Preparation of a radiobiology beam line at the 18 MeV proton cyclotron facility at CNA

Depósito de investigación de la Universidad de Sevilla https://idus.us.es/ “This is an Accepted Manuscript of an article published by Elsevier in Physica Medica-European Journal of Medical Physics on June 2020, available at: https://doi.org/10.1016/j.ejmp.2020.04.022 .” 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Preparation of a radiobiology beam line at the 18 MeV proton cyclotron facility at CNA Anna Baratto-Rold´ana,b,∗ , Mar´ıa del Carmen Jim´enez-Ramosa, Sonia Jimenoc,d, Pablo Huertasc,d, Javier Garc´ıa-L´opeza, Mar´ıa Isabel Gallardob, Miguel Antonio Cort´es-Giraldob, Jos´e Manuel Espinoa,b aCentro Nacional de Aceleradores, Sevilla, 41092, Spain bDepartamento de F´ısica ´ Atomica, Molecular y Nuclear, Universidad de Sevilla, Sevilla, 41012, Spain cDepartmento de Gen´etica, Universidad de Sevilla, Sevilla, 41012, Spain dCentro Andaluz de Biolog´ıa Molecular y Medicina Regenerativa-CABIMER, Sevilla, 41092, Spain Abstract Proton therapy has gained interest in recent years due to its excellent clinical outcomes. However, the lack of accurate biological data, especially in the Bragg peak region of clinical beams, makes it difficult to implement biophysically optimized treatment plans in clinical practice. In this context, low energy proton accelerator facilities provide the perfect environment to collect good radiobiological data, as they can produce high LET beams with narrow energy distributions. This study presents the radiobiology beam line that has been designed at the 18 MeV proton cyclotron facility at the National Centre of Accelerators (CNA, Seville, Spain), to perform irradiations of mono-layer cell cultures. To ensure that all the cells receive the same dose with a suitable dose rate, low beam intensities and broad and homogeneous beam profiles are necessary. To do so, at the CNA an unfocused beam has been used, broadened with a 500 µm thick Aluminium scattering foil. Homogeneous dose profiles, with deviations lower than 10% have been obtained over a circular surface of 35 mm diameter for an incident average energy of 12.8 MeV. Further, A Monte Carlo simulation of the beam line has been developed with Geant4, and benchmarked towards ex- ∗Corresponding author Email address: [email protected] (Anna Baratto-Rold´an) article.pdf Click here to view linked References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 perimental measurements, with differences generally below 1%. Once validated, the code has been used, together with an ionization chamber, for dosimetry studies, to characterize the beam and monitor the dose. Finally, cultures of Human Bone Osteosarcoma cells (U2OS) have been successfully irradiated at the radiobiology beam line, investigating the effects of radiation in terms of DNA damage induction. Keywords: proton beams; Monte Carlo simulations; dosimetry; radiochromic films; radiobiology. 1. Introduction Radiobiological investigations using particle accelerators have gained interest in the last decades, coinciding with the worldwide spread of particle therapy centres and with the establishment of proton and carbon ion therapy as recognized treatment techniques for different types of tumours, with excellent clinical outcomes. The prescription of the physical dose in particle therapy goes through the knowledge of the relative biological effectiveness (RBE), which in the specific case of proton therapy is usually considered to be a constant value equal to 1.1 [1]. However, the conventional use of this value has been questioned for some time, as numerous studies suggest that the RBE varies with parameters such as tissue type, dose, biological endpoint and linear energy transfer (LET) [2, 3, 4, 5, 6, 7, 8]. But, despite the fact that some biophysical models have been proposed to predict proton RBE [9, 10, 11], none of them is currently applied in clinical proton therapy treatment planning, due to the uncertainties associated to both models and experimental data [12]. In order for clinics to use biologically optimized proton therapy treatment plans, more research is required to obtain accurate biological data, especially for proton energies typically found at the Bragg peak region of clinical beams (below 40 MeV) where the RBE is expected to increase. In this context, low energy proton beams are ideal for radiobiological experiments, since they can produce high LET values with narrow energy distributions [13]. Indeed, radiobiological 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 experiments with low energy protons should be ideally performed at facilities providing proton beams with nominal energy below the limit mentioned, in order to minimize the straggling related to passive degradation [13, 14, 15, 16, 17]. As described in [18], an experimental beam line installed at the 18 MeV proton cyclotron facility of the Spanish National Centre of Accelerators (CNA) in Seville, has already been adapted for the irradiation of mono-layer cell cultures placed vertically with respect to the beam direction. In this work, we present the improvements made to the beam line and its characterization with a dedicated Monte Carlo (MC) code developed with the simulation toolkit Geant4 [19, 20, 21]. This study aims at optimizing the beam line and validating the MC application developed, in order to further use MC simulations for dosimetry applications. Finally the first results obtained with U2OS Human Bone Osteosarcoma cells in terms of DNA damage induction are reported. 2. Materials and Methods 2.1. Beam Line The cyclotron facility at CNA (Cyclone 18/9 model, Ion Beam Applications, Louvain-La-Neuve, Belgium) is equipped with an external beam line for multipurpose research. The cyclotron accelerates protons and deuterons up to 18 and 9 MeV, respectively; the beam is then guided to the experimental room through the beam transport system and extracted in air, as depicted in Figure 1. A detailed description of the beam line can be found in [18]. As can be inferred from Figure 1, the beam dynamics of the cyclotron experimental line counts on a variable collimator consisting on a set of graphite slits, an XY magnetic steerer and a set of quadrupole magnets for the beam shaping and guidance. With the purpose of irradiating cell cultures, we implemented some modifications in the beam line to improve the irradiation conditions and meet the requirements for cell sample irradiation, i.e. homogeneous dose distribution throughout the biological sample and dose rate comparable to that conventionally used in clinics (2 Gy/min). These modifications were partially 3 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Figure 1: Schematic representation of the beam line. described in [18] and have been extended and improved in this work. First, the extracted beam current has been lowered to the minimum achievable and all the magnets for beam optics have been turned off in order to have a completely unfocused beam. With these conditions, the control of the position of the beam maximum intensity becomes harder, but can be achieved by resorting to the set of steering magnets placed immediately after the cyclotron exit port, as described in Section 2.3. Second, a retractable aluminium scattering foil of 500 µm thickness has been inserted in the vacuum pipe upstream the pneumatic isolation valve and the circular fixed aluminium collimator, of 15 mm diameter, depicted in Figure 1, at a distance of 2 m from the exit flange; this allows to broaden the beam profile, taking advantage of the angular spread given by its distance from the samples. Finally, we specifically designed an exit flange for the irradiation of biological samples, consisting of a 100 µm thick Mylar window mounted on a circular graphite collimator of 40 mm diameter. With these modifications, a higher degree of homogeneity can be achieved at the position of the samples, reducing the distance needed between the exit window and the samples and increasing the energy range available if compared to the previous setup presented in [18], where the beam was scattered in air using tungsten scattering foils of different thickness. 4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 2.2. Monte Carlo simulation In the cyclotron external beam line, information about the beam energy distribution and profile must be derived from measurements in air and their comparison with MC simulations. To this end, we implemented the geometry of the beam line in an MC code developed with Geant4 (version 10.5.0), and we built a dedicated and versatile simulation to reproduce the different experimental setups. Figure 2: Geometry of the Geant4 simulation developed to reproduce the cyclotron beam line. Figure 2 illustrates the geometry of the simulation. The primary protons are created 2 m far from the exit flange, immediately upstream the aluminium scattering foil. The various elements of the beam line have been implemented with an accuracy better than 2 mm in what respects their thickness and positioning. Then, we performed different experimental measurements to adjust the initial beam parameters, such as energy spread and shape, and validate the code. Since the aim of these measurements was to infer the initial characteristics of the beam, to be eventually used in other experimental applications, they were carried out removing the aluminium scattering foil from the beam line. Futhermore, depending on the type of experiment, the detector represented in Figure 2 and its position could be changed by a simple macro command, as all the detectors described in the following sections have been modelled independently. 5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 All the simulations have been carried out using a linux cluster (CentOS Linux 6.4) consisting of 24 servers, having each 24 cores and 64 GB minimum of RAM. 2.2.1. Determination of beam energy distribution A lithium-drifted silicon detector (L-040-075-5, ORTEC) has been used to measure the energy and energy distribution of the beam. This detector, powered with a bias voltage of 800 V, has an active thickness of 5 mm, sufficient to stop the 18 MeV protons coming from the cyclotron. To calibrate the detector and its electronic chain, a triple alpha source of 239Pu – 241Am – 244Cm was employed. Once calibrated, we measured the energy distribution of the beam in different configurations, by placing the detector in front of the exit window together with a graphite collimator of 1 mm diameter, and degrading the beam with tungsten scattering foils of six different thicknesses, ranging from 50 to 320 µm. A picture of the experimental setup is shown in Figure 3. Figure 3: Setup for the measurement of proton energy with a 5 mm thick silicon detector Finally, we compared the resulting experimental spectra with the simulations, to find the optimal values for the initial mean energy and energy spread to be inserted in the MC. To this end, a point-like proton source with Gaussian energy distribution has been simulated, setting the mean energy equal to 18 MeV (nominal value) and changing the standard deviation from 0.10 to 0.18 MeV in steps of 10 keV. The optimal value for the energy spread has been then chosen 6 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 as that which minimized the differences between experimental and simulated data for all the experimental configurations considered. 2.2.2. Determination of beam profile After determining the energy distribution of the beam, we carried out various measurements to determine its spatial distribution, using Gafchromic EBT3 films previously calibrated under known photon doses at the ”Virgen Macarena University Hospital” in Seville. A medical linear accelerator in 6 MV photon mode was used for this calibration, irradiating stripes of a sheet of EBT3 film from the same lot used at the cyclotron, with doses in the range 0-20 Gy, under standard conditions, i.e: 10x10 cm2field size, Source to Surface Distance equal to 10 cm and at a depth of 1.5 cm in solid water. After irradiation, we scanned the films with an Epson Perfection V700 photo scanner (Suwa, Nagano, Japan), and converted the acquired images from colour value to net optical density (netOD). The dose-to-optical density calibration curve obtained has been then used to convert in a first approximation the experimental netOD profiles measured at the cyclotron to dose, with the purpose of comparing them with MC ones. These profiles have been measured in air at seven different distances from the exit flange, from 0 to 66 cm, corresponding to an incident average beam energy ranging from 17.2 to 15 MeV. It must be noticed that, while the EBT3 response to photon radiation is generally independent from the energy, this is not true for proton radiation, where a saturation in the blackening, or quenching effect, emerges as the LET increases [22]. However, in the energy range considered this effect is negligible and the same calibration curve can be used for all the experimental configurations. Given the characteristics of the experimental beam produced, i.e. completely defocused and collimated with a 15 mm diameter collimator, we tested two initial spatial configurations in the MC code: uniform circular, with a radius of 8 mm, and two dimensional Gaussian, with a sigma varying from 8 to 20 mm independently in both directions on the transversal plane. 7 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 2.3. Setup for the irradiation of cell samples The setup for dosimetry and for the irradiation of biological samples consist of a PMMA holder, made of two supports of 1 cm thickness each, screwed to a plastic board placed on a moving table in front of the exit window. Each side of the sample holder has a circular collimator of 35 mm diameter (dimension of a standard Petri culture dish), to allow the insertion of a parallel plate ionization chamber (IC) between the two supports. This chamber operates at 400 V and it is connected to an electrometer for charge readout. It is made of three thin parallel kapton electrodes of 7.5 µm with two air gaps in between, and was especially designed at the GSI laboratory for radiobiological experiments. On the back of the holder, a custom made guide is mounted to allow for the positioning of Petri culture plates in the vertical position, with one plate aligned with the PMMA collimator and the beam each time. This position, finally, is also used for dosimetric studies by means of Gafchromic EBT3 type films. Considering the dimension of the wells of the Petri plate, beam homogeneity over a surface of at least 35 mm diameter must be achieved to ensure that all the cells of the sample receive the same dose. To do so, besides using a completely defocused beam and inserting the 500 µm thick aluminium scattering foil in the beam line, we decided to place the samples at a distance of approximately 25 cm from the exit window, to allow for a further broadening of the beam. However, these measures alone are insufficient, since the beam needs also to be centred so that the maximum of the intensity reaches the sample. Therefore, previously to the calibration, different pieces of EBT3 have been irradiated for approximately the same time to centre the beam, changing by small steps the current of the steering magnet coils to slightly move the beam and find the conditions that maximized beam homogeneity at the sample position. Figure 4 depicts the displacement of the maximum of beam intensity as a function of the steering magnets parameters. Following the identification of the best steering magnet configuration, corresponding to Figure 4d, a proton dose calibration curve has been measured, attaching pieces of EBT3 film to the back of the sample holder, as shown in 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Table 2: Experimental configurations used for the determination of the spatial characteristics of the beam to be compared with MC simulations. Configuration Window - Film distance [cm] Proton Energy [MeV] a0.5 17.2 b10.9 16.9 c15.7 16.7 d20.5 16.5 e25.3 16.4 f51.5 15.5 g65.9 15.0 As can be appreciated from Figure 10, the two-dimensional Gaussian elliptical beam having σx= 12 mm and σy= 20 mm is the one that best reproduces the experimental data. However, some small differences can be observed between the experimental and simulated profiles, which most probably are related to the initial in-homogeneity of the beam that is not reproduced in the MC simulations. These differences, nevertheless, become negligible as the distance from the exit window increases and multiple coulomb scattering in air becomes predominant, as shown in Figure 11. 3.2. Irradiation of cell samples Previously to the irradiation of the cell cultures, we built a proton dose calibration curve (to be used instead of the clinical photon calibration curve) and checked the dose homogeneity at the position of the samples, with the experimental configuration described in Section 2.3. Different EBT3 films were placed in the sample position, and irradiated with doses ranging from 1 to 17 Gy. These doses were computed by means of the validated MC simulation, from the charge collected by the IC, with Equations (1) and (2). Figure 12 shows the distributions of energy deposited respectively in the IC and in the active layer of the film, from which the mean values ∆EIC and ∆EAL have been extracted. 15 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 (a) (b) (c) (d) Figure 9: (a) Beam spot measured at a distance of 0.5 cm from the exit window (configuration aof Table 2). The colour scale is in dose units (Gy). (b) Beam spot simulated with a circular uniform distribution having a radius of 8 mm. (c) Beam spot simulated with a 2D circular Gaussian distribution having σx=σy= 12 mm. (d) Beam spot simulated with a 2D elliptical Gaussian distribution having σx= 12 mm and σy= 20 mm. In the conditions just presented, the proton beam arrives to the active layer of the film having a mean kinetic energy of 12.8 MeV and Gaussian distribution with 1.7% standard deviation. Experiments have been carried out with an approximate dose rate of 5-6 Gy/min: we observed maximum deviations from the mean netOD value lower than 10% in all the irradiated samples and all over 16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 (a) (b) (c) (d) (e) Figure 10: Lateral dose profiles along x (left) and y (right) directions, extracted from the measured (red) and MC (blue) beam spots plotted in Figure 9: (a) and (b) beam simulated with a circular uniform distribution with radius of 8 mm; (c),(d) and (e) beam simulated with a two dimensional Gaussian distribution having σx= 12 mm and σy= 12 mm (d) and σy= 20 mm respectively. x [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 scaled dose 0 0.2 0.4 0.6 0.8 1MC Measured (a) x [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 scaled dose 0 0.2 0.4 0.6 0.8 1MC Measured (b) x [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 scaled dose 0 0.2 0.4 0.6 0.8 1MC Measured (c) Figure 11: Lateral dose profiles measured (red) and simulated (blue) in some of the configurations reported in Table 2: (a) configuration c;(b) configuration eand (c) configuration g. 17 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Figure 12: Distribution of energy deposited per individual proton track in the ionization chamber (∆EIC) and in the active layer of the film (∆EAL). the irradiated surfaces, with relative standard deviations lower than 5%. The proton dose calibration curve for the red colour channel at the energy considered is plotted in Figure 13 and compared with the photon calibration curve. Table 3: Results obtained for the parameters of Equation (3) from the fit of the data in Figure 13. Only red channel is reported. parameter 12.8 MeV proton 6 MV photon a[Gy] 6.0±0.9 5.4±0.2 b[Gy] 67 ±10 57 ±2 c3.1±0.3 3.03 ±0.07 Following the indications in [23], the curves drawn in Figure 13 have been fit with the following function: D=a·netOD + b·netODc,(3) . where a,band care the fitting parameters, whose values are collected in Table 3. The measurements performed for the calibration of the EBT3 films have been also used to build a charge-to-dose calibration curve for the irradiation of the 18 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Figure 13: Dose to net optical density calibration curve for the red colour channel. Experimental data are presented together with their fit, whose parameters are reported in Table 3. The dotted line corresponds to protons reaching the film with 12.8 MeV and the continuous line corresponds to the photon calibration. biological samples. To this end, a dedicated Monte Carlo simulation has been developed to derive the energy deposited in the layer of cells irradiated, assuming water as the cells material. A thickness of 11 µm has been considered for the mono-layer cell culture, corresponding to the mean value of the thicknesses of cells in different phases of their cycle, measured with a confocal microscope. The calibration curve obtained is depicted in Figure 14, where the points have been fit with a linear function. Figure 15 shows the result of the quantitative analysis of γH2AX foci and the comparison between photon (average LET in water ≈0.35 keV/µm [24]) and proton (average LET in water ≈3.7 keV/µm) irradiation. Only positive cells, i.e. cells with a number of foci greater than 10 have been taken into account. In Figure 15a the variation of the median number of γH2AX foci per nucleus as a function of the dose is depicted, while Figure 15b reports the variation with dose 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Figure 14: Charge-to-dose calibration curve. Experimental points have been fit with a linear function of the form p0+ p1·x: where p0=−0.003 Gy and p1= 0.0026 Gy/nC. of the median total intensity of foci per nucleus. It must be noticed that proton doses are slightly higher than photon doses due to the difficulties associated to manually stop proton irradiation at an exact amount of charge collected in the IC. The asymmetric uncertainty intervals associated to the experimental points correspond to the 25th (lower) and 75th (upper) percentile of the distributions of the number of foci per nucleus and of the total intensity of foci per nucleus. Experimental points have been normalized to their controls (results obtained at 0 Gy) and fit with a linear function. No evident difference between the two radiation qualities emerges from the comparison of the number of foci per nucleus. However, a dependence of the total intensity of foci per nucleus on the incident particle can be observed, which could probably be related to the more clustered damage caused by protons for the same level of absorbed dose. 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 (a) (b) Figure 15: (a) Median number of γH2AX foci per positive nucleus (more than 10 foci) as a function of the dose after proton (red) and photon (blue) irradiation. The asymmetric uncertainty intervals associated to the experimental points correspond to the 25th (lower) and 75th (upper) percentile of the distributions of the number of foci per nucleus. (b) Median total intensity of γH2AX foci per positive nucleus (more than 10 foci) as a function of the dose after proton (red) and photon (blue) irradiation. The asymmetric uncertainty intervals associated to the experimental points correspond to the 25th (lower) and 75th (upper) percentile of the distributions of the total intensity of foci per nucleus. The experimental points are reported normalized to the their controls and fit with a liner function. 4. Discussion In this work, we have presented an improved experimental setup for the irradiation of mono-layer cell cultures at the cyclotron accelerator facility at the CNA, and reported the preliminary results of the first radiobiology experiment performed. One major modification, with respect to what we published in 2018 [18], has been the insertion of a removable scattering foil directly in the vacuum pipe and 2 m upstream the exit flange, which contributed to reach the same degree of homogeneity while increasing the energy range available for radiobiology experiments and reducing the air distance between the samples and the exit window. A 500 µm Aluminium foil has been initially used for this purpose, as it was easy to retrieve and mount. However, even better results in terms of homogeneity and energy could probably be obtained by changing the 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 scatterer material, increasing the atomic number and reducing the foil thickness. Further, we have implemented the geometry of the beam line and cell irradiation setup in Geant4, developing a dedicated MC application for testing purposes, and validating and benchmarking it towards comparison with dosimetric data. The initial MC beam energy and spatial distributions have been determined by running multiple simulations, changing by small steps one parameter at a time, such as energy spread and shape of the spatial distribution, in order to find the set of parameters that better reproduced experimental data obtained from measurements with a silicon detector and EBT3 films. An initial proton beam energy of 18 MeV with a Gaussian energy spread of 0.14 MeV gave the best agreement between experimental and simulated results, with percentage deviations generally lower than 1%. The highest discordances between measurements and MC, relatively to the energy distribution, have been observed for the thicker tungsten foils used to degrade the beam, as can be seen in Figure 7. Two different factors might have influenced this comparison: (1) the lower statistics collected experimentally for these points, (2) the uncertainty associated to the thickness of the foils, which the manufacturer reports to be up to 10%. Regarding the reproducibility of lateral dose profiles, a 2D elliptical Gaussian MC beam spot, having σx= 12 mm and σy= 20 mm, gives the best agreement with the experimental profiles measured with EBT3 films. The asymmetry in the measured dose profiles, especially in the horizontal (x) direction for profiles detected at a closer distance from the exit window, does not emerge from MC simulations. This effect is probably related to an initial misalignment of the beam coming from the cyclotron, which can be partially corrected by resorting to the magnetic steerers. However, as the distance from the exit window increases, this effect fades away, and the broadening due to multiple coulomb scattering in air becomes predominant. Finally, small differences between MC and experimental profiles could also arise from the calibration curve used to scale measured netOD profiles to dose. Indeed, a photon dose calibration curve (green solid line in Figure 13) has been used to this purpose in the first approximation. However, since the quenching effect observed in EBT3 films under 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 proton radiation decreases as the proton energy increases, the ”true” calibration curves for the energies considered (between 15 and 18 MeV) are expected to lay between the 12.8 MeV proton curve (dotted line in Figure 13) and the photon one, and the effect to be negligible. A homogeneous lateral dose profile is beneficial in cell experiments, to ensure that the dose stays constant throughout the dish at a given depth. Thanks to the insertion of the 500 µm Aluminium foil in vacuum, maximum deviations from homogeneity lower than 10% have been observed in all the irradiated samples, at a distance of 25 cm from the exit window and for an approximate dose rate of 5-6 Gy/min; with this setup, the mean energy of protons is 12.8 MeV at the sample position. However, homogeneity seemed to worsen as the irradiation proceed in time, probably due to the overheating of the steering magnets coils with subsequent displacement of the beam maximum intensity. This effect is shown in Figure 16, where the map of the deviation from mean netOD computed pixel by pixel is represented for two pieces of film irradiated with the same dose (approximately 10 Gy) respectively at the beginning and at the end of the experimental campaign. Even if this effect can be considered negligible for the film calibration and computation of the dose, it must be taken into account and minimized when it comes to the irradiation of biological samples. Therefore a check on beam homogeneity and settings of the steering magnet parameters is strongly recommended before any irradiation of biological material. This study shows the potential for performing cell irradiation for exploring the RBE of protons, and presents the feasibility of our setup to produce proton beams for radiobiology experiments. To test the performances of our setup, we conducted the first irradiation of cell cultures and studied the damage response of U2OS cells to proton and photon radiation. From Figure 15 no evident dependence emerges on the number of γH2AX foci generated after proton and photon irradiation. On the other side, a slight difference emerges in the total intensity of foci per nucleus, being the intensity of foci higher for protons at the same level of absorbed dose, probably indicating a higher and more localized concentration of DSBs per incident particle. These results are not unexpected, since in the 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 (a) (b) Figure 16: Map of the relative difference, computed pixel by pixel, of netOD from the mean netOD value (netODµ) in the irradiated surface. The two samples have been irradiated with the same dose (approximately 10 Gy) (a) and at the end (b) of the irradiation time after approximately two hours. range of LET considered for this preliminary experiment (≈0.35 keV/µm for photons and ≈3.7 keV/µm for protons in water), no significant differences are expected in the dependence of the biological effect on the LET [25]. To extract significant conclusions from this comparison, however, other experimental data should be collected, considering more dose points, increasing the statistics and comparing different cell lines. Furthermore, the dependence on the LET could be explored by further degrading the incident beam energy. 5. Conclusion An experimental beam line for the irradiation of mono-layer cell cultures at the 18 MeV proton cyclotron facility at the CNA has been presented. This line, based on a completely unfocused beam scattered with a 500 µm thick scattering foil placed 2 m upstream the exit flange, provides a homogeneous proton beam (maximum deviation from homogeneity lower than 10%) over a circular area of 35 mm diameter, with an incident narrow energy distribution on the sample having 12.8 MeV average value and a spread of 1.7%. Indeed, this is the main advantage of using this kind of facilities for radiobiology experiments, since they 24 x [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 scaled dose 0 0.2 0.4 0.6 0.8 1MC Measured 157profx-eps-converted-to.pdf y [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 scaled dose 0 0.2 0.4 0.6 0.8 1MC Measured 157profy.eps 20.png Click here to download high resolution image 22.png Click here to download high resolution image 24.png Click here to download high resolution image x [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 scaled dose 0 0.2 0.4 0.6 0.8 1MC Measured 253profx.eps x [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 scaled dose 0 0.2 0.4 0.6 0.8 1MC Measured 253profx-eps-converted-to.pdf y [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 scaled dose 0 0.2 0.4 0.6 0.8 1MC Measured 253profy.eps 26.png Click here to download high resolution image 28.png Click here to download high resolution image 62_g_20_x.png Click here to download high resolution image 62_g_20_y.png Click here to download high resolution image 62_u_8_x.png Click here to download high resolution image 62_u_8_y.png Click here to download high resolution image x [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 scaled dose 0 0.2 0.4 0.6 0.8 1MC Measured 659profx.eps x [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 scaled dose 0 0.2 0.4 0.6 0.8 1MC Measured 659profx-eps-converted-to.pdf y [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 scaled dose 0 0.2 0.4 0.6 0.8 1MC Measured 659profy.eps 7_10_r.png Click here to download high resolution image calebt3.png Click here to download high resolution image calexp.jpg Click here to download high resolution image x [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 scaled dose 0 0.2 0.4 0.6 0.8 1MC Measured y [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 scaled dose 0 0.2 0.4 0.6 0.8 1MC Measured exp_mc_05_profile-eps-converted-to.pdf x [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 y [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 D [Gy] 0 2 4 6 8 10 12 14 Measured x [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 y [mm] -50 -40 -30 -20 -10 0 10 20 30 40 50 D [Gy] 0 2 4 6 8 10 12 14 MC exp_mc_05-eps-converted-to.pdf MCA channel 200 300 400 500 600 700 [MeV] p E 4 6 8 10 12 14 16 18 Experimental Geant4 g4_exp_cal.eps MCA channel 200 300 400 500 600 700 [MeV] p E 4 6 8 10 12 14 16 18 Experimental Geant4 g4_exp_cal-eps-converted-to.pdf ifoci.png Click here to download high resolution image layout.jpg Click here to download high resolution image mcsim.png Click here to download high resolution image meanenergy.png Click here to download high resolution image nfoci.png Click here to download high resolution image sidet_setup.jpg Click here to download high resolution image