DNA is the primary target of radiation damage, whether caused by phonons, protons, heavy ions, or neutrons (14, 15). γH2AX serves as a key marker for DNA double strand breaks (DSBs), initiating the recruitment of DNA repair proteins and playing a crucial role in maintaining genomic stability after irradiation (16). Masutani et al. reported an increase in γH2AX at 6 h after BNCT in a lymphosarcoma model. γH2AX and poly (ADP-ribosylation) (PAR) staining persisted at 20 h after BNCT (17). DNA damage increases with increasing radiation LET (18). High-LET particle components principally induce direct damage to DNA, causing irreparable DSBs, referred to as ‘complex DSBs’. Low-LET radiation primarily causes indirect, reparable DNA single-strand breaks (SSBs) (1, 14, 19–21). Epithermal neutrons can cause more than 50% of the DNA strands to break, and with increasing ¹⁰B concentration, more DNA strands break. Compared with photon RT, BNCT produces larger and more complex micronuclei in tumor cells (22, 23). BNCT produced a significantly larger focus size of γH2AX than phonon treatment did in a thyroid follicular cancer cell line (12).

DNA damage activates the DNA repair system (24). In mammalian species, DSBs are repaired through nonhomologous end joining (NHEJ) in most cases. Natsuko Kondo et al. reported that BNCT-induced DNA damage can be partially repaired by the NHEJ repair protein DNA ligase IV (25). Ku70 is crucial for NHEJ, a faster but less accurate repair pathway primarily active in the G1 phase. In contrast, Rad51 and Rad54 are integral to homologous recombination (HR), a high-fidelity repair mechanism active in the S and G2 phases of the cell cycle (12, 14). Rodriguez et al. reported that the mRNA expression of Rad51 and Rad54 increased, but that of Ku70 did not significantly change (12). Perona et al. reported that Ku70 expression increased at different times after irradiation with neutrons but decreased after BNCT (neutrons plus BOPP). This decrease in Ku70 expression after BNCT explains the increase in sensitization to radiation in the BNCT treatment group (26).

DNA damage naturally initiates cell cycle checkpoints, which provides cells with the time required for repair or to decide on programmed cell death if damage is irreparable (1, 27, 28). Cell cycle analysis revealed that BNCT induced G2/M arrest at 24 and 48 hours after irradiation (26). G2/M arrest has been associated with specific regulatory cyclin B1 (proteins associated with G2 arrest), and an inhibition or a delay in the activation of CDK1 (29, 30). Similarly, Sun et al. found a decreased expression of cyclin B1 and CDK1 proteins after BNCT (31). Fujita et al. and Kamida et al. also reported that Wee1, cdc2, and cyclin B1 were altered in the oral squamous cell carcinoma (OSCC) cell line SAS. Caspase 3 induces both G1 and G2 arrest, whereas apoptosis is related to G1 arrest (18, 32). BNCT inhibits OSCC cells in both p53-dependent and p53-independent manners. P53 is necessary for G1 arrest-associated apoptosis (1, 18, 32, 33). However, Seki et al. reported that DNA damage induced by BNCT was not dependent on p53 function (33). Wang et al. reported that the BNCT-induced apoptosis of glioma cells may be associated with Bax activation and Bcl-2 downregulation (34). However, Aromando et al. suggested that apoptosis may not play a significant role in BNCT-induced tumor control (35). To date, the role of apoptotic machinery after BNCT is still being explored.

Cells in S/G2/M phase having higher uptake of boronophenylalanine (BPA) than that in the G1/S phase due to metabolic activity (36). This effect was stronger with BPA than with borocaptate sodium (BSH), as BPA relies on cellular uptake, whereas BSH is a diffusion drug (36–38). Hypoxia-inducible factor 1α (HIF-1α) mediates adaptive responses to hypoxia and controls L-type amino acid transporter (LAT1) expression in hypoxic tumor cells (39, 40). In a lymphosarcoma model, high mobility group box-1 (HMGB1) levels increased 6 hours after BNCT and decreased at 20 hours (17). Unexpectedly, BNCT increased the metastatic potential of high-grade gliomas. This effect occurs because of bystander effects in adjacent cells; that is, BNCT can induce mutations in normal cells near boron-containing tumor cells, and NF-κB may be involved in the response (41). BNCT altered the extracellular matrix by decreasing collagen synthesis and elevated the levels of the tumor necrosis factor (TNF) receptor and cleaved caspases 3, 7, 8 and 9 in melanoma. These findings suggest that multiple pathways associated with cell cycle arrest and apoptosis are involved in the treatment of tumors by BNCT (42).

High-grade glioma is a category of aggressive primary brain tumors with limited therapeutic options and poor prognosis (88–90). Although glioblastoma (GBM) rarely metastasizes to other organs, it exhibits a highly infiltrative growth pattern (44). Traditional photon RT cannot kill infiltrating GBM cells, as the radiation dose required to eliminate tumor cells would also induce necrosis in the surrounding healthy brain tissue (44, 91, 92). Gliomas, account for most attempts to use BNCT in clinical settings, using the terminally differentiated nature of neurons to the advantage of a therapy based on lethal-upon-replication genome damage (13). GBM was chosen as the initial clinical target for phase I and II trials of BNCT (44). In 1951, Sweet and Javid reported the first case at the Brookhaven Graphite Research Reactor for primary brain cancer treated with BNCT (93). Forty brain tumor patients subsequently participated in the clinical trial. Unfortunately, patients experienced serious side effects, including scalp radiation damage, brain radionecrosis, cerebral edema and intractable shock, because the penetration force of the neutron sources used for treatment at the time was weak and the targeting of the boron agents was poor (94–99). Hence, the US completely discontinued clinical trials of BNCT in 1961. Hatanaka continued this research in Japan. In 1990, he reported that 120 patients with Grade III-IV gliomas whose tumors were within the limits of maximum therapeutic depth had a very satisfying 5-year survival rate of 58% (100). Four years later, he reported that 9 patients had lived longer than 10 years (101). This was an unexpected result and encouraged researchers to proceed with BNCT studies. In the 1990s, the USA initiated several clinical trials of BNCT with BPA and epithermal neutron beams (44–46). From September 1994 to May 1999, fifty-three primary GBM patients at the Brookhaven National Laboratory received BNCT after surgery via one, two or three irradiation fields. The median survival times (MST) were 14.8, 12.1 and 11.9 months, respectively. Extended exposure to thermal neutron beams was linked to increased neurotoxicity but was not positively correlated with improved local control or survival. This is indirect proof that BNCT has a greater advantage at low neutron irradiation doses (44, 45). The EORTC trial 11961 was launched in Germany in 1997 and included 26 GBM patients treated with BSH-based BNCT. This trial demonstrated the safety of BSH for clinical application at a dose rate of 1 mg/kg/min and a dose of 100 mg/kg. However, cerebral radiological changes, such as cerebral atrophy and white matter changes, appeared in half of the patients within the first year after BNCT (48–50). In the same period, intraoperative NCT and external beam NCT were compared in Japan. They found the MST of the two groups were 23.3 and 27.1 months, respectively (59). Twenty-two selected malignant glioma (MG) patients with progression after surgery and traditional RT entered a phase I study (NCT00115440) in Finland between 2001 and 2008. The MST after BNCT was 7 months, and the 1-year overall survival (OS) was 36% (52). In 2001, Sweden carried out two clinical studies in which the infusion of BPA was increased to 900 mg/kg body weight and was administered via a 6-hour intravenous infusion to increase the boron concentration in tumor cells. One study included 30 GBM patients, 27 of whom underwent debulking surgery, and reported that the boron concentration in the blood during irradiation ranged from 15.2-33.7 µg/g. Although the efficacy of BNCT was comparable to that of conventional photon RT, it worsened quality of life. Interestingly, patients treated with temozolomide (TMZ) at recurrence had a longer survival rate (17.7 months) than did those treated with BNCT alone (11.5 months) (53). The results of the other trial, which included 12 patients with recurrent GBM, were more encouraging. BNCT was reported to be as effective as concentration RT for recurrent GBM (54). Later, radiotherapy plus concomitant and adjuvant temozolomide was shown to be an optimal therapy for GBM (102). J.W. Hopewell et al. and Anja Sander et al. attempted to compare the OS between BNCT and RT+TMZ by reanalyzing the published data. Regrettably, no high-confidence results were found because of high patient heterogeneity across different trials (103, 104). Twenty-one newly diagnosed glioblastoma patients treated with BNCT had an MST of 15.6 months after diagnosis, which was significantly better than that of patients treated with postoperative radiotherapy and chemotherapy. There was a significant prolongation of survival in the BNCT+XRT boost group (MST, 23.5 months after diagnosis) compared with the BNCT alone group (MST, 14.1 months after diagnosis) (63). On the basis of these findings, a multicentric phase II clinical study was performed in Japan (NCT00974987), which planed enrolled 32 participants treated with BNCT, X-ray radiation treatment and TMZ. The outcome was not reported. In addition to TMZ, bevacizumab (BV) is an efficient antitumor drug for GBM. Twenty-five GBM patients with recurrent malignant glioma who were treated with BNCT and BV achieved prolonged OS and progression-free survival (PFS) compared to those achieved in prior BNCT-only studies. In addition, combining bevacizumab with BNCT may mitigate adverse effects such as pseudo-progression and radiation necrosis (64, 65). A multi-institutional, open-label, phase II clinical trial for 27 recurrent MG patients was conducted with the abovementioned accelerator-based BNCT system (JG002). BV-naïve MG patients who experienced recurrence after standard treatment were enrolled between February 2016 and June 2018. In that study, the 1-year survival rate was 79.2% and the MST was 18.7 months in recurrent GBM patients, whereas those of JO22506 patients were 34.5% and 10.5 months, respectively (https://meetings.asco.org/abstracts-presentations/190090). These results are exciting, but the monitoring efficacy of magnetic resonance imaging cannot provide sufficient information on the biological features of the tumor to identify pseudo-progression. Therefore, the use of ¹⁸F-BPA-PET to monitor treatment efficacy and evaluate patient prognosis is promising. If the trial is successful, recurrent MG may have a new indicated therapy (JPRN-UMIN000022850).

With successful clinical trials for the treatment of MG underway in the 1990s, researchers began to focus on other cancers. BNCT has achieved great success in treating HNC. Before the advent of immunotherapy, therapeutic approaches for recurrent or locally advanced HNC included only RT, platinum drugs and cetuximab. Thus, there is much room for the development of BNCT to treat HNC. The first patient with recurrent HNC received BNCT in 2001. In the 2000s, several clinical trials were performed in Finland. In a prospective, single-center phase I-II study (NCT00114790), twelve patients with locally advanced inoperable HNC were treated with BNCT. The outcomes included partial response (PR), complete response (CR) in 7 patients, and stable disease (SD) in 2 patients, and the 1-year OS was 66.7%. BNCT was thus shown to be an effective and safe treatment modality for locally advanced inoperable HNCs that recur at previously irradiated sites (67). Another clinical trial (NCT00114790) involving 30 patients with inoperable, locally recurrent HNC also utilized BNCT. The MST after BNCT was 13 months, and the 2-year OS was 30%. BNCT has shown efficacy in treating patients with cancer recurrence at previously irradiated sites, although recurrence remains common (51, 68, 69). The EORTC 11001 study explored the feasibility of BNCT for head and neck squamous cell carcinoma (HNSCC). Prior to the planned tumor resection, three patients received BSH, and three received BPA (71, 72). At that time, several trials were performed in Japan that combined BSH and BPA. There was a cohort of 62 patients, and primary severe Grade 3 or 4 toxicities were manageable (75, 81). In 2008, the dream of “from reactors to accelerators” came true. One interesting trial, including 9 patients with recurrent HNC, was performed in Taiwan, China, in the era of intensity-modulated radiation therapy (IMRT) and suggested that BNCT combined with compensated IMRT can increase treatment homogeneity and conformity compared with BNCT alone, particularly for tumor volumes exceeding 100 cm³, and may improve local tumor control (86). The JHN002 trial included patients with recurrent SCC or with recurrent/locally advanced non-SCC (R/LAnSCC). The ORR for all patients was 71%, and the 2-year OS rates for R-SCC and R/LAnSCC were 58% and 100%, respectively (81). Because of the success of JHN002, the Japanese government approved accelerator-based BNCT equipment, boropharan and health insurance coverage of BNCT for HNC in 2020 (4, 105). Under the Japan National Health Insurance System, a retrospective analysis investigated the first 47 patients treated with BNCT between May 2020 and February 2021 in Japan. All patients had undergone RT. The minimum dose administered to the tumor was 27.4 Gy-Eq, with a range of 13.3–45.2 Gy-Eq. The overall survival rates at 1 and 2 years were 86.1% and 66.5%, indicating high degrees of efficacy and safety (82). Similarly, the other retrospective study included 36 hypopharyngeal/laryngeal cancer patients with prior head and neck irradiation. The CR rate was 72%, and the objective response rate (ORR) was 84%. The MST was 15.5 months, and the 2-year OS was 79.8%. No acute G4–5 adverse events (AEs) were observed except for hyperamylasemia, and no late-phase G3 or higher AEs occurred. This finding demonstrates again that BNCT can achieve a good tumor response while preserving the larynx without severe AEs (83). In addition, a prospective phase I/II trial enrolled 14 patients, and 12 patients received combined treatment. The median BNCT average dose for the gross tumor volume (GTV) was 21.6 Gy-Eq, and the median image-guided intensity-modulated radiotherapy (IG-IMRT) dose for the PTV was 46.8 Gy/26 fractions. The 1-year OS and local PFS rates were 56% and 21%, respectively. Although the trial showed a high response rate (64%), it also experienced a significant incidence of in-field and marginal failure. Future research could explore combining BNCT with non-radiation modalities (87).

Proton therapy is a promising radiotherapy modality for treating deep-seated and unresectable tumors. However, its biological advantages have not yet been addressed. In 2014, Do-Kun et al. first proposed the idea of proton-boron capture therapy (PBCT), a modality that combines the concepts of proton RT and BNCT (106). Four years later, Cirrone et al. experimentally tested this idea for the first time by using the p+¹¹B→3α nuclear fusion reaction to increase the biological effectiveness of protons (107). Two subsequent reports demonstrated that osteosarcoma cells (108), prostate cancer cells and glioblastoma cells (109) treated with PBCT exhibited reduced survival and increased chromosomal aberrations compared with those treated with protons alone. Furthermore, Manandhar et al. first reported the use of DSBs as a surrogate measure of the dose enhancement effect of alpha particles arising from the proton–boron reaction. They discovered that BSH radiosensitized cells to protons, but this effect was independent of DNA damage (110).

CIRT is a therapeutic modality that relies on the Bragg peak of carbon ions to achieve precise and conformal dose deposition in tumors (111). Theoretically, the combined BNCT–CIRT modality can offer a more homogeneous tumor dose distribution and lower normal tissue toxicity by integrating the biological targeting capabilities of BNCT with the intensity modulation capabilities of CIRT. Han et al. assessed the feasibility and potential advantages of integrating BNCT with CIRT. BNCT–CIRT ensures uniform delivery within the clinical tumor volume (CTV) via the reversed gradient effect from the CIRT component. BNCT–CIRT can thus minimize damage to normal brain tissue and skin (112).

GdNCT is another neutron capture therapy (NCT). BNCT and GdNCT have their own merits and limitations. Higher levels of DNA damage are caused by the release of secondary high-LET particles during BNCT, but improved dose uniformity is expected for the GdNCT technique because secondary particles with lower LET values (electrons and photons) are released during the GdNCT, and ¹⁵⁷Gd has a higher neutron capture cross-section than ¹⁰B, allowing the use of a lower neutron flux for the NCT technique. In addition, the secondary particles released after neutron capture by ¹⁵⁷Gd have long ranges inside the target volume. Therefore, a combination of ¹⁰B and ¹⁵⁷Gd may improve treatment efficiency in terms of dose uniformity and the relative biological effectiveness (RBE) of DNA damage. Shamsabadi et al. reported that the combined Gd/BNCT technique increases tumor coverage at relatively high doses but reduces the RBE model of DSB induction, potentially affecting the clinical efficacy of NCT (113).

The mechanism of ultrasound differs from that of RT (direct DNA damage). First, ultrasound enhances the sensitivity of tumor cells to radiotherapy (114–116). Second, ultrasound with microbubbles is expected to increase BPA uptake in tumor cells, thereby increasing the boron adsorption capacity. Ultrasound-induced cavitation can compromise the integrity of endothelial and tumor cells by disrupting cell–cell junctions and causing leakage of transport molecules from blood vessels into the tumor microenvironment (TME) (117). Notably, previous study has shown that focused ultrasound (FUS), when combined with a microbubble agent, has ability to temporarily disrupt blood-brain barrier (118). Microbubble-based sonoporation has been shown to reduce the expression of P-glycoprotein in the blood-brain barrier in rats (119). This would increase drug penetration and accumulation in the central nervous system (120), which is one of the excellent advantages for the treatment of glioma using BNCT.