Boron Neutron Capture Therapy: a technological journey to bring neutrons in hospital

1 Introduction

Boron Neutron Capture Therapy (BNCT) is a form of radiotherapy based on the enrichment of tumour with 10-boron (¹⁰B) atoms and on the subsequent irradiation with low-energy neutrons. The therapeutic effect is due to the neutron capture in ¹⁰B, which releases highly ionizing particles (an alpha particle and a lithium ion) causing non-reparable damages to the DNA. This nuclear reaction occurs with a high cross section: the probability of neutron capture in boron is orders of magnitude higher than the probability of interacting with other elements in biological tissue. Moreover, the charged particles have a combined range in tissues comparable to the average size of cells (14 $\mu$m); this means that the damages due to boron neutron capture are confined into the cell where the reaction takes place, with limited or no impact on the surrounding healthy tissues (fig. 1). In 93% of cases, lithium is in an excited state, and it releases a 478 keV de-excitation photon which is potentially useful for SPECT imaging on the reactions taking place in the patient.

The possibility to carry boron inside tumour cells makes BNCT the only radiotherapy whose selectivity is biological-based rather than beam-based: it is in fact the tumour targeting of the drug which triggers a therapeutic dose absorption in the malignancy. Moreover, BNCT treatment needs one session, only in some cases it is delivered in two sessions, the second around one months after the first.

A necessary element for BNCT is the availability of a suitable neutron beam, with adequate flux and spectral characteristics. These beams, in the past, were available only in nuclear research reactors, which have proven the BNCT therapeutic potential for orphan tumours, such as Glioblastoma Multiforme, nodular melanoma and advanced or recurrent head and neck cancer. Today, a new era of BNCT is a reality since neutron beams can be produced by proton accelerators coupled to beryllium or lithium targets and proper Beam Shaping Assembly (BSA). This last element is an ensemble of different materials arranged in proper shapes that tailor the spectrum of neutrons and collimates the beam. Figure 2 shows the principal components of an accelerator based BNCT facility.

The design of the BNCT facility is a challenge in many aspects: accelerators must deliver high proton currents, targets must resist to high power densities and must be permeable to gases, and BSA requires development and test of new materials. Stemming on basic research, the technological advancements in these areas have led to the opening of a new clinical perspective: for the first time BNCT will be a therapy in hospital environment with hi-tech equipment marked as medical devices.

Accelerator-based BNCT is already a clinical reality, especially in Japan, where the boron drug Borofalan, a variant of 10-Boronophenylalanine (BPA), the typical formulation used in BNCT, has been approved by the Ministry of Health, Labour and Welfare. There are currently about 30 projects related to the installation, commissioning and use of accelerators for clinical BNCT in the world (https://isnct.net). In Italy, INFN has a long history of basic research in BNCT: projects on the clinical application and on feasibility studies in different tumours, from the point of view of dosimetry, boron uptake, effectiveness, have been funded since the Eighties. Moreover, a 5 MeV, 30 mA accelerator has been designed and manufactured at the National Laboratories of Legnaro. This machine is able to generate a suitable neutron beam for patients when coupled to a Be target and a BSA. Both the target and the BSA have been designed and studied at INFN, and the overall ensemble constitutes the core of a clinical facility which will be installed at the University of Campania “Luigi Vanvitelli”. This is one of the goals of the project ANTHEM, funded in the PNC-PNRR program in 2022. The BNCT project foresees the construction of a building with two irradiation rooms, one for clinical treatment and one for research, the installation of the INFN technology to obtain the neutron beam and the development of preclinical research.

2 The technology for the BNCT beam: the accelerator and the target

Different types of accelerators can be used to obtain the neutron beam via nuclear interaction in the target: linear electrostatic machines, radiofrequency quadrupole accelerators (RFQ) or cyclotrons. The choice depends on different factors, with the final goal of obtaining a neutron flux of at least 10⁹ cm^–2s^–1 after the BSA. This ensures an irradiation time of the order of 1 hour.

Sumitomo in Japan has produced a cyclotron delivering 30 MeV, 1 mA protons impinging on a Be target. The high proton energy, compared to the final neutron beam which should be peaked between 1 and 10 keV, allows a lower current, but it requires a larger BSA to decrease the high-energy component of the emerging neutron beam. Neutron Therapeutics, for the clinical BNCT in Finland, has produced an electrostatic Tandem accelerator delivering 2.5 MeV, 30 mA protons and a Li target.

The same target has been chosen for the Neuboron clinical BNCT centre located in Xiamen (China), using an electrostatic accelerator from TLS, as well as for the CNAO expansion in Pavia. Electrodynamic machines such as the RFQ have been chosen by INFN in Italy, by Ibaraki Neutron Medical Research Center, University of Tsukuba, in Japan, by the Gachon University Gil Medical Center, Songdo, South Korea (produced by Dawonsys) and by Dongguan People’s Hospital, China. The RFQ is a Radio Frequency linear accelerator particularly suited for the acceleration of a low-energy beam with high intensity. An important innovation connected with the construction of the BNCT system in Italy is the fact that the RFQ is not operated by klystrons, but by eight solid- state amplifiers of 125 kW each in parallel. This solution, developed in collaboration with an Italian industry, has many advantages: lower operating costs (cost and duration of components), availability and reliability (non stop operation in case of components failure) and absence of high voltages, which is very important for the installation in medical environments.

The choice of combining a proton beam with a beryllium target is motivated by the fact that beryllium is easier to handle than lithium (which melts at lower temperature, must be kept in vacuum and becomes radioactive because of the production of ⁷Be). However, beryllium targets still pose some challenges because of the needed power dissipation and because of blistering, which is the rupture of the Be surface due to low gas permeability. For this reason, it is necessary to engineer the target in such a way that the Be layer is thin enough that protons are absorbed in another material, characterized by good gas permeability (vanadium or palladium, for example) and that this double-layer structure is backed on copper, for heat removal. At the Legnaro Labs, a first target prototype had been designed in the past and currently a new version, more efficient in terms of power dissipation, is under study. A prototype of the new Be target will soon be experimentally tested under a 5 MeV proton beam at the CN accelerator (LNL).

These two components of the technology necessary to generate the neutron beam constitute an innovation in the BNCT field: other existing facilities, in fact, produce neutron via proton interaction in Be at lower proton current but at higher energy, as for example in Japan by Sumitomo (30 MeV) or in Korea by Dawonsys (8 MeV). Higher proton energies have the drawback of generating a neutron spectrum of higher energy, which are the cause of unselective, unwanted dose absorbed in patient due to scattering in hydrogen. The capacity to design and manufacture a machine able to maintain 30 mA and a Be target able to dissipate a large power density are a significant improvement in the neutron beam quality.

3 The technology for the BNCT beam: the Beam Shaping Assembly

The neutron field produced at the target has a double-differential spectrum that is not suitable for direct patient irradiation, because the maximum energy is 3.2 MeV and the emission of neutrons spans the whole solid angle. To obtain a thermal neutron field in the tumour volume, the neutron spectrum should be epithermal, with energy peaked between 1 and 10 keV. These neutrons lose energy mainly by neutron scattering in hydrogen and thermalize crossing the biological tissues. The epithermal energy is the right compromise between the need to slow down neutrons and to spare the healthy tissue by an excessive dose deposition. For this reason, the neutron spectrum must be characterized by a low thermal component (which is absorbed almost entirely in the skin) and a low fast component (which deposits much energy by scattering) if the tumour to be treated is located deeply in the patient. Moreover, the beam must be adequately collimated to spare the peripheral organs. Finally, the photon contamination, generated in the nuclear reaction of protons and neutrons with the surrounding materials, must be filtered to protect the healthy tissues. To shape the spectrum and to collimate, the Beam Shaping Assembly is interposed between the target and the patient. This ensemble of materials arranged in proper geometries and proportions, slows down the fast component of the spectrum, absorbs the thermal one and lets neutrons emerge in the desired directions. This of course depresses the total neutron flux, which should be higher than 10⁹ cm^–2s^–1 to allow for treatment times of about one hour. The combination of a high neutron intensity obtained in the target and a smart design of the BSA is a guarantee of an effective and safe neutron beam for the clinical application. In Pavia, we have demonstrated by simulation that aluminium fluoride (AlF₃) is the best candidate as the neutron moderator for deep-seated tumours. The cross section of aluminium and fluor allows tailoring the epithermal beam, lowering the fast component. Further materials such as lithiated polyethylene, lithium fluoride and lead are used to reflect neutrons to increase the neutron flux at the beam port, shield the unwanted gamma radiation and absorb the thermal neutrons. However, aluminium fluoride is not commercially available in monolithic form. Only one material, Fluental, obtained by cold pressing mixtures of powders of AlF₃ and metallic aluminium has been used to obtain neutron beams, but the ratio of Al and F is not optimal. A technological development was started in collaboration between the INFN Unit of Pavia and the Departments of Chemistry and Physics of the Pavia University to densify powders of AlF₃ by sintering. The sintering of AlF₃ is complicated by its volatility at high temperatures, but, for the first time, samples with a density close to the nominal one (about 3 g/cm³) were obtained starting from commercial powders. The process was characterized by a high temperature sintering in the presence of uniaxial pressure, with high heating rates and overall times of few minutes. The obtained material was named Alliflu, it displayed good mechanical properties, and can be easily machined. The need to produce elements of large volume then required the design and construction of a prototype machine capable of producing monolithic samples of relatively large dimensions, while maintaining the characteristics of the process. The new machine, designed and built at the mechanical workshop of Pavia INFN, was funded with an R4I (Research for Innovation) grant by the National Commission of Technology Transfer of INFN. A high current (up to 5000 A) and low voltage (6-12 V) AC power supply system has been adopted which is much simpler and less expensive than the high-frequency pulsed current systems generally used in commercial equipment. Particularly innovative is the use of composite moulds, which allow to optimize current flows to obtain high temperatures and good temperature uniformities while significantly reducing the maximum current required in the case of large samples. The machine, called TT_Sinter, is shown in fig. 3.

Care was dedicated to sensors and safety: interlocks and controls are integrated, monitored by a PLC, allowing to control the operational conditions and simple tuning of the sintering set-up. TT_Sinter is composed of a water-cooled vacuum chamber, vacuum system, vertical single-axis hydraulic press, liquid-cooled punch electrodes, AC generator, temperature, pressure and other sensors, all controlled by the PLC. The PLC allows the control of all operational parameters and implements all the necessary safety controls, such as enable or disable the vacuum pump and the hydraulic unit, as well as to manually define the piston position and the working temperature. Several operational parameters can be set through the interface, such as the values of the PID (proportional-integrative-derivative) controllers or the maximum heating power in the different temperature ranges. Recently the machine has obtained the CE certification, and it represents a good example of technology that was developed to respond to a precise need and then used by other groups for other purposes. For example, it is presently used to produce multi-layer targets to obtain innovative radioisotopes for medicine at the National Laboratory of Legnaro.

4 How to tailor and evaluate a BNCT clinical beam?

Using TT_Sinter, we will build the BSA for the clinical centre in Caserta. To design the proper shape and arrange the proper materials for the BSA, we tested different solutions. The traditional way to evaluate a neutron beam for clinical applications had been the computation of physical characteristics such as flux, unwanted spectral components, gamma contamination in air and to compare the results with some threshold values considered adequate for a safe and effective BNCT. However, a new concept of BNCT requires that the beam performance is inserted in the wider perspective of the clinical effect. The beam should be tested on its ability to ensure an advantageous radiation dose in the tumour, keeping the dose to the normal tissues below the tolerance level. BNCT treatment planning, in fact, consists of prescribing the maximum tolerated dose to the most radiosensitive tissue involved in the irradiation: the therapeutic effect in the tumour is due to the differential boron concentration which causes a deposition of a higher dose. The dose calculation in BNCT is a complex task, as it entangles the formation of a mixed radiation field, determined by the interaction of neutrons with the elements of the biological tissues. This calculation is only possible via Monte Carlo methods; the Treatment Planning System (TPS) must produce a voxelized model of the patient (using the medical images) and construct an input file for the transport of the radiation in the tissues once the irradiation position has been established. This calculation delivers the absorbed dose in the relevant Regions of Interest (ROIs). The designed beams were then used to calculate the dose distribution in a real clinical case, treated in Finland. The criterium established was as follows: the best beam was the one which ensured the highest Tumor Control Probability (TCP) combined with the lowest Normal Tissue Complication Probability (NTCP). This figure of merit is called Uncomplicated Tumour Control Probability (UTCP), and it was calculated using proper radiobiological models. This evaluation is based on relevant clinical criteria rather than on physical characteristics, and it represents a substantial improvement in our comprehension of the effectiveness of a facility. In fact, even if the spectrum of the neutron beam does not fully comply with the traditional recommendation, for example having a higher fast neutron contamination, it may still ensure a therapeutic advantage to the patient. The beam obtained through this design process produced an UTCP comparable with the one obtained in Finland with the clinical beam used for the treatment of 300 patients. A more refined evaluation process also led to consider the safety of a clinical beam: we employed the out-of-field dosimetry, i.e., the dose delivered to all the other organs not included in the irradiation field, as a quantity to minimize. Figure 4 shows a section of the final BSA and the obtained neutron spectrum, peaked in the epithermal region and collimated as much as possible to lower unwanted peripheral dose.

Innovation in the BNCT begins with the beam design and optimization, not only for the new materials conceived to shape the spectrum and the collimation, but also in the new methods to evaluate and understand the relation between the beam quality and the clinical effects.

5 How to prescribe the radiation dose to patients?

Each component of the radiation produced by BNCT has a different biological effect due to the characteristics of energy deposition in matter of radiation with high and low linear energy transfer (LET). For this reason, the dose in BNCT is expressed in photon-equivalent units using proper models fed by radiobiological experimental data. Photons, in fact, are the reference radiation since the dose-response is well known from convention radiotherapy for almost all tumours and healthy tissues. Radiobiological models, such as cell cultures and animal models, are used to compare the effects obtained by photon reference radiation and by BNCT and are a pivotal tool to deepen our knowledge on the relation between the dose absorbed due to BNCT and the biological effect obtained. Dosimetry in BNCT is a field where BNCT still requires basic research: the Tumour Control Probability (TCP) and Normal Tissue Complication Probability (NTCP) mentioned above are not known as a function of BNCT dose with the same level of precision as in photon therapy. This will undoubtedly change with the production of more clinical data in the next years, with the activation of trials at the accelerator-based facilities. However, innovation is necessary in our capacity to express BNCT dose and being able to predict a clinical outcome using figures of merit as TCP and NTCP. The way to address this need is to express BNCT dose in photon-equivalent units, using suitable computational models, then, the clinical figures of merit can be calculated based on the experience in conventional radiotherapy. The Pavia group has adopted a new concept of photon-equivalent dosimetry which is based on the Argentinean model called photon-isoeffective dose. This model has been demonstrated to be more reliable to retrospectively explain the clinical outcomes of different trials. In fact, the traditional way to express the BNCT dose gives an artificially high photon-equivalent dose in tumours, which does not justify the tumour control obtained in the clinics. We have worked to implement this model and to make it available for different types of tumours. In fact, the model needs radiobiological data as an input, allowing the comparison of the BCNT effects with those of conventional photon therapy. For this reason, an important task of the BNCT research group is to produce dose-effect curves. One of the most used biological models is represented by cell lines cultivated in vitro, supplemented with borated formulations and irradiated in thermal neutron fields. In Pavia, we have set a well-established protocol. Figure 5 shows a sketch of the typical curves obtained by irradiating cell cultures with photons (from a clinical radiotherapy beam or for ⁶⁰Co source), with neutrons and with neutrons in the presence of boron (BNCT). These curves are used to calculate the dose of the reference radiation as a function of the different components of the BNCT dose, which produce the same effect as the BNCT treatment. This is the definition of photon isoeffective dose, which can be used to calculate TCP, NTCP and UTCP.

The $x$-axis of the dose-survival curves requires a precise dosimetry of the radiobiological models, which, in turn, demands a precise knowledge of boron present in cells during irradiation (for the BNCT curve). Our protocol involves the preparation of dedicated cultures used to measure boron concentration by neutron autoradiography. The influence of precise dosimetry in biological models on the final assessment of in-patient dose and on our capacity to predict a clinical outcome by calculation of a TCP has been described.

Another field in which we expect a major improvement is the development of detectors and methods to characterize the beam quality with higher level of precision. For example, the microdosimetric characterization of the beam, giving information on the dose delivery at subcellular level, will allow deepening our knowledge on the dose-effect relationship in BNCT, bridging the gap between the physical process of dose deposition and the biological effect. This will offer a further instrument to conceive models for in-patient dosimetry, improving the quality of the treatment in terms of safety and effectiveness. These aspects highlight once again a need for a better integration of the different areas of investigation in BNCT towards a holistic view of the process: to obtain a successful treatment and to gain prediction power, the beam engineering, the computational dosimetry, the radiobiological evaluations and the treatment planning must be closely interconnected.

The Treatment Planning System (TPS) is the software that allows the positioning of the patient and the calculation of the dose distribution in the tumour and in the surrounding tissues. Commercial TPS are starting to be distributed, with the features typical of traditional BNCT. Research- based TPS have the potential to include all the mentioned innovations, integrating more advanced models, radiobiological data, clinical figures of merit. This is being pursued at INFN in a project called IT_STARTS funded by the National Scientific Committee 5 of INFN as a young researcher grant. This has been recently enlarged with another project, AI_MIGHT, which intends to apply Artificial Intelligence techniques to elaborate medical images and automatically contour relevant ROIs. This would enable researchers to study treatment plans and evaluate the possible results of the treatment faster and for a larger number of patients. The possibility to employ such TPS in clinical trials requires a CE mark and this is time consuming and complex. A good strategy is to collaborate with companies that already provide commercial TPS and transfer into the software the models and data produced in BNCT research. A synergy between research groups providing innovative dose engines and companies providing solid software and more structured procedures for authorization is a virtuous example of knowledge transfer, bringing innovation in the clinical trials.

Another aspect necessary for a modern BNCT in hospitals concerns the development of more selective borated drugs, ideally labelled with probes which make them measurable in vivo with MRI or PET scans. The possibility to know the precise boron concentration and distribution in the patient the day of the irradiation would allow a more precise treatment planning, and thus a more realistic relation between the dose delivered and the clinical outcome. Many efforts are being devoted to the development of such boron carriers. Despite many encouraging results have been published so far in preclinical biological models, no borated drug has been tested and approved for human treatment, and BNCT still relies on borophenylalanine (BPA), an amino acid carrying one ¹⁰B atom, developed and used since the Eighties. The innovation in this field will likely happen when more pharmaceutical industries are involved in BNCT, and this will naturally occur with the establishment of in-hospital BNCT in different countries. Meanwhile, BNCT mediated with BPA is possible, safe and effective as demonstrated by the clinical trials carried out so far with nuclear reactors and with the first accelerator- based beams. While this aspect is under development, the application of more advanced methods for beam delivery, dosimetry and treatment planning will improve the quality of BNCT.

6 conclusions and perspectives

What described in this article are only some examples of how the technological innovation can improve BNCT, a therapy that was conceived just after the discovery of neutrons and that is now entering its clinical era with medical devices such as accelerators.

The possibility of offering BNCT as an established therapy to patients who have no other clinical option has been granted by the innovation in accelerator physics and by many other aspects which have been improved over decades of research in medicine, biology, chemistry, pharmacology, engineering, and physics. Since the Fifties, the journey from the basic principles of the radiobiological effects of boron neutron capture to the application of BNCT in clinics is a paradigm of inter-disciplinary science and technological evolution. During these decades, much knowledge has been constructed in the named fields, and new tools such as accelerators, targets, neutron moderators, software, models are now available.

Today, a leap forward is possible by integrating all these aspects, as a further step of the interdisciplinarity. Radiobiological experiments and analysis cannot be separated by computational dosimetry and treatment planning, and beam design cannot be separated by clinical figures of merit anticipating the effects of that beam on the tumour and on the healthy tissues.

The BNCT in the ANTHEM project aims at including the innovation described above and at proposing a new vision of this therapy based on integrating the knowledge produced in many years of BNCT research in different areas of science, as a complex process responding to the complex problem of orphan tumours. It will be the first BNCT centre among the ones that are currently under construction completely based on public research, with the spirit of interacting and welcoming companies for a fruitful knowledge transfer and a smooth translation to clinics. The technology designed and built to face this challenge must reflect a holistic view of the tumour treatment with neutrons, with the final goal to exploit at best the therapeutic advantage of the simple yet elegant concept of BNCT selectivity.

Acknowledgments

This work was partially funded by the National Plan for NRRP Complementary Investments (PNC, established with the decree-law 6 May 2021, n. 59, converted by law n. 101 of 2021) in the call for the funding of research initiatives for technologies and innovative trajectories in the health and care sectors (Directorial Decree n. 931 of 06-06-2022) - project n. PNC0000003 - AdvaNced Technologies for Human-centrEd Medicine (project acronym: ANTHEM). The BNCT research has been funded for many years in Italy by the National Scientific Committee 5 of INFN. This work reflects only the authors’ views and opinions, neither the Ministry for University and Research nor the European Commission can be considered responsible for them.