RT is known to prompt the release of neoantigens and activate antigen-presenting cells (APCs). At 30 to 60 hours after irradiation with moderate doses (e.g., 10Gy), tumor cells undergo immunogenic cell death, which leads to release of tumor antigens (5). Irradiated tumors can express somatic mutations that generate neoepitopes, which can serve as targets of more robust immune responses (6, 7). One case report of a patient with non-small cell lung cancer (NSCLC) who experienced a complete response to RT and CTLA4 blockade demonstrated a rapid in vivo expansion of CD8⁺ T-cells that recognized a neoantigen encoded in a gene upregulated by RT (8). Thus, RT may increase the expression of immunogenic mutations in the irradiated tumor, which subsequently leads to the formation of neoantigens and activation of tumor specific CD8⁺ T-cells.

During immunogenic cell death, radiation-damaged tumor cells subsequently activate APCs by translocating or releasing endogenous damage-associated molecular patterns (DAMPs) such as, calreticulin, ATP, and high-mobility-group-box-1 (HMGB1) that activates TLR4. These signals collectively lead to a potent inflammatory cytokine response that promotes DC maturation, upregulation of costimulatory signals that facilitate cross-priming of cytotoxic CD8⁺ T-lymphocytes, and upregulation of chemokine receptors (4). This cascade of events will eventually activate and attract effector adaptive immune cells to elicit antitumor functions. Moreover, RT-induced damage results in production of cytoplasmic DNA fragments, the presence of which activates the cGAS/STING pathway (9). Specifically, the damaged nuclei of tumor cells release double-stranded DNA fragments into the cytosol. The cytosolic protein cGAS converts those fragments into cGAMP, which activates the protein STING. This leads to the production of type-I interferons (IFNs), which promote the recruitment and activation of DCs and lead to adaptive immune activation (9–11).

Tumors are known to use several mechanisms to evade immune surveillance, including downregulation of MHC-I expression, suppression of effector cytokines and chemokines, and production of inhibitory soluble factors such as TGF-β (4, 12). Poorly immunogenic tumors can also impair effector T-cells by expressing inhibitory checkpoint ligands, such as PD-L1, leading to exhausted/nonfunctional phenotype. In addition, since tumors lack the expression of costimulatory molecules (CD80, CD86, OX40L, 4-1BBL) amongst others, the T-cells become anergic and get eliminated. Moreover, inhibitory immune cells, such as M2 macrophages, myeloid-derived suppressor cells (MDSCs), regulatory T-cells (Tregs), and stromal fibroblasts, can also drive T-cell suppression and apoptosis (13–15). Optimized RT at the right dose and fractionation schedule, could promote an inflammatory response in tumors by inducing IFNs and chemokines that attract T-cells (16, 17). Low-dose RT (<2Gy), for example, leads to repolarization of tumor-associated macrophages from M2 to iNOS⁺ M1 cells (18), and promotes the infiltration of effector CD4⁺ T- and NK-cells, accompanied by reduction in TGF-β cytokine (19). The RT induction of chemokines such as CXCL9, CXCL10, CXCL11, and CXCL16 facilitates the homing of T-cells to tumors (18, 20–22). The upregulation of MHC-I molecules, Fas, and natural killer group 2 member D (NKG2D) on residual tumor cells in response to higher doses of RT allows recognition by incoming T-cells, with subsequent release of effector cytokines and killing of tumor targets (4).

Given the ability of metastatic tumors to disguise and evade immunity by armoring themselves with suppressive cellular and physical barriers, especially by forming an inhibitory stroma, the abscopal effect remains quite rare in the clinic (12). This indicates that RT by itself is insufficient to activate and maintain antitumor immunity, thus there has been increased interest in combining RT with immunotherapy. Our group previously showed that a single cycle of SBRT (12Gy×3 fractions) combined with a costimulatory OX40 agonist produced abscopal responses in a murine model of PD1-resistant lung adenocarcinoma while RT or OX40 agonist alone did not (23). Additionally, immune checkpoint inhibitors such as anti-CTLA4 and anti-PD1/PD-L1 liberate T-cells from immunosuppression and reduce exhaustion (24–26). Alternative immunotherapies (TIM-3, LAG-3, and TIGIT) are also under investigation to overcome T-cell exhaustion (27). Clinical findings have provided some evidence of added benefits of combining RT and immunotherapy (28–33). For example, in a phase II single-arm study, patients with oligometastatic NSCLC given pembrolizumab after completing locally ablative RT to all known sites of disease showed a median PFS of 19.1 months, which exceeded the historical median of 6.6 months (P=0.005), with no reduction in quality of life (34). In another randomized trial, the combination of pembrolizumab and SBRT (to one metastatic site) in patients with oligometastatic NSCLC did not extend PFS or OS (29). These findings suggest collaboration between RT and immunotherapy, but the inconsistencies underscore the need to explore new treatment strategies, such as increasing the number of RT cycles or irradiating additional metastatic lesions.

To demonstrate that pulsed-RT could enhance macrophage and T-cell stimulation and effector functions, we have developed a bilaterally transplanted murine model of lung adenocarcinoma. The right hind legs of 129Sv/Ev mice were subcutaneously implanted with 2.5×10⁵ 344SQ-P cells to establish primary tumors, while the left legs were concurrently implanted with 1×10⁵ cells on day 0 to establish secondary tumors. Primary tumors were irradiated with 12Gy×2 on days 9 and 10, while secondary tumors were irradiated on days 15 and 16 with the same dose as primary. We also included control groups where only primary tumors were irradiated, and additional groups where anti-CTLA4 was given i.p. (50µg/injection). Sera were collected on day 19 post-tumor inoculation and analyzed by multiplex cytokines panel (BioRad #M60009RDPD). When comparing RT to pulsed-RT ( Figure 1A ), IL-1a (P=0.21) and IL-1b (P=0.04) proinflammatory cytokines were upregulated with pulsed-RT, which is indicative of innate immune-cell activation. IL-12(p70) cytokine was also upregulated (P=0.09) along with IFN-γ (P=0.13), which shifts the balance towards Th1 antitumor responses. Moreover, the monocyte chemoattractant protein-1 (MCP-1/CCL2) was increased (P=0.07), which usually helps attract monocytic precursors to the TME, that can further differentiate into M1 or M2 macrophages, therefore making MCP-1 act as a double-edged sword. At last, TNF-α cytokine was elevated with pulsed-RT treatment (P=0.10) which is previously shown by others to mediate abscopal responses (41) and favor M1 macrophage polarization (42). We next sought to investigate if using systemic anti-CTLA4 might augment the efficacy of pulsed-RT ( Figure 1B ). Indeed, IL-12 (40) cytokine (P=0.03) and KC (CXCL1) chemokine (P=0.01) were significantly upregulated, while there was no change detected in the Th2-polarazing IL-6 cytokine (P=0.73) (43).

RT-induced antitumor immune responses are predominantly attributed to MHC-I−restricted CD8⁺ cytotoxic T-cells, but MHC-II−restricted B- and NK-cells are also important drivers of antitumor immunity. Tumor-infiltrating B-cells are heterogeneous and have distinct functional subsets that can be either tumor-promoting or suppressing (44). For example, IL-10- and TGF-β-producing regulatory B-cells exert several protumor functions, while B-cells that are associated with tertiary lymphoid structures (TLSs), seem to be associated with more favorable clinical outcomes (44–47). TLSs are ectopic structures arising in inflamed, infected, or cancerous tissues that can generate an adaptive immune response (45–48). In one study of patients with melanoma, chemokine gene-expression signatures predicted the presence of TLSs and linked the presence of tumor TLSs with better OS (49). B-cells role can be summarized as 1) sampling local tumor antigens for presentation, 2) maturation to plasma cells and production of antibodies that mediate antibody-dependent cellular cytotoxicity (ADCC), 3) promote antitumor immunity by secreting IFN-γ and IL-12 as well as Granzyme B and TRAIL, which directly kill tumor cells (50). In the hypothesized process of pulsed-RT, a portion of effector B-cells differentiate into memory B-cells with each cycle ( Figure 2A , blue curve), and can rapidly become effector cells again upon re-exposure to the same tumor-specific antigen. This translates into faster antibody production, facilitated ADCC through NK-cells engagement, and enhanced production of antitumor cytokines with subsequent RT cycle(s). Additional studies are warranted to further dissect how pulsed-RT may impact memory B-cell generation. On the other hand, NK-cells can also be activated via the NKG2D receptor, and RT has been shown to upregulate its ligand on various tumors (51, 52). Gathering the above, pulsed-RT treatment may enhance the crosstalk between B- and NK-cells and reprogram the TME in favor of antitumor outcomes.

As noted previously, RT acts to prime and enhance the adaptive immune system. Pulsed-RT might enhance RT-immune memory effects, leading to the production of long-lived immune memory cells, which could be amplified by the addition of immunotherapy ( Figure 2A ). Tumor-rechallenge experiments in previous preclinical studies showed complete tumor rejection as well as increased antigen-specific memory CD8⁺ T-cells in mice that cleared initial tumors after combination therapy of RT and anti–PD-1/PD-L1 (26, 53). One supporting study with PULSAR radiation (given in two cycles, 10 days apart) showed better immunological memory upon MC38 tumors rechallenge, that was medicated by CD8⁺ T-cells (40). Moreover, the existence of an RT-immune memory effect is supported by a secondary analysis of KEYNOTE-001, suggesting that patients previously treated with RT had better response to anti-PD1 (pembrolizumab) given afterwards (54). Other immunotherapeutic agents, such as IL-2 with RT also seemed to produce synergistic immune-memory effects, especially in patients with a history of previous RT (median OS 8.8 vs. 7.34 months without previous RT, P=0.0116) (55). Although these findings from clinical studies are encouraging, the mechanisms underlying these observations have yet to be elucidated. One possibility is the presence of a T-cell subtype with tissue-resident memory (TRM), which is correlated with favorable prognosis in several types of cancer (56, 57). In one recent study, a proportion of T-cells showed increased motility and higher production of IFN-γ after RT and shared transcriptional-level similarities with TRM T-cells (58). These findings led the investigators to speculate that T-cells capable of surviving irradiation are more likely to have a TRM-like phenotype (58). With that rationale, pulsed-RT regimen may lead to the accumulation of more TRM T-cells with each cycle and prevent future tumor recurrence.

Cancer vaccines aim to activate and stimulate the proliferation of tumor antigen-specific T-cells, however, this approach is limited by the fact that every patient’s tumor has a unique set of mutations. A melanoma study for patients who responded to epitope-loaded DC vaccine showed expansion of T-cells specific to the epitope, but also several other nonvaccine tumor antigens, suggesting that mobilization of a broad T-Cell Receptor (TCR) repertoire is required for tumor control (59). Preclinical work has demonstrated that RT-mediated tumoricidal effects largely hinge on both the presence and the specificity of cytotoxic T-cells (60). Addition of checkpoint inhibitors, such as anti-CTLA4 has been shown to augment RT to expand T-cell populations and increase diversity in TCR clonotypes (24, 53). For example, in a mouse model of triple-negative breast cancer showed that RT combined with anti-CTLA4 controlled both primary and metastatic tumors via tumor-infiltrating CD8⁺ T cells with a diverse TCR repertoire (61). Also, a recent clinical study revealed that a diverse and persistent TCR repertoire was associated with abscopal responses to RT and ipilimumab in patients with NSCLC (8). Similarly, our phase I-II clinical trial of SBRT plus ipilimumab in patients with metastatic tumors in lung or liver, partial response or stable disease (≥6 months) was associated with a higher number of peripheral CD8⁺ T-cells, higher CD8⁺/CD4⁺ ratios, and an increased number of either activated (4-1BB expressing) or potentially exhausted (PD1 expressing) CD8⁺ T-cells (62). RT has shown to induce the production of potential tumor antigens that leads to a phenomenon known as ‘epitope spreading’ (63–65), activating an array of T-cells, which in turn target the tumor to release another wave of antigens, creating a positive feedback loop (66, 67). These findings suggest that multiple rounds of RT (directed to a different lesion at each round) would increase the quantity and diversity of tumor-specific T-cells and further enhance this loop. Moreover, given the antigenic diverse nature of metastatic lesions, irradiating those tumors with a second or third pulse of RT will not only release tumor antigens shared with primary irradiated tumor, but will also release neoantigens specific to the metastatic site(s). The release of any identical antigens from the first and subsequent cycles of RT will contribute to the TCR repertoire clonality and the building of immunological memory in an incremental fashion (68), while releasing of neoantigens at metastatic sites will contribute to TCR repertoire diversity. Both clonality and diversity are important to prevent future recurrence and to generate systemic antitumor response.