There is a positive correlation between tumor responses to therapy and the extent of radiation-induced tumor cell death [33,34]. RT is able to induce apoptosis, primary and secondary necrosis, all leading to alterations in immunomodulatory pathways [35,36,37,38]. Cell death can lead to immunological priming in a process called immunogenic cell death (ICD), which is defined as a form of regulated cell death sufficient to activate an adaptive immune response [39]. Activation occurs through release or expression of damage-associated molecular patterns (DAMPs). DAMPS are usually not presented to the immune system, but can be found upon cytoprotective stress responses or cell death and are able to bind to pattern recognition receptors expressed by immune cells [40,41]. A multitude of DAMPs are encountered upon RT-related ICD, e.g., calreticulin translocation to the membrane, heat shock protein (HSP)-70, HSP-90 and high mobility group box (HMGB)-1 secretion and ATP passive release [42,43,44,45]. Upon single-dose RT delivery (2, 5, 10 or 20 Gy) to tumor cells, a dose-dependent increase in calreticulin and ATP could be appreciated after 24 h and for HMGB1 72 h after irradiation [46]. RT fractionation with 2 Gy during five consecutive days in glioblastoma (GBM) models led to enhanced secretion of HSP-70 and model-dependent HMGB1 secretion [47]. In patients experiencing esophageal squamous cell carcinoma (SSC) and rectal cancer, elevated HMGB1 plasma levels after chemoradiotherapy (administration weekly or on day 1 of RT, respectively) were associated with a RT enhancement effect [48,49]. A single-dose RT (18 Gy) to cancer-associated fibroblasts did not induce ICD molecules, which may indicate that tumor cells are the main source for obtaining ICD immunogenicity [50]. Hence, RT is capable of evoking an immune stimulation through cancer cell kill, although to what extent it contributes to the success of IO has not been determined.

RT engages both the innate and adaptive arms of the immune system and functions as an in situ vaccine that attracts T cells and natural killer (NK) cells through DC induction. Radiation promotes the release of existing tumor antigens displayed on a tumor cell surface [51] and can even further elevate antigen expression to levels sufficient for cross-presentation, thus increasing the number of DCs presenting antigens [52]. It has been demonstrated that beside antigens, radiation can upregulate the neoantigen pool, which is attributed to the formation of novel mutations owing to error-prone radiation-induced DNA damage repair [44]. Neoantigens are highly specific per tumor type and even for each individual tumor, however low levels of neoantigen (uptake) fail in evoking a significant immune response. It was demonstrated that RT induced the expression of proteins on tumor cells that were normally not presented by major histocompatibility complex (MHC) molecules or were indicated as “silent”, including MHC-I expression [53,54], prior to radiation [55]. Similar to existing antigens, radiation-induced cell death can drive the release of existing (neo)antigens towards a detectable level for cross-presentation by DCs. It has been recently demonstrated that radiation of tumor cells (8 Gy) also enriched extracellular vesicles (EVs) with a variety of antigens and DAMPs, serving as a carrier of antigens to prime cytotoxic lymphocytes and causing significant anti-tumor response, both local and systemic, as compared to non-irradiated EVs [56]. Altogether, RT can both increase the reservoir of (neo)antigens and enhance their presentation; however, this might depend on the dose of radiation [24,57].

What advocates for using single dose RT is that prolonged dose administration induces matrix deposition and long-term fibrotic reactions [69], apart from the evidence of reduced lymphopenia with decreasing number of fractions, for example, in breast cancer patients [70]. However, different pro- and anti-immunogenic effects can be induced depending on the level of the dose administered. Interestingly, a moderate radiation dose between 1 and 10 Gy was able to reprogram macrophage type 2 (Mɸ-2) towards a Mɸ-1 like antitumor phenotype and only a high dose of >10 Gy led to Mɸ-2 polarization [71]. High dose RT (25 Gy) drove the accumulation of CD11b+Gr-1+ neutrophils in the center of the necrotic area and CD11b-F4/80+ tumor associated macrophages at the junctions between necrotic areas and surrounding hypoxic regions [72], although the function of these cells is not fully understood yet. Furthermore, high dose RT (15–20 Gy) may permanently reduce blood flow leading to the additional formation of hypoxic immunosuppressive microenvironment, limiting infiltration of immune cells [73]. A dose of 8–16 Gy led to upregulation of acid sphingomyelinase (ASMase), which induced endothelial cell apoptosis and triggered immune responses through increased production of cytokines/chemokines that induce immune cell recruitment [74]. Moreover, a single dose of 15 Gy was more effective than consecutive 5 × 3 Gy at priming antitumor T cells to antigen ovalbumin, but priming was seen with both regimens [75]. RT dose of 6 Gy showed equal to better abscopal responses than 10 Gy and 15 Gy with an anti-CD40 agonistic antibody [76]. Low dose RT can suppress the proliferation of T regulatory (Treg) cells, specifically at a dose of 0.94 Gy more than high dose RT with 15 Gy [77]. Thus a high dose of >15 Gy appears not to be preferred in context of IO, whereas an intermediate dose range seems to provide a more balanced stimulation of pro-inflammatory and inhibition of anti-inflammatory signals.

Fractionated radiotherapy (fRT) remains the most frequently applied (curative) RT strategy in the clinic, e.g., 5 × 2 Gy per week over a 3–7 week period [78]. The immunological responses induced by fRT differ from those induced by single dose RT. For example, induction of gene expression of the TGF and IFN family was higher upon fRT delivery versus single dose RT [79]. Sustained IFN signaling, however, can contribute to the expression of the immune checkpoint programmed death-ligand (PD-L)1 and to PD-L1-independent adaptive resistance leading to tumor relapse in an in vivo study combining RT and immune checkpoint inhibitors (ICI), anti-CTLA4 and anti-PD-L1 [80]. It has also been demonstrated that fRT with 2 × 7.5 Gy resulted in similar tumor growth inhibition as 15 Gy single dose irradiation, while showing the tendency for lower Treg numbers in spleens as compared to a single dose [81]. In breast and seminoma cancer patients, fRT with 50 Gy delivered in 25 fractions in five weeks promoted the formation of memory and cytotoxic T cells whilst decreasing Tregs [82]. The seminal work of Vanpouille-Box and colleagues showed that single dosage of 12–18 Gy enhanced DNA exonuclease Trex-1 expression in different cancer cells and attenuated their immunogenicity by degrading DNA that accumulates in the cytosol upon radiation in vitro, whereas upon fRT this effect was not seen. A hypofractionated schedule of 3× 8 Gy was identified as the optimal schedule in inducing efficient local and abscopal responses in combination with ICI [83], which was subsequently used to test other immunotherapeutic agents [84]. Others also showed that a radiation dose of at least 5 Gy is required to activate inflammatory pathways, suggesting that doses below 5 Gy might not be optimal to combine with IO [85].

Recently, certain patient populations have undergone hypofractionation RT to shorten treatment duration (reduce patient treatment burden and enhance cost effectiveness) and it resulted in a similar or better toxicity profile [86,87,88]. Stereotactic ablative radiation (SABR or SBRT) delivers a higher dose per fraction in a smaller number of fractions [89,90]. It was shown that a SABR schedule of 52 Gy in 8 consecutive daily fractions (6.5 Gy per fraction) versus intensity-modulated radiation (IMRT) hypofractionated treatment of 60 Gy in 25 fractions (2.4 Gy per fraction) induced different plasma cytokine changes in NSCLC patients, with the latter schedule having a more limiting therapeutic effect [91]. SABR administered at a high dose (>7 Gy on no more than five consecutive days), increased the local production of IFN-γ, which enhances MHC-I and antigen presentation in cancer cells. Similarly, studies involving human melanoma cells confirmed that SABR with 10–25 Gy per fraction increased MHC-I expression, rendering cancer cells more susceptible to T cell attack [92]. SABR provided excellent local control rates (98% at three years and 87% at five years) in patients with NSCLC [93], but 10–20% distant metastasis and 10–15% regional lymph node recurrence was encountered [94], highlighting the need for combinatorial treatment approaches.

The type of IO adds another level of complexity in guiding the choice of radiation dose and schedule. For example, fRT, but not single-dose RT, induced an immune-mediated abscopal effect when combined with anti-CTLA-4 antibody in mouse tumor models [95]. In contrast, fRT with 5 × 2 Gy or 5 × 5 Gy combined with the immunocytokine L19–IL2 triggered only growth delay of distant tumors, while a single dose of 15 Gy resulted in complete remission of 20% of the non-irradiated tumors despite the fact that both schedules resulted in 100% cure of the primary tumor [96,97]. Similarly, fRT (3 × 4 Gy) and single dose (7 Gy) in combination with an anti-PD-L1 antibody resulted in the same therapeutic effect on the primary murine tumor CT26 [13].

Vaccination strategies are often applied in relation to DCs, since antigen presentation is a crucial step in the initiation of an adaptive immunity. Typically, personalized DC vaccines, produced by isolating monocytes or hematopoietic stem and progenitor cells from peripheral blood of the patient, are subsequently treated with recombinant cytokines to induce differentiation, stimulated to induce maturation, and loaded with tumor-associated antigens (TAAs) of various forms [103]. In addition, DCs can be modulated to downregulate inhibitory molecules such as zinc-finger protein A20, which is a negative regulator of toll-like receptors (TLRs) and TNF-ɑ receptor pathways [104]. Several factors have a direct impact on DC biology and the quality and potency of the ensuing T cell responses, such as route of administration and frequency of injection, delivery system, type of adjuvants, nature of DC vaccine formulations and nature of cancer cell lysates/antigen cargo [105,106]. The difficulty in orchestrating the aforementioned factors is underlined by data showing that engineered DCs provide a shorter long-term memory effect than autologous cells [104]. However, success has already been achieved in a clinical trial, where a CD141+ subset of blood-derived myeloid DC (mDC) had superior capacities at cross-presenting TAAs to CD8+ T cells [107]. The disadvantage of vaccination is that, when applied alone, it will not overcome an immune-restrictive and suppressive TME. Herein, use of RT could empower vaccination strategies. This is supported, for example, by the results of a clinical study, where relapsed Non-Hodgkin’s B-cell lymphoma (NHL) patients benefited from pulsing DC vaccines with cancer cells dying after exposure to γ-ray, heat shock and ultraviolet-C ray associated with ICD [43]. In a Phase II clinical trial, patients with GBM received treatment with surgery, RT (60 Gy fractionated over 6 weeks) and chemotherapy followed by DC-based therapy 1–2 months post treatment initiation [108]. Increased short-term (1–3 years) survival rates were seen compared to control group receiving conventional therapy without DC vaccination. In a pilot Phase I study, six patients with diverse cancers received two four-week cycles of four intradermal daily doses of DC vaccine matured with polyinosinic-polycytidylic acid (poly-ICLC), TNF-ɑ and IFN-ɑ [109]. In addition to intratumoral injections of poly-ICLC on days 8 and 10 of each cycle and cyclophosphamide administration one week before the first cycle, SABR (3 × 8 Gy) was delivered to the selected tumor lesions between the first and the second DC vaccine cycle. DC vaccination enhanced serum IL-12 and IL-1β concentrations, confirming efficient immune stimulation. The combined treatment approach demonstrated signs of preliminary clinical efficacy. Another clinical study in patients with soft tissue sarcoma demonstrated that radiotherapy (28 × 1.8 Gy) combined with concomitant autologous intratumoral injection of DCs resulted in increased T cell responses as compared to the baseline levels prior to this therapy [110]. There are currently two active (NCT01818986, NCT01833208) and two completed clinical trials (NCT02232230, NCT01807065) in patients with prostate cancer assessing the use of RT as adjuvant in combination with sipuleucel-T. The latter is a cancer vaccine consisting of autologous antigen-presenting cells, activated ex vivo and loaded with PAP (prostatic acid phosphatase) antigen, which is expressed by the majority of prostate cancer cells, and granulocyte-macrophage colony-stimulating factor (GM-CSF) to promote DC maturation [111]. Similar to previous research, fRT is selected over single-dose radiation, which concurs with the clinical standard of care. Overall, the data suggest that combining RT with DC vaccines is a safe and promising strategy to improve patient outcomes, but requires further evidence of clinical efficacy derived from randomized clinical trials.

Currently the main research lines in adoptive transfer therapy lie in treatment with autologous in vitro expanded NK and T cells or engineered T cells. Autologous NK or T cells that have acquired antitumor activity are expanded in vitro before reinfusing the cells back into the patient [112]. T cells have been effectively isolated and expanded in treatment of naïve NSCLC tumors, even showing reactivity against tumor digests in 13 out of 17 patients [113]. Either naïve T cell or effector memory T cell populations can be used, although there is an indication that naïve T cells have superior antitumor immunity than effector memory T cells [114]. Alternatively, T cells can be engineered before infusing them into patients as either chimeric antigen receptor (CAR) T cells or T cell receptor (TCR) engineered T cells with retroviral TCRs. The difference is that retroviral TCRs can recognize intracellular and surface antigen, whereas CAR T cells can solely recognize surface antigens.

Therapy with solely adoptive transferred T or NK cells, however still encounters impairment in trafficking and persistence in suppressive solid tumors [115]. Conventional RT was able to enhance adoptive OT-I and lymph-node-derived T cell function through improved cross-priming, homing and cytotoxicity [116,117]. More specifically, it has been shown that RT induced tumor cell secretion of the cytokine IL-8 that enhanced trafficking of modified CAR T cells applying a single dose of 1 Gy in vitro and 3–4.5 Gy in vivo [118]. In addition, RT sensitized tumor cell lines MC38-OVA and EG7-OVA to the cytotoxic effects of OT-I T cells via induction of the Fas pathway. Additionally, adoptive T cells derived from splenocytes, NKG2D expressing CAR T cells, showed a synergistic therapeutic benefit in an in vitro and in vivo orthotopic GBM model when combined with RT (4 Gy) [61]. Similarly to T cells, adoptive NK cell migration and infiltration in murine tumor sites were enhanced upon RT (single-dose) and long-term survival was more pronounced as compared to NK or RT treatment alone [119]. Low dose radiation (2 Gy) sensitized heterogeneous orthotopic pancreatic tumors consisting of 25% antigen negative cells to immune rejection by locally activated CAR T cells in a TRAIL (TNF-related apoptosis-inducing ligand) dependent manner [120]. Additionally, the authors demonstrated that palliative RT (5 × 4 Gy) followed by CD19 CAR T cells in a patient with B cell lymphoma, bearing a large proportion of CD19- cells, resulted in disease-free local control one year after treatment. In another lymphoma patient study covering 11 patients, RT was used as temporary treatment for lymphoma until manufacturing of autologous T cells and CAR T cell administration had been completed. At follow-up, a complete response was attained in 5 (out of 11) patients [121]. Taken together, these studies indicate the high potential of RT not only to enhance infiltration of engineered T and NK cells, but also to increase the sensitivity of tumor cells to adoptive immunity.

Immune co-stimulation through agonist receptors provides a potent signal to promote the expansion and proliferation of CD8+ cytotoxic and CD4+ helper T cells. A range of agonists is currently under investigation, such as inducible costimulatory (ICOS) agonists, TLR agonists, OX40, CD27, anti-glucocorticoid-induced TNF receptor family-related gene (GITR), 4-1BB and the cytosolic DNA sensor STING [122,123,124,125]. The GITR agonist monoclonal antibody DTA-1 has been shown to co-stimulate activated effector CD4+ and CD8+ T cells and to inhibit the suppressive activity of Tregs [126]. Also, STING agonists increase immunogenicity in non-immunogenic tumors and improve the efficacy of IO through reversing resistance to anti-PD-1 agents in mouse tumor models [127]. Agonistic agents have also been shown to significantly amplify RT-induced immune responses. It has been demonstrated in a preclinical study that fRT (3 × 8 Gy) delivered during the course of topical TLR-7 agonist imiquimod treatment resulted in significant growth inhibition of both irradiated and non-irradiated mouse breast tumors compared with either treatment [128]. This therapeutic effect was associated with increased tumor infiltration by DC and T cells. Similarly, it has been shown experimentally in a CT26 mouse colorectal cancer model that therapeutic efficacy of the small molecule TLR-7 agonist DSR-29133 was augmented by combination with fractionated irradiation (5 × 2 Gy), which was dependent on the activity of CD8+ T-cells but independent of CD4+ T-cells and NK/NKT cells [129]. Importantly, administration of the agonist either 24 h prior to, or on day 1 of fRT, but not on the last day of RT, significantly improved antitumor response. This highlights the importance of the identification of the optimal scheduling of IO and RT to achieve maximal therapeutic benefit. Another agonist, the anti-OX40 antibody, providing a co-stimulatory signal for T cell proliferation upon interaction with the OX40 receptor, has been tested in combination with RT in a mouse lung cancer model [130]. Single-dose irradiation (20 Gy) delivered on the first day of the treatment with the agonistic antibody, cured a significant proportion (50%) of the tumors, which could not be achieved with either therapy alone, which was associated with strong recruitment of antigen-specific CD8+ T cells. In line with these preclinical findings, it has been shown that single-dose RT (10 Gy) synergized with synthetic oligonucleotides (ODN) containing unmethylated cytosine-guanine (CpG) motifs which bind to TLR-9, and stimulated both innate and adaptive immunes responses, in fibrosarcoma and mammary carcinoma mouse models [131]. Another study confirmed the ability of radiation (2 × 8 Gy) to boost antitumor effects of CpG ODN in poorly immunogenic mouse breast carcinoma [132]. Combining fRT (12 × 3 Gy) with CpG ODN significantly prolonged survival of rats bearing 9L glioma, increasing the percentage of animals rejecting the tumor from ~30% after single modalities to 70% [133]. The therapeutic efficacy of this combination did not depend on scheduling, i.e., injection of CpG ODN before or after RT. The combination of low-dose fRT (2 × 2 Gy) to a tumor lesion with intratumoral injection of CpG ODN has also demonstrated the ability to induce regression of systemic disease in clinical trials [134,135]. One important limitation of the latter treatment combination is that delivery of CpG ODN relies on intratumoral injections and thus is restricted to accessible tumors. Despite the promising preclinical results exploring the therapeutic potential of RT with several co-stimulatory agents, clinical experience with this approach is limited.

Treatment with cytokines in the past have achieved active immune responses [136], but also caused toxicities due to its unspecific targeting [137,138]. To guide cytokines selectively to the location of the target, immune-modulating cytokines can be conjugated to whole immunoglobulins (Ig), to the Fc fragment of an antibody, or fused to antigen-binding fragments specific to tumor antigens. Most cytokines are composed of a single monomeric protein domain (e.g., IL-2, IFN-ɑ) due to size advantages. Others are homodimeric (e.g., IL-10 and IFN-γ) or homotrimeric molecules (e.g., TNF, TRAIL) and some cytokines are composed of two different polypeptide chains, thus are heterodimeric (e.g., IL-12 and IL-27) [139]. The advantage of conjugated cytokines has been proven with IL-2 therapy dramatically reducing toxicity, while extending half-life from minutes to hours [140,141]. Several IL-2 conjugations have been tested, such as L19-IL2 and TNF-ɑ-IL2 [142,143]. However, even if immune cells are actively attracted to the tumor site, stimuli are still needed to target tumor cells. The combination with RT can provide this stimuli effectively [144]. The IL2-NHS immunocytokine consisting of a human NHS76 (antibody specific for necrotic DNA) fused to genetically modified human IL2, administered three days after the end of fRT (5 × 3.6 Gy), resulted in greater growth inhibition than either therapy alone in the mouse lung tumor model LLC [145], supporting the use in lung cancer patients. Interestingly, the same combination with cisplatin did not change the tumor growth but the triple combination (IL2-NHS, RT, cisplatin) resulted in the largest antitumor effect, controlling 83% of these aggressive tumors. Next, the authors tested the combination of IL2-NHS with RT (5 × 4 Gy to a single pulmonary nodule) in patients with metastatic NSCLC and showed that this approach is not only safe, but can also lead to long-term survival. IL2 was also genetically fused with a disialoganglioside D2 antibody—D2 is expressed in neuroblastoma and melanoma cells—and demonstrated complete regression of established tumors in most animals when it was combined with 12 Gy single-dose RT [146,147]. A combined therapy of ablative radiation dose (20 Gy) with a fusion protein of humanized anti-CEA (carcinoembryonic antigen) and human IL-2 (M5A-IL-2) in a transgenic murine breast and colon tumor model expressing human CEA, had significant antitumor affects as compared with RT alone [148]. It appeared that this immunocytokine generated greater therapeutic effect in combination with fRT (4 × 2.5 Gy) as compared with a single dose of 10 Gy. We have demonstrated in several preclinical studies the ability of RT to synergize with L19–IL2 and to induce durable systemic antitumor responses [96,97,149]. Our promising results prompted us to conduct clinical trials investigating the toxicity, which is deemed safe or tolerable, and therapeutic efficacy of this combination [102,150] (NCT02735850, NCT03705403, NCT04604470).

The cytokine L12 is another promising pro-inflammatory immune mediator to boost radiation-induced immune responses, which has been conjugated to tumor-specific antibodies to enhance targeted delivery, thus reducing toxicity. IL12 might be advantageous over unmodified IL2 since it might elicit TH1 responses without the risk of the expansion of Treg population unlike IL-2 [151]. fRT (5 × 3.6 Gy) combined with targeted delivery of cytokine IL12 using the NHS-IL12 immunocytokine resulted in significant growth delay compared to either treatment alone in colorectal and lung mouse tumor models [152]. In humanized mice bearing rhabdomyosarcoma xenografts, neither tumor irradiation with single dose 8 Gy, nor NHS-IL12 alone, but the combination demonstrated highly efficient systemic antitumor activity, identifying senescence and differentiation as underlying mechanisms for the observed efficacy [153]. Taken together, promising preclinical results indicate that immunocytokines in combination with RT can generate significant durable antitumor response, but clinical trials are required to proof the therapeutic benefit in clinical settings.

The processes of immune cell, especially T cell, activation and suppression are tightly regulated by various cytokines as well as immune checkpoints [154]. Radiotherapy and ICI combinations have been extensively studied both in preclinical and clinical settings and showed promising results [155]. Tumor-specific T cells can upregulate expression of immune checkpoint molecules leading to T cell exhaustion; likewise, cancer cells can upregulate the expression of ligands of immune inhibitory receptors potentiating immune escape. Apart from direct induction of immune checkpoint molecules on tumor cells by RT, it has been shown that ovarian cells can express higher levels of HLA-G and PD-L1 upon mitotic arrest, indicating even more the relevance of combining RT with ICI [156]. A number of preclinical studies have shown that blockade of the PD-1/PD-L1 axis or CTLA-4 acts synergistically with RT in murine models of breast, lung, colon, skin, head and neck cancer and pleural mesothelium [101,157,158,159,160,161,162,163]. In patients with chemo-refractory metastatic NSCLC, RT (5 × 6 Gy or 3 × 9.5 Gy) and CTLA-4 blockade induced systemic antitumor T cells with 18% of patients showing objective responses and 31% disease control [164], confirming previous observations in a patient with lung cancer [165] and melanoma [166]. In Phase I clinical trials, concomitant administration of SABR and ipilimumab (anti-CTLA-4) has been well tolerated (NCT02239900), with clinical benefit associated with an increase in peripheral CD8+ T cells and the proportion of CD8+ T cells expressing 4-1BB and PD-1 [167,168]. In a prospective cohort study in patients with advanced melanoma, hypofractionated radiotherapy (26 Gy in 3–5 fractions) to one site combined with anti-PD1 therapy induced long-lasting responses with confirmed complete response in 24% of patients [168]. The significant correlation between responses in irradiated and non-irradiated areas suggested an abscopal effect. Treatment with fRT (3–5 fractions 10 Gy at 2–4 tumor sites) and pembrolizumab (anti-PD-1) against metastatic tumors showed acceptable toxicity profiles [169]. A clinical trial is ongoing to determine the effect of atezolizumab (anti-PD-L1) with hypofractionated SABR (3 fractions of 15 Gy) on metastatic tumors (NCT02992912). In a Phase 3 randomized clinical trial in patients with unrespectable stage III NSCLC (PACIFIC trial), durvalumab (anti-PD-L1) significantly improved progression-free survival as well as overall survival compared with placebo [170,171] with manageable side effects consistent with other immunotherapies. These data formed the basis for subsequent approval of durvalumab as a treatment for patients with unresectable, stage III NSCLC cancer, whose disease had not progressed following platinum-based chemoradiotherapy [172]. Moreover, in a currently ongoing Phase I/II clinical trial, patients with extracranial metastatic lesions will receive two ICIs, namely, durvalumab and tremelimumab (anti-CTLA-4) and SBRT in 3–5 fractions (total dose range 30–50 Gy) (NCT03283605) between ICI cycles. The combination of multiple ICIs with RT could provide new potential, which is supported by a recent preclinical study where combination treatment with anti-PD-1 and Indoximod, an ICI-like tryptophan, together with 2× 12 Gy RT induced rapid tumor regression [173]. This triple therapy only significantly retarded the tumor growth and led to eventual tumor relapse accompanied by increased apoptosis of intratumoral T cells. The tumors were re-irradiated (2 × 10 Gy) at a late tumor regression phase or after relapse. Only late tumor regression re-irradiation cured the majority of mice bearing melanoma or strongly delayed relapse in mice having poorly immunogenic mammary carcinoma, in agreement with more memory T cells found in the tumor draining lymph nodes and spleen.

Despite the breakthrough in treatment of cancer patients with ICIs, the overall efficiency remains unsatisfactory. More recently, malignancies, in which ICIs have shown efficacy in metastatic setting such as head and neck cancer, were investigated in combination therapies consisting of ICI and RT. However, no added benefit was found after combined treatment compared to standard RT combined with chemotherapy or targeted drugs (cetuximab) in locally advanced setting, e.g., GORTEC 2015-01 “PembroRad” or JAVELIN Head and Neck 100 trials [174,175]. Another category, where the addition of ICI has not shown any added benefit, e.g., CheckMate 548 (NCT02667587), are stroma-rich but inherently immune-cold tumors, such as GBM and pancreatic cancer. These results highlight the urgent need of novel and more successful strategies for combination therapy. New immune checkpoint molecules are being identified and explored, such as lymphocyte activation gene-3 (LAG-3), T cell immunoglobulin and mucin-domain containing-3 (TIM-3), T cell immunoglobulin and ITIM domain (TIGIT), V-domain Ig suppressor of T cell activation (VISTA) [176,177]. It has been also suggested to combine anti-PD-L1 treatment with anti-exosomal treatment to enhance antitumor response as critical evidence was found that the presence of PD-L1 in exosomes was able to exhaust T cells in lymph nodes and to reduce splenic size [178]. This exosomal PD-L1 appeared resistant to current anti-PD-L1 treatment. Upon reversal of exosome biogenesis and exosomal PD-L1 load, an effective tumor response was generated, even against at a later stage injected tumor mocking distant tumor sites. Along with the discovery of new immune checkpoint targets and the design of specific inhibitors, multiple immune modulation therapies are currently combined to enhance therapeutic outcome. Moynihan et al. described that a maximal antitumor efficacy requires four components: a tumor antigen targeting antibody, an extended half-life of recombinant IL-2, anti-PD-1, and a powerful T-cell vaccine [179]. In a currently ongoing clinical trial, GBM patients receive a personalized neoantigen vaccine together with RT and pembrolizumab (NCT02287428). In a recent study, we also combined RT with the immunocytokine L19-IL2 and the immune checkpoint inhibitor anti-PD-L1 [180]. Notably, this triple combination was most effective in the poorly immunogenic mouse lung LLC tumor model, in contrast to other two colon carcinomas, which highlights the need of biomarkers to predict which patient would benefit from this and potentially other currently tested combined approaches.

In recent years, attention has been paid to the identification and development of predictive biomarkers for the efficacy of ICIs, including biomarkers of the tumor genome and neoantigens, the tumor immune microenvironment phenotype, liquid biopsy biomarkers and host-related factors [181,182]. Once therapies are more readily available for clinical implementation, therapy selection based on biomarkers will play an important role in treatment success. This is underlined by an example of a prediction model for RT-ICI combinations, where a RT sensitivity signature based on advanced mismatch repair-deficient cancers across 12 different tumor types was overlaid to PD-L1 status. Radiosensitive patients were associated with a higher PD-L1 status compared to radioresistant patients, thus confirming the benefit from the combination [183].

While there is convincing preclinical and clinical evidence on significant therapeutic benefit from combining photon RT with IO, only a limited number of comprehensive preclinical studies have investigated whether PRT elicits a different immune response compared to photons or modifies tumor cells into a more immunogenic phenotype to increase their sensitivity to immune surveillance (Figure 2).

There are indications that PRT has the potential to inflict higher immunogenicity than photons, especially CIRT. It has been reported that high-LET CIRT (>70 keV/μm) significantly increased HMGB1 levels in the culture supernatant of three human cell lines 72 h after irradiation compared with lower-LET CIRT (13 keV/μm) [204]. The comparison with photon RT was not performed in this study, but it can be assumed, based on the comparison with 13 keV/μm, that high-LET CIRT has a greater potential to enhance the release of the immunogenic molecules as compared to lower-LET RT such as linac X-rays (6–16 MeV with LET of only 0.3 keV/μm) or protons (~5 keV/µm) [205]. In another study, comparable levels of HMGB1 were found after exposure to isoeffective doses of carbon ions and photons, where HMGB1 was also measured after 72 h but with carbon ions with LET of 50 keV/μm [206]. In murine SCC, a dose-dependent release of HMGB1 and surface calreticulin expression was found after both γ-rays (¹³⁷Cs) and CIRT, but only calreticulin expression was significantly higher after CIRT with a biologically equivalent dose [207]. A similar radiation-induced (8 Gy) upregulation of surface molecules involved in immune recognition (HLA, ICAM-1, and the tumor-associated antigens CEA and MUC-1) was seen regardless of the type of RT, in several human cancer cell lines [208]. Both photon RT and PrRT induced cytotoxic T lymphocytes lysis mediated by calreticulin cell-surface expression. In addition, PrRT increased translocation of calreticulin in both cancer stem-cell- and non-stem-cell-like populations, potentially facilitating the immune attack on cancer stem cells, which are regarded as being more radioresistant; however, this potential was not tested. Whether induction of calreticulin translocation in cancer stem cells is a unique effect observed after PrRT remains to be elucidated. Calreticulin expression was compared 48 h after photon RT, PrRT, and CIRT in four different human cancer cell lines [209]. The study concluded that proton RT and PrRT were equally effective in inducing calreticulin expression, whereas CIRT had significantly stronger effects on increasing calreticulin levels compared to PrRT and photon RT at 2 and 4 Gy. In this study, however, despite reporting different RBE values for different cell lines, comparison of calreticulin expression at isoeffective doses for each individual cell line was not performed, which affects the data interpretation. Moreover, it should also be noted that for evaluation of the immunogenic potential of a cytotoxic insult upregulation of multiple ICD molecules is required (HMGB1, calreticulin, HSPs, ATP) to trigger a solid immune response [41,210]. In a recent study, it was shown that isoeffective doses of carbon ions induced higher levels of phosphorylated mixed lineage kinase domain-like (MLKL) protein, which mediates and regulate necroptosis, providing an important source of DAMPs and ICD molecules [211].

It may be expected that PRT leads to a higher release of antigens and greater diversity of T cell repertoire (neo-antigens) than photons because clustered DNA lesions, defined as two or more individual lesions within one or two helical turns of the DNA, commonly observed with charged particles, might result in more unrepaired DNA damage and genomic mutations or instability [212,213]. DNA damage in human pancreatic cancer cells, as measured by γH2AX foci up to 24 h after irradiation with isoeffective doses, was more persistent 24 h after CIRT (50 keV/μm) as compared to 200 kVp X-rays especially in cancer cells that had a stem cell phenotype [214], which was paralleled by the larger foci size [212]. Similarly, PrRT resulted in more and persistent DNA damage in glioma stem cells, as assessed using the DNA comet assay up to 48 h after radiation exposure as compared to photon RT, supporting higher probability of antigen accumulation after PRT [212].

The concept of increased genomic instability is supported by recent studies demonstrating that PRT induces chromothripsis in cells, i.e., a form of genomic instability characterized by massive, but highly localized chromosomal rearrangements following generation of more shattered chromosome domains along the particle track in the nucleus [215]. This phenomenon leads to a new genome configuration and the formation of complex chromosomal alterations by provoking inaccurate rejoining of chromosome fragments [10]. While this event can lead to the development of secondary malignancy in the case of normal tissue exposure, it can result in the formation of neoantigens provoking stronger immune reaction, particularly when combined with IO. However, it still remains to be determined whether neoantigens induced by PRT are immunogenic. Furthermore, micronuclei, by-products of mitotic progression, formed in response to radiation, may result in the initiation of chromothripsis (the micronuclei hypothesis [216]) and can trigger activation of inflammatory signaling and are a repository for the pattern-recognition receptor cyclic GMP–AMP synthase (cGAS) [85]. CIRT and PrRT are more effective in the induction of micronuclei than photon RT, suggesting that PRT may have greater potential to activate immune response via the aforementioned mechanisms [217,218,219]. The other consequence of the complex DNA damage induced by high-LET radiation including PrRT (12 keV/μm) is specific induction of histone H2B ubiquitylation on lysine 120, which is essential for repair of clustered DNA damage. This was not observed for low-LET PrRT (1 keV/μm) and photon RT [220]. It has been shown that this process is catalyzed by E3 ubiquitin ligases MSL2 and RNF20/RNF40 complex, which modulates inflammation and inflammation-associated cancer in mice and humans. It has been shown that downregulation of RNF20 augmented the TNF-ɑ response in mouse colonocytes and innate immune cells, suggesting that induction of RNF20 would improve TNF-ɑ profile [221]. In addition, RNF20-deficient mice had more myeloid-derived suppressor cells (MDSCs) than wildtype mice, hence increased RNF20 in response to clustered DNA damage might enable an antitumor immune reaction via decreased MDSC activation. Along with the indirect immune modulating effects of PRT via induction of pro-immunogenic processes in tumor cells, CIRT (170 KeV/μm) has been shown to directly enhance phagocytic activity of murine macrophages as compared to 250 keV photon RT, while there was no difference between the effects of CIRT and photon RT on macrophage vitality [63]. Moreover, an enhanced immunogenic effect of CIRT was found to augment in vivo antibody-dependent cellular toxicity [222,223].

Numerous cytokines are involved in the regulation of radiation-induced inflammatory response by recruiting immune cells, which are crucial for both local and systemic tumor responses to radiation. The impact of PrRT on the profile of inflammatory cytokines has been investigated in several studies. Specific genes coding pro-inflammatory molecules such as IL-6, IL-8, CCL2 were less expressed in HNSCC cells following PrRT as compared to photon RT especially after fractionated irradiation with three fractions one week apart either with 2 Gy or 8 Gy per fraction [224]. In line with this result, a panel of pro-inflammatory factors was also less expressed after PrRT in fibroblasts and in blood samples of irradiated mice as compared to photon RT [225,226,227]. While lower expression levels of molecules regulating inflammation following PrRT would decrease the risk of late radiation-induced tissue injury such as fibrosis, it still remains to be determined how this would impact the immune-stimulating effects of PrRT. Whether CIRT differentially affects the expression of pro-inflammatory cytokines as compared to photon RT is yet unknown.

It has been reported and summarized in several reviews that tumor hypoxia creates an immunosuppressive (tumor) microenvironment [228,229,230,231]. Hypoxia-inducible factor 1ɑ (HIF-1ɑ) is one of the mediators of the immune escape, for example, via upregulation of immune checkpoint protein PD-L1 on dendritic cells, macrophages and tumor cells [232,233]. Although hypoxia is the main factor contributing to HIF-1ɑ stabilization, radiation-induced reactive oxygen species have been shown to stabilize HIF-1ɑ also in the presence of oxygen [234,235]. The latter effect was observed only after photon RT, whereas CIRT attenuated HIF-1ɑ signaling. This phenomenon observed for carbon ions does not only prevent adaptive prosurvival cellular processes in response to HIF-1ɑ induction, but may also result in better recognition of tumor cells by immune cells in contrast to photon radiation, supporting the potentially superior immunogenicity of CIRT. It is well established that the efficacy of PRT is less dependent on the presence of oxygen as compared to photons [236,237]. It has been shown that CIRT (5 × 1.5 Gray equivalent (GyE)) eradicated patient-derived GBM stem cells in an orthotopic xenograft tumor model, while isoeffective photon irradiation led to cancer stem cells enrichment resulting in a worse outcome as compared with CIRT [237]. Given that cancer stem cells reside in the hypoxic niche, these data suggest that immune modulatory effect of CIRT may be less dependent on hypoxia-associated immunosuppression, which is in line with the observation that the eradication of the glioma cancer stem cells correlated with the switch of transcriptome signatures related to hypoxia-associated pathways in the latter study. The same authors performed a series of experiments in murine tumor models and suggested that prolonged survival after CIRT was attributed to the reduction of Mɸ2-like macrophages and MDSCs, increased influx of CD8+ T cells, generation of an immunopermissive niche, as opposed to photon RT. Overall, it can be expected that PRT will be especially effective in combination with immunotherapies in hypoxic tumors, which requires further investigations.

The greater immunogenic potential of PRT, as compared to photons observed in some of the aforementioned in vitro experiments, has been translated in some animal studies. Mouse osteosarcoma cells irradiated with carbon ions formed less distant metastases compared to photons after exposure to isoeffective single dose of 10 Gy (5 GyE) [238]. The difference was explained by decreased metastatic capabilities of CIRT as compared to photon RT assessed by cell migration and invasion and the expression of integrins. However, it cannot be excluded that CIRT triggered immune responses protecting animals from metastatic spread, which is also supported by the greater response of primary tumors to the RBE-weighted CIRT dose. It was suggested that patients with localized prostate cancer treated with CIRT have a lower risk of subsequent primary cancers than those treated with photon RT [239]. CIRT also resulted not only in efficient elimination of the primary tumor, but also in a dramatic reduction of tumor formation after secondary tumor challenge at a contralateral site in the murine SCCVII model, indicating the development of the protective immunological memory [240]. In a different mouse SCC (NR-S1), the incidence of lung metastases decreased with increasing dose but this effect did not depend on radiation quality [241]. Clinically, a case has been reported, where a patient with an inoperable metastatic retroperitoneal sarcoma was treated with palliative PrRT and had complete regression of non-irradiated metastases consistent with the abscopal effect [242], supporting the potency of PrRT to induce systemic antitumor immune response. Taken together, preclinical and clinical data provide some evidence that PRT can trigger superior antitumor immune response than conventional photon RT, but further thorough investigations are warranted.