Facts and Hopes in Radioimmunotherapy for Localized Stages of Cancer.

Introduction
Radiotherapy (RT) promises to be an ideal partner to immunotherapy to enhance cancer immunogenicity. RT-induced DNA damage both in the nucleus and mitochondria leads to cytosolic DNA accumulation, activating cGAS/STING to release IFNβ, a cytokine that activates and helps recruit BATF3+ dendritic cells, for cross-presentation and cross-priming of CD8+ T cells (1–5).
RT induces additional immunogenic effects, including augmented trafficking of activated CD8+ T cells by the CXCL16–CXCR6 interaction (6) and exposure of neoantigens through the expression of mutated genes, in response to repair and damage response pathways (7, 8). These mechanisms allow RT to convert the irradiated tumor into an in situ vaccine, contributing to the response of the irradiated tumor and, potentially, of unirradiated metastases, a mechanism originally defined as the abscopal effect (9, 10).
Our group originally linked RT-induced immunogenicity to the abscopal effect (11, 12), heralding a new area of research on the immune effects of radiobiology, and over the past 20 years has investigated how to exploit the combination of focal RT with immunotherapy. Multiple factors are emerging as relevant to potentiate RT immunogenicity, including the choice of fields, dose/fractionation, and sequencing with distinct immunotherapy agents.
Timing for combining focal RT with systemic immunotherapy is also crucial. In general, immune checkpoint inhibitors (ICI) work best when the tumor is on site as demonstrated preclinically in different syngeneic murine tumor models across different genetic backgrounds (13). Similarly, preceding vaccination by anti–PD-1 treatment has shown to prevent successful immunization compared with anti–PD-1 treatment after vaccination (14).
In support of the concept of radiation converting the tumor into an in situ vaccine, preclinical findings have been successfully translated to the clinic when RT was applied to primary cancers in the neoadjuvant setting. Success was also observed after primary therapy for unresectable cancers, for which ongoing immunogenic cell death after chemoradiation seems to derive benefit from anti–PD-L1 immunotherapy.

Adapting RT to Immune Checkpoint Blockade
Modifying standard RT practice seems necessary to the success of radioimmunotherapy. Preclinical data support the importance of sparing draining nodes, when RT is combined with immunotherapy (15). Also, preclinical and clinical evidence supports RT hypo-fractionated regimens, with three to five RT doses of 8 to 12 Gy (2, 16), to preserve the viability of circulating immune cells (17), expand the proliferating fractions of peripheral T cells (18), and prevent the induction of exonuclease Trex-1. Notably, these hypo-fractionated regimens are usually chosen to match RT doses given by standard fractionation through the biologically effective dose conversion formula, a commonly applied method in radiation biology (19). Trex-1 is induced at higher doses per fraction (generally >12 Gy) as it removes cytosolic DNA, preventing its activation of the cGAS–STING pathway to induce type 1 interferon, hence generating a barrier to RT-mediated immune stimulation (2). Sequencing of administration of radiation and immunotherapy is also relevant as preclinical evidence suggests that the optimal timing of anti–PD-1 is after radiation (20). Consideration of these variables is relevant to best combining radiation and immunotherapy in the clinic.
This review examines key clinical settings in which trials have tested the immunogenicity of RT. The strategies explored include combinations of RT with ICIs versus RT, ICI versus ICI + RT, or chemoradiotherapy (chemoRT), followed by ICI versus placebo or chemo + ICI versus the triplet of chemotherapy, RT, and ICI. We highlight phase III and selected phase II trials testing these comparisons. Although the landscape of RT/ICI trials is extensive, as reviewed by Zhang and colleagues (21), a complete summary lies beyond the scope of this article. Likewise, detailed mechanistic insights into RT/ICI synergy are available in recent comprehensive reviews (22–24).

RT versus RT + ICI
Recent trials investigating ICI added to RT have yielded mixed results in early-stage non–small cell lung cancer (NSCLC) and primary soft-tissue sarcoma (STS). In a randomized controlled phase II trial of medically inoperable early-stage NSCLC (iSABR trial), adding nivolumab to stereotactic ablative RT (SABR), 50 Gy to the planning target volume, significantly improved event-free survival (EFS) and reduced both locoregional and distant recurrences, incidence of new tumors, and death events (6% for SABR + nivolumab vs. 12% for SABR alone; ref. 25). Three subsequent phase III trials have tested SABR in combination with ICI in early-stage NSCLC (NCT04214262, NCT03924869, and NCT03833154).
KEYNOTE-867 randomized patients with stage I to II NSCLC, medically inoperable or refusing surgery, to SABR with or without approximately a year of pembrolizumab. In contrast to the results of iSABR, no differences in EFS were observed (hazard ratio, 0.92; P = 0.2932). Patients assigned to pembrolizumab had a higher incidence of grade ≥3 adverse events (20.4% vs. 3.7%) and a greater number of treatment-related deaths [5 (2.2%) vs. 0; ref. 26]. The discrepant outcomes between iSABR and KEYNOTE-867 likely reflect differences in trial design, patient selection, and treatment strategy. iSABR tested short-course concurrent nivolumab with SABR in carefully selected early-stage NSCLC and showed improved EFS, whereas KEYNOTE-867 used prolonged pembrolizumab with SBRT in a broader medically inoperable population, failing to improve outcomes and increasing toxicity. These distinctions highlight how ICI choice, treatment timing, and patient heterogeneity can critically influence results.
In localized high-risk STS, few trials have evaluated immunotherapy associated with standard preoperative RT. The randomized phase II SU2C-SARC032 trial showed improved disease-free survival (DFS) with pembrolizumab (anti–PD-1) added to RT, in the preoperative setting, although it was underpowered for assessing overall survival (OS; ref. 27). Importantly, toxicity was acceptable when ICI was added to SABR.
Notably, this approach differs from common clinical practice at many centers, in which patients typically receive cytotoxic chemotherapy (e.g., doxorubicin/ifosfamide), followed by RT and then surgery. This divergence in treatment paradigms raises important questions about the optimal sequencing and integration of cytotoxic chemotherapy, RT, and immunotherapy in localized STS. Future trials may explore whether a sequential strategy such as chemo-immunotherapy followed by radioimmunotherapy and surgery could outperform existing regimens. Alternatively, chemo-immunotherapy prior to surgery could be reserved for patients with suboptimal responses to initial radioimmunotherapy treatment.
Importantly, across trials of NSCLC and STS, the combination of RT with ICI has demonstrated a manageable safety profile, generally comparable with that of chemo-immunotherapy regimens in terms of the frequency of treatment-related grade ≥3 adverse events (28).

ICI versus RT+ICI
Although multiple trials have extensively investigated ICIs in medically operable NSCLC as standalone (29–31), or in combination with chemotherapy (32–38), in the neoadjuvant setting, our group tested immuno-RT preoperatively. A regimen of RT delivered at a non-ablative dose of 24 Gy in three fractions of 8 Gy in combination with two cycles of preoperative durvalumab was compared with durvalumab alone. This approach translated our preclinical work (39), demonstrating immunogenicity of a regimen of 8 Gy x 3, and was specifically selected to mitigate the risk of combined pulmonary toxicity from RT and ICI in highly curable, operable patients with NSCLC.
Main endpoints of this single-center phase II randomized controlled trial (NCT02904954; ref. 40) were safety and activity, as measured by pathologic response, of the combination of RT plus durvalumab compared with durvalumab alone in patients with operable NSCLC. The trial met its primary outcome of safety, demonstrating comparable and very limited toxicity in both arms. It also showed preliminary efficacy of the combination, with 53.3% major pathologic responses (MPR) in the RT + durvalumab arm versus 6.7% in the durvalumab-only group (P < 0.0001; ref. 40). In the trial, only the primary tumor was treated in patients assigned to the combination of RT + durvalumab, deliberately excluding the draining nodes even if biopsy-proven to be involved before treatment. Among patients with pretreatment positive nodes, four of six patients (66%) in the RT + durvalumab arm versus one of seven (14%) in the durvalumab alone arm achieved a pathologic response. The finding indirectly supports the hypothesis of a radiation induced abscopal effect that yielded a superior response in involved nodes, compared with durvalumab alone. Moreover, in EGFR wildtype carriers, MPR occurred in 15 of 21 (71%) patients in the RT + durvalumab arm versus in two of 21 (9%) in the durvalumab alone arm.
In our study (40), the pathologic complete response (pCR) rate in the RT + durvalumab group, which included both EGFR-mutant and EGFR wild-type tumors, was eight of 30 patients (26%). In comparison, pCR rates of 17.2% and 18.1% were reported in molecularly unselected populations in the AEGEAN and KEYNOTE-671 trials, respectively (34, 35), which tested ICI in combination with chemotherapy in localized NSCLC. When restricting the analysis to patients with EGFR wild-type cancers, our group observed a pCR in eight of 21 patients (38%) with only two cycles of durvalumab. pCRs in comparable EGFR wild-type carriers in two chemo-immunotherapy trials, CM816 and CM77T, were 29% (43 of 149 resected patients; ref. 32), and 32% (58 of 178 resected patients; ref. 36), respectively.
Although these cross-trial comparisons must be interpreted with caution given differences in study design and size, eligibility criteria, and treatment approaches, the neoadjuvant non-ablative RT + ICI strategy seems to be well tolerated and effective. Although pCR and MPR reliably capture early antitumor activity, they remain imperfect surrogates for survival outcome measures in NSCLC (41). Notably, although recurrence-free survival was not a prespecified outcome measure and our trial was not powered for defining it, at a median follow-up of 49 months, a recurrence-free survival trend favoring the RT + durvalumab group was observed, with half as many patients treated by combined durvalumab and RT developing a recurrence compared with the durvalumab-only group (15% vs. 30.7%; ref. 42).
Based on the preliminary, hypothesis-generating results of this trial, a multicenter, randomized phase II clinical trial (NCT06623656) has been initiated to evaluate neoadjuvant non-ablative RT (8 Gy x 3 fractions) plus perioperative cemiplimab (anti–PD-1) versus the current standard of neoadjuvant platinum-based chemotherapy combined with perioperative cemiplimab. The primary outcome measure is pCR, whereas key secondary outcome measures include quality of life and EFS.

ChemoRT ± ICI
The setting of inoperable, locally advanced NSCLC offers examples of how timing seems to be essential for integrating ICI with standard-of-care chemoradiation approaches as suggested by comparing the results of the phase III PACIFIC (43, 44) and PACIFIC-2 (45) trials.
The PACIFIC trial randomized patients with unresectable stage III NSCLC, who had completed definitive concurrent chemoRT administered as 60 to 66 Gy in 30 to 33 fractions, to sequential durvalumab or placebo every 2 weeks as consolidation therapy for up to 12 months. Patients who suspended therapy or progressed during chemoradiation were excluded. At median follow-up times of 34.2 months (all randomized patients) and 61.6 months (censored patients, i.e., patients alive at last follow-up), the trial demonstrated a statistically significant improvement of OS favoring the experimental arm (median 47.5 vs. 29.1 months, hazard ratio = 0.72; ref. 44), establishing a new treatment paradigm for unresectable stage III NSCLC.
The recognition that up to 30% of patients progress during chemoRT or do not adequately recover from treatment toxicity, which precluded accrual to PACIFIC, provided the rationale for the PACIFIC-2 trial, which randomized upfront patients with unresectable stage III NSCLC to receive concomitant chemoRT with or without durvalumab from the inception of chemoradiation. At a median follow-up of 30.5 months, the trial showed no differences in OS although a trend toward improved progression-free survival (PFS) favoring the durvalumab + chemoRT arm was observed. It is likely that treatment-related lymphopenia, a common consequence of definitive chemoradiation (46), has hampered the efficacy of immunotherapy in PACIFIC-2, but not in PACIFIC, in which patients started durvalumab only once blood counts had recovered.
Analogous considerations can be made for limited-stage small cell lung cancer (LS-SCLC), in which durvalumab administered after completion of concurrent chemoRT resulted in improved PFS and OS (NCT03703297; ref. 47), whereas a parallel study that tested atezolizumab during chemoRT failed to improve OS (NCT03811002; ref. 48).
In locally advanced squamous cell carcinomas of the head and neck (HNSCC), adding ICI to standard-of-care chemoRT has failed to demonstrate benefit across multiple trials (49–53). Notably, in most of these studies, ICIs were administered concurrently with chemoRT (49, 52) or with RT alone (50, 51). Conversely, a study by Zandberg and colleagues (54) recently demonstrated that sequential administration of pembrolizumab following chemoRT resulted in significantly improved locoregional control compared with concurrent administration (96% vs. 64%), along with a trend toward improved PFS and OS.
As in NSCLC and LS-SCLC, it is possible that concurrent treatment-related lymphopenia may have contributed to trial failure, particularly for pembrolizumab, an agent that has demonstrated efficacy against HNSCC in the recurrent and metastatic setting (55).
Collectively, evidence from NSCLC, LS-SCLC, and HNSCC—derived from both direct and indirect comparisons of concurrent versus sequential ICI integration with chemoRT—highlights the critical importance of treatment sequencing. Concurrent administration seems to be compromised by treatment-induced lymphopenia—or at least depletion of locoregional secondary lymphoid organs—whereas sequential approaches may allow for immune recovery, thereby enhancing the efficacy of immunotherapy. Evidence from the PACIFIC and ADRIATIC trials suggests that a delay of 1 to 6 weeks between completion of definitive chemoRT and ICI start is sufficient to restore immune competence and enable therapeutic efficacy.
Recent preclinical work by Saddawi-Konefka and colleagues (15) elucidates another mechanism underlying the failure of chemoRT–ICI trials in locally advanced HNSCC. In a syngeneic murine model of tongue cancer, exquisitely sensitive to ICI, surgical dissection or ablative irradiation of tumor-draining lymph nodes abolished the efficacy of immune checkpoint blockade. As demonstrated by time-of-flight mass cytometry of tumor samples harvested 10 days after neck dissection and after CTLA4 inhibitor start, this effect was mediated by loss of CD8+ and CD4+ T cells and increase of myeloid-derived cells and M2 macrophages. Preclinically, irradiation of regional lymphatics recapitulated the detrimental effects of surgical lymphadenectomy, demonstrating that regimens that include RT to draining lymph nodes during immunotherapy can be a barrier to effective immune activation.
Although cross-disease comparisons point to a shared RT-induced pathophysiologic mechanism influencing immunotherapy efficacy, tumor-intrinsic factors remain critical determinants of outcomes. For instance, the recent phase II CONTINUUM trial showed a significant improvement in EFS with the addition of sintilimab to chemoRT in patients with locoregionally advanced nasopharyngeal carcinoma (56). Nasopharyngeal carcinoma is biologically distinct from other HNSCC. It is generally Epstein–Barr virus–driven and characterized by a dense lymphoid infiltrate and high PD-L1 expression, both characteristics that may confer increased susceptibility to PD-1 blockade.
A further example of how tumor-intrinsic factors—and particularly viral etiology—influence the response to combined chemoRT and immunotherapy comes from the experience in uterine cervical cancer. Platinum-based chemoRT and brachytherapy have been the standard of care for locally advanced cervical cancer (LACC) for more than 2 decades. Two recent trials have tested the integration of ICI with chemoRT. The phase III CALLA trial randomized patients with LACC—classified as International Federation of Gynaecology and Obstetrics (FIGO) 2009 stage IB2 to IIB lymph node positive, or stage ≥ III any lymph node status—to durvalumab (anti-PD-L1) concurrently with and after chemoRT for up to 24 cycles. Despite an acceptable toxicity profile of the experimental arm, the trial failed to improve PFS versus placebo at a median follow-up of 18.5 months (57). By contrast, the phase III, placebo-controlled, ENGOT-cx11/GOG-3047/KEYNOTE-A18 study tested chemoRT plus five-cycle pembrolizumab (anti–PD-1), followed by 15 cycles of pembrolizumab maintenance in high-risk LACC classified as FIGO 2014 stage IB2 to IIB with node-positive disease or stage III to IVA regardless of nodal status. KEYNOTE-A18 significantly prolonged the OS (HR for death, 0.67) at a median follow-up of 29.9 months, with a manageable safety profile and high brachytherapy adherence (58).
Several factors may underlie the divergent results of CALLA and KEYNOTE-A18. First, the trials tested different ICIs: durvalumab, used in CALLA, blocks PD-L1, whereas pembrolizumab, used in KEYNOTE-A18, blocks PD-1. Although PD-L1 blockade has been shown to enhance the activation of human T cells engaging with antigen-presenting cells, anti–PD-1 antibodies—which inhibit interactions with both PD-L1 and PD-L2—have a theoretical advantage in promoting the activation and priming of T cells, as well as in rescuing exhausted CD8+ T cells. In the context of radiation-induced antigen release and dynamic checkpoint upregulation, broader inhibition of PD-1 signaling with pembrolizumab in cervical cancer may have enhanced immune priming and effector function, potentially translating into stronger clinical benefit compared with selective PD-L1 blockade with durvalumab in cervical cancer. The caveat is that in other tumor types such as NSCLC, durvalumab has been shown to be highly effective when administered after chemoRT, underscoring the complexity of biology and mechanism of action in disease-specific settings.
The lack of detailed information on what percentage of patients in the immunotherapy arm received para-aortic (PA) irradiation in the KEYNOTE-A18 trial limits interpretation on whether variations in baseline nodal burden and in the extent of nodal irradiation may have affected treatment efficacy. KEYNOTE-A18 enrolled a higher risk population; 3% of patients had isolated PA involvement and 20% had both pelvic and PA involvement. In CALLA, relatively fewer (11%) patients had PA involvement; however, a disproportionately high number of patients in the immunotherapy arm (34.5%) received a PA boost. The high degree of nodal burden in KEYNOTE-A18 may have enhanced antigen release and immune priming, and the high percentage of patients receiving a PA boost in CALLA may have disproportionately exposed a large volume of blood and lymphoid tissue and hampered immune fitness. Along with differences in checkpoint inhibitor mechanism and patient risk profiles, trial power also varied: KEYNOTE-A18 enrolled a larger, higher-risk cohort (n = 1,060) and was adequately powered to detect survival benefit, whereas CALLA enrolled fewer patients (n = 770) across a broader spectrum, with favorable control outcomes (relative to historical controls), which reduced its ability to demonstrate clinical benefit. Taken together, these factors potentially explain why KEYNOTE-A18 was a positive trial whereas CALLA was not. In future trials of immunotherapy and RT, radiation dosimetric data and biomarkers are required to understand the role of reduced irradiation volumes in preserving systemic immunity and amplifying the benefit of PD-1 blockade.
Additional technical factors may account for the relative success of concurrent RT + pembrolizumab in cervical cancer relative to other tumor types. Indeed, cervical cancer RT commonly includes brachytherapy, a form of radiation delivery that results in high-dose gradients. Such gradients are emerging as an immunologically relevant variable, as described in a recent publication highlighting the preclinical value of dose heterogeneity in eliciting distinct tumor microenvironment (TME) effects, influencing immune responses (59).

Long-Course versus Hypo-fractionated ChemoRT ± ICI: The Case of Rectal Cancer
A multimodal treatment combining chemotherapy, RT, and surgery is the standard of care for most locally advanced rectal cancers (LARC). In the neoadjuvant setting, the National Comprehensive Cancer Network Guidelines recommend two main evidence-based radiation strategies, long-course chemoRT (LC-chemoRT), delivering ∼50 Gy in 25 to 28 fractions with concurrent fluoropyrimidine, and short-course RT (SCRT), consisting of 25 Gy in five daily fractions before chemotherapy, followed by surgery. Total neoadjuvant therapy (TNT) has also emerged for selected patients with clinical complete response (cCR) after neoadjuvant therapy, with avoidance of surgery or delayed surgical salvage. Although the two approaches of neoadjuvant chemoRT are generally equivalent in terms of DFS and OS (60), the choice is now driven by physician preference, anatomic location, and patient logistics. Importantly, many centers have shifted to LC-chemoRT because of at least one study demonstrating higher rates of locoregional failure observed with SCRT (61).
Interestingly, recent clinical results from the PRIME-RT trial suggest potential superiority of SCRT over LC-chemoRT when combined with ICI (62). The PRIME-RT trial randomized patients with LARC to SCRT with durvalumab, followed by 5-fluorouracil + oxaliplatin (FOLFOX) + durvalumab, vs LC-chemoRT with capecitabine and durvalumab, followed by FOLFOX + durvalumab. The primary outcome measure was a composite of cCR or pCR. The median follow-up time was 13.3 months. At 1 year after the end of treatment, 11 of 18 (61%) patients in the durvalumab + SCRT and eight of 21 (38%) in the durvalumab + LC-chemoRT had a sustained cCR or pCR. Although the difference is not statistically significant, the trend toward higher proportion of cCR/pCR rates in the SCRT favors the hypo-fractionated regimen in combination with ICI through lesser exposure of circulating blood and secondary lymphoid organs. Caveats include the brief follow-up, given the natural history and timing of when rectal cancers locoregionally recur (frequently after 2 years; ref. 61), and small sample size.
Similarly, the phase II TORCH trial randomized patients with stage II to III microsatellite-stable rectal cancer to SCRT, followed by immunochemotherapy [capecitabine and oxaliplatin (CAPOX) plus toripalimab], or by the reverse sequence. Notably, both SCRT + ICI sequences achieved pCR rates substantially higher than historical controls [50% vs. <30% in historical controls (63, 64) with an acceptable safety profile (65)]. A separate phase II trial evaluating SCRT, CAPOX, and camrelizumab in LARC reported a 48.1% pCR rate, including 46.2% in mismatch repair–proficient tumors (66). Immune-related toxicities were mild, with no grade 4 to 5 events. Along the same lines, another trial comparing neoadjuvant chemoRT with or without sintilimab (anti–PD-1) reported increased complete response rates favoring the chemoRT + ICI arm versus chemoRT, at 44.8% versus 26.9%, respectively (67).
In contrast, the results of the NRG-GI002 platform trial remain controversial. In this study, the addition of pembrolizumab to TNT showed improved 3-year OS (95% vs. 87%, HR = 0.35; P = 0.04) compared with the control TNT arm, yet no difference in 3-year DFS was observed (68).
Collectively, positive signals from these small clinical trials in LARC suggest that hypo-fractionated RT regimens, particularly when integrated with immunochemotherapy, invigorate interest in SCRT-based platforms as a backbone for neoadjuvant immunotherapy strategies in rectal cancer. Shorter RT courses may mitigate lymphoid and hematologic toxicity, thereby preserving immune competence and enhancing the effect of ICIs. Although the results of the aforementioned trials supporting the ICI/SCRT approach are encouraging, we acknowledge that in the context of relatively high rates of pCR reported in trials such OPERA, a multicenter, phase III randomized controlled trial demonstrating the benefit of contact X-ray boost versus traditional external beam RT boost (56% and 79% rate of organ preservation at 5 years) when added to standard neoadjuvant chemoRT in early-stage rectal adenocarcinoma, larger trials and prolonged follow-up are required to determine whether or not short-course chemoRT is superior to LC-ChemoRT when combined with IO- and FOLFOX-based chemotherapy (69).
Importantly, the choice of immunotherapy partners is another area of active investigation in the rectal radioimmunotherapy field. Ongoing trials such as PANTHER (NCT05024097) further explore these strategies. The PANTHER trial, part of the NCI-funded Radiation Oncology-Biology Integration Network (NCT05943210), is a phase I/II study investigating SCRT combined with chemotherapy and immunotherapy (AB928, a dual A2aR/A2bR antagonist, and AB122, a PD-1 inhibitor). The trial targets adenosine-mediated immunosuppression in the TME, particularly in irradiated and hypoxic tissues. Preliminary results indicate that the combination is safe, with early signs of activity, as demonstrated by a pCR/cCR rate of 82% (nine of 11 patients; ref. 70). These findings support further exploration of adenosine axis blockade as a strategy to enhance immune responses in non-ablative radiation settings.

Chemotherapy–ICI ± RT: The Case of Breast Cancer
ICIs have been successfully integrated with neoadjuvant chemotherapy for triple-negative breast cancer (TNBC) since the completion of the KEYNOTE-522 trial, with demonstrated improvements in pCR, EFS, and OS (71–73).
A preoperative approach of RT directed to the primary breast tumor in combination with anti–PD-1, followed by standard chemotherapy, in the neoadjuvant setting was tested in the PEARL trial (74). This single-arm, phase I/IIb trial, conducted in the pre–KEYNOTE-522 era, tested preoperative pembrolizumab, followed by pembrolizumab and focal RT (8 Gy x 3), in patients with TNBC or hormone receptor–positive (HR+), HER2-negative (HER2–) breast cancer. The regimen was well tolerated and patients with TNBC and HR+ breast cancer achieved pCR rates comparable with those observed in landmark trials testing neoadjuvant chemotherapy and ICI (71, 75, 76). Importantly, the PEARL trial produced robust and high-quality evidence of ICI–RT immunomodulatory effects on breast tumors through serial biopsies. Spatial analysis of cancer and TME cells identified two patterns of responders deriving different effects from RT (77). These preliminary results suggest that some tumors that were immune-desert at baseline became immune-enriched following treatment with pembrolizumab and RT. These findings for the first time established the concept of “adaptive clusters of response” in the radiation immuno-oncology field and highlight the role of RT in recruiting activated T cells to the tumor bed of a subset of patients.
The promising results of PEARL led to the design of the randomized P-RAD trial (NCT04443348), which tests different doses of RT (9 and 24 Gy, each regimen being administered in three consecutive fractions) compared with no RT during neoadjuvant pembrolizumab/chemotherapy in patients with TNBC or high-risk HR+/HER2− breast cancer who have biopsy-proven, node-positive disease. In this trial, the presence of a positive node represents a surrogate setting to study the distant or “abscopal” effect of induction ICI ± RT, followed by neoadjuvant ICI + chemotherapy. Patient accrual was completed in January 2025, and results will be reported at San Antonio Breast Cancer Symposium 2025. The co-primary endpoints of P-RAD were nodal response rates (ypN0), as well as change in T-cell infiltration using a modified Breast Immunoscore; the study was powered to the latter endpoint. Correlative studies will include spatial transcriptomics of activated immune cells in the primary tumor and the lymph node.
Lastly, the NeoCheckRay trial investigated neoadjuvant RT (8 Gy x 3) alone, with durvalumab, or with durvalumab plus oleclumab (anti-CD73) in high-risk HR+/HER2– breast cancer, all of whom were Mammaprint classified as high risk. In contrast to P-RAD, the control arm in NeoCheckRay was RT alone. Preliminary data confirmed safety (78) and suggested higher-than-expected activity of the experimental regimens as measured by the primary endpoint, residual cancer burden class 0/1 (79), upon treatment completion at 37.8%, 46%, and 51.1% across the three arms, respectively (80), compared with 23.6% in the placebo–chemotherapy arm of KEYNOTE-756 (76). Together with PEARL, NeoCheckRay results support the safety and activity of RT–ICI although the trial did not demonstrate a statistically significant result with IO–RT combinations compared with RT alone.

Insights and Future Perspectives
While acknowledging that many early trials integrating RT and immunotherapy in localized tumors have failed to meet clinical endpoints to demonstrate the superiority of the combination, these outcomes must be interpreted in light of the substantial knowledge gaps that existed at the time of their design. Figure 1 summarizes the design of pivotal trials in the field by fractionation type and sequence and classifies them as “positive” versus “negative,” based on comparison of outcomes with the control arm. Moving forward, we identify three key priorities for the future development of radiation immuno-oncology in nonmetastatic settings: optimization of RT dose and sequencing, rational selection of immunotherapy partners, and immune-based patient stratification, particularly through a deeper investigation of how irradiated normal tissues modulate the immune response.
Hypo-fractionated RT has the advantage of reducing the risk of lymphopenia, a known predictor of systemic tumor control (81, 82), by limiting the exposure of immune cells and lymphoid structures. Non-ablative dosing combined with surgery has emerged as a safe and effective option in operable cancers. In locally advanced disease, conventionally fractionated RT has improved outcomes when followed by ICI in NSCLC and LS-SCLC—while providing no benefit when administered concurrently—whereas concurrent approaches have succeeded in nasopharyngeal carcinoma and cervical cancer. Although cross-disease comparisons are limited by differences in radiosensitivity, immunogenicity, and biology, available evidence suggests that when conventional fractionation is used, a sequential administration of PD-(L)1 blockade may be preferable.
The potential of RT–ICI combinations remains insufficiently explored. Notably, few studies have addressed the hypothesis that targeting innate immunity could enhance the efficacy of RT–ICI strategies. Although preclinical data support the antagonism of the adenosine pathway (83) and indicate a role for macrophage-targeted strategies (84), the full spectrum of innate immunity interventions in the context of immunomodulatory RT remains to be explored, particularly in the context of resident macrophages in the TME and adjacent normal tissue, which is inevitably exposed to a significant RT dose.
Lastly, although there is a clear need for immune biomarker–based patient stratification in trials combining RT with immunotherapy, as exemplified by some of the breast trials (74, 77), limited current evidence precludes a direct application in the clinic. Much needed information from studies that obtain tumor and normal tissue biopsies before and after RT as a single-treatment modality is rapidly emerging and will enlighten the immunomodulatory mechanisms of radiation on human tissues. To this end, our group is conducting two studies that leverage sequential biopsies and blood sample collections to track immune responses across different RT + immunotherapy combinations. The first one is a randomized, four-arm clinical trial in HR+ breast cancer during neoadjuvant aromatase inhibition (see trial design in Fig. 2). The objective of the trial is to compare different immunomodulatory interventions (anti– PD-1 and FLT3 Ligand) combined with RT versus RT alone, during aromatase inhibition, with the endpoint of pathologic response. This trial enables the application of spatial transcriptomics to biopsies before and after treatment and at surgery.
The second study (NCT05024097) is a multicenter clinical trial in rectal cancer, conducted under the Radiation Oncology-Biology Integration Network initiative, an international consortium dedicated to generating multiomic datasets to advance research in radiation oncology. The clinical trial specifically aims at obtaining blood and tumor tissue before and after SCRT (5 Gy x 5). Ongoing single-cell RNA sequencing and spatial analyses will define the distinct immune effects of RT on malignant versus normal tissues, an indispensable tool to interpret the findings from any RT/ICI combination.

Conclusions
Radiation immuno-oncology is an emerging field shaped by early trials that revealed key challenges and opportunities. Variability in tumor biology, patient selection, and treatment parameters likely explains the conflicting results of radioimmunotherapy trials in disease sites such as cervical cancer, NSCLC, and rectal cancer, underscoring the need for biomarker-driven, context-specific approaches. Moving forward, optimizing RT parameters, selecting effective immunotherapy partners, and directing preclinical and clinical data to implement TME-based stratification will be essential to unlock the full potential of this field.