specific dose regimens and fractionation.

Finally, while radiation can convert the tumor into an in situ vaccine, its effects on the tumor microenvironment persist and evolve long after the time of radiation exposure. The degree and duration of these effects may depend on the degree of pre-existing immunosuppression and the balance between signals that promote tumor rejection versus signals that hinder it. In most cases, radiation therapy alone is unable to induce effective immune-mediated tumor rejection: however it may potentiate immunotherapy, as exemplified by preclinical and clinical work with immune checkpoint blocking agents. We will discuss below the combination of radiation with anti-CTLA-4 antibodies.

2. Cytotoxic T lymphocyte antigen-4, a negative regulator of T-cell activation

An array of co-stimulatory and co-inhibitory molecules regulates T cells activation, balancing the need to eliminate pathogens with the prevention of autoimmunity [71]. T cell activation requires two signals, the first is delivered by TCR binding to MHC-I/antigen. The second is delivered by CD28 costimulatory receptor that binds to CD80 (B7-1) and CD86 (B7-2) on the surface of antigen presenting cells (APC), resulting in abundant secretion of IL-2 and T-cell proliferation [72].

After TCR engagement CTLA-4 is rapidly recruited to the immune synapse where it competes with CD28 for binding to CD80 and CD86. Because of its greater affinity, when co-stimulation is suboptimal interactions between CTLA-4 and costimulatory molecules prevail, thus impeding T cell proliferation [73]. Importantly, CTLA-4 not only hinders the interaction between CD28 and CD80/86 but also impairs T cell activation by dephosphorylating key effector molecules required for TCR signaling [74, 75]. Additionally, when constitutively expressed on Tregs, CTLA-4 has been shown to reduce expression of CD80 and CD86 on APCs as well as to increase the secretion of TGFβ, thus fostering T cell tolerance to tumors by multiple mechanisms [76, 77].

2.1. Effects of anti-CTLA-4 antibodies on priming of anti-tumor T cells

The role of CTLA-4 in limiting the development of anti-tumor immune responses in cancer patients has been demonstrated by the therapeutic success of anti-CTLA-4 monoclonal antibody (mAb) in metastatic melanoma [9]. The mechanisms by which anti-CTLA-4 mAbs unleash anti-tumor immunity remain incompletely understood. Preclinical studies have shown that CTLA-4 blockade enhances anti-tumor immunity by reducing the threshold for T cell activation [78], resulting in monoclonal and oligoclonal expansion of CD4 T cells responding to cognate peptide antigens or super-antigens [79]. Importantly, work by Jim Allison's lab demonstrated that CTLA-4 plays a major role in shaping the breadth of reactivity of a primed T cell population, by constraining the “best-fit” population [80]. Thus, when CTLA-4 is blocked, the lower threshold required for T cell activation allows proliferation and expansion of tumor antigen-specific T cells, leading to tumor rejection. Importantly, Kvistborg et al. have recently provided evidence for a similar broadening of melanoma-reactive T cell responses in patients treated with anti-CTLA-4 [81].

2.2. Effects of anti-CTLA-4 antibodies on the effector phase of tumor rejection

The ability of anti-CTLA-4 mAb to promote anti-tumor immunity is not only confined to the priming phase. A common feature associated with anti-CTLA-4 mediated tumor rejection is an increase in the ratio of effector T to Tregs cells within the tumor [82, 83]. While effector T cells are increased due to the pro-proliferative effect of CTLA-4 blockade, recent preclinical data demonstrated that depletion of intratumoral Tregs also plays a major role in tumor rejection. Anti-CTLA-4 mAbs capable of mediating antibody-dependent cellular cytotoxicity (ADCC) effectively and selectively eliminated intratumoral Tregs, which express higher levels of CTLA-4 compared to circulating Tregs [84-86]. CTLA-4 ligation on effector T cells has also been shown to enhance motility of both CD4 and CD8 T cells [20, 87, 88]. Using CD8 T cells expressing a Pmel-1 transgenic TCR specific for a melanoma antigen Pentcheva-Hoang et al. showed that increased motility was associated with tumor rejection, and hypothesized that anti-CTLA-4 may act by reversing motility paralysis of exhausted intratumoral T cells [87].

3. Synergy of radiotherapy with anti-CTLA-4 antibody

Previous work in pre-clinical models of melanoma and breast cancer showed that tumors insensitive to anti-CTLA-4 treatment as monotherapy became responsive upon vaccination with modified autologous tumor cells [89, 90]. We hypothesized that in situ vaccination by radiation could also convert a poorly immunogenic tumor, unresponsive to anti-CTLA-4 into a responder. This hypothesis was confirmed in three different murine tumor models, 4T1 and TSA mammary carcinomas syngeneic to BALB/c mice and MCA38 colorectal carcinoma syngeneic to C57BL/6 mice [16, 17]. Importantly, the anti-tumor T cells elicited by the combination not only did reject the irradiated tumor but also inhibited spontaneous 4T1 lung metastases or synchronous unirradiated tumors (abscopal effect) in mice bearing TSA and MCA38 carcinomas [16, 17].

The mechanisms of synergy between radiation and anti-CTLA-4 mAb were studied in more details in the 4T1 model. These studies showed that anti-CTLA-4 therapy and radiation interact at multiple levels, each contributing to tumor rejection (Figure 1). First, development of CD8 T cell responses to the tumor was required for tumor rejection [16, 18]. Priming of CD8 T cells was monitored in tumor-draining lymph nodes by measuring IFNγ production to the AH1 gp70-derived CD8 T cell epitope. Radiation and anti-CTLA-4 antibody induced measurable T cell priming only when used in combination, indicating that each treatment provided critical non redundant signals. The number of DCs available locally to cross-present the antigens released by radiation determined the magnitude of the elicited anti-tumor T cell responses [91]. We found that in 4T1 tumor-bearing mice the number of DCs present within the tumor and draining lymph nodes was regulated by invariant natural killer T (iNKT) cells. Blocking CD1d-mediated interaction of iNKT cells with DCs led to an increase in DCs and improved CD8 T cell priming and overall tumor response to the combination of radiation therapy and anti-CTLA-4 treatment [91].

Figure 1. Mechanisms of synergy between radiotherapy and anti-CTLA-4 treatment.

Schematic illustration highlighting the critical changes induced by each treatment in the tumor-draining lymph nodes (dLNs) and tumor. Untreated tumors: Priming of T cells in dLN is limited by negative signals delivered by CTLA-4. In the tumor, tumor rejection by CD8 T cells is hampered by low MHC class I and immune-stimulatory ligands (e.g., RAE-1) on tumor cells, and by the immunosuppressive tumor microenvironment, rich in TGFβ, Tregs and tolerogenic DCs. Anti-CTLA-4 treatment: Activation and expansion of tumor-specific T cells in dLNs is improved by blocking CTLA-4-mediated negative signal, but it remains suboptimal in the setting of poorly immunogenic tumors due to low antigen availability and low activation of DC. Clearance of intratumoral CTLA-4^hi Tregs is mediated by anti-CTLA-4 mAb via ADCC if FcγR-expressing myeloid cells are present. At the same time, the pro-motility effects of CTLA-4 ligation hinder the formation of a stable immune synapse between activated CD8 T cells and tumor cells. Radiation: induction of ICD provides tumor antigens and activation signals to DCs, but cross-presentation of tumor-derived antigens to CD8 T cells in dLNs is limited by inhibitory signaling via CTLA-4. Radiation combined with anti-CTLA-4: priming of tumor-specific CD8 T cells is markedly enhanced by complementary effects of radiation and anti-CTLA-4. In addition, activated T cell homing to the tumor is facilitated by increased levels of chemokines (CXCL16 and CXCL10) released by tumor cells in response to radiation. Inside the tumor, co-engagement of TCR and NKG2D on CD8 T cells by radiation-induced MHC-I and RAE-1 allow formation of stable immune synapses with tumor cells. Overall, radiation and anti-CTLA-4 therapy have complementary effects that underlie their synergistic interaction in inducing tumor rejection.

In addition, radiation-induced molecular signals were required for tumor rejection. For instance, CXCL16 upregulation by radiation was required for effector CD8 T cell homing to 4T1 tumors. CXC chemokine receptor (CXCR) 6-deficient mice, whose T-cells are unable to respond to this chemokine, failed to show increased CD8 T cells infiltration post-treatment and response to radiation and anti-CTLA-4 [19].

Finally, intravital imaging using two photon laser scanning microscopy was used to study in vivo how radiation affected the interaction of CD8 T cells with 4T1 cancer cells. To track the polyclonal population of CD8 T cells activated by treatment with radiation and anti-CTLA-4 we used CXCR6^+/gfp mice. In these mice, GFP expression within the tumor was largely confined to activated tumor-specific CD8 T cells, as determined by their expression of activation markers, ex vivo IFNγ production and by the fact that blocking MHC-I disrupted the interaction of GFP⁺ T cells with CFP⁺ 4T1 cells [20]. Some activated CD8 T cells were present in all tumors, although over time they only increased in tumors of mice treated with radiation and anti-CTLA-4. Motility of these T cells was mildly enhanced in irradiated tumors and markedly enhanced in tumor of mice treated with anti-CTLA-4 monotherapy. In contrast, when these two treatments were combined, activated CD8 T cells displayed reduced motility, forming stable contacts with 4T1 cells. Interactions between NKG2D receptor and its ligand retinoic acid early inducible-1 (RAE-1), which was induced by radiation on 4T1 cells, were required to achieve the stable contacts between tumor and CD8 T cells. Blocking NKG2D with an antibody increased CD8 T cell speed and abrogated the tumor inhibition achieved with radiation and anti-CTLA-4 [20]. Thus, in the presence of anti-CTLA-4 antibody radiation is providing a signal to promote the formation of an effector immune synapse between CD8 T and tumor cells. The requirement for co-engagement of TCR and NKG2D to achieve a stop signal when CTLA-4 is ligated was supported by additional in vitro data [20]. Intriguingly, while a similar motility enhancement by anti-CTLA-4 treatment was recently reported in intratumoral CD8 T cells expressing a transgenic TCR specific for a melanoma antigen, in the latter experimental system enhanced motility resulted in more effective tumor rejection [87]. This difference may reflect the different requirements for formation of an effector immune synapse for a homogeneous T cell population expressing a TCR with relatively high affinity for an abundant tumor antigen. In our experimental system we studied a polyclonal T cell population with TCRs of varying affinity for antigens that may be expressed at very low levels within the tumor. Low affinity/avidity TCR interactions with tumor cells will be dependent on NKG2D co-engagement and killing may require a more protracted interaction [92].

Overall, this data supports the multiplicity of effects of radiotherapy on the efficacy of immunotherapy targeting the checkpoint receptor CTLA-4. It is likely that dominant mechanism(s) of interaction between these two modalities mainly depend on the pre-existing tumor microenvironment and host general immune status.

However the type of radiation regimen used may also be a determinant of success. In a comparison of two fractionated radiation regimens (8 Gy X 3 and 6 Gy X 5) and a single large dose (20 Gy) in two tumor models, TSA and MCA38, we found that effective anti-tumor immunity leading to rejection of the irradiated and non-irradiated synchronous tumors (abscopal effect) was only achieved by the fractionated regimens, while the single 20 Gy dose failed to show synergy with anti-CTLA-4 [17]. While the reasons for this difference are being actively investigated, it is intriguing that the two most notable cases of abscopal responses seen in patients with melanoma and lung cancer treated with radiotherapy and ipilimumab occurred with the use of similar fractionated radiotherapy regimens, 9.5 Gy X 3 in the melanoma case, and as 6 Gy X 5 in the lung cancer case [21, 23].

4. Clinical translation

Since 2011, after the approval of ipilimumab for patients with metastatic or unresectable melanoma, a few dramatic abscopal responses have been reported after radiation of one metastasis in patients who were unresponsive or had ceased to respond to ipilimumab [21, 22, 93]. These reports have sparked several retrospective analyses of outcome in melanoma patients receiving radiation while treated with ipilimumab, with an excellent review of these studies recently published by Barker and Postow [94]. Most of the patients in these retrospective series received radiation to the brain, and in most cases the analysis suggested a survival benefit associated with the combination. A retrospective study in a cohort of 21 patients who received radiation to the brain or extracranial sites after progression on ipilimumab reported abscopal responses in 62%, which were associated with increased survival [95]. While these cases seem to confirm the preclinical synergy of radiation with ipilimumab, a caveat is the fact that since melanoma is known to respond to ipilimumab the findings could represent late responses attributable to ipilimumab alone. Several prospective trials are ongoing to determine the benefits of the combination of radiation with anti-CTLA-4 antibodies in melanoma (Table 1).

Results of a phase I study in 22 melanoma patients with escalation of the radiation dose were recently reported and confirmed that radiation did not worsen the toxicity expected with ipilimumab alone. There were no complete responses and overall disease control rate was not much higher than what would be expected with ipilimumab alone in melanoma [96]. While this study does not provide conclusive evidence about the benefits of radiation combined with ipilimumab it raises the question of whether two doses of either 6 or 8 Gy, which were received in 12 out of 22 patients, may be suboptimal in melanoma to achieve in situ vaccination.

A large randomized trial in metastatic castrate-resistant and docetaxel-refractory prostate cancer compared radiation given as a single 8 Gy dose to a bone metastasis with radiation plus ipilimumab. Overall, there was no significant difference in survival between the two arms, but a benefit in the combination arm was seen among patients with good prognostic features [97]. This data suggests that the degree of immune competence may influence the likelihood of response, and again raises the question whether a single 8 Gy dose, may explain the limited success of this study [17]. We have reported a complete abscopal response in a patient with non small cell lung cancer treated with radiation given in 5 fractions of 6 Gy each and ipilimumab [23]. Since lung cancer has been shown to be insensitive to anti-CTLA-4 monotherapy [98] this case supports the hypothesis that radiation could be used to sensitize unresponsive tumor types to anti-CTLA-4 treatment. This hypothesis is currently being tested in a clinical trial (NCT02221739).

5. Conclusions

The ability of radiation to elicit anti-tumor immune responses has been unequivocally demonstrated in experimental models, and many of the mechanisms involved have been identified. However, more work is required to define the dose(s) and fractionation that optimally induce anti-tumor T cells, and identify the tumor characteristics that predict which tumors will respond to a given combination of radiation and immune checkpoint blockade. While the growing number of reports of occasional abscopal responses in patients receiving radiation therapy and anti-CTLA-4 antibody has generated a lot of interest, the results of the first two completed prospective trials testing radiation and anti-CTLA-4 highlight the need to carefully design future studies. Recent data have emphasized the importance of T cell responses to unique individual tumor antigens [60]. In this context, local radiation deserves to be thoroughly explored for its potential to offer an attractive, easy to use and cost-effective intervention for personalized tumor vaccination.