Introduction

Despite ongoing progress in early diagnosis as well as in treatment methods, cancer represents the second most frequent cause of deaths in developed countries. For most cancer patients, their treatment involves radiotherapy, either alone or in combination with surgery or chemotherapy. Conventional radiotherapy is based on the use of photon beams which after an initial build-up lead to dose profiles exponentially decreasing with increasing penetration depth. Side effects from inevitable irradiation of healthy tissues surrounding the tumour frequently limit the applicable dose and hence cure rates. On the other hand, heavy charged particles such as protons or heavier ions possess an inversed dose profile, depositing most energy in the so-called Bragg peak just before their stopping. Consequently, proton radiotherapy offers highly conformal dose distributions which deliver high dose to the tumour and largely spare surrounding healthy tissues and organs at risk. Paediatric patients or those with head-and-neck tumours are among those for whom the largest benefits from proton radiotherapy are expected1.

A few years ago the idea of proton-boron capture therapy (PBCT) was proposed2,3 as a strategy to amplify the benefits of proton therapy. PBCT can be viewed as a modality combining the concepts of proton radiotherapy and boron-neutron capture therapy (BNCT). It utilizes the proton-boron nuclear capture reaction, p + 11B → 3α. The produced alpha particles possess a short range in tissue and represent a densely ionizing type of radiation, i.e. provide a very localized and highly effective irradiation, similarly to the 10B + n → α + 7Li reaction that underpins BNCT4. Contrary to sparsely available and hardly focusable neutron beams required for BNCT, PBCT combines the ballistic precision of proton irradiation with the localized and highly effective nature of the alpha component and offers prospects for the treatment of hypoxic or radioresistant tumours3. Effectively enhancing the proton dose and its conformity to the tumour, PBCT would also help reduce proton therapy costs by reducing the number of fractions needed and increasing patient throughput.

The effectiveness of PBCT has recently been demonstrated in pioneering experiments3,5,6. In these in vitro studies, the authors reported boron-mediated enhanced effectiveness in killing DU145 prostate and PANC-1 pancreatic cancer cells as well as in inducing chromosome aberrations in breast epithelial MCF-10A cells. They used proton beams of energies ranging from 1 to 165 MeV and boron concentrations of 40 or 80 ppm of 11B in the form of sodium mercaptododecaborate (BSH), a compound widely used and concentrations reached in BNCT applications4,7,8. Originally, PBCT was thought to result in dose enhancement of the order of 50–100% via the produced alpha particles, leading to increased biological effects2,9. However, later studies10,11,12 pointed out that under clinically relevant boron concentrations, this alpha-related dose is several orders of magnitude smaller than the dose from the proton beams and thus virtually negligible, questioning the mechanisms proposed to underpin PBCT. A potential role for secondary neutrons captured by 10B in PBCT was suggested13. Furthermore, a number of experimental studies have been published recently14,15,16 that provide negative or inconclusive data on PBCT effectiveness.

To help address the validity and generality of PBCT, in this work we independently tested PBCT effectiveness for DU145 cells originally used in the pioneering PBCT experiments3,5,6 and expanded the tests for U-87 MG glioblastoma cells. The results provide the first independent validation of PBCT and demonstrate its validity for U-87 MG glioblastoma. Further, we performed dedicated experiments to help elucidate PBCT mechanisms, in particular to test the potential role of secondary neutrons captured by 10B suggested as an alternative to p + 11B capture13. We alternatively used natBSH, i.e. BSH with the natural isotopic ratio of boron (natB containing 80% 11B and 20% 10B), or 10BSH, i.e. BSH with highly enhanced content of 10B (virtually 100% 10B). As an independent benchmark on the role of neutrons in PBCT, we irradiated the U-87 MG cells with clinical 6 and 18 MV photon beams, the latter including and the former virtually absent of secondary neutrons. The results do not suggest an important role of secondary neutrons.

Results

Independent validation of PBCT in DU145 cells

To verify the PBCT concept, we first irradiated DU145 prostate cancer cells, the cell line for which the PBCT was observed originally3,5,6. The cells were irradiated by monoenergetic 190.6 MeV proton beams (range in water of 23.9 cm) in plateau (entrance) or Bragg peak positions (2.1 and 23.6 cm water-equivalent depths, respectively). Cells were irradiated in the absence or presence of natBSH at 40 ppm natB (32 ppm 11B), a concentration within ranges reached in clinical BNCT applications4,7,8. Cell survival fractions were obtained via the standard colony formation assay, scoring colonies of at least 50 cells 12 days post irradiation. Comparing plating efficiencies in unirradiated samples with and without boron, 0.303 ± 0.061 vs 0.341 ± 0.057 mean values and standard deviations from n = 4 independent experiments, revealed no signs of boron toxicity in terms of cell killing, in agreement with the experience from BNCT applications. As shown in Fig. 1, the presence of natBSH did not modulate the survival of DU145 cells in the plateau region (blue squares). Cell survival in the Bragg peak region (empty red triangles) was notably reduced compared to the plateau region. It was further reduced upon the presence of natBSH (filled red triangles), by factors of 2–3 at the highest studied doses. The measured data were fitted by the classical linear-quadratic (LQ) model of cell survival, SF = exp(-αD – βD2); the fits are included in Fig. 1 (lines), and the corresponding model coefficients are listed in Table 1. Due to biological variabilities and the low number of independent experiments (in particular at the highest dose), the data and consequently the reported LQ coefficients possess large uncertainties, and the observed boron enhancement did not reach statistical significance (p = 0.125). Yet, these results provided a preliminary validation of the PBCT concept.

Figure 1
figure 1

Survival of DU145 prostate cancer cells following irradiation by 190.6 MeV proton beams in plateau or peak regions without or with the presence of natBSH at a concentration of 40 ppm natB (i.e., 32 ppm 11B). Points represent mean values from at least 2 independent experiments, error bars depict standard errors of the mean, and lines show the corresponding LQ fits.

Full size image
Table 1 Coefficients α and β of the LQ fit to survival data for DU145 cells irradiated by 190.6 MeV protons under the absence or presence of natBSH (40 ppm natB corresponding to 32 ppm 11B and 8 ppm 10B). Uncertainties are reported as standard deviations.
Full size table

PBCT in U-87 MG cells

To further test the PBCT concept, its generality and potential underpinning mechanisms, we measured the survival of U-87 MG glioblastoma cells irradiated using the same proton beams and positions as described above, with the absence or presence of 40 ppm natB or 10B in the form of BSH. There were no signs of boron toxicity to U-87 MG cells at the applied concentration (plating efficiencies in unirradiated samples of 0.189 ± 0.049 without vs 0.196 ± 0.059 with natBSH, from n = 7 independent experiments; 0.271 ± 0.021 without vs 0.254 ± 0.058 with 10BSH, n = 2). As depicted in Fig. 2a, the presence of BSH did not modify the killing of U-87 MG cells by the proton beam in the plateau position, either with the natural isotopic mixture of boron or with 10B. Irradiation in the Bragg peak position (Fig. 2b) was more effective in killing the cells than in the plateau region. Cell survival was slightly reduced upon the presence of 10BSH (not reaching statistical significance, p = 0.067), and notably reduced by natBSH (highly statistically significant, p = 2 × 10−5), by a factor of about 2 at the highest doses. The data were fitted by the LQ model (lines); the resulting coefficients are listed in Table 2.

Figure 2
figure 2

Survival of U-87 MG glioblastoma cells following irradiation by 190.6 MeV proton beams in (a) plateau or (b) peak regions without BSH (black) or with the presence of natBSH (red) or 10BSH (green) at a concentration of 40 ppm natB or 10B, respectively. Symbols represent mean values from at least 3 independent experiments, error bars depict standard errors of the mean, lines show linear-quadratic fits to the data.

Full size image
Table 2 Coefficients α and β of the LQ fit to survival data for U-87 MG cells irradiated by 190.6 MeV protons under the presence or absence of 40 ppm natB or 10B in the form of BSH. Uncertainties are reported as standard deviations.
Full size table

These results support the concept of PBCT and demonstrate its validity for U-87 MG glioblastoma cells, at proton energy as high as 190 MeV and boron levels of 40 ppm natB (11B of 32 ppm) only. Furthermore, the results demonstrate that the key player involved in the observed effects is 11B likely capturing protons, and not 10B capturing secondary neutrons or another component of the BSH molecule such as the thiol (SH) group.

To additionally assess the potential role of thermal neutron capture by 10B, we compared killing of U-87 MG cells by therapeutic 6 and 18 MV photon beams. This was motivated by the well-known fact that while secondary neutrons are abundant in 18 MV beams, they are virtually absent in 6 MV beams for which the photon energies are below the threshold for photoneutron production reactions17. Thus, if the observed boron-mediated enhancements of proton effects were due to n + 10B capture rather than due to p + 11B, similar enhancements should be present also for 18 MV photons, but not for 6 MV ones. Despite some variability typical for radiobiological experiments, the results presented in Fig. 3 do not reveal any statistically significant differences between 6 or 18 MV beams without BSH or with natBSH or 10BSH (p > 0.14 for all pairwise comparisons). The LQ model coefficients are listed in Table 3. In agreement with the aforementioned data, these results point against the potential key contribution of secondary neutrons in the observed PBCT in U-87 MG cells.

Figure 3
figure 3

Survival of U-87 MG glioblastoma cells following irradiation by (a) 6 MV or (b) 18 MV photon beams without BSH (black) or with the presence of natBSH (red) or 10BSH (green) at a concentration of 40 ppm natB or 10B, respectively. Points represent mean values from 4 independent experiments, error bars depict standard errors of the mean, lines depict linear-quadratic fits to the data.

Full size image
Table 3 Coefficients α and β of the LQ fit to survival data for U-87 MG cells irradiated by 6 or 18 MV photons under the presence or absence of 40 ppm natB or 10B in the form of BSH. Uncertainties are reported as standard deviations.
Full size table

Dose-modifying factor

To capture the magnitude of PBCT effects, a dose-modifying factor (DMF) was introduced3, defined as the ratio of doses without and with the presence of boron that produce a given effect, e.g. 10% cell survival, DMF10%. For the present data, the DMF10% calculated from their LQ fits amounts to 1.14 ± 0.34 for DU145 cells and to 1.18 ± 0.08 for U-87 MG cells, respectively, for the Bragg peak position and with 40 ppm natB (32 ppm 11B); the reported uncertainties in DMFs refer to standard deviations. In other words, in the presence of 40 ppm natB, cell killing by protons in the Bragg peak increases as if the applied dose increased by 14% for DU145 and 18% for U-87 MG cells, respectively.

In Fig. 4, these results are compared with the data from the pioneering PBCT experiments3,5,6. DMFs generally decrease towards unity with increasing proton energy. DMFs at distal edges of spread-out Bragg peaks (SOBPs) are comparable to those from monoenergetic beams, as only protons started with the highest energy are present in distal SOBPs. DMF at mid-SOBPs (empty symbols) are smaller than at their distal edges, since the beam at the mid-SOBP position contains not only protons stopping there but also ones with higher energies and penetration depths, which are less effectively captured by 11B. As expected, DMFs decrease with decreasing 11B concentration. A comparison of DMFs for DU145 prostate and PANC-1 pancreatic cancer cells illustrates the biological variability between cell lines. The present results for DU145 and U-87 MG cells in the Bragg peak position of 190 MeV protons are in line with the previously published DMF values and these trends.

Figure 4
figure 4

Dose-modifying factor at 10% cell survival DMF10% for DU145 (blue) and U-87 MG cells (red) from the present study compared with values (black symbols) from pioneering PBCT experiments3,5,6. The legend items state the cell line, irradiation type, and concentrations of boron added in the form of BSH. Error bars represent standard deviations.

Full size image

Discussion

We have verified that the presence of 11B enhances cell killing by proton beams. For the DU145 prostate cancer cell line originally used in the pioneering PBCT experiments3,5,6, statistical significance was not reached (p = 0.125), in particular due to only n = 2 independent experiments at the highest dose. For the U-87 MG glioblastoma cell line, on which the present study was focused and for which correspondingly more experiments were performed, the enhancement was statistically highly significant (p = 2 × 10−5). For both cell lines, the presence of boron decreased cell survival after proton irradiation by about a factor of 2 at the highest doses studied, selectively in the Bragg peak region. At 10% cell survival, enhanced cell killing was observed as if the dose increased by about 14 or 18% for the studied cell lines. The reported results provide the first independent validation of PBCT in vitro and suggest it is not limited to a few cell lines but likely represents a rather general phenomenon.

Contrary to natBSH containing the natural boron isotopic mixture of 80% 11B and 20% 10B, there was little to no effect of 10BSH with virtually 100% 10B. Little to no modulations were observed for 6 or 18 MV photon beams with or without either form of BSH. Taken together, the results strongly suggest that the observed results are due to 11B, not due to other components of the BSH molecule such as sodium or the thiol group, and not due to 10B capturing secondary neutrons. The potential role of secondary neutrons could not have been excluded a priori but had to be assessed directly: Secondary neutrons including thermal as well as high-energy ones are known to be present in therapeutic proton fields, especially if the beams are spread out using passive modulation techniques but to a lesser extent also in active scanning techniques such as the one used in the present work18,19. The cross section for proton capture by 11B reaches about 1 b in a sharp peak around 0.7 MeV. The cross section of 10B for capturing thermal neutrons, 3840 b, may partially compensate for the several-orders-of-magnitude difference between secondary neutron and primary proton fluences, so that the number of boron-neutron capture events may become roughly comparable to that of proton-boron reactions. Also high-energy photon beams such as 18 MV ones used in this work contain considerable fluences of secondary neutrons. The performed dedicated experiments with 10BSH provided negative data, demonstrating that while neutron capture by 10B may contribute to the observed enhancement of cell killing by proton beams in the presence of natBSH, its role is clearly not dominant, at least for the studied U-87 MG cells. Further, the reported data provide no support for the previously suggested role of BSH acting as a general radiosensitizer14. The present data do not provide indication for a protective effect of the thiol (SH) group in the BSH molecule, either, contrary to observations in plasmid DNA20.

Obviously, the reported data do not provide further insight into the underpinning mechanism regarding whether the key process is indeed the p + 11B → 3α reaction. However, its involvement is consistent with the selectivity of PBCT to proton beams and their Bragg peak region, given the sharp maximum of the p + 11B cross section at very low energies. The proton-boron capture reaction is a very rare event, occurring once per about a million of protons at clinically relevant boron levels as used here10. Being able to observe such rare events manifest at the cell population level implies that cells (or sub-cellular organelles such as mitochondria) are extremely sensitive radiation detectors. These initial events likely need to be amplified by neighbouring cells that did not directly participate in the interaction with the densely ionizing alpha particles. We have recently hypothesized that intercellular signalling is responsible for this amplification21.

The present results extend the PBCT validity to high-energy (190 MeV) protons and boron levels as low as 32 ppm 11B (40 ppm natB). Both aspects are clinically relevant: High-energy beams are needed for irradiating deep-seated tumours; typically, clinical proton therapy centres are equipped with beam energies up to 250 MeV. While boron concentrations of 80–100 ppm can be reached in BNCT applications8, having observed enhanced effectiveness at 32 ppm 11B (40 ppm natB) suggests PBCT is robust to potential issues with boron distribution in tissue e.g. due to a reduced blood perfusion.

The reported in vitro results demonstrated an enhanced effectiveness of proton beams in the Bragg peak region under the presence of boron not only for DU145 prostate cancer cells but also for U-87 MG cells derived from glioblastoma, an aggressive brain tumour. This suggests PBCT might increase the effectiveness of proton therapy for these tumours. This would be highly relevant, since glioblastomas are linked with expected patient survival of less than a year only22. Obviously, in vitro studies such as the present one are valuable in helping elucidate the underpinning mechanisms, but it is up to preclinical and clinical studies to judge the validity of clinical methods in general and PBCT in particular. Importantly, in the period of preparing this manuscript, a pre-clinical study was published that used a xenograft model of glioblastoma generated by subcutaneously injecting U-87 MG cells to mice23. The authors reported enhanced cell killing, reduced metabolic tumour volume and total lesion glycolysis, as well as gene expression changes related to cellular communication, in line with the results reported in this work and supporting the potential clinical applicability of PBCT.

On the other hand, our own data24 show large variations in the boron-mediated enhancement of proton effectiveness among cell lines, including the lack of enhancement in some cells at 40 or 80 ppm natB. This might be a consequence of the proposed key role of intercellular communication in the PBCT phenomenon: Radiation-induced signalling and its effects have been shown to depend on cellular characteristics, timing, and medium or serum composition including e.g. the presence of antioxidants or serotonin level25,26. Likely, variations in these factors may also help explain inconclusive or negative PBCT data reported in the literature14,15,16. The PBCT phenomenon, its underpinning mechanisms and potential clinical applications thus clearly deserve further research.

Materials and methods

Cell culture and boron treatment

Adherent DU145 prostate cancer cells (kindly provided by Prof. L. Manti, University of Naples, Italy) or U-87 MG human glioblastoma astrocytoma cells (ECACC) were used, as non-synchronized cell populations in their exponential phase. The cells were grown in tissue culture flasks (TPP, Techno Plastic Products AG, Trasadingen, Switzerland) in Minimum Essential Medium Eagle (MEME, Sigma-Aldrich, Germany) supplemented with 10% Fetal Bovine Serum (FBS, Biosera, France), 2 mM glutamine, 1% non-essential amino acids solution, 1 mM sodium pyruvate, 100 U/ml penicillin and 0.1 mg/ml streptomycin (all: Sigma-Aldrich, Germany) in an incubator maintained at 37°C in 5% CO2 atmosphere. Under these culture conditions, the doubling time was approximately 30 h or 33 h and the plating efficiency 60 % or 20% for DU145 or U-87 MG cells, respectively.

The cells were seeded in tissue culture T25 flasks two days before irradiation at a concentration of 4000–10,000 cells/cm2 in 4 ml media. One day before irradiation, 1 ml media without or with boron was added. As boron delivery agents, two forms of sodium mercaptododecaborate (BSH, mercapto-undecahydro-closo-dodecaborate, also called sodium borocaptate, Na2B12H11SH): either natBSH with natural boron (natB) isotope ratio of 20% 10B and 80% 11B, or 10BSH with virtually 100% 10B (Katchem, Czech Republic). Stock BSH solution was prepared shortly before experiments by dissolving BSH in sterile distilled water. The final boron concentration in the total 5 ml media was 0.04 mg/ml natB or 10B, i.e. 40 ppm natB or 10B, corresponding to 0.083 mg/ml natBSH or 0.070 mg/ml 10BSH, respectively; control samples received no boron supplement.

Irradiations

Proton irradiations were performed at the Proton Therapy Center Czech in Prague using monoenergetic irradiation maps of 190.6 MeV (range in water of 23.9 cm), covering an irradiation area of 20 cm × 20 cm at the isocentre using the pencil beam scanning mode with spot spacing of 4 mm. The gantry was positioned at 180°, irradiating the sample from below, through the patient couch. The cells with or without BSH were irradiated at two positions of the Bragg curve27. The first position was at 2.1 cm of water equivalent thickness (WET) corresponding to the plateau area (also called entrance region) with a dose-averaged linear energy transfer (LET) of 0.44 keV/µm; the second position at 23.6 cm WET corresponded to the Bragg peak area with a dose-averaged LET value of 3.11 keV/µm. The WET was adjusted using solid water (RW3) and polymethyl methacrylate slabs below the samples (Fig. 5a). The applied doses were calculated for both positions in the treatment planning system XiO (ELEKTA, Sweden) as done in the standard clinical routine at the facility. The calculated doses were benchmarked by a plane parallel ionization chamber (PPC05, IBA dosimetry, Belgium). The beam range was verified using a multi-layer ionization chamber ZEBRA (IBA Dosimetry, Belgium). Doses up to 8 Gy were applied (specifically, 0, 0.94, 1.88, 2.82, 3.76, 4.7, 5.64 and 7.52 Gy in the plateau position, and 0, 0.97, 1.94, 2.91, 3.88, 4.86, 5.82 and 7.77 Gy in the Bragg peak position; for simplicity, these doses are approximately referred to as 0, 1, 2, 3, 4, 6 and 8 Gy below). The dose uncertainty due to range or positioning errors was estimated not to exceed 2%.

Figure 5
figure 5

(a) Proton irradiations. Left: Percentage depth-dose curve, acquired using a ZEBRA multi-layer ion chamber device, with marked irradiation positions of 2.1 cm (plateau) and 23.6 cm (Bragg peak). Right: Experimental setup at the Proton Therapy Center Czech in Prague. The samples were irradiated from below. The depth was realized using slabs in both plateau (top) and peak regions (bottom). (b) Photon irradiations. Left: Experimental setup using Siemens ARTISTE linac. The samples were irradiated from below. Right: Percentage depth-dose curves with marked irradiation positions for 6 MV (top) and 18 MV photon beams (bottom).

Full size image

All irradiations were repeated several times using independent cell cultures. For DU145 cells, two independent repeats were performed measuring survival curves (methodology described below) by irradiating the cells with 0, 1, 2, 3 and 4 Gy, and then further two repeats with the dose range extended to 5 or 6 Gy in the peak or plateau positions, respectively, so that in total there were 2–4 independent repeats per datum. For U-87 MG cells, three repeats were performed with 0, 1, 2, 3 and 4 Gy without or with natB, followed by four repeats with doses up to 8 Gy without or with natB or 10B (only 3 repeats with 10B in plateau position), providing in total 3–7 independent repeats per datum. Here, ‘datum’ includes dose level, absence or presence of 40 ppm natB or 10B, and plateau or peak positions.

Photon irradiations were performed at the Thomayer University Hospital in Prague using linac (Siemens ARTISTE, Germany) photon beams of 6 or 18 MV. For both energies, the source-to-axis distance was 90 cm and the irradiation field size was 20 cm × 20 cm at the isocentre. The delivered doses were determined using the treatment planning system XiO by ELEKTA as done in routine clinical operation of the machine. The dose delivery uncertainty was 0.5%, estimated from previous dosimetric assessments. Samples were positioned horizontally on the treatment couch. Solid water slabs were used to assure the dose rate of 2 Gy/min. The gantry was positioned at 180°, irradiating the sample from below, through the patient couch. Additional solid water slabs were placed above the sample to include backscattered radiation. Doses up to 8 Gy were applied. The irradiation scheme and the percentage depth-dose curves for the used beams are shown in Fig. 5b. All irradiations were repeated four times using independent cell cultures.

Cell incubation and clonogenic cell survival assay

Immediately after returning from the irradiation facilities, the cells were washed twice by 1X Phosphate Buffered Saline (Sigma-Aldrich, Germany) and trypsinized using 1X Trypsin–EDTA (Biosera, France) for 3 min in the incubator. Then 4 ml of medium were added, and cell concentrations were counted by a MUSE cell analyzer (EMD Millipore) using the Muse Count & Viability Assay Kit (Merck Millipore/Luminex, MCH100102) according to the manufacturer’s instructions. Cells were then re-seeded to 6-well plates (providing 6 technical replicates per datum and repeat) with fresh, boron-free medium at appropriate densities estimated to yield about 30 colonies per well, given the expected survival levels. Following incubation for 12 days at 37°C in 5% CO2 atmosphere, the formed colonies were fixed and stained using Crystal violet dye (Sigma-Aldrich, Germany) diluted in water and methanol. The colonies were counted manually, considering only those containing at least 50 cells. Cell survival fraction SF(D) at dose D was calculated as the ratio of plating efficiencies PE(D) at this dose and PE(0) in unirradiated controls, the PE being given by the ratio of the number of colonies observed to the number of cells seeded.

Statistical analysis of survival data

Cell survival fraction (SF) in dependence on applied dose (D) was fitted by the Linear-Quadratic (LQ) model, SF = exp(-(αD + βD2)), with parameters α and β, using the maximum likelihood fitting for Poisson-distributed data implemented in CFAssay tool28. In this step, the total number of cells seeded and colonies scored in the 6 replicates (6 wells of a plate) were considered, for each irradiation type (plateau or peak position for protons, or 6 or 18 MV photons), boron level (controls without boron or samples with natB or 10B) and dose level. Each experiment was independently repeated several times as specified above. Statistical significance of the observed differences for pairs of survival curves, in particular those with versus without the presence of boron, was assessed by ANOVA-based tests for one-way designs in CFAssay28. The number of repeats per datum was reflected in its uncertainty (presented by error bars in the survival plots) as well as in the statistical power of differences between curves (p-values) estimated by CFAssay.

Dose modifying factor (DMF) of boron was calculated as the ratio of doses without versus with boron presence, for protons in the Bragg peak and 10% survival level, chosen throughout this work for consistency with the literature3,5,6. The doses leading to 10% survival were estimated from LQ fits to the data. Standard error propagation techniques were used to estimate uncertainties in the DMF from those of the LQ parameters.