Safety, efficacy, pharmacokinetics, and mechanisms of novel 10B‐boronophenylalanine combined with a radio‐frequency quadrupole accelerator in boron neutron capture therapy: A preclinical study in head and neck squamous cell carcinoma Authors: Zhaoyuan Zhang, Junyan Li, Minhua Fan, Huafu Ouyang, Qiubo Chu, Huijie Yan, Yannan Zheng, Linxuan Huang, Gui Bao, Zhenting Ma, Shaobo Huang, Lin Li, Xiao Xu, Zhigang Liu Published: 2026 ABSTRACT AbstractBackground and PurposeBoron neutron capture therapy (BNCT) is an emerging binary‐targeted radiotherapeutic modality with promising potential for treating recurrent or refractory head and neck squamous cell carcinoma. We comprehensively evaluated the safety, efficacy, pharmacokinetics, and mechanisms of action of a novel 10B‐boronophenylalanine (10B‐BPA) combined with a radio‐frequency quadrupole (RFQ) accelerator‐based neutron source using systematic preclinical experiments.MethodsThis systematic preclinical study began by evaluating the drug's in vivo pharmacokinetics, tumor accumulation specificity, maximum tolerated dose, and systemic toxicity to verify its safety profile. Subsequently, its antitumor efficacy was assessed through dose‐dependent experiments, and the cellular and molecular mechanisms of BNCT treatment were investigated.ResultsThe novel 10B‐BPA exhibited rapid and highly specific tumor accumulation, achieving a stable tumor‐to‐blood ratio of approximately 3.5. The treatment resulted in dose‐dependent tumor regression without inducing systemic toxicity, with a maximum tolerated dose of more than 1000 mg/kg. Mechanistic investigations revealed that BNCT triggered a hypoxic stress response, leading to a substantial accumulation of reactive oxygen species and induced DNA double‐strand breaks. This metabolic crisis subsequently activated the hypoxia‐inducible factor‐1 alpha and nuclear factor kappa‐B signaling axes during apoptosis.ConclusionsThe novel 10B‐BPA/RFQ accelerator‐based neutron source BNCT system demonstrates favorable safety, potent antitumor activity, and multi‐target molecular regulatory mechanisms. These findings provide a solid foundation for its clinical translation. INTRODUCTION Boron neutron capture therapy (BNCT) is a dual‐model precision radiotherapy technology that integrates biological targeting with physical destruction, making it particularly suitable for treating recurrent head and neck squamous cell carcinoma (HNSCC). 1 , 2 , 3 , 4 Its core advantage lies in leveraging L‐amino acid transporter 1 (LAT1)‐mediated boron drug accumulation, combined with high linear energy transfer (LET) particles (α and 7Li) generated by neutron irradiation, to precisely damage tumor cells at the single‐cell level while sparing surrounding normal tissues. 5 , 6 Notably, the unique high LET characteristics of BNCT render its cytotoxicity independent of the oxygen enhancement effect, enabling it to effectively overcome microenvironmental hypoxia‐induced radioresistance, which is a major challenge in conventional radiotherapy. 7 Clinical application of BNCT depends on advancements in efficient boron carriers and neutron irradiation technologies. Following the early exploration of first‐generation inorganic boron compounds, the second‐generation agent, boronophenylalanine (BPA), has emerged as a clinical standard owing to its favorable safety profile and targeting capability. 8 , 9 BPA achieves active uptake and specific accumulation in tumor cells by utilizing the highly expressed LAT1 on the tumor cell membrane. 10 Concurrently, neutron source technology has undergone a transformative shift from reliance on nuclear reactors to accelerator‐based neutron sources (AB‐BNCT). 11 Accelerator‐based systems offer distinct advantages, including enhanced safety, compact design, and suitability for hospital‐based clinical deployment. The successful development of medical accelerator‐based neutron irradiation systems in Japan and China in 2020 has marked the entry of BNCT into widespread clinical promotion. 12 Benefiting from these technological breakthroughs, international clinical studies since the 1990s have provided initial evidence of the remarkable efficacy of BNCT in treating recurrent head and neck tumors. 13 , 14 Notably, a 2019 clinical trial based on a cyclotron neutron source demonstrated that BNCT achieved an objective response rate as high as 71% in patients with recurrent head and neck cancer, with a manageable safety profile, significantly promoting the clinical acceptance of this technology. 15 However, as an emerging radiotherapeutic modality, the cellular mechanisms of BNCT, beyond physical DNA damage, have not yet been comprehensively elucidated. More critically, systematic and comprehensive preclinical biological evaluation data are still lacking for the specific combination of the domestic radio‐frequency quadrupole (RFQ) accelerator‐based neutron source and the 10B‐BPA formulation employed in this study. 16 The development of boron agents aims to achieve higher purity and convenience. This involves the evolution from infusion bags to lyophilized powder formulations, which offer enhanced stability, safety, and compatibility with hospital‐based neutron sources. To achieve this, establishing a robust preclinical dataset is essential. Compared with the conventional BPA–fructose solution widely used in Japanese BNCT systems, the lyophilized 10B‐BPA formulation employed in this study provides several potential pharmaceutical and practical advantages. 17 A traditional BPA–fructose formulation is typically prepared as an aqueous infusion that requires strict storage conditions and fresh preparation before clinical use. 18 In contrast, the lyophilized powder form offers improved physicochemical stability and longer shelf life during storage. In addition, this formulation avoids the use of large amounts of fructose as a solubilizing excipient, thereby simplifying formulation composition and potentially reducing the metabolic burden associated with high fructose exposure. 19 From a clinical implementation perspective, freeze‐dried formulations can be conveniently transported, stored, and rapidly reconstituted before administration, possibly enhancing the logistical flexibility and compatibility with hospital‐based accelerator BNCT platforms. 17 Collectively, these features highlight the pharmaceutical innovation and translational potential of the lyophilized 10B‐BPA formulation in the development of next‐generation BNCT systems. To address this gap, we conducted a systematic assessment of the pharmacokinetics, safety, and anti‐tumor efficacy of this domestic system, along with an in‐depth investigation of its molecular mechanisms, thereby providing critical theoretical and experimental support for subsequent clinical trials. METHODS Storage and utilization of 10B‐BPA The high‐purity (≥99%) 10B‐BPA powder (Borofalan (10B), C9H12 10BNO4, molecular weight: 208.21; Hainan Poly Pharm Co., Ltd., China) was stored at 2–8°C. Solutions were prepared 24 h before use by dissolving the compound in phosphate‐buffered saline (PBS) for in vitro assays or sterile injectable water for in vivo studies, followed by sterilization via a 0.22 µm filter. Cell lines Human tongue squamous cell carcinoma cell lines (SAS and CAL27) were obtained from Procell, and human oral keratinocytes (HOK) were obtained from Yaji Biological. Cells were maintained in Dulbecco's modified Eagle medium (DMEM) (cat no. PM150210, PROCELL) supplemented with 10% fetal bovine serum. Cells were subjected to routine mycoplasma testing and authenticated using short‐tandem repeat sequencing. Animal sources All animal experiments were conducted in accordance with the protocols approved by the Animal Management and Use Committee of Anling Biopharmaceutical Co., Ltd. (Shenzhen, China). The protocols were approved under the IACUC numbers 2023‐11‐ALSHZH‐19 (efficacy study) and 2024‐04‐ALSHZH‐17 (single‐agent acute and irradiation‐toxicology studies in rats). Four‐week‐old BALB/c nude mice and six‐week‐old Sprague Dawley (SD) rats were obtained from Guangzhou Beicheng Biotechnology Co., Ltd. Human tumor xenograft For xenograft models, SAS or CAL27 cells in the logarithmic growth phase were resuspended in PBS. Nude mice received a subcutaneous injection of 120 µL cell suspension containing either 5 × 105 SAS cells or 1 × 106 CAL27 cells into the right groin region. Tumor volume, body weight, and food intake were recorded every 3 to 4 days until the endpoint. In vitro and in vivo incubation with 10B‐BPA Cell Treatment: SAS and CAL27 cells in the logarithmic growth phase were incubated with fresh complete DMEM containing 10B‐BPA. Following incubation, cells were harvested and resuspended in the complete medium supplemented with varying concentrations of 10B‐BPA (10, 25, or 40 ppm). The suspensions were sealed in 1.5 mL tubes and positioned for neutron irradiation. Animal Treatment: Tumor‐bearing mice were intravenously injected with 10B‐BPA (500 mg/kg) 1 h before irradiation. The animals were anesthetized with pentobarbital (50 mg/kg, intraperitoneally) 20 min before, immobilized in custom molds, and subjected to neutron irradiation. Neutron irradiation dosimetry methodology All neutron irradiation experiments were conducted using an accelerator‐based neutron source manufactured by Guoke Neutron Medical Technology Co., Ltd. and deployed at the Boron Neutron Capture Therapy Center of Dongguan People's Hospital. The irradiation dose was uniformly quantified and controlled using Monitor Units (MU), where 1 MU is defined under standard conditions (0°C, 760 mmHg) as equivalent to the ionization produced by 5.2 × 108 thermal neutrons in 1 cubic centimeter of air. To precisely control the physical dose delivered to tumor‐bearing mice, we established an individualized dose calculation scheme based on simulations using the Monte Carlo N‐Particle (MCNP) transport code and correction with measured boron concentrations. First, MCNP simulations were used to calculate the physical dose rates contributed by hydrogen capture reactions (1H(n,γ)2H), nitrogen capture reactions (14N(n,p)14C), boron capture reactions (10B(n,α)7Li), and accompanying gamma rays in tumors and critical normal tissues. The boron dose rate was simulated using a reference concentration of 30 ppm 10B. Subsequently, pharmacokinetic studies determined the critical therapeutic time window of 1–2 h post‐administration, during which the average boron concentration in tumors was measured. Next, the boron dose rate was individually corrected using the following formula: corrected boron dose rate (Gy/s) = measured tumor boron concentration (ppm) × MCNP‐simulated boron dose rate (Gy/s)/reference boron concentration (30 ppm). The total dose rate was obtained by adding the hydrogen, nitrogen, gamma, and corrected boron dose rates. The irradiation time required to achieve the target dose (e.g., 2, 4, and 8 Gy) was calculated using the following formula: irradiation time (s) = preset physical dose (Gy)/total dose rate (Gy/s). Finally, based on a pre‐calibrated device conversion factor of 2.22 MU per second, the required MU was determined as required MU = irradiation time (s) × 2.22 (MU/s). For In vitro cell experiments, to streamline the process and ensure reproducibility, cells uniformly received a fixed neutron irradiation dose of 2,610 MU following co‐incubation with boron‐containing drugs. All in vivo and In vitro irradiations strictly followed the above‐calculated or preset MU protocols, thereby ensuring the accuracy and comparability of the delivered neutron dose across different experimental groups. Live/dead cell staining Post‐treatment cells were seeded into six‐well plates (5 × 104 cells/well) and cultured for 48 h. Cell viability was assessed using a live/dead cell staining kit (cat. no. L0465, Beyotime). After 40 min of incubation in the dark at room temperature, the cells were visualized and imaged using an inverted fluorescence microscope. Cell cycle analysis Forty‐eight hours after treatment, cells were fixed in 70% ethanol overnight at 4°C, stained with PI/RNase A for 60 min at room temperature, and analyzed using flow cytometry (488 nm excitation) (cat no. C1052, Beyotime). CCK‐8 assay The cells were seeded in 96‐well plates (500 cells/well, n = 6). At the indicated time points, 10 µL of the CCK‐8 reagent was added to each well. Following a 2 h incubation in the dark, the absorbance at 450 nm was measured using a microplate reader (cat no. A311‐01, Vazyme). Plate clone formation assay Cells were seeded into six‐well plates (500 cells/well) and cultured for 14 days, with medium replacement every three days until macroscopic colonies appeared. The colonies were fixed with 4% paraformaldehyde (cat no. G1101, Servicebio) for 30 min and stained with crystal violet (cat no. GC307002, Servicebio) for 30 min. Colonies containing > 50 cells were counted to calculate the colony formation efficiency. Immunohistochemistry and hematoxylin and eosin staining (HE) Formalin‐fixed paraffin‐embedded tissue sections (4 µm) were deparaffinized and rehydrated. Antigen retrieval was performed for immunohistochemistry, followed by blocking and incubation with primary antibodies (cleaved caspase‐3, Ki67) and horseradish peroxidase (HRP)‐conjugated secondary antibodies. Signal was detected using a DAB substrate kit with hematoxylin counterstaining. For hematoxylin and eosin (HE) staining, parallel sections were processed using hematoxylin (cat. no. G1004, Servicebio) and 0.5% eosin (cat. no. G1002, Servicebio), dehydrated, cleared in xylene, and mounted with neutral gum. Additional reagent details are provided in the Supplement Reagents or Resources. Inductively coupled plasma mass spectrometry (ICP‐MS) Samples were digested with 65% nitric acid at 70°C overnight, filtered (0.22 µm), and analyzed for 10B via ICP‐MS (Thermo Fisher Scientific) using 10B as an internal standard. Quantification was performed using the standard sample bracketing method, with a linear range of 0.01–100 µg/L (R2 > 0.999). High‐performance liquid chromatography (HPLC) 10B‐BPA concentration was quantified via reverse‐phase high‐performance liquid chromatography (HPLC; Thermo Fisher Vanquish) using a C18 column (250 mm × 4.6 mm, 5 µm) at 40°C with isocratic methanol–water (70:30 v/v, 1.0 mL/min). It was detected using fluorescence (Ex 275 nm/Em 313 nm) with external standardization (range: 0.5–50 mg/L; LOD: 0.1 µg/L; LOQ: 0.3 µg/L). Western blotting Proteins were extracted in the radioimmunoprecipitation assay buffer (cat no. FD009, Fudebio) and separated by 10% sodium dodecyl‐polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes (Millipore). After blocking with 5% bovine serum albumin (BSA) (cat no. H1130, Solarbio), the membranes were incubated with primary antibodies overnight at 4°C and HRP‐conjugated secondary antibodies for 1 h at room temperature and subsequently visualized using ECL (cat no. FD8020, Fudebio). Immunofluorescence assay Cells were fixed with 4% paraformaldehyde and permeabilized with 1% Triton X‐100 (cat no. T8200, Solarbio), and blocked with 5% BSA (cat no. H1130, Solarbio). Following overnight incubation with primary antibodies at 4°C, samples were stained with Alexa Fluor 488‐conjugated secondary antibodies and DAPI (cat no. C1002, Beyotime). Images were acquired using a Zeiss confocal microscope and analyzed using the ImageJ software. Additional reagent details are provided in the Supplement Reagents or Resources Transmission electron microscopy (TEM). Specimens were prepared using standard protocols for fixation, dehydration, and embedding in an epoxy resin. Ultrathin sections were subsequently stained with uranyl acetate and lead citrate to enhance contrast. Sections were examined using a Hitachi TEM system operating at an accelerating voltage of 80.0 kV. Representative images were acquired at a magnification of ×8.0 k. Transcriptome sequencing Transcriptome sequencing and analysis were performed by Shanghai Bioprofile Biotechnology Co., Ltd. Briefly, the cells were washed with PBS and lysed directly in TRIzol reagent (cat no. 15596‐026, Invitrogen) without trypsin digestion to prevent RNA degradation. Lysed samples were snap‐frozen in liquid nitrogen, stored at −80°C, and transported on dry ice for subsequent library construction and sequencing. Statistical analysis Quantitative data are presented as mean ± standard deviation (SD) or mean ± standard error of the mean (SEM), as specified in the respective figure legends. All statistical analyses were performed using GraphPad Prism version 9.5 (GraphPad Software, USA). Comparisons between two groups were performed using a two‐tailed unpaired Student's t‐test, whereas comparisons among multiple groups were conducted using one‐way analysis of variance (ANOVA). Statistical significance was set at *p < 0.05, **p < 0.01, and ***p < 0.001. The specific sample sizes (n) and statistical details are provided in legends. RESULTS Pharmacokinetic properties and selective accumulation of 10B‐BPA in vivo We evaluated the pharmacokinetics of 10B‐BPA to delineate the research framework (Figure 1). In SD rats (250–1000 mg/kg), the drug reached its peak plasma concentration (Cmax) within 1 h, followed by rapid metabolism (Figure 2A). 10B‐BPA exhibited selective accumulation in the tumors, pancreas, and kidneys of SAS xenograft mice (500 mg/kg), with near‐complete clearance within 24 h and no gender‐related differences (Figure 2B). Excretion studies in SD rats confirmed renal clearance as the primary route (within 168 h, 62.2% of the total excreted BPA and 82.9% of the total excreted 10B were eliminated via urine) (Figure 2C). Collectively, 10B‐BPA demonstrated rapid tumor accumulation and predominant renal clearance, supporting its translational feasibility for BNCT in head and neck malignancies. Research flowchart. 10B‐BPA exhibits selective accumulation in HNSCC and demonstrates rapid metabolic kinetics in vivo. (A) Plasma concentration–time curve of 10B‐BPA in rats after a single intravenous injection at 250, 500, and 1000 mg/kg. (B) Distribution of 10B‐BPA and 10B‐containing metabolites in different tissues of the SAS HNSCC xenograft model after a single intravenous injection of 500 mg/kg 10B‐BPA. (C) Cumulative excretion of 10B‐BPA and 10B in urine, feces, and bile of SD rats within 168 h after a single intravenous injection of 500 mg/kg 10B‐BPA. (D) Changes in 10B accumulation in HOK, SAS, and CAL27 cells after treatment with different concentrations of 10B‐BPA (one‐way ANOVA, n = 5). Data are presented as mean ± standard deviation. ***p < 0.001. Therapeutic efficacy and safety of RFQ accelerator‐based neutron source BNCT combined with 10B‐BPA in vivo Following confirmation in vivo, we compared uptake of 10B‑BPA (2–8 mM) in normal (HOK) versus tumor (SAS, CAL27) cells In vitro. One hour incubation with 2 mM 10B‑BPA achieved intracellular boron levels in tumor cells exceeding the BNCT threshold (16.67 ng/g); we used this concentration for further analysis. Furthermore, 10B accumulation was consistently higher in tumor lines than in HOK cells (Figure 2D), confirming the specific tumor‐targeting of 10B‑BPA. Subsequently, cytotoxicity was evaluated in cells treated with 10B‑BPA (40 ppm), neutron irradiation (8 Gy‑Eq), or their combination. Live/dead cell staining (Figure 3A) and CCK‐8 assays (Figure 3B) demonstrated that the combination treatment significantly increased cell death and cytotoxicity, whereas monotherapies demonstrated negligible effects compared to the controls. Colony formation assays further revealed the potent suppression of clonogenic capacity by the combination treatment (Figure 3C). Cell cycle analysis indicated pronounced G2/M arrest (Figure 3D), suggesting that BNCT‐induced DNA damage disrupts cell cycle progression. This was confirmed by immunofluorescence showing substantial γ‑H2AX foci aggregation only in the combination group (Figure S1). Collectively, these results demonstrate that BNCT significantly induces DNA damage and inhibits proliferation, with minimal impact from monotherapy. Evaluation of the cytotoxic effect of RFQ accelerator‐based neutron source combined with 10B‐BPA on HNSCC i n vitro. (A) Live/dead cell staining of SAS and CAL27 cell lines under different treatment conditions (calcein‐AM for live cells, green; propidium iodide for dead cells, red). (B) Cell proliferation ability of SAS and CAL27 cell lines after different treatments, measured by absorbance at 450 nm (one‐way ANOVA, n = 5). (C) Colony formation assay of SAS and CAL27 cells in different treatment groups to evaluate long‐term proliferation capacity (one‐way ANOVA, n = 5). (D) Cell cycle distribution of SAS and CAL27 cells in each treatment group was analyzed by flow cytometry, with DNA content quantification distinguishing G0/G1, S, and G2/M phases. Data are presented as mean ± standard deviation. *p < 0.05, ***p < 0.001, nonsense (ns). Therapeutic efficacy and survival benefits of RFQ accelerator‐based neutron source BNCT combined with novel 10B‐BPA in vivo To evaluate in vivo efficacy, SAS and CAL27 xenograft models were established in BALB/c nude mice. As the tumor volume reached 100 mm3, mice were randomized into six groups (n = 10/group, equal sex distribution): the control group, the BNCT irradiation group, the 10B‐BPA (500 mg/kg) group, and three combination groups (10B‐BPA 125, 250, 500 mg/kg + BNCT 2, 4, 8 Gy‐Eq). Macroscopic examination 14 days post‐treatment revealed significant, dose‐dependent tumor regression in combination groups, whereas monotherapies demonstrated no difference from the controls (Figure 4A). The BNCT combination therapy significantly inhibited tumor growth in a dose‐dependent manner (Figure 4D, E), with no sex‐based differences in response. Apart from a slight body weight reduction in the SAS‐bearing mice after BNCT, no significant weight changes were observed in other groups (Figure S2). Survival analysis further demonstrated that the high‐dose combination group (500 mg/kg + 8 Gy‐Eq) had significantly improved overall survival compared to the control group (Figure 4B, C). Histological analyses were performed on tumors from the control and high‐dose groups. HE staining revealed increased vacuolization, improved cellular organization, and reduced nuclear atypia in the combination‐treated tumors, indicating effective tumor cell death and structural disruption. Immunohistochemical analysis revealed a marked decrease in Ki67‐positive cells, reflecting suppressed proliferation alongside upregulated Caspase‐3 expression, consistent with the activation of apoptosis following BNCT treatment (Figure 5A, B). These molecular‐level changes were consistent with the macroscopic results of reduced tumor volume and increased tumor inhibition rate, further confirming the significant anti‐tumor effect of the combination therapy. In vivo therapeutic efficacy of RFQ accelerator‐based neutron source BNCT combined with 10B‐BPA against head and neck squamous cell carcinoma.(A) Gross appearance of subcutaneous tumors in SAS tumor‐bearing mice after treatment (n = 6). (B–C) Survival curves of female and male mice implanted with SAS (B) or CAL27 (C) tumors following respective treatments. (D–E) Dynamic changes in tumor volume and corresponding statistical analysis in female and male mice bearing SAS (D) or CAL27. (E) tumors across different treatment groups (one‐way ANOVA, n = 5). Data are presented as mean ± standard deviation. *p < 0.05, **p < 0.01, ***p < 0.001, nonsense (ns). In vivo antitumor efficacy evaluation of RFQ accelerator‐based neutron source BNCT combined with 10B‐BPA against head and neck squamous cell carcinoma. (A) Representative images of H&E staining and immunohistochemical staining for Ki‐67 and Caspase‐3 in subcutaneous xenograft tumors from control and 10B‐BPA (500 mg/kg) combined with BNCT (8 Gy‐Eq) treated mice bearing SAS or CAL27 tumors. (B) Quantitative analysis of the percentage of Ki67‐positive cells and cleaved caspase‐3‐positive cells in SAS and CAL27 subcutaneous xenograft tumors across different groups (two‐tailed unpaired t‐test, n = 5 per group). Data are presented as mean ± standard deviation. ***p < 0.001. These findings collectively demonstrate that BNCT using the GKZS neutron source combined with 10B‐BPA achieves favorable therapeutic efficacy and survival benefits in vivo. Safety evaluation of RFQ accelerator‐based neutron source BNCT combined with 10B‐BPA in vivo Following efficacy validation, we further evaluated the safety of neutron irradiation combined with 10B‐BPA. In SD rats, 10B‐BPA administration (250–1000 mg/kg) induced no significant abnormalities in body weight, food consumption (Figure 6A, B), or comprehensive parameters, including hematology, biochemistry, necropsy, and histopathology. The MTD was determined to be > 1000 mg/kg. An i n vitro rabbit hemolysis test demonstrated no hemolysis or agglutination at concentrations of 0.6–3 mg/mL (Figure 6C). Collectively, both 10B‐BPA and accelerator‐based neutron irradiation demonstrated controllable safety profiles in the tested systems. Safety assessment of 10B‐BPA. (A) Changes in food intake and quantitative analysis in SD rats after treatment with different doses of 10B‐BPA (one‐way ANOVA, n = 5). (B) Changes in body weight and quantitative analysis in SD rats after treatment with different doses of 10B‐BPA (one‐way ANOVA, n = 5). (C) Results of rabbit red blood cell hemolysis assay: negative control (0.9% sodium chloride solution), positive control (sterile water for injection), and test tubes containing different concentrations of 10B‐BPA (0.6–3 mg/mL). All data are presented as mean ± standard deviation. ***p < 0.001, nonsense (ns). Investigation of molecular mechanisms underlying BNCT‐induced tumor cell death Transcriptomic sequencing analysis of BNCT‐treated versus control cells revealed the activation of apoptosis‐related pathways, hypoxia‐associated signaling, and nuclear factor kappa‐B (NF‐κB) signaling post‐treatment (Figure 7A, B), suggesting that BNCT may induce hypoxic stress and trigger NF‐κB activation, ultimately leading to apoptosis. RFQ accelerator‐based neutron source BNCT combined with 10B‐BPA induces tumor cell apoptosis through ROS‐mediated NF‐κB pathway activation. (A–B) Transcriptome sequencing results of SAS and CAL27 cells in the control group and BNCT treatment groups: (A) KEGG pathway enrichment analysis. (B) Gene set enrichment analysis (GSEA) plot for the NF‐κB signaling pathway. (C) Transmission electron microscopy (TEM) observation of cellular ultrastructure. Representative TEM micrographs of CAL27 and SAS cells at 24 h post‐BNCT treatment. (D) Expression of proteins related to the NF‐κB and apoptosis pathways in SAS and CAL27 cells after BNCT treatment. (E–F) Assessment of p‐NF‐κB nuclear translocation. (E) Representative immunofluorescence images displaying p‐NF‐κB distribution in SAS and CAL27 cells following BNCT. (F) Quantitative analysis of the rate of p‐NF‐κB nuclear translocation. All data are presented as mean ± standard deviation. ***p < 0.001. TEM provided ultrastructural evidence of this damage, demonstrating that the mitochondria in BNCT‐treated cells were swollen with disordered or dissolved cristae and were vacuolized (Figure 7C), consistent with the onset of intrinsic apoptosis and metabolic stress. Western blotting further demonstrated significant upregulation of hypoxia‐related factors (hypoxia‐inducible factor‐1 alpha HIF‐1α and p53), p‐NF‐κB, and pro‐apoptotic markers (BAX, cleaved caspase‐3), alongside downregulation of the anti‐apoptotic protein BCL‐2 (Figure 7D). Immunofluorescence assays corroborated these findings, revealing enhanced fluorescence intensity for p‐NF‐κB, p65, and HIF‐1α in treated cells (Figure 7E, F; Figure S3A, B). In addition, intracellular reactive oxygen species (ROS) levels increased substantially after BNCT treatment (Figure S3C, D). In summary, BNCT induces a hypoxic microenvironment and mitochondrial dysfunction, upregulates HIF‐1α/p‐NF‐κB, promotes ROS accumulation, and ultimately executes tumor cell death through apoptosis. DISCUSSION The treatment of recurrent HNSCC remains a formidable challenge in oncology. 20 Resistance can easily develop after conventional photon‐based radiotherapy (X‐rays/electrons) due to previous irradiation and the subsequent hypoxia and therapy‐resistant microenvironment within the tumor. 21 Validating a pioneering BNCT system with a domestic RFQ accelerator and a lyophilized 10BPA agent, this study also identified a key radiobiological mechanism underlying the distinct advantages of BNCT over conventional radiotherapy. 22 Conventional radiotherapy depends on “geometric precision,” using physical collimation (e.g., intensity‐modulated radiation therapy [IMRT]) to conform the dose distribution to the three‐dimensional shape of the tumor. 23 , 24 However, this approach inevitably irradiates normal tissue along the beam path. In contrast, the 10B‐BPA/RFQ accelerator‐based neutron source BNCT system evaluated here achieves “cellular‐level precision.” It specifically delivers a high concentration of boron to tumor cells (achieving a tumor‐to‐blood [T/B] ratio of approximately 3.5) by leveraging the high expression of LAT1 transporters in HNSCC cells. 25 , 26 This biological targeting mechanism confines the release of therapeutic alpha particles strictly within the diameter of a single cell (<10 µm), thereby effectively sparing adjacent normal cells that are highly vulnerable within the penumbra of conventional radiotherapy. 10 In addition to comparing BNCT with conventional photon therapy, it is important to consider other high LET modalities, such as proton and carbon ion therapies. 27 , 28 Although protons offer precise physical targeting, and carbon ions provide enhanced biological effects owing to their high LET properties, BNCT uniquely combines both biological and physical selectivity. BNCT delivers cytotoxic alpha particles at the single‐cell level by exploiting tumor‐specific uptake of 10B‐BPA via LAT1 transporters, thereby confining damage strictly within tumor cells while sparing surrounding normal tissue. 6 , 29 This cellular‐level precision is difficult to achieve with proton or carbon ion therapy. Moreover, BNCT offers minimal systemic radiation exposure, can be implemented with relatively compact accelerator‐based neutron sources suitable for hospital deployment, and may entail lower operational costs than carbon ion therapy facilities. These advantages underscore the unique therapeutic potential of BNCT as a clinically feasible, high LET modality that complements existing radiotherapy options. 30 Our data further highlight the core theoretical advantage of BNCT in overcoming hypoxia‐induced radioresistance, which is a key limitation of conventional therapy. Conventional photon therapy is a form of low LET radiation, whose cytotoxic effect depends heavily on oxygen to fix DNA damage (the oxygen enhancement effect). 31 In hypoxic tumor cores, such damage is more easily repaired, leading to treatment failure. Conversely, the high LET alpha particles and 7Li nuclei generated in BNCT induce complex, clustered DNA double‐strand breaks via direct ionization. 32 This damage is not only lethal and difficult to repair, but also critically independent of the cellular oxygenation status. Our in vivo data empirically confirmed the potent cytotoxicity of high LET radiation in hypoxic environments, as demonstrated by significant tumor regression in xenograft models typically refractory to standard radiotherapy. 32 In addition to confirming the established physical advantages, this study adds a new dimension to BNCT radiobiology. At the mechanistic level, this challenges the conventional view that BNCT depends primarily on physical killing by high LET particles. Notably, the ROS/HIF‐1α/NF‐κB pathway represents a key mechanistic novelty of BNCT. 33 In conventional photon‐based radiotherapy, ROS are primarily generated by the radiolysis of water, and their cytotoxic effects are strongly dependent on oxygen availability. In contrast, BNCT induces a rapid and localized burst of ROS directly within boron‐enriched tumor cells, even under hypoxic conditions. 34 This acute ROS accumulation creates a pseudo‐hypoxic tumor microenvironment that aberrantly activates HIF‐1α and downstream NF‐κB signaling. Consequently, BNCT not only promotes DNA damage via high LET particle interactions but also amplifies oxidative stress‐mediated apoptotic signaling, a mechanism largely absent in conventional radiotherapy. ROS‐induced stress leads to acute mitochondrial dysfunction and triggers an irreversible caspase‐3‐mediated apoptotic program. 35 , 36 This mechanistic synergy, combining physical particle‐induced damage with metabolic stress, represents a novel mode of tumor cell killing that may effectively overcome hypoxia‐associated radioresistance and prevent adaptive resistance commonly observed after radiotherapy. Although this study confirmed the therapeutic potential of the domestic RFQ accelerator‐based neutron‐source BNCT system combined with lyophilized BPA, its limitations warrant careful consideration. First, existing xenograft mouse models do not comprehensively recapitulate the complex microenvironment of clinically refractory/recurrent tumors. 34 Therefore, the efficacy observed in these models cannot be extrapolated to patients with recalcitrant or recurrent diseases. Secondly, this study focused solely on HNSCC, whereas BNCT is theoretically applicable to a broader spectrum of malignancies. In the future, its efficacy needs to be validated in additional tumor models. Finally, regarding BNCT's killing mechanism, the ROS/HIF‐1α/NF‐κB apoptotic axis we revealed is likely just the tip of the iceberg of its complex radiobiological effects. 37 Unlike conventional radiotherapy, it remains largely unknown how high LET particles generated in BNCT within milliseconds of physical energy deposition trigger a cascade of events ranging from DNA damage repair and metabolic reprogramming to immunogenic cell death. Furthermore, whether BNCT concurrently induces other novel programmed cell death pathways, such as ferroptosis or pyroptosis, along with apoptosis, requires further exploration. 38 Deciphering these unknown molecular networks is crucial for the future optimization of BNCT‐based combination therapeutic strategies. In summary, this study provides solid preclinical evidence confirming that a domestic RFQ accelerator‐based neutron‐source BNCT system combined with lyophilized BPA powder possesses a favorable safety profile and potent anti‐tumor efficacy against HNSCC. We have preliminarily elucidated that its therapeutic mechanism involves not only direct DNA damage but also the induction of hypoxic stress and activation of apoptotic pathways. These findings provide critical theoretical and experimental support, laying the groundwork for imminent clinical translation of this technology. Based on these results, a phase I clinical trial (no. ChiCTR2500104608) was initiated to evaluate its safety and preliminary efficacy in human patients. CONFLICT OF INTEREST STATEMENT The authors declare no competing interests. ETHICS STATEMENT All animal experiments in this study were approved by the Institutional Animal Care and Use Committee and strictly followed national and institutional animal welfare guidelines. The cell lines used in this study were sourced from commercial channels or obtained through legal authorization, complying with relevant ethical regulations. Supporting information Supporting Information Supporting Information Supporting Information Supporting Information Supporting Information