Ideal neutrons for BNCT should be radiated at high flux rates, designated energy levels and targeted volumes without unwanted heavy particles. Although the characteristics and purity of the neutrons were subpar with respect to modern standards, experimental nuclear reactors provided the first thermal neutrons that contributed to the proof of concept of BNCT. One decade after World War II, Farr LE and Sweet WH used the Massachusetts Institute of Technology nuclear reactor (MITR) as neutron sources (11). In the 1960s, Hatanaka initiated clinical trials on brain malignancies via the Hitachi Training Reactor and the Musashi Institute of Technology reactor (12). As mentioned above, the results of these trials were far from fruitful. The slow neutrons had a critical disadvantage. Owing to their low energy level, thermal neutrons can only pass through an average of 2 cm in the tissue before they lose the capacity to induce a nuclear reaction. This resulted in intracranial tumor patients often needing debulking resection and the surgical cavity being left open when receiving the external beam, which might have led to brain necrosis by increasing the boron concentration at exposed incisions (13). In addition, the short path length of thermal neutrons has limited the noninvasive application of BNCT to melanomas and other superficial neoplasms.

Reactors remained the mainstream neutron source until the early 21^st century as the physical characterization of their neutrons improved. The epithermal neutrons from MITR II and Brookhaven Medical Research Reactor (BMRR) with modified beam parameters came into clinical use for melanoma and GBM patients in the 1990s (14, 15). The relatively high-energy epithermal neutrons (0.5 eV < E_n < 10 keV) from reactors had better penetration of approximately 6 cm beneath the tissue surface, comparing to 2 cm for thermal (E_n < 0.5 eV) neutrons, which significantly increased the body volume in reach and made craniotomy unnecessary (14, 16). However, several limitations of reactor-based BNCT exist. First, the rareness of a nuclear reactor is a major impediment to BNCT research. The technical requirements and financial burdens of building and maintaining a nuclear reactor dwarf the benefits of this promising solution as definitive precision medicine. In addition, the reactors need to be shared with basic physics researchers and shut down for maintenance for one third of a whole year (17). Finally, nuclear accidents are always a concern.

Alternative neutron sources have been under development since the 1980s (18). Accelerators at that time for physical research had a neutron density far less than the requirement of BNCT, which is more than 1×10¹¹ to 10¹² (n/cm²·s) (19). In recent years, accelerator-based (AB) systems have provided comparable intensities of epithermal neutrons with the whole system in a compact form suitable for in-hospital installation (20). In short, protons are accelerated by cyclotrons (21) or linear accelerators (22) to hit the target metal materials, such as lithium (23) or beryllium (24). Nuclear reactions, e.g., ⁷Li(p, n)⁷Be or ⁹Be(p, n)⁹B, occur, and then, epithermal neutrons are emitted from the target. By 2023, AB-BNCT has become standard clinical therapy covered by medical insurance in 5 hospitals in Japan (25). Currently, several clinical studies in China (26) and Korea (27) in which several manufacturers use accelerator-based neutron sources (ABNSs) are in progress. The clinical application of ABNS in the new era warrants further investigation.

The development of boron carriers has spanned over 70 years and remains critical to the success of BNCT. The general requirements for boron agents include low systemic toxicity, high selectivity, and durable persistence in tumors during BNCT (28). The challenge lies in achieving at least 20 μg/g ¹⁰B in tumors for effective radiation, while less than one-third of the boron is present in blood or normal tissue. Advances in synthetic techniques and biological targeting have led to the emergence of several promising boron delivery strategies.

The first boron carrier used in early BNCT studies was borax. In Farr’s procedure, sodium tetraborate was infused intravenously, followed by 30 min of irradiation of given neutron counts at the skin surface, and a few derivatives of it were used in subsequent trials (11, 29). These first-generation compounds have poor selectivity, a low tumor tissue ratio, short persistence in brain tumors and high systemic toxicity (29). BSH (30) and BPA (31) are second-generation boron agents that are more selective and have been widely used in clinical trials. BSH is a polyhedral borane anion commonly used in Japanese clinical trials. The EORTC 11961 study revealed that the tumor-to-brain ratio of BSH was greater than 40 in GBM patients (32). BPA is a hydrophobic dihydroxyboryl derivative of phenylalanine that can be actively transported into cells via the L-amino acid transport system and is upregulated in most tumors (33). The safety of intravenous administration of BSH and BPA has been clinically tested in Japan, the USA, Europe, and Argentina (34), although neither meets all of the requirements. The main limitations of BSH are its lack of receptor-mediated tumor selectivity, low tumor-to-blood ratio, and significant side effects during treatment (35). While some researchers have addressed this by conjugating BSH with specific ligands (36, 37), its high cost remains a barrier to its use as an ideal boron carrier. The low boron content of BPA requires high doses, increasing treatment costs and straining liver and kidney metabolism. Its poor solubility and instability lead to a reduced boron concentration at the tumor site (38), whereas its short retention time, possibly influenced by LAT1’s anti-transport mechanism, further limits its effectiveness.

The next generation of boron compounds features the conjugation of boron agents to tumor-targeting molecules to achieve the selective accumulation of boron in tumors. The targeting relies on the binding of linked ligands to corresponding receptors that are overexpressed by tumor cells. These novel boron agents generally consist of a stable boron cluster or moiety of high boron concentration and tumor-targeting molecules linked together. Amino acids, peptides, carbohydrates, nucleosides, antibodies, and nanoparticles as targeting or carrier moieties have attracted considerable attention.

Low-molecular-weight agents as targeting moieties. In addition to BPA, boronated amino acids, along with their derivatives, have been widely investigated. These include natural amino acids and unnatural cyclic amino acids (39, 40). However, none of these compounds possess better in vitro characteristics than BPA derivatives do (35). Peptide ligands have been demonstrated through the functionalization of carriers with high affinity and selectivity (41). Michiue et al. reported that when connected to arginine repeats, BSH has increased intracellular accumulation in both the cytoplasm and nucleus (42). In vivo positron emission tomography (PET) confirmed the selective retention of BSH-3R in tumor tissue. Nagasawa et al. developed and evaluated a novel boron carrier by conjugating BSH to cell-membrane penetrating peptides (CPPs) (43). Compared with BSH, the CPP-conjugated form presented a greater intracellular concentration and better eradication of T98G cells when exposed to neutrons. Barth’s group proposed a strategy to target vascular endothelial growth factor receptor (VEGFR) and epidermal growth factor receptor (EGFR), which are often overexpressed in tumors (44, 45). However, indirect tumoricidal activity via ischemic necrosis and the expression of EGFR on normal glial cells have limited their success.

Cancer cells have an increased rate of glucose uptake and glycolysis, known as the Warburg effect (46), which renders carbohydrates an ideal ligand for boron conjugates. Aoki et al. synthesized highly hydrophilic 2-borylsugars that were actively transported into cancer cells by glucose transporter 1 (GLUT1). Then, 2-borylsugar is phosphorylated and stored intracellularly, as confirmed by in vitro assays (47). Ekholm et al. developed 6-O-carboranylmethyl glycoconjugates with high affinities for GLUT1 (48). Compared with second-generation boron agents, this compound has a 40-fold greater boron delivery capacity. Tsurubuchi et al. reported the synthesis of an α-D-mannopyranoside derivative named MMT1242 with prolonged intracellular retention compared with that of BPA (49). Notably, both mannose receptors (MRs) and GLUT1 mediate the uptake of MMT1242. Given that MRs are upregulated on the cell membranes of tumors and correlate with tumorigenesis (50), mannose conjugates hold potential for improving the therapeutic effects of BNCT. In summary, despite their high water solubility and rapid clearance from metabolically active tumor cells, leading to poor intracellular boron retention, these studies underscore the potential of carbohydrate-based boron carriers. Their enhanced uptake, high tumor affinity, and growth inhibition offer valuable insights for developing new boron agents for BNCT.

Other small molecules, such as nucleosides, porphyrins, folic acid (FA), hyaluronic acid (HA), and COX-2 substrates, have also been widely investigated in recent years. Thymidine kinase 1 (TK1) is highly expressed in the cytosol of proliferating tumor cells (51). To target TK1, Barth et al. synthesized a 3-carboranyl thymidine analog (3CTA), N5–2OH, with high tumor-to-brain and tumor-to-blood ratios (52). In vivo studies demonstrated a significantly prolonged mean survival time (MST) in rats bearing RG2 gliomas, whereas validation of N5–2OH in an F98 glioma model revealed minimal MST improvement (53). The underlying cause of this phenomenon needs further investigation. FA receptors are highly expressed in tumor cell lines and are correlated with a metastatic tendency (54, 55). In 2020, Nakagawa’s group synthesized several hydrophilic FA derivatives with IC₅₀ values of less than 3 mM in the U87 MG glioma cell line (56). The HA receptor CD44 is upregulated in cancers of numerous origins (57). In 2008, Crescenzi’s group designed an HA-conjugated carborane, named HapACB, with high affinity for CD44 in vitro (58). COX-2 is highly expressed in oral squamous carcinoma cells (59). Boronated COX-2 inhibitors demonstrated great radiosensitization capacity in CAL27 cells by suppressing the PI3K/Akt and MAPK signaling pathways (60). The conjugation of boron-enriched moieties with well-validated targeting molecules might be a productive avenue in the future.

High-molecular-weight agents as targeting moieties. Monoclonal antibodies possess unparalleled targeting selectivity, and antibody-based therapies are achieving significant clinical success across various cancers (61). With respect to BNCT, the first antibody-assisted localization was achieved by conjugating benzenediazonium ions to antibodies against carcinoembryonic antigen (CEA) in 1984 (62). This conjugate contained 30 boron atoms per IgG molecule and demonstrated a high degree of selective accumulation in CEA-positive human colonic carcinomas grown in hamsters. Recently, EGFR and CD133 have been investigated as targets for antibody conjugates. Barth and colleagues developed dendrimers with high boron concentrations linked to cetuximab, a widely used monoclonal antibody that targets EGFR (63); or to L8A4, which targets EGFR_vIII (64); or to EGF (45) per se. These EGFR-binding boron agents demonstrated significant tumoricidal effects in a rat model bearing F98 gliomas transfected with the corresponding genes (63–66). Nakase’s group also targeted EGFR via the conjugation of dodecaborate to cetuximab through the Z33 peptide (67). EGFR-expressing cells internalize the conjugates via the macropinocytotic pathway, but further validation of BNCT therapeutics involving this compound is necessary. Sun’s group reported that with CD133 on SU2 glioma cells targeted by boron−antibody conjugates, the elongation of experimental mouse survival time was remarkable (68). To date, the key challenges of antibody-based boron delivery strategies include (1) coping with variable expression of the target across tumors (2); elucidating the mechanisms for entering cancerous cells; and (3) achieving uniform boron distribution or selective enrichment within critical compartments of the tumor.

As the burst analysis indicated, extensive investigations have recently been conducted on the use of nanocarriers for boron delivery. Nanoparticles are materials with specialized functions formed by atoms or molecules at the nanoscale. Recent progress in materials chemistry, biology, and related areas has rapidly advanced the use of nanoparticles in fields such as industry and medicine (69, 70). These nanoparticle-based systems have demonstrated numerous advantages in drug delivery, e.g., high stability, water solubility, tumor accumulation, and low preparation requirements (71), due to the selective accumulation of nanoparticles in tumors, known as the enhanced permeability and retention (EPR) effect (72). Additionally, nanocarriers can easily deliver other therapeutic agents alongside boron compounds to tumors. As a result, nanostructures have emerged as a focus of BNCT research.

Polymers represent a prominent class of nanocarriers that have been extensively studied in recent years, such as dendrimers (73) and polymer micelles (74), which have the ability to extend drug retention in tumors and enhance the hydrophilicity of drugs. The boron-containing compounds are either conjugated to or encapsulated by the polymers. In 2012, Sumitani et al. reported significant suppression of colon-26 tumor (CT26) growth in mice treated with carboranes embedded by their poly(ethylene glycol)-block-poly(lactide) copolymer (PEG-b-PLA) (75). Five years later, Makino and coworkers reported that poly(L-lactide-co-glycolide) (PLLGA)-encapsulated carboranes maintained a tumor-to-blood ratio of boron concentration over 5 for more than 8 hours after administration (76). In 2019, Chen’s group developed a polymer-based tumor-targeting boron delivery system, named iRGD-PEG-PCCL-B (77). When wrapped by this polymer, BSH had six times higher uptake by A549 cells than its original form. In addition, iRGD-PEG-PCCL-B could simultaneously pack the BSH with doxorubicin, enabling highly targeted BNCT combined with chemotherapy for synergistic tumor suppression. A year later, Nomoto and colleagues prepared a novel polyvinyl alcohol-BPA complex that demonstrated improved uptake mediated by LAT1, prolonged cellular retention, and substantial CT26 growth suppression with high biocompatibility (78). In 2023, Dai et al. synthesized novel BPA-containing polydopamine (B-PDA) nanoparticles that had significant glioma ablation capacity in mice exposed to neutron radiation, with a tumor-to-brain ratio of 6.83 ± 1.75 (79). Overall, the use of polymers as boron carriers has provided substantial preclinical evidence, demonstrating their potential as promising candidates for further investigations.

Liposomes and inorganic boron-containing nanoparticles have undergone extensive investigation, as examined in two recent reviews (80, 81). These inorganic nanoparticles can be classified into boron nitride (82, 83) (BN) nanotubes (84–86), boron-doped carbon dots (87), magnetic nanomaterials (88, 89), gold nanoparticles (90, 91), and mesoporous silica nanoparticles (92, 93), although a more detailed review is beyond the scope of this study. Owing to their high biocompatibility, low systemic toxicity, improved physiochemical stability, controlled release and EPR effect, the majority of these multifunctional nanocomposites demonstrated improved selectivity for tumor cells and increased boron uptake in both in vivo and in vitro experiments through ligand conjugations, surface modifications or structural alterations, rendering these categories of nanocarriers promising blueprints for innovative boron delivery materials.

Traditional imaging techniques have offered limited insight into the distribution of boron compounds and the dose delivered to tissues. To analyze the boron concentration, various methods are available, including colorimetry, prompt γ-ray analysis (PGRA), inductively coupled plasma atomic emission spectrometry (ICP−AES), direct-current plasma atomic emission spectroscopy (DCP-AES), inductively coupled plasma−mass spectrometry (ICP-MS), quantitative neutron capture radiography (QNCR), electron energy loss spectroscopy (EELS), flow injection combined with ESI-MS/MS (FI/ESI-MS/MS), sputter-initiated resonance ionization microprobe (SIRIMP), laser atomization resonance ionization microprobe (LARIMP), secondary ion mass spectrometry (SIMS), single-photon emission computed tomography (SPECT), PET, nuclear magnetic resonance (NMR), and magnetic resonance imaging (MRI). All of the current boron drugs and their imaging techniques are presented in Table 6 .

Currently, not all of these technologies are utilized in clinical settings. Many remain at the cellular level or in research stages, such as QNCR and SIRIMP, whereas newer methods such as nano-SIMS require additional development before clinical use. Presently, four primary methods are employed in clinical practice, all of which are based on modern imaging techniques: positron emission tomography (PET) and magnetic resonance imaging (MRI) provide enhanced visualization of the boron distribution and tumor response. Additionally, advanced dosimetry tools now enable precise measurement of radiation doses, ensuring accurate treatment planning.

Early clinical trials had mixed results due to technology and compound limitations. More recent clinical trials have demonstrated promising results, particularly with new boron compounds and improved neutron sources. Trials are exploring the effectiveness of BNCT for various cancers, including those resistant to conventional therapies. The evidence suggests better outcomes and fewer side effects in some cases.

Glioblastoma multiforme: The use of BNCT for treating brain gliomas can be traced back to 1951-1961 (111). During this period, Brookhaven National Laboratory conducted a clinical trial using 10B-enriched borax and recruited 28 patients with brain cancers, including GBM and other types of gliomas. However, these trials failed because the first generation of boron drugs could damage normal tissue and be cleared from tumor cells too quickly, resulting in insufficient boron concentrations in tumors. With the subsequent development of borocaptate sodium (BSH) and boronophenylalanine (BPA) compounds, a series of clinical findings emerged after the 1990s. Five trials conducted in Japan from 1998 to 2007 (112), 1998 to 2008 (113), 1999 to 2002 (114), and 2002 to 2007 (115) reported median survival times (MSTs) of 25.7 months, 19.5 months, 23.2 months, and 10.8 months, respectively. These trials involved 15, 23, 9, and 21 patients with brain cancer, and they used various medication regimens and radiation doses. This series of studies revealed that BNCT combined with X-ray and temozolomide could prolong the median survival time of patients with GBM. In the past 20 years, with the development of new boron drugs and advancements in dosimetry, seven clinical studies on GBM have been registered with the JRCT and NCT. A summary of the results is shown in Table 7 . Two studies have reported preliminary results, namely, BNCT with temozolomide (TMZ) for GBM, with an MST of 15.6 months in patients with newly diagnosed GBM and 8.6 months in patients with recurrent GBM, with the remaining studies ongoing. Other studies are in the recruiting stage, and it is worthwhile to follow the results of these studies to better guide BNCT in treating GBM.

Head and neck cancers: BNCT has been used for various head and neck cancers, including squamous cell carcinoma (SCC). It provides a viable alternative to conventional therapies, especially for patients who have not responded well to traditional treatments or those with recurrent head and neck cancer (HNC). In the past 20 years, approximately 10 clinical trials have investigated BNCT for HNC. The earliest trial, EORTC 11011 (NCT00062348) (121), began in Essen, Germany, in 2003 and involved six patients with advanced SCC of the head and neck. These patients received an injection of either BPA or BSH, and the researchers assessed the distribution of boron-10 (B10) in both tumor and normal tissues. The mucosa and skin were identified as the most critical organs at risk.

Between 2010 and 2015, researchers in Taiwan conducted studies involving 17 patients who underwent BNCT with BPA at a dose of 400 mg/kg, delivering a prescribed dose of 12–35 Gy. Among these patients, six achieved a complete response, and six achieved a partial response, resulting in a 2-year overall survival (OS) rate of 47%. This study demonstrated that fractionated BNCT, administered at 30-day intervals with adaptive planning, is both effective and safe (122). Another study at the same center (NCT02004795) (135) focused on combining BNCT with image-guided intensity-modulated radiotherapy (IG-IMRT). The trial involved nine participants, and the combined BNCT+IMRT plan showed a significantly better conformity index for gross tumor volume (GTV) than did the BNCT-alone plan (P = 0.003). This improvement was particularly notable for tumors larger than 100 cm³, indicating that combining BNCT with IG-IMRT enhances homogeneity and conformity in treating larger tumors. This study aimed to enroll 28 patients, and the final survival-related results are anticipated.

A phase I/II trial (NCT00114790) from Finland involved 12 patients with locally advanced (rT3, rT4, or rN2) head and neck cancer that had recurred and was inoperable (124). This study revealed that 83% of patients responded positively to treatment, and 17% experienced tumor growth stabilization for 5.5 to 7.6 months. The median duration of response was 12.1 months, highlighting the effectiveness and favorable safety profile of BNCT for treating inoperable, locally advanced head and neck carcinomas, including those with recurrence at previously irradiated sites.

In Japan, BNCT research for HNC has also been substantial. Over the past 20 years, five trials have been registered. Hirose et al. conducted a trial (jRCT2080224571) using BNCT with a cyclotron-based epithermal neutron source (C-BENS). The study included 21 patients, with eight recurrent SCC (rSCC) and 13 non-SCC (r/la-nSCC) cases. The response rates were 42.9% at one month, 57.1% at two months, and 71.4% at three months (126). The two-year OS rates were 58% for the rSCC group and 100% for the non-SCC group. Additionally, two small-sample phase II studies initiated in 2019 have yet to publish any results. By 2023, two other trials (UMIN000044118 and ChiCTR2200066473) established by Japanese and Chinese researchers were registered for HNC and BNCT. The Japanese study is expected to include 120 patients with SCC of the head and neck, providing valuable insights from larger clinical research outcomes.

Melanoma: Boron–Neutron capture therapy (BNCT) has demonstrated efficacy in treating melanoma, particularly in advanced stages. Research suggests that BNCT is effective in reducing tumor size and controlling metastases. Since 2013, five studies have investigated BNCT for skin tumors, but only one study has published results. A phase I/II trial conducted by Kamitani (136) (jRCTs061180066) and completed in June 2020 included three patients with recurrent skin malignancies who had not undergone surgical treatment. The patients were intravenously administered the BPA-F complex (200 mg/kg body weight) for 2.5 to 3 hours before irradiation. The results revealed a 100% tumor control rate (CR+PR) with no adverse effects exceeding Grade 3. These findings indicate the effectiveness of BNCT against recurrent skin cancers and malignancies. However, the study included only three patients. The largest ongoing study is being conducted by researchers from Xiangya Third Hospital in China. This study aimed to enroll 30 patients, use 350 mg/kg BPA and deliver 20.0 Gy of radiation biological effectiveness (RBE). The results of this study are anticipated and may provide further insights into the efficacy of BNCT for the treatment of melanoma.

In addition to the tumors mentioned above, BNCT has also been investigated in other cancers, such as liver cancer, sarcoma, angiosarcoma, and refractory breast cancer, demonstrating objective efficacy. For example, a phase I/II clinical study published in 2015 (137) reported the treatment of malignant peripheral nerve sheath tumors (MPNSTs) with 500 mg/kg of boronophenylalanine (BPA) and 24.3 Gy-Eq, resulting in a two-year stable disease (SD) period. Other nonregistered studies are not discussed here.