1. Introduction
Medulloblastoma (MB) is the most common malignant brain tumor in children. The European annual incidence rate is 6.5 per million children [
1]. There are several types of MB, which are distinguished based on histological classification or on genetic alterations. MB, genetically defined, is classified into WNT-activated, SHH-activated and
TP53-mutant, SHH-activated and
TP53-wild-type, and non-WNT/non-SHH groups [
2]. In particular, about one-third of all MBs show aberrant activation of the SHH signaling pathway, a developmental axis in which the tumor suppressor gene
PATCHED1 (
PTCH1) normally imposes an inhibitory effect [
3]. Currently, multimodal treatment—surgery, radiotherapy, and chemotherapy—is the most effective strategy against MB. The conventional doses of RT delivered to the craniospinal axis and to the posterior fossa are 54–56 Gy. Using such doses, a high proportion of MB survivors have significant long-term consequences, including marked losses of Intelligence Quotient and endocrine dysfunction [
4]. Many attempts have been made to control tumor growth while trying to reduce the radiation-related long-term neurocognitive effects, especially in young children. Recent strategies in pediatric oncology have also included the use of 3D conformal radiotherapy, as well as proton therapy, to minimize late effects. Protons are now increasingly used to treat pediatric MB patients, with important and clinically relevant differences compared to photon radiation treatments [
5,
6].
TOP-IMPLART (Intensity Modulated Proton Therapy Linear Accelerator for Radiotherapy) [
7] is a pulsed proton linear accelerator (linac) for proton therapy applications developed at the ENEA Frascati Research Center. The machine, which consists of a sequence of accelerating modules, each increasing the maximum proton energy in steps of a few MeVs, is built in the framework of a national project funded by Lazio Innova-Regione Lazio and led by ENEA in collaboration with the Italian National Institute of Health (ISS) and Regina Elena Hospital [
8,
9]. The modularity of this type of accelerator allows extraction of the proton beam at the end of the last installed module for application in the bio-medical field and in particular for “in vitro” and “in vivo” radiobiology experiments. The output energy can be decreased from the maximum value, combining active and passive energy degradation techniques, to satisfy the requirements of different experiments in terms of penetration depth in the target volume.
The stability, reproducibility, and quality of the beam have been validated by radiobiological experiments combined with dosimetric characterization [
10]. The entire process of characterization of TOP-IMPLART, supported by the results obtained with radiobiological in vitro experiments of cell killing and clonogenic survival experiments conducted with V79 and CHO cells (unpublished observation), allowed us to address the efforts toward an in vivo radiobiological experiment campaign in order to verify the suitability/potentialities of this facility for these kinds of studies. In fact, despite the steady worldwide increase in the number of proton therapy centers, proton radiobiology still has many open questions requiring either basic or translational preclinical research. Small animal proton therapy research may contribute to the basic understanding of in vivo radiation effects, but systems for proton radiotherapy research in these models are still rather limited, so far [
11]. For proton research on small animal models, for instance, the position of the target volume and dosimetry are critical factors demanding specific solutions. Generally, murine models offer many advantages for mechanistic radiobiological investigations on normal tissue and tumor response, including short lifespan and the availability of genetically engineered mice to study the relevance of specific genes on radiation responses.
In this study, the proton therapy linear accelerator TOP-IMPLART was employed for in vivo radiobiological investigations on the effects of proton irradiation on a Shh-dependent MB mouse model, in both tumor and normal brain tissues. Proton beam parameters were successfully adapted to preclinical studies on mice as shown by the homogeneity of the dose distribution. Apoptotic response in MB tumor allografts after irradiation with protons or photons was compared to gain information on their biological effectiveness. In addition, by combining active and passive energy degradation techniques, we had reproducibly restricted the depth at which the proton radiation dose is delivered, carrying out the exposure of neonatal mouse brain at postnatal days 5 (P5) with two different collimated Spread-Out Bragg Peaks (SOBP) of 3 mm and 8 mm in depth. These results support TOP-IMPLART as an in vivo radiotherapy facility for in vivo proton beam research on mouse models, besides its use in proton therapy. This has implications for the open mechanistic research questions on proton radiobiology that limit the advancement in proton therapy research.
3. Discussion
Preclinical in vivo studies are of paramount importance for translational research in radiation oncology. In particular, proton radiobiology still calls for basic and translational preclinical research [
15]. Access to proton facilities and the availability of relevant in vivo animal models for proton research are both among the main bottlenecks limiting the advancements in the field of proton radiobiology and related clinical implementation to improve patient treatment.
Although access to proton facilities has progressively improved over the years, there is a paucity of published preclinical in vivo studies for protons, strongly soliciting the implementation of proton radiobiological facilities. In fact, radiobiological experiments mostly rely on clinical facilities, where the high beam energies used for patient treatments shall be degraded for the irradiation of small animals or even directly used in the plateau region of the pristine depth–dose curve [
16,
17,
18,
19,
20]. Moreover, in clinical facilities, beam priority is correctly given to patient treatments.
The aim of this study was to investigate the feasibility of the proton beam generated by TOP-IMPLART, a prototype of a proton-therapy linear accelerator developed at ENEA Frascati Research Center, for irradiation of small targets represented by preclinical mouse models. Target sizes for in vivo radiobiological experimentation ranged from tumor mass of about 2 cm3 down to the size of the newborn mouse brain at P5 (9 mm). To the best of our knowledge, this is the first time that a proton-therapy system based on a full-linear accelerator has been employed for radiobiological in vivo experiments. For the experimental setup, a close interdisciplinary collaboration, involving engineers, physicists, and biologists, was required. The specific mode of proton energy deposition in penetrating materials, characterized by a maximum release (the Bragg peak) close to the end of the particles’ range, poses several critical issues that have to be addressed. In particular, the positioning of the target, indeed small positioning errors can cause irradiation of healthy tissue. Currently, at the TOP-IMPLART facility, positioning is performed manually, but it implies a very time-consuming procedure; therefore, improvements in the mouse housing are under development.
Another important aspect to be considered in the small animal irradiation is the homogeneity of the dose distribution; to this aim, the uniformity of the beam, in terms of flatness and symmetry, was investigated for both collimators used (20 mm and 8 mm in diameter). Results showed values of flatness <6% and symmetry <0.5% in the transverse profiles in x and y-directions for both collimators. These findings evidenced suitable characteristics of the TOP-IMPLART proton beam for small animal irradiation. Additionally, the agreement between the results obtained with the two collimators highlights that, on one side, a reduction in the collimator down to 8 mm in diameter does not affect the beam, and on the other side, a good alignment of the collimators in respect to the beam axis was achieved.
Several dosimetry systems, specifically dedicated to this proton beam, were designed and built inside this collaboration, for beam monitoring and dose assessment (2D-IC and iIC). These systems were complemented by commercial dosimetry systems such as mD and EBT3. Even if this approach provides redundant pieces of information, we consider it particularly relevant in the case of this accelerator (prototype of first proton linac characterized by high dose per pulse) used for in vivo experiments. It is extremely important to have immediate (real-time) information about the delivered dose or other irradiation parameters useful to define/check the good behavior of the system, which was achieved through the online detectors (2D-IC) developed within the TOP-IMPLART collaboration. Nevertheless, we consider an important step for the robustness of the results is the independent verification/checking of those quantities with more conventional dosimetry systems (i.e., Gafchromic films), suitable for these kinds of measurements. Unfortunately, data provided by films are not available during the irradiation session (real-time), but, as for all passive dosimetry systems, they can be provided after a period of time successive to the irradiation, which depends on the protocol used for the detector calibration.
From a biological point of view, the choice of the experimental model was made to reflect clinically relevant endpoints. In fact, we developed a model system involving subcutaneous transplantation into the mouse flank of a brain tumor developed in the
Ptch1 heterozygous mouse model [
13,
14]. This is relevant to clinics because SHH-dependent MB represents one-third of all MB cases [
3], affecting mainly children less than 3 years of age, and because proton therapy for pediatric MB is currently used as an innovative approach with the potential to enhance the outcome of radiotherapy [
21,
22]. In addition, as an MB normal tissue counterpart, we employed the newborn mouse brain at P5, for its peculiar sensitivity to radiation effects at neonatal age [
23]. On both these mouse model systems, to investigate tumor and normal tissue response, we employed an experimental approach to visualize the radiation-induced apoptosis through the assessment of the apoptotic marker caspase-3-activated by IHC. In addition, Western blot analysis was employed to quantify the apoptotic levels in tissue regions within and outside the proton irradiation field. The experimental data we produced unambiguously established that, by combining active and passive energy degradation techniques, we could reproducibly restrict the depth at which the proton radiation dose is delivered both in the tumor mass and in the newborn mouse brain, thus irradiating definite tissue volumes of organs and tumors, depending on the research question. Future investigations on the local control of proton therapy evaluating late local toxicities and pattern of tumor relapse in MB allografts after photon- or proton-irradiation are needed and clinically relevant.
Regarding the evaluation of the tumor response, xenografted tumor models are frequently used for in vivo proton relative biological effectiveness (RBE) estimation [
24,
25,
26], while for normal tissue responses in vivo, the skin and the intestinal crypts are considered good model systems and are often employed to determine early and late tissue response to protons [
27,
28,
29,
30,
31,
32]. In particular, similar to our study, the endpoint of apoptosis was exploited to investigate the damage induced by protons in vivo in mouse intestinal tissue [
33]. Further mouse experiments using the TOP-IMPLART accelerator and suitable mouse models might be foreseen to contribute to in vivo RBE calculation and to improve the mechanistic understanding of proton radiobiology.
The lack of dedicated small animal particle beam irradiators is recognized as a main limitation to advances in proton radiobiology and attempts are currently being made to develop experimental settings for preclinical in vivo studies, mostly based on adaptation of the technology for clinical particle beam irradiators [
11,
34,
35]. Altogether, our findings demonstrated that the TOP-IMPLART accelerator and the related methodologies here developed, from the manual positioning of the target volume to the delivery of prescribed doses, through dosimetry and radiobiological endpoints, successfully fulfilled the requirements for a research platform for small animal studies.
4. Materials and Methods
4.1. Beam Line Design and Characteristics
The TOP-IMPLART accelerator is a radiofrequency (RF) pulsed fully linear machine consisting of a commercial injector operating at 425 MHz and an S-band section operating at 2997.92 MHz. The injector, consisting of a duoplasmatron source, an RFQ (Radio Frequency Quadrupole), a DTL (Drift Tube Linac), and their RF power source, accelerates the beam at 7 MeV. The S-band section consists of a sequence of Side-Coupled DTL (SCDTL) accelerating modules, powered by two 10 MW peak power klystrons. The measurements described in this paper were performed with the beam exiting from SCDTL-6 structure with a maximum energy of 55.5 MeV, as reflected by the schematic layout depicted in
Figure 7A.
The proton beam propagates in vacuum in the accelerator and exits in air through a 50 µm titanium window. An AC current transformer (ACCT3) installed immediately after the vacuum window monitors the output current during the irradiation. Before the ACCT3, an aluminum foil of 1 mm of thickness is inserted to suppress the secondary electrons. The proton beam current used for this experiment was set to 10 µA in pulses of 2.4 µsec duration at a repetition frequency of 25 Hz. To obtain a homogeneous irradiation area at the target position, the beam is diffused first through a lead scattering foil (210 µm of thickness), placed in air at 18.5 mm after the vacuum window, then the protons move in air and reach the target position set at 176.5 mm from the linac exit. The irradiation end station, where the target is positioned (detailed in
Figure 7B,C), is provided with online monitoring devices (ionization chambers), dosimetry diagnostics, collimators, a range shifter, and a custom-made range modulator, consisting of 7 plexiglass sectors of different thickness organized in a rotating wheel for generating the SOBP. Range shifters of different thicknesses were used to select different penetration depths.
The proton beam field was shaped through the use of aluminum collimators of different sizes. A first collimator (20 mm diameter) was employed in all the irradiation settings, and a second (8 mm diameter) was added to further reduce the beam size for irradiation of neonatal mouse brain. Irradiation was driven by an integral ionization chamber (iIC), placed close to the target position, immediately after the primary collimator. A second ionization chamber (2D-IC) acted as an independent dose monitor, calibrated by the mD.
LINAC [
36] and SRIM [
37] codes were employed for the simulation of the beam dynamics inside the accelerator and the simulation of the irradiation line up to the target position, respectively: the beam coordinates describing positions, divergence, and energy of the beam before the titanium window, produced by the LINAC code, constitute the input data for SRIM to continue the propagation of the beam up to the target. As an example of this methodology, in
Figure 8A,B, we show the computed beam spot at the target for the setup with a RS 1.5 mm thick. The final distribution has a Gaussian shape with a standard deviation of about 40 mm on x and y-axis (
Figure 8C,D), corresponding to a homogeneity of 3% on the area defined by the collimator of 20 mm diameter, evaluated as the maximum variation. The pristine energy at the target surface, due to the interaction of protons with different materials, is decreased from 55.5 to 46.4 MeV.
4.2. Dosimetry
Several dosimetry systems were considered to experimentally characterize the TOP-IMPLART proton beam from a dosimetric point of view.
A diamond detector and two IC monitors provide the online control of the delivered dose, uniformity of transverse beam profile, and short and long-term stability of the beam. mD was also used for dose assessment and the preliminary calibration of the IC monitor. Offline EBT3 Gafchromic films and LiF crystals were used to complement/integrate/validate the data obtained with online systems, including the beam energy measurement.
The mD detector, model 60019 (PTW-Freinburg, Germany) connected to a Keithley electrometer (mod. 6517A with no polarizing voltage) was calibrated at the Italian Primary Standard Dosimetry Lab (ENEA-INMRI) in terms of absorbed dose to water in 60Co ɤ-rays.
The 2D-IC, a real-time 2D IC, developed within the TOP-IMPLART project, provides a pulse-by-pulse response. The 2D segmented IC in Micro Pattern Gaseous Detector technology is able to simultaneously acquire x/y strip readout with 0.3 mm spatial resolution, 100 fC sensitivity, and dynamic range greater than 104. It is equipped with dedicated electronics that automatically adjust gain to the input collected charge through a feedback sample and hold capability.
A preliminary calibration of the 2D-IC was carried out using an mD detector. For this purpose, several irradiations at different dose values, in defined experimental configurations, were performed with the 2D-IC placed with its center on the beam axis at a fixed distance from the beamline exit and mD placed at the target position. The total charge values measured by the 2D-IC were recorded vs. mD dose; a good linearity response of the 2D-IC was verified in all configurations. An example of calibration performed using the 8 mm collimator is shown in
Figure 9. The maximum deviation from linearity of the 2D-IC response was 4% in the (2.7–8) Gy dose range. This favorable behavior of the 2D-IC allows it to be used to monitor the stability of the output charge delivered by the accelerator during irradiation when the mD was removed and replaced by the actual target. In addition, 2D-IC provides accurate x and y profiles of the beam and its position, allowing, in such a way, the alignment of all systems (detectors, RS, RM, etc.) placed along the beamline to be checked; this piece of information is also relevant for a preliminary target positioning.
The iIC, an integral ionization chamber, is used to drive the irradiation. It operates at a bias voltage of 250 V (variable) and is realized with aluminized Mylar electrodes (12 μm Mylar, 4 μm aluminum) spaced by 2 mm of air. This chamber measures the dose in terms of monitor units and turns off the beam when a preset number of MUs (corresponding to the prescribed dose) is reached. The calibration of monitor units in terms of absorbed dose to water is performed by using the mD positioned in the place of the target (
Figure 7C).
Gafchromic EBT3 (Ashland Advanced Materials) are self-developing dosimetry films with a symmetrical construction made of two 125 µm thick matte-polyester external layers and a 28 µm thick active middle layer. These radiochromic films have a dynamic dose range between 0.1 and 20 Gy, and a high spatial resolution, suitable to control beam uniformity. They have been digitized with an EPSON Expression 10,000XL/PRO flatbed color scanner. EBT3 films were calibrated at the Italian Primary Lab in 60Co source in terms of absorbed dose to water. In our experiment, they are specifically used during mice irradiations and placed in front of the target to evaluate the transverse beam profiles at the target positioning, and they also measure the absorbed dose of water delivered at the mouse surface.
Imaging exploiting the visible photoluminescence of color centers (CCs), created in LiF crystals by the proton beam, was employed to characterize the energy of the proton beam on the target and dose distribution with the depth of pristine Bragg peak and SOBP. Commercially available 15 × 15 mm
2 LiF crystals with polished faces were irradiated with one of the 1 mm thick sides exposed perpendicularly to the impinging protons. For doses below a saturation threshold, the PL intensity is proportional to CCs concentration, which, in turn, increases linearly with the absorbed dose, and a luminescent image of the Bragg curve can be analyzed by means of a fluorescence microscope [
38]. The high intrinsic spatial resolution of LiF crystals allows dose curve reconstruction with micrometric precision and determination of the energy beam parameters [
39].
4.3. In Vivo Allograft and Study Design
Small fragments of murine MB tumors developed by a C57Bl/6/
Ptch1+/− female mouse [
13,
14] were transferred into the flank of immunocompetent WT C57Bl/6 females, and once the tumors reached 1400–2400 mm
3, the mice were randomized to (i) sham, (ii) protons, and (iii) photons. Both proton and photon irradiations were single fractions.
4.4. Animal Irradiation
For tumor irradiation, tumor-bearing mice were anesthetized (65 mg/kg sodium pentobarbital i.p.), shielded by covering the whole mouse body, but for the tumor mass, with lead bars. Irradiation in photons was performed using a Gilardoni CHF 320G X-ray generator (Gilardoni S.p.A., Mandello del Lario, Italy; HVL = 1.6 mm Cu) operated at 250 kVp, 15 mA, with filters of 2.0 mm Al and 0.5 mm. For proton irradiation, the beam was delivered through a collimator of 20 mm in diameter to the flank tumors. Two different RS were used with thicknesses of 1.5 mm and 12.9 mm, respectively corresponding to maximum energies in water of 46.4 MeV (range in water 19.1 mm) and 25.8 MeV on target (range 6.6 mm). Additionally, the presence of the RM generated a different SOBP for each RS used with widths of 19 mm and 7 mm, respectively (
Table 4, SRIM evaluations). The irradiation duration was around 95 s.
For neonatal irradiation, the proton dose was delivered to P5-neonatal mice restrained into an ad hoc plexiglass holder. SOBP widths of 3 and 8 mm were generated to deliver the dose through 8 mm diameter collimator at different tissue depths by using RS with thicknesses of 15.2 mm and 11.2 mm, corresponding to maximum energies in tissue of 16.2 MeV (range in water 2.9 mm) and 29.5 MeV (range in water 8.4 mm) (
Table 5). It is worth noting that the amplitude of the SOBP coincides, within the uncertainty, with the range of the proton’s maximum energy, indicating that the entire thickness of the target is covered by the SOBP. The irradiation time was around 30 s.
4.5. Histological and Immunohistochemical Analysis
After irradiation, tumors and cerebella of C57Bl/6 pups were processed for histology by standard techniques, and tissue sections were cut (4 μm) for H&E staining and microscopic morphological examination. Fixed tissue sections were immunostained as described [
40] using an antibody against cleaved-caspase-3 (Asp 175) (dilution 1:150 overnight; Cell Signaling Technology CS9664, Danvers, MA, USA). Samples were analyzed by light microscopy using the software NIS-Elements BR 4.00.05 (Nikon Instruments Europe B.V.; Florence, Italy).
4.6. Western Blot
Brain and tumor samples were collected at 4 h post-treatment, soaked in RNA later, and conserved at 5 °C. Proteins from tumors and cerebella at P5 were extracted, normalized, separated, and immunoblotted as described [
41]. To evaluate apoptosis, cleaved-caspase-3 (Asp 175) rabbit polyclonal antibody was used (dilution 1:1000 overnight; Cell Signaling Technology, Danvers, MA, USA). Monoclonal antibody against HSP70 (dilution 1:10.000 30 min, Sigma-Aldrich H5147, St. Louis, MI, USA) was used as loading control. Specific proteins were visualized using ECL™ Prime Western Blotting Detection Reagent (Cytiva RPN2232) with ChemiDoc system XRS+ Biorad and quantified using ImageJ software version 1.8.0.
4.7. Statistical Analysis
Data were given as mean ± standard error of the mean (SEM). All statistical analyses were carried out using GraphPad Prism 6 statistical software (GraphPad, San Diego, CA, USA). The Kolmogorov–Smirnov test was used to verify that data were sampled from populations following the Gaussian distribution. Comparisons between groups were performed using the parametric t-test (significance taken as p < 0.05).
5. Conclusions
Combining the TOP-IMPLART proton-therapy accelerator with pre-clinically relevant mouse models provides an important preclinical setting to test a variety of questions on proton radiobiology on both tumor and normal tissues. Open mechanistic research questions on proton radiobiology—including normal tissue toxicity, differences in biological responses after proton and photon irradiation, and a reduction in uncertainties of the proton RBE at the end of the SOBP—might greatly benefit from the results of the research here presented. In fact, these results extend the experimental possibilities and pave the road for further mechanistic radiobiological investigations to compare in vivo the oncogenic effect of protons delivered by a pulsed fully linear accelerator vs. a comparable dose of photons in Patched1 heterozygous (Ptch1+/−) mice, a mouse model of radiation hypersensitivity with a predisposition to cancer and non-cancer radiation-induced pathologies, including MB and lens opacity.
Despite the current limitations of the dose delivery system (manual positioning, custom range modulator, and range shifter), which are under improvement in different directions (laser-controlled positioning, implementation of magnetic proton beam scanning for a more efficient utilization of the beam), the TOP-IMPLART has demonstrated adequate performance for small animal in vivo studies by using a dose rate of about 0.2 Gy/sec for neonatal mice and 0.08 Gy/sec for MB mice.
However, the flexibility of the machine in terms of output pulse current and repetition frequency combined with the replacement of the actual passive beam delivery with an active modality could provide an increase in the dose rate of up to a factor of sixty. This will be exploited for the investigation of unconventional in vivo irradiation modalities.