Beyond radiation: immunological potentials of boron neutron capture therapy

Abstract

Boron neutron capture therapy (BNCT) is a targeted radiotherapeutic modality that employs boron-10 (¹⁰B) to capture thermal neutrons, thereby releasing high-linear energy transfer α-particles and lithium ions that selectively eradicate tumor cells while sparing the surrounding normal tissues. Immune checkpoint inhibitors (ICIs), including inhibitors of PD-1/PD-L1 and CTLA-4, enhance antitumor immunity by alleviating immunosuppression within the tumor microenvironment. As immunotherapy continues to expand across multiple tumor types and becomes integral to systemic cancer treatment, attention has increasingly focused on combining systemic immunomodulation with localized therapies. BNCT functions primarily as a local therapeutic approach, whereas ICIs elicit systemic immune activation. Whether the two modalities can produce synergistic therapeutic effects remains an important question.

In this narrative review, we summarize current mechanistic insights, preclinical and clinical developments, and ongoing challenges associated with the combination of BNCT and ICIs, highlighting the potential of this strategy to achieve durable, synergistic antitumor responses.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00262-026-04424-w.

Background

Boron neutron capture therapy

The ability of boron-10 nuclei to capture neutrons and undergo a nuclear reaction that yields helium-4 (α-particles) and ⁷Li nucleus was first identified by H. J. Taylor in 1935 [1]. Boron neutron capture therapy (BNCT) was subsequently proposed in 1936 as a potential therapeutic approach for tumors, and its development has since progressed through several primary stages, particularly in the refinement of neutron sources and boron-containing pharmaceuticals [2] (Fig. 1). BNCT selectively targets tumor cells through the administration of boron-containing compounds that preferentially accumulate within malignant tissues. Upon irradiation with neutrons, boron-10 atoms capture thermal neutrons and undergo nuclear fission, producing a helium-4 nucleus (α-particle) and a ⁷Li nucleus and releasing gamma radiation (γ). These charged particles traverse less than 10 μm, approximately the diameter of a single cell, to cause highly localized, lethal radiation damage to tumor cells while largely sparing adjacent healthy tissues [3].

Fig. 1.

Timeline of BNCT therapy

An optimal BNCT neutron source should fulfill several requirements: it should generate predominantly epithermal or thermal neutrons, achieve a neutron flux of at least 10⁹ n/(cm²·s), deliver maximal thermal neutron flux within tumor regions enriched with ¹⁰B, minimize the production of fast neutrons and gamma rays, and maintain sufficient beam collimation to ensure precise targeting. Likewise, an ideal boron delivery agent should demonstrate low intrinsic toxicity, high tumor uptake, strong tumor selectivity, high bioavailability, and an appropriate lipid–water partition coefficient. Guided by these development goals, and following the introduction of the first hospital-based BNCT system in Japan during the 2010s, BNCT has entered a period of global expansion, with increasing clinical adoption and active ongoing investigation.

Current clinical applications of BNCT

BNCT has been explored for the treatment of multiple malignancies, most notably glioblastoma multiforme, head and neck cancer, and melanoma. Continued progress in boron delivery agents and neutron source technology has renewed clinical interest and contributed to a growing number of clinical trials worldwide.

The clinical exploration of BNCT for brain gliomas began between 1951 and 1961, when Brookhaven National Laboratory conducted a trial involving 28 patients. These early studies demonstrated limited efficacy due to inadequate tumor retention of the boron compounds available at that time [4]. The subsequent development of borocaptate sodium (BSH) and boronophenylalanine (BPA) enabled a new generation of clinical investigations in Japan from 1998 to 2007, with reported median survival times ranging from 10.8 to 27.1 months [5, 6]. In the past two decades, improvements in boron agents, image-guided dosimetry, and neutron beam technology have led to the initiation of several registered clinical trials. Historical data indicate median survival times exceeding 20 months in newly diagnosed glioblastoma multiforme and approximately 10 months in recurrent disease [7]. Ongoing studies are expected to clarify the therapeutic role of BNCT further and optimize its application in glioblastoma multiforme management [8].

BNCT has been investigated as a therapeutic option for squamous cell carcinoma and recurrent head and neck cancers that are refractory to conventional treatments. The earliest exploratory study, EORTC 11011 (2003), evaluated the biodistribution of boron-10 in six patients [9]. From 2010 to 2015, the clinical findings demonstrated a 2-year overall survival rate of 47% using fractionated BNCT [10]. Additional work integrating BNCT with image-guided intensity-modulated radiotherapy (IG-IMRT) improved dose conformity and facilitated the treatment of larger or anatomically complex tumors [11]. A Phase I study demonstrated a high overall response rate of 83% among 12 patients with inoperable, recurrent head and neck cancer, underscoring the therapeutic potential of BNCT in heavily pretreated patients [12]. Building on these findings, a subsequent Phase II trial including 29 patients reported a median PFS of 7.5 months (95% CI, 5.4–9.6 months). The 2-year progression-free survival (PFS) and overall survival (OS) rates were 20% and 30%, respectively, with 27% of patients achieving 2-year locoregional recurrence-free survival [13]. The Phase II JHN002 trial from Japan further refined patient stratification by histological subtype, including 21 patients with recurrent head and neck squamous cell carcinoma (SCC) or recurrent/locally advanced non-squamous cell carcinoma (nSCC). The ORR (objective response rate) was 71%, and complete response/partial response were 50%/25% in SCC and 8%/62% in nSCC. The 2-year OS for SCC and nSCC were 58% and 100%, respectively [14]. Collectively, the transition from early-phase feasibility studies to histology-stratified Phase II trials reflects a maturation of the clinical evidence base for BNCT. As a result of accumulating evidence, BNCT has been incorporated into Japanese clinical guidelines for unresectable, locally recurrent, or locally progressive head and neck tumors and has been covered under Japan’s national medical insurance system since 2020.

BNCT has shown encouraging efficacy in advanced melanoma, particularly in the management of metastatic or recurrent disease. Since 2013, five clinical studies have examined the use of BNCT for skin tumors, although only one has published results to date. A phase I/II trial (jRCTs061180066), completed in 2020, enrolled three patients with recurrent cutaneous malignancies and reported a 100% tumor control rate without severe treatment-related adverse effects [15]. A larger clinical study currently underway in China aims to enroll 30 patients. It is employing a regimen of 350 mg/kg BPA with a prescribed dose of 20.0 Gy-Eq (biologically equivalent dose). Results from this trial are expected to clarify further the therapeutic benefit of BNCT in melanoma [16]. Beyond melanoma, BNCT has also been investigated in liver cancer [17], sarcoma [18], angiosarcoma [19], and refractory breast cancer [20], with early studies indicating promising antitumor activity. Continued clinical research will be essential to define the broader therapeutic potential of BNCT across diverse malignancies.

Mechanism of immune checkpoint inhibitors

Immune checkpoint inhibitors (ICIs) are therapeutic agents that restore antitumor T cell activity by blocking inhibitory immune checkpoint pathways. These drugs primarily target the programmed death receptor-1 (PD-1)/programmed death ligand-1 (PD-L1) axis and the cytotoxic T lymphocyte antigen-4 (CTLA-4) pathway to enhance antitumor immune responses.

PD-1 is an inhibitory receptor expressed on the surface of T cells [21], whereas PD-L1 is a ligand expressed on tumor cells and antigen-presenting cells [22]. Under physiological conditions, PD-1 engagement by PD-L1 transmits an inhibitory signal that attenuates T cell activation and proliferation, serving as a mechanism to prevent autoimmunity.

CTLA-4, another inhibitory receptor, functions primarily during the early phases of T cell activation [23]. Its ligands, CD80 and CD86, are expressed on antigen-presenting cells. Under normal circumstances, CD80 and CD86 bind to the costimulatory receptor CD28 on T cells, delivering the stimulatory signals required for full T cell activation. CTLA-4 has a substantially higher affinity for CD80 and CD86 than it does for CD28. When expressed on activated T cells, CTLA-4 competes with CD28 for ligand binding, thereby dampening T cell activation and reducing antitumor immune responses.

Both the PD-1/PD-L1 and CTLA-4 pathways serve to limit T cell activation. Therapeutic blockade of these checkpoints—through PD-1 or PD-L1 inhibitors and CTLA-4 inhibitors—disrupts these inhibitory interactions, enhancing T cell activity and strengthening antitumor immunity.

Immunological basis for combining BNCT with ICIs

To improve clarity, the mechanisms described in this section are organized into sequential immunological processes: (i) induction of immunogenic cell death, (ii) activation of innate immune sensing pathways, (iii) enhancement of antigen presentation, (iv) modulation of immune checkpoint signaling, and (v) remodeling of the tumor microenvironment. Numerous studies have shown that the combination of radiotherapy and ICIs can produce synergistic antitumor effects. Radiotherapy enhances antitumor immunity while simultaneously inducing specific immunosuppressive responses. Importantly, an immune-mediated antitumor effect occurring outside the irradiated field, referred to as the abscopal effect, has been repeatedly observed following radiotherapy. Conversely, immunotherapy can modulate the tumor microenvironment (TME), including its hypoxic state, thereby influencing the efficacy of radiation treatment. These bidirectional interactions have enabled a clearer understanding of the mechanistic basis underlying the combined effects of conventional radiotherapy and ICIs. High-linear energy transfer (high-LET) radiation may exert even stronger immunomodulatory effects, as it induces pronounced oxidative damage and increases endoplasmic reticulum stress within tumor cells [24]. Despite these observations, the immunological consequences of combining BNCT with ICIs remain poorly characterized. The following section summarizes the potential mechanisms through which BNCT may interact with immunotherapy to enhance antitumor responses.

Similar mechanisms of BNCT and ICIs compared with conventional radiotherapy

Promotion of antigen release and cross-presentation

As described above, boron-10 undergoes neutron capture and subsequently splits into two highly energetic particles—helium-4 nucleus (α) and ⁷Li nucleus—accompanied by gamma-ray (γ) emission. The resulting high-LET ionizing radiation generates dense, localized DNA double-strand breaks that induce immunogenic cell death (ICD) [25]. ICD initiates the exposure and release of tumor antigens, as well as damage-associated molecular patterns (DAMPs). Key DAMP-related processes include surface expression of calreticulin (CRT) and heat shock proteins, the release of high-mobility group box 1 protein (HMGB1), and the active secretion of adenosine triphosphate (ATP) [26]. These DAMP-associated signals are expected to promote dendritic cell maturation and enhance antigen uptake. Within regional lymph nodes, tumor antigens are cross-presented on MHC class I (MHC-I) molecules, activating tumor-specific CD8⁺ cytotoxic T lymphocytes (CTLs). Concurrent presentation on MHC-II molecules supports CD4⁺ T-helper cell activation, facilitating the clonal expansion of antitumor effector subsets, particularly CTLs and natural killer (NK) cells.

Evidence of ICD following BNCT has been demonstrated in multiple studies. CRT positivity is observed in tumor specimens 7 days after BNCT, confirmed by immunofluorescence analysis [27]. Heat shock proteins, including HSP90, are also upregulated after BNCT, and h-¹⁰BN-PG-mediated BNCT induces robust HSP90 expression accompanied by cytoplasmic HMGB1 translocation (Fig. 2). Notably, HMGB1 release, a hallmark of ICD, is markedly increased in human squamous carcinoma and melanoma cells 24 h after BPA-mediated BNCT irradiation. Plasma HMGB1 levels remained elevated up to 3 days post-treatment, even when tumor volume had already declined [28]. Another study reported significant increases in intratumoral HMGB1 within 6 h of BPA-BNCT in a rat lymphosarcoma model, with subsequent release into the extracellular environment by 20 h [29]. The extracellular levels of ATP, a crucial “find-me” signal released by dying tumor cells, also markedly and dose-dependently increase after BNCT, underscoring its importance in dendritic cell recruitment and activation [30] (Fig. 2).

Fig. 2.

BNCT-induced immune activation and remodeling of the tumor microenvironment. (i) Boron neutron capture therapy (BNCT) generates high-linear energy transfer (LET) α-particles and ⁷Li nucleus through the ¹⁰B thermal neutron capture reaction, producing dense DNA damage and inducing immunogenic cell death (ICD). ICD is characterized by the exposure or release of calreticulin (CRT), heat shock proteins (HSPs; e.g., HSP90), high-mobility group box 1 (HMGB1), ATP, and tumor antigens, collectively functioning as damage-associated molecular patterns (DAMPs) [1, 25, 27–30]. (ii) Radiation-induced DNA damage activates the cGAS–STING–TBK1–IRF3 pathway, driving type I interferon (IFN-α/β) production and inflammatory cytokines (e.g., IL-6), which in turn support IFN-γ–JAK–STAT1-dependent upregulation of MHC-I and PD-L1 [33–35, 41]. (iii) The ATM/ATR–Chk1–IRF1 signaling cascade further enhances PD-L1 transcription in irradiated tumor cells [40, 41]. (iv) HSPs, tumor antigens, HMGB1, and sEV-associated cargo (including MHC-I molecules components) cooperate with type I IFN to promote dendritic cell maturation and CCR7-mediated migration to draining lymph nodes, where DCs cross-present antigens on MHC-I to prime CD8⁺ cytotoxic T lymphocytes (CTLs) and present antigens on MHC-II to support CD4⁺ T-helper activation [65, 66]. PD-1/PD-L1 blockade reverses T cell exhaustion, promotes M1 macrophage polarization, enhances NK and CTL trafficking into tumors, and facilitates systemic antitumor immunity with potential abscopal effects [27]. Tumor microenvironment (TME) reshaping: BNCT-induced type I/II interferons and DAMP signaling reprogram the TME toward a proinflammatory, Th1-skewed state, characterized by increased M1 and reduced M2 macrophages, contraction of immunosuppressive MDSCs and Tregs, heightened recruitment and retention of effector lymphocytes via chemokines such as CXCL9, CXCL10, CXCL11, CXCL16, and CCL5, enhanced antigen presentation (increase in MHC-I on tumor cells and in mature DCs), and activation of innate effectors, including NK cells with increased GZMB and IFN-γ production [44, 56]. Abscopal effect: Effector T cells primed in draining lymph nodes migrate to distant, non-irradiated tumors, where they secrete IFN-γ, IL-12, and TNF-α, converting immunologically “cold” tumors into “hot” inflamed lesions capable of mounting effective antitumor responses [63, 64]. Together, these processes reinforce both local and systemic antitumor immunity. BNCT, boron neutron capture therapy; DAMPs, damage-associated molecular patterns; sEVs, small extracellular vesicles; DC, dendritic cell; CTL, cytotoxic T lymphocyte; MDSC, myeloid-derived suppressor cell; Treg, regulatory T cell;HSP, heat shock protein.

Collectively, these findings demonstrate that BNCT robustly induces ICD through the coordinated release of multiple DAMPs (Fig. 2). This process enhances the immunogenicity of tumor cell death, promotes tumor-associated antigen presentation, and strengthens inflammatory signaling, thereby amplifying antitumor immune responses. These effects represent a key mechanism by which high-LET radiation, including BNCT, potentiates immune activation.

Activation of the STING pathway

The cGAS–STING signaling pathway functions as a cytosolic DNA surveillance system. When DNA damage leads to the formation of micronuclei or cytoplasmic chromatin fragments, cyclic GMP–AMP synthase (cGAS) recognizes these aberrant DNA structures and catalyzes the production of cyclic GMP–AMP (cGAMP). cGAMP subsequently activates the stimulator of interferon genes (STING), an adaptor protein located on the endoplasmic reticulum membrane that plays a central role in innate immune signaling [31]. Upon activation, STING recruits and activates TANK-binding kinase 1 (TBK1), which phosphorylates the transcription factor interferon regulatory factor 3 (IRF3), leading to the induction of type I interferons (IFN-α/β), a signaling cascade often referred to as the STING–TBK1–IRF3 axis [32]. In parallel, STING activation stimulates the NF-κB pathway, resulting in the release of proinflammatory cytokines such as tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6). Through these coordinated mechanisms, the cGAS–STING pathway elicits both cytotoxic and immune-stimulatory effects and constitutes a central sensor system for detecting cytosolic DNA and initiating antitumor immune responses.

Both BNCT and conventional radiotherapy induce extensive double-strand DNA breaks and may increase mitochondrial outer membrane permeability, thereby releasing mitochondrial DNA into the cytoplasm. Cytosolic mitochondrial DNA and double-stranded DNA serve as potent activators of the cGAS–STING pathway (Fig. 2). The resulting type I interferon response may further amplify the immunogenic cascade initiated by BNCT-induced antigen release and cross-presentation. However, the activation of this pathway is highly dependent on both the radiation dose and the biological context. At higher radiation doses, the DNA exonuclease Trex1 becomes upregulated and degrades cytosolic DNA, thereby reducing STING activation and dampening downstream immune responses [33–35]. Optimizing radiation parameters to maximize immunogenic DNA accumulation while avoiding excessive Trex1 induction is therefore critical for fully harnessing radiotherapy-induced immune activation. As a highly localized, high-LET radiation modality in the context of BNCT, such DNA-sensing responses are biologically plausible because of the highly localized and complex DNA damage induced by high-LET particles; however, BNCT-specific evidence regarding the magnitude, kinetics, and Trex1 dependence of this pathway remains limited. Although activation of the cGAS–STING pathway is well established in conventional radiotherapy, direct experimental evidence in the context of BNCT remains limited. Therefore, the extent and kinetics of STING activation specific to BNCT require further investigation.

The STING pathway has emerged as a major focus in contemporary tumor immunotherapy research and may ultimately serve as a novel radiosensitizing strategy analogous to immune checkpoint inhibition. In the setting of BNCT, clarifying whether and how cGAS–STING signaling contributes to immune activation may improve mechanistic understanding of its synergy with immunotherapy. Further BNCT-specific studies are needed to determine the extent to which this pathway can be effectively and safely leveraged for therapeutic benefit.

Enhancement of tumor cell MHC-I expression and antigen visibility

MHC-I molecules are essential for presenting tumor-derived antigens to CTLs. However, many tumor cells downregulate or lose MHC-I expression to evade immune recognition. Conventional radiotherapy has been shown to upregulate MHC-I surface expression [36]. In addition, radiation stimulates the release of small extracellular vesicles, nanoscale vesicles enriched in MHC molecules, tumor-associated antigens, and heat shock proteins [37]. These small extracellular vesicles serve as additional antigen carriers, thereby enhancing the immune system's visibility of antigens.

BNCT, characterized by its high-LET and dense ionization tracks, induces robust DNA double-strand breaks and more potent activation of DNA damage response (DDR) pathways. The activation of signaling cascades, such as the ATM/ATR-p53 axis, has been shown to promote MHC-I upregulation [38]. Furthermore, BNCT-induced activation of the cGAS–STING pathway and subsequent IFN-β–JAK–STAT signaling also contribute to increased MHC-I expression on tumor cells [35]. Given the dense DNA damage induced by BNCT, enhanced antigen visibility through increased MHC-I expression is biologically plausible (Fig. 2). However, much of the mechanistic rationale currently derives from broader radiotherapy and interferon-signaling literature, and direct BNCT-specific validation remains limited.

In summary, BNCT enhances antigen presentation and improves tumor antigen visibility by strongly inducing DNA damage and activating immunostimulatory signaling pathways, potentially increasing tumor cells' susceptibility to cytotoxic T cell-mediated clearance (Fig. 2).

Upregulation of immune checkpoint molecules

Beyond the DNA damage and interferon-associated events described previously, BNCT may also influence immune checkpoint regulation [25, 39]. In broader radiotherapy and immunotherapy, PD-L1 upregulation has been linked to several pathways, including the DDR–ATM/ATR–Chk1–IRF1 axis [40] and the IFN-γ–JAK–STAT1 signaling pathway [41] (Table 1). In contrast, PD-1 expression is predominantly induced on T cells by persistent T cell receptor (TCR) stimulation and other chronic activation signals [42, 43]. In the context of BNCT, direct mechanistic evidence connecting these pathways to the checkpoint remains limited. However, studies have reported PD-L1 upregulation after BNCT and have shown enhanced antitumor effects when BNCT is combined with anti-PD-1 or anti-PD-L1 therapy in resistant tumor models in the B16F10 melanoma mouse model [44–46] (Fig. 2). While PD-1 and PD-L1 upregulation is generally associated with immune escape and immunosuppression, it may also indicate the emergence of a targetable adaptive resistance state, thereby providing a rationale for combining BNCT with immune checkpoint blockade. Additional regulators, such as hypoxia-inducible factor-1α (HIF-1α) and forkhead box O1 (FoxO1), appear to influence PD-1/PD-L1 regulation only indirectly or in specific biological contexts. For example, HIF-1α, activated under hypoxic conditions, can enhance PD-L1 transcription by binding to hypoxia-responsive elements within the CD274 promoter, whereas FoxO1-mediated PD-1 expression is largely driven by reactive oxygen species (ROS) [47, 48]. Although BNCT also generates ROS, its cytotoxicity primarily arises from direct high-LET ionization rather than oxygen-dependent oxidative damage. Therefore, current evidence supports the biological rationale for combining BNCT with PD-1/PD-L1 blockade, but the BNCT-specific causal pathways governing checkpoint regulation still require direct experimental validation.

Table 1.

Representative signaling pathways related to PD-1 and PD-L1

Pathway	Cell type	Upregulated molecule	Mechanistic summary
IFN-γ–JAK–STAT1 [41]	Tumor and immune cells	PD-L1	STAT1 activation promotes CD274 transcription
DDR (ATM/ATR–Chk1–IRF1) [40]	Tumor cells	PD-L1	DNA damage activates IRF1, enhancing PD-L1 expression
TLR pathway [49]	Tumor/immune cells	PD-L1	TLRs, cytokines, or ROS activate NF-κB signaling
MAPK pathway [50]	Tumor cells	PD-L1	MAPK signaling induces AP-1–mediated PD-L1 transcription
PI3K–AKT pathway [51]	Tumor cells	PD-L1	PI3K/AKT/mTOR pathway activation enhances PD-L1 expression
HIF-1α [52]	Tumor cells	PD-L1	Under hypoxia, HIF-1α upregulates PD-L1
TCR–NFAT [42]	T cells	PD-1	Persistent antigen stimulation activates PD-1 expression
IL-6–STAT3 [43]	Tumor/immune cells	PD-1	Inflammatory cytokines activate STAT3 and induce PD-1
ROS–FoxO1 [52]	T cells	PD-1	Oxidative stress promotes PD-1 through FoxO1 signaling

Modulation of the tumor microenvironment

The immune microenvironment, a component of the broader TME, comprises immune cells, cytokines, vasculature, and extracellular matrix components surrounding tumor tissue. This microenvironment plays a central role in determining the immune system's ability to recognize and eliminate malignant cells. The importance of the TME was first emphasized by Hanahan in 2000 [53] and has since become a major research focus, particularly following the advent of ICIs.

Tumors with strong immune activity, characterized by substantial T cell infiltration and elevated PD-L1 expression, are commonly described as “hot” tumors and are generally more responsive to immunotherapy. Conversely, tumors dominated by immunosuppressive cell populations, such as regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs), and by metabolic dysregulation, are referred to as “cold” tumors and exhibit weaker responses to ICIs [53]. Conventional radiotherapy in some settings promotes the transition of cold tumors toward a more inflamed state by inducing proinflammatory chemokines and immune cell recruitment, as a hot tumor by inducing chemokines such as CXCL9, CXCL10, and CXCL16, thereby enhancing effector T cell recruitment and local immune activation [50, 54]. However, this effect is often context-dependent and may be counterbalanced by collateral immune cell damage and treatment-induced immunosuppressive responses [55].

In this context, BNCT may offer a distinct immunological advantage rather than broadly damaging both malignant and immune cells. Recent work by Sun et al. [56]. provided strong support for this concept in immunocompetent cold-tumor B16F10 and hot-tumor MC38 models, showing that BNCT can reshape the immune microenvironment and induce systemic antitumor immunity in both cold and hot tumors, although immune-mediated tumor control appears stronger in hot tumors. Single-cell transcriptomic analysis revealed that BNCT reshaped the tumor microenvironment by enhancing dendritic-cell antigen processing and presentation, increasing CD8 + T cell activity, and augmenting NK-cell cytotoxicity and activating markers, including Gzmb, Ifng, Cd69, and Stat4, indicating not only enhanced infiltration but also functional reprogramming toward cytotoxic, antitumor phenotypes (Fig. 2). Another study also confirmed that treated tumors exhibit increased infiltration of CD3 + , CD4 + , and CD8 + T lymphocytes, as well as NK cells [44], accompanied by reduced frequencies of immunosuppressive cell types, including Tregs, MDSCs, and M2-polarized macrophages. Notably, BNCT leads to a significantly higher ratio of proinflammatory/antitumor (M1) to immunosuppressive/ tumor-promoting (M2) macrophages compared with conventional radiotherapy [27], indicating a more favorable immune polarization profile. Preclinical studies further support BNCT's ability to reshape the TME. (Fig. 2) In a rat model of intracerebral gliosarcoma, BNCT combined with tumor cell-based immunoprophylaxis (administration of live or X-irradiated gliosarcoma 9LGS cells) resulted in markedly improved long-term survival and induced robust immunological memory that protected animals from subsequent tumor rechallenge [57]. Additionally, a study analyzing peripheral blood following BNCT demonstrated that irradiation did not damage peripheral blood mononuclear cells (PBMCs), even though these cells contained boron. In typical tumor settings, infiltrating macrophages are often reprogrammed to reduce IL-12 secretion and increase IL-10, thereby fostering an immunosuppressive microenvironment. IL-10 further supports tumor progression by inducing macrophages to release protumorigenic mediators. However, under BNCT irradiation, cytokine profiling revealed a phenotypic shift in PBMCs toward an antitumor state, marked by increased IL-12 and reduced IL-10 expression [58] (Fig. 2). These findings suggest that BNCT may actively reprogram macrophages toward an immunostimulatory phenotype.

Collectively, these preclinical findings suggest that BNCT can remodel the tumor microenvironment toward a more immunostimulatory state and may support more effective integration with immune checkpoint blockade.

BNCT-induced bystander effects

Radiation-induced bystander effects, first reported by Nagasawa [59], are defined as biological responses in non-irradiated cells adjacent to irradiated cells. The phenomenon was observed when irradiation of only a small fraction of cells with α-particles led to increased chromosomal exchanges in neighboring non-irradiated cells. Since all cells along the radiation path receive direct exposure during conventional radiotherapy, bystander responses are often regarded as a double-edged sword. In BNCT, theoretically, only cells containing sufficient ¹⁰B are expected to generate high-LET α particles and ⁷Li nucleus, resulting in localized DNA double-strand breaks. However, because boron uptake is heterogeneous and often incomplete, many tumor cells do not accumulate sufficient ¹⁰B to undergo direct BNCT-induced nuclear reactions. The bystander effect may help compensate for uneven boron distribution.

Direct experimental evidence for BNCT was provided by Kinashi et al., who demonstrated that even cells not directly targeted by BSH-mediated BNCT exhibited point mutations attributable to bystander signaling. Molecular analysis further indicated that these bystander-associated mutations were dominated by point mutations rather than the larger deletions typically associated with direct neutron-induced damage, suggesting that the underlying mechanisms may differ [60]. More recently, Shi et al. [27] reported that PEG-B-COF-mediated BNCT combined with the TLR7 agonist R837 led to widespread nuclear shrinkage throughout the entire tumor mass in a B16F10 melanoma model, including regions with low boron accumulation, demonstrating a robust bystander effect that contributed to tumor regression. Although this finding does not directly establish a classical bystander mechanism, it is consistent with the possibility that BNCT-associated responses may extend beyond areas of high boron uptake, thereby broadening tumor injury.

Overall, current evidence remains insufficient to conclude that BNCT-associated bystander effects are intrinsically stronger than those observed in conventional radiotherapy. Rather, their distinctive importance in BNCT lies in their potential to complement the highly selective but distribution-dependent cell-killing mechanism of this modality.

Induction of the abscopal effect

The abscopal effect, first described in 1953, refers to tumor regression at sites distant from the irradiated field [61]. With the emergence of radiotherapy–immunotherapy combinations, this phenomenon has regained considerable attention. Early in the twenty-first century, Demaria and colleagues proposed that radiotherapy could induce abscopal effects through immune activation. This concept gained clinical support in 2012 when Postow et al. [62] reported the first case of melanoma in which combined radiotherapy and ICI (ipilimumab) resulted in a systemic abscopal response. These findings significantly strengthened the hypothesis that combining radiotherapy with immunotherapy could enhance distant tumor control.

An early investigation of an abscopal effect induced by BNCT was conducted by Trivillin in an immunocompetent rat model of colon cancer [63]. Tumors were established bilaterally, and only the right-sided tumor received BNCT. Weekly measurements over 7 weeks demonstrated significant tumor reduction not only in the irradiated tumor but also in the contralateral non-irradiated tumor compared with untreated controls. Subsequent work evaluated BNCT combined with Bacillus Calmette–Guérin (BCG). Both BNCT alone and BNCT + BCG produced strong local antitumor effects, whereas BCG alone showed only modest efficacy. Importantly, BCG and the BNCT + BCG combination induced significant tumor regression at non-irradiated sites, with the combination group exhibiting reduced metastatic spread to draining lymph nodes [64]. Although the studies were preliminary and the mechanisms remained incompletely defined, these studies provided early evidence that BNCT may induce abscopal responses. More recently, this concept is substantially strengthened based on research BNCT inhibited not only the irradiated tumor but also the distant non-irradiated tumor, whereas dose-equivalent X-ray irradiation failed to suppress distal tumor growth, importantly, the researcher found that BNCT-treated tumor cells could function in a tumor vaccination-like manner and that BNCT-induced durable immune memory in rechallenge experiments, Memory-associated CD8 + T cell populations were increased in blood, lymph nodes, and spleen, suggesting that BNCT may prime sustained systemic immunity rather than merely transient distant tumor suppression. These findings indicate that BNCT-associated abscopal effects may be linked to a broader process of systemic immune priming, and that the magnitude of this response may still depend on the baseline immune context of the tumor.

Additionally, studies suggest that immunotherapeutic co-interactions may further strengthen this systemic potential. Fujimoto et al. [44] further examined BNCT in combination with anti-PD-1 therapy in a melanoma mouse model and assessed immune responses across multiple tissues and time points. The BNCT–immunotherapy group demonstrated increased intratumoral CD8⁺ T cell infiltration and elevated serum HMGB1 levels, indicating enhanced ICD-associated immune signaling. In addition, increased numbers of tumor-specific memory T cells and early activated T cells were detected in distant tumors, consistent with systemic immune activation. However, macroscopic tumor regression at distant sites was not observed within the study period, potentially due to suboptimal immune checkpoint blockade or insufficient time to capture delayed abscopal responses. In another study employing bilateral and distant metastatic B16F10 melanoma tumor models, BNCT combined with localized release of the immune adjuvant imiquimod elicited robust systemic antitumor immunity, resulting in significant regression of non-irradiated lesions and prolonged survival [27]. Compared with BNCT monotherapy, the combination therapy induced markedly higher serum levels of proinflammatory cytokines, including IL-6, IL-12p70, and TNF-α, and produced clear macroscopic evidence of abscopal tumor control (Fig. 2). Taken together, current evidence suggests that controlled release of immune stimulants at the irradiated site can amplify immune activation and enable abscopal effects that are rarely observed with BNCT alone. Collectively, these studies offer compelling preliminary evidence that BNCT can induce abscopal immune responses, particularly when paired with immunotherapeutic strategies. Further investigation is warranted to determine which immunotherapy classes synergize most effectively with BNCT to promote systemic immune activation and distant tumor regression (Fig. 2).

BNCT combined with immunotherapy: unique mechanisms

Compared with other radiotherapy modalities used in combination with immunotherapy, BNCT occupies a distinct biological position. Similar to carbon-ion radiotherapy, BNCT generates high-LET damage and is less oxygen-dependent than conventional photon radiotherapy, making both approaches attractive for immunologically resistant or hypoxic tumors.

Previous studies have shown that CIRT can remodel the tumor immune microenvironment by modulating immunosuppressive cells and increasing immunostimulatory molecules [67–70]. Briefly, it reduced MDSCs and M2-like polarization, enhanced CD8 + T cell infiltration, and increased IFN-γ/IL-2 expression. However, unlike CIRT, whose immunomodulatory effects depend primarily on physical radiation properties, BNCT introduces an additional layer of selectivity through boron-mediated drug delivery. This selectivity allows BNCT to deliver high-LET energy preferentially to boron-accumulating tumor cells while largely sparing circulating immune cells and normal tissues—a feature that proton radiotherapy, despite its favorable dose distribution, does not offer. Table 2 would provide a more intuitive comparison of key characteristics across photon, proton, carbon-ion, and BNCT combined with immunotherapy.

Table 2.

Comparison of key characteristics across photon, proton, carbon-ion, and BNCT combined with immunotherapy

Comparison item	Photon RT	Proton RT	Carbon ion RT	BNCT
Physical and dosimetric characteristics	Most widely used modality in clinic; with greater entrance and exit dose to normal tissues	Bragg peak allows improved dose conformity and reduced exit dose	Bragg peak plus reduced lateral scattering; superior dose localization with high-LET properties	Selective boron uptake followed by neutron irradiation generates high-LET short-range particles, producing highly localized cell killing at the microscopic level
LET / RBE features	Low LET; clinical RBE≈ 1	Low-to-moderate LET; clinical RBE≈ 1.1	High LET with increased RBE≈2.08	Very high LET at the microdosimetric RBE≈1.4–4.1
DNA damage	Primarily induces sparsely distributed DNA damage that is relatively repairable	Similar to photons but dose distribution is more favorable	Induces more complex and clustered DNA damage and more difficult to repair	Produces dense, localized high-LET damage in boron-containing cells
Effects on the TME	Can convert some “cold” tumors into a more inflamed state, but may also enhance immunosuppressive signaling	Potentially beneficial because of better normal tissue sparing, but microenvironmental data remain limited	May promote greater CD8 + T cell infiltration and reduce immunosuppressive components	May improve local immune conditions and preserve immune-related components, but current data are still sparse
Relationship with hypoxia	More affected by hypoxia; oxygen enhancement effect is substantial	Still influenced by hypoxia, although dosimetric advantages may improve target coverage	Less dependent on oxygen than photons and protons; retains efficacy better under hypoxic conditions	Less oxygen-dependent; one of its major theoretical advantages, but highly dependent on boron distribution
Impact on lymphocytes	More likely to irradiate circulating lymphocytes and surrounding normal tissues, contributing to lymphopenia	Better normal tissue sparing may reduce unnecessary immune cell irradiation	More conformal dose delivery may better preserve circulating lymphocytes and immune function	Highly localized effect may help preserve systemic immune cells, although this depends on boron distribution and treatment design
Immune escape related changes	Can increase PD-L1 expression, providing a rationale for combination with immune checkpoint blockade	PD-L1-related effects have been reported, but the evidence base is smaller	May also upregulate PD-L1, in some settings more prominently than photons, supporting combination with ICIs	PD-L1 modulation after BNCT has been reported in early studies, but evidence is far more limited

Relative efficacy of BNCT under hypoxic conditions

Conventional radiotherapy is highly dependent on oxygen, as X-rays predominantly induce DNA damage through the fixation of free radicals by molecular oxygen. This dependence is reflected in an oxygen enhancement ratio (OER) of approximately 2–3 [71], meaning that radiotherapy becomes significantly less effective under hypoxic conditions [72]. Hypoxia not only reduces the formation of ROS-mediated DNA damage, but also promotes immunosuppressive signaling, including HIF-1α, which enhances PD-L1 transcription, contributing to immune evasion in hypoxic tumors [73]. These factors collectively create substantial challenges for combining conventional radiotherapy with immunotherapy in the treatment of poorly oxygenated tumors.

In contrast, BNCT is generally considered less constrained by hypoxia because its biological effectiveness does not rely on oxygen-mediated free radical generation. The high-LET α-particles and ⁷Li nucleus produced by the ¹⁰B(n,α)⁷Li reaction exhibit a LET of approximately 100 keV/μm, placing them in a region characterized by high RBE and low OER (≈1.2) [48] Supplementary Fig. S1. Similar advantages have also been described for carbon-ion radiotherapy, which likewise shows high RBE and low OER; however, unlike carbon ions, it additionally depends on boron carrier uptake and microdistribution. Harada et al. provided important mechanistic insight into the influence of hypoxia on BPA-mediated BNCT, hypoxia upregulates HIF-1αand reduces LAT1 expression, suppresses intracellular boron accumulation, and attenuates BNCT-induced cell killing in multiple tumor cell lines. Importantly, HIF-1α knockdown restored LAT1 expression, and HIF-1 inhibition with YC-1 sensitized hypoxic tumor cells to BNCT. Sanada et al. further showed that disruption of HIF-1α prevented the hypoxia-induced reduction in BPA uptake and SLC7A5 expression, thereby restoring BNCT sensitivity under hypoxic conditions, suggesting that the limitation imposed by hypoxia is not just oxygen-independent radiation biology but also impaired boron delivery.

Accordingly, the potential advantage of BNCT in hypoxic tumors may be considered conditional. In tumors where sufficient boron delivery can still be achieved, BNCT may preserve antitumor efficacy more effectively than photon radiotherapy under hypoxic conditions. This provides a rationale for exploring BNCT-based combination strategies in hypoxic, immunotherapy-refractory tumors, while also underscoring the importance of improving boron delivery in poorly oxygenated tumor regions.

BNCT induces less lymphopenia and depletion of immune effector cells

Lymphopenia is a frequent adverse effect of conventional radiotherapy. Hematopoietic stem and progenitor cells in the bone marrow are highly radiosensitive; thus, even relatively low radiation doses can result in transient bone marrow suppression, whereas higher doses may cause irreversible impairment of hematopoietic function [64]. Bone marrow toxicity is one of the major limiting factors in delivering high-dose external-beam radiotherapy and presents a challenge when radiotherapy is combined with immunotherapy, as the depletion of circulating lymphocytes reduces systemic immune competence. In contrast, BNCT is associated with substantially lower rates of lymphocytopenia and immune effector cell depletion due to its highly localized dose deposition. Thermal neutrons selectively cause high-LET damage only in cells that have accumulated sufficient ¹⁰B, thereby minimizing exposure to the surrounding normal tissues, including bone marrow. Clinical data support this favorable toxicity profile. In the JHN002 trial [14], grade 3 adverse events occurred in 5 of 21 patients (24%), with grade 3 lymphopenia observed in only two patients (10%). Similarly, in a real-world analysis of 150 patients with head and neck cancer treated with BNCT, acute anemia occurred at grade 1–2 in only 0.6% of patients and grade 3 in 1.2%, with no cases of late-onset anemia reported [74]. Mechanistically, this concept is further supported by recent preclinical evidence showing that BNCT has minimal effects on immune cell viability compared with dose-equivalent X-ray irradiation, while preserving immune populations and promoting systemic antitumor immunity. Although these findings primarily reflect immune cell sparing within the tumor context rather than direct measurement of marrow toxicity or circulating lymphocyte kinetics, they are consistent with the broader hypothesis that BNCT may preserve immune effector cells more effectively than non-selective radiotherapy [56]. These findings suggest that BNCT may better preserve systemic immune competence than conventional radiotherapy, although direct evidence specifically addressing circulating lymphocyte kinetics and marrow exposure remains limited and this advantage critically depends on minimal boron accumulation in normal tissues. Bone marrow, in particular, remains highly sensitive to radiation, and inadvertent boron uptake could negate the sparing effect of BNCT. Careful optimization of boron pharmacokinetics and biodistribution, therefore, remains essential for maximizing the immunological compatibility of BNCT.

Potential role of BNCT in overcoming immunotherapy resistance

As outlined previously, both conventional radiotherapy and BNCT induce the cytosolic accumulation of double-stranded DNA, which activates the cGAS–STING pathway. Activation of this pathway promotes IFN-β secretion, maturation of dendritic cells, enhanced antigen presentation, and subsequent recruitment and activation of T cells, events that collectively contribute to antitumor immunity.

However, persistent or low-dose fractionated irradiation can chronically activate the cGAS–STING pathway, shifting its role from immunostimulatory to immunosuppressive [75]. Chronic, low-grade production of type I interferons and NF-κB–dependent proinflammatory cytokines creates a tumor-promoting inflammatory milieu that facilitates cancer progression and metastasis [76]. In parallel, sustained endoplasmic reticulum stress may disrupt intracellular calcium homeostasis, priming T cells for apoptosis, increasing the accumulation of MDSCs, and suppressing IL-10 production [77]. In addition, the ectoenzyme ENPP1 (CD203a), which degrades extracellular cGAMP, the second messenger produced by cGAS, promotes metastasis and immune escape and is associated with resistance to ICIs [78]. Complicating this picture, IFN-γ produced by TAMs can activate cGAS–STING signaling in cancer cells; in the absence of effective immune checkpoint blockade, this macrophage-derived IFN-γ can paradoxically drive tumor-promoting inflammation rather than antitumor immunity [79].

Collectively, these mechanisms highlight chronic cGAS–STING-associated resistance in broader RT literature. This reinforces the need for therapeutic strategies that produce strong, transient, and immunogenic activation of cGAS–STING rather than persistent low-level signaling. These features position BNCT as a unique radiotherapy modality capable of overcoming resistance to ICIs by balancing type I interferon dynamics, promoting ICD, and reprogramming the tumor immune microenvironment. As we mentioned in the previous section, BNCT activates cGAS/STING signaling with high intensity and short duration rather than a chronic “high-on” state. However, direct evidence specifically linking BNCT-induced cGAS-STING kinetics to the reversal of ICI resistance is still preliminary. Even so, available preclinical studies have demonstrated the ability of BNCT to reinvigorate antitumor immunity and reverse key features associated with ICI resistance. These findings provide a compelling rationale for integrating BNCT with ICIs to overcome intrinsic or acquired immunotherapy resistance (Supplementary Fig.S2). However, direct evidence specifically linking BNCT-induced cGAS–STING dynamics to reversal of ICI resistance remains preliminary and requires further experimental validation.

Current preclinical research on BNCT combined with immunotherapy and future clinical perspectives

Research on BNCT combined with immunotherapy remains largely preclinical, with most studies published within the past 3 years. A growing body of evidence supports the synergistic potential of integrating BNCT with ICIs, TME modulators, and immune-stimulatory agents. Initial efforts have focused on PD-1/PD-L1 blockade. Fujimoto et al. demonstrated that BNCT combined with anti-PD-1 therapy overcame ICI resistance in melanoma, elicited abscopal responses, and required CD8⁺ T cells for therapeutic efficacy. Chiu et al. showed that a polyboronate ester micelle delivering ¹⁰B significantly enhanced CD3⁺, CD4⁺, and CD8α⁺ T cell infiltration and prolonged tumor growth delay when combined with anti-PD-L1 therapy. More recently, Deng et al. [46] developed ¹⁰B/siPD-L1 nanoparticles capable of silencing both basal and radiation-induced PD-L1 expression. These nanoparticles augmented BNCT-induced DNA damage and ICD, preserved T cell populations, and inhibited metastatic progression. Collectively, these findings indicate that PD-1/PD-L1 blockade, whether achieved through monoclonal antibodies or RNA interference, synergizes with BNCT by enhancing antigen exposure, restoring T cell functionality, and strengthening systemic antitumor immunity.

In addition to adaptive immune checkpoint blockade, several studies have explored the immunosuppressive TME. Colony-stimulating factor 1 receptor (CSF1R), a key regulator of myeloid cell survival and recruitment, has been inhibited to reduce tumor-associated macrophages (TAMs) and myeloid-derived suppressor cells (MDSCs) using PLX-3397 (pexidartinib, a selective CSF1R inhibitor), Chang et al. reported reductions in monocytic MDSCs and TAMs, decreased CD68⁺/F4/80⁺ cell populations, increased CD8⁺ T cell infiltration, and extended survival by up to 30 days [80]. Other investigators utilized CD47-SIRPα blockade, a strategy targeting innate immune checkpoints, together with BNCT, employing nucleus-targeted DOX–CB liposomes to co-deliver CRISPR/CD47 constructs. This approach tripled nuclear boron accumulation compared with BSH and significantly increased dendritic cell maturation (elevated CD80/CD86), CD8⁺ T cell infiltration, and proinflammatory cytokine production (IL-1β, IL-12, TNF-α) while reducing the levels of cancer stem cell markers such as Nestin and CD133.

These effects collectively relieved phagocytic inhibition, enhanced macrophage-mediated tumor cell clearance, and promoted M1 macrophage polarization [81].

Additional strategies have centered on direct immune stimulation. Toll-like receptor 7 (TLR7, a key pattern recognition receptor that triggers innate immune responses) agonists and “BNCT-shocked tumor cells” (BTCs, i.e., tumor cells exposed to boron neutron capture therapy that undergo immunogenic cell death) have been employed as in situ vaccines to promote dendritic cell maturation. Chen et al. [81] demonstrated that TLR7 stimulation increased ICD hallmarks, including CRT exposure, ATP release, and HMGB1 translocation, and robustly enhanced dendritic cell activation. Integration of immune modulation with next-generation nanoplatforms has shown particular promise. PEG-B-COF, combined with the TLR7 agonist R837, enabled the burst release of the immunostimulant upon neutron capture, thereby amplifying both dendritic cell maturation and BNCT-associated bystander effects. [27]

Classical immunoadjuvants, such as Bacillus Calmette–Guérin BCG [64], oligo-fucoidan, and glutamine [82, 83], have each shown the capacity to enhance antitumor immune responses combined with BNCT in preclinical settings. These adjuvants increased CD8⁺ T cell levels (e.g., glutamine), enhanced NK-cell infiltration, and generated notable abscopal responses (with BNCT + BCG producing ~ 30% higher abscopal activity than controls). Extracellular vesicle-based approaches further inhibited both primary and metastatic lesions and induced durable immune memory.

Together, these preclinical studies highlight several complementary strategies, including checkpoint inhibition, TME modulation, immune stimulation, and adjuvant-assisted vaccination, that synergize with BNCT to enhance ICD, reshape the tumor immune landscape, and strengthen systemic antitumor immunity. These findings provide a compelling rationale for advancing BNCT–immunotherapy combinations toward clinical translation. To date, no published clinical trials have combined BNCT with immunotherapy; all existing evidence is based on preclinical studies. While these findings are promising, this combination is still in the early stages of clinical translation. Key challenges will be discussed in the discussion section.

Discussion

This review highlights the growing body of evidence indicating that BNCT has substantial potential as a partner for immunotherapy. Several recurring mechanistic themes have emerged across the preclinical and early clinical findings. First, the high-LET microlesions generated by the ¹⁰B(n,α)⁷Li nucleus reaction produce dense, spatially confined DNA damage and robust features of ICD. These include the exposure or release of CRT, heat shock proteins (e.g., HSP90), HMGB1, and ATP, signals that promote dendritic cell maturation, antigen cross-presentation, and the recruitment of CTLs. Second, under conditions where sufficient boron delivery can be achieved, BNCT retains advantages over photon radiotherapy in hypoxic tumors because of its low oxygen enhancement ratio (OER ≈1.2) and track-structure characteristics, enabling more efficient tumor cell killing and immune priming even in hypoxic regions that typically resist conventional radiotherapy. However, this advantage is conditional, as hypoxia can upregulate HIF-1α and reduce LAT1 expression, thereby suppressing boron accumulation via BPA and attenuating BNCT cytotoxicity. Third, multiple clinical reports describe a markedly lower incidence of lymphocytopenia and immune effector cell depletion following BNCT compared with photon radiotherapy, consistent with BNCT's highly selective microdosimetry, which limits marrow exposure and preserves circulating immune cells—an essential requirement for effective immune checkpoint blockade. Collectively, these observations support the exploration of BNCT-based combination strategies in hypoxic, immunotherapy-refractory tumors, while underscoring the need to improve boron delivery in poorly oxygenated tumor regions.

A central mechanistic question concerns how BNCT remodels the TME. BNCT can induce cytosolic accumulation of double-stranded DNA and activate the cGAS–STING pathway, leading to type I interferon production and downstream immune priming. However, cGAS–STING biology is highly context-dependent. Dose per fraction, LET, number of fractions, and, critically, the timing of BNCT relative to immune checkpoint blockade all influence whether downstream responses favor productive immune activation or lead instead to adaptive resistance. High-LET radiation delivered in fewer, larger fractions may induce cGAS–STING signaling in a manner that generates an “inflamed-but-suppressed” state. In addition, BNCT-induced cellular stress can increase PD-L1 expression and recruit immunosuppressive myeloid populations, including TAMs and MDSCs. Taken together, these findings provide a mechanistic basis for combining BNCT with immunotherapy while underscoring the need for careful immunodynamic optimization. Despite BNCT's ability to induce ICD, release tumor antigens, and reshape the TME, clinical evidence on integrating BNCT with ICIs remains scarce.

The translation of these promising preclinical concepts into clinical practice requires several bottlenecks to be addressed. The first relates to patient selection. Although BNCT is currently available at roughly 100 facilities worldwide (approximately 40 in the United States, 24 in Japan, 7 in Germany, and 5 in Russia), access remains limited, and most clinical experience focuses on head and neck cancers and melanoma. This constrains the pace at which BNCT–ICI trials can be conducted. Biomarker heterogeneity presents an additional challenge. In routine ICI use, PD-L1 is evaluated using multiple scoring systems, including the Tumor Proportion Score (TPS), Combined Positive Score (CPS), and immune cell scoring approaches, with indication-specific thresholds (e.g., TPS ≥ 1% in non-small cell lung cancer, CPS ≥ 1 in recurrent/metastatic head and neck cancer, CPS ≥ 10 in advanced esophageal cancer). Whether these established criteria are meaningful in the BNCT–ICI setting is unknown. As BNCT expands into new disease types, it will be essential to determine whether existing PD-L1 scoring systems should be adopted or whether BNCT-specific criteria, potentially incorporating boron microdistribution and immune contexture, are required. Prospective clinical datasets are needed to clarify this issue.

A second challenge concerns treatment timing and sequencing. The optimal scheduling of BNCT and ICIs remains undefined, and the relevant immune biology is inherently time-sensitive. After BNCT, the kinetics of DAMP release, type I interferon signaling, dendritic cell activation, and T cell clonal expansion likely evolve over days to weeks; however, comparative data across synchronous versus sequential strategies are lacking. Safety considerations add further complexity. BNCT-related mucosal and cutaneous toxicities may overlap with, and even potentiate, immune-related adverse events such as thyroiditis, pneumonitis, hepatitis, and colitis. Developing approaches to manage irAEs without blunting antitumor immunity will be critical, and the comparatively favorable toxicity profile of BNCT could represent a clinical advantage if integrated appropriately with ICIs.

The third challenge concerns what to deliver. Transitioning next-generation boron carriers from conceptual development to clinical application exposes a substantial translational gap. Key obstacles include manufacturability, achieving a sufficient and quantifiable boron payload, ensuring uniform particle size distribution, maintaining physicochemical stability, and controlling release profiles. Although all boron agents share the fundamental objective of delivering ¹⁰B to tumor cells, differences in payload capacity, molecular structure, targeting specificity, tissue retention, and release kinetics may differentially influence cellular responses and immune regulation after neutron irradiation. This raises the compelling, but still unvalidated, hypothesis that distinct boron delivery platforms may modulate the radiobiological and immunological outcomes of BNCT in clinically meaningful ways. Identifying which classes of boron carriers are most suitable for combination with immune checkpoint inhibitors—without generating antagonistic interactions—remains an essential prerequisite for rational clinical development.

It is important to acknowledge the limitations of the current synthesis. Much of the mechanistic understanding summarized here derives from in vitro and animal models, and extrapolation to human disease should be undertaken cautiously. For clarity, Supplementary TableS1 compiles key relationships among boron delivery systems, their radiobiological and immunological effects, and the experimental contexts in which they were evaluated. Nonetheless, prospective clinical validation is indispensable.

In summary, these interdependent constraints—patient selection, treatment sequencing, and boron carrier optimization—substantially hinder the advancement of BNCT–immunotherapy combinations. Addressing these barriers will require coordinated progress in several domains: refined patient stratification grounded in boron pharmacodynamics and microdistribution; timed immune profiling aligned with clinically actionable decision points; and clinically robust boron delivery platforms that ensure verifiable, on-target ¹⁰B exposure in vivo. Only through such integrated efforts can the full therapeutic potential of combined BNCT and immunotherapy be realized.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ATP

Adenosine triphosphate

BNCT

Boron neutron capture therapy

BPA

Boronophenylalanine

BSH

Borocaptate sodium

cGAMP

Cyclic GMP–AMP

cGAS

GMP–AMP synthase

COF

Covalent organic framework

CRT

Calreticulin

CTLA-4

Cytotoxic T lymphocyte antigen-4

CTLs

Cytotoxic T lymphocytes

DAMPs

Damage-associated molecular patterns

DC

Dendritic cell

DDR

DNA damage response

EVs

Extracellular vesicles

FoxO1

Forkhead box O1

HMGB1

High-mobility group box 1

HNSCC

Head and neck squamous cell carcinoma

ICD

Immunogenic cell death

ICIs

Immune checkpoint inhibitors

IFN-α/β

Type I interferons

IG-IMRT

Image-guided intensity-modulated radiotherapy

IL-6

Interleukin-6

IRF3

Interferon regulatory factor 3

LET

Linear energy transfer

MDSCs

Myeloid-derived suppressor cells

MHC-I

Major histocompatibility complex class I

NK cells

Natural killer cells

nSCC

Non-squamous cell carcinoma

ORR

Objective response rate

OER

Oxygen enhancement ratio

PBMCs

Peripheral blood mononuclear cells

PD-1

Programmed cell death protein 1

PD-L1

Programmed death ligand-1

RBE

Relative biological effectiveness

ROS

Reactive oxygen species

SCC

Squamous cell carcinoma

STING

Stimulator of interferon genes

TAMs

Tumor-associated macrophages

TBK1

TANK-binding kinase 1

TCR

T cell receptor

TEMs

Tumor-specific memory T cells

TNF-α

Tumor necrosis factor-α

Author contribution

Conceptualization, M.A.; methodology, software, and data curation, M.A.; writing—original draft preparation, M.A. and Y.S.; writing—review and editing, Y.M., Y.L.; supervision, H.S.; funding acquisition, M.A. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the Japan–China Sasakawa Medical Fellowship and JST SPRING (grant number JPMJSP2124).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Conflict of interest

The authors declare no competing interests.

Ethics approval

Not applicable.

Consent for publication

Not applicable.