Significant past work has identified unexpected risks of central nervous system (CNS) exposure to the space radiation environment, where long-lasting functional decrements have been associated with multiple ion species delivered at low doses and dose rates. As shielding is the only established intervention capable of limiting exposure to the dangerous radiation fields in space, the recent discovery that pions, emanating from regions of enhanced shielding, can contribute significantly to the total absorbed dose on a deep space mission poses additional concerns. As a prerequisite to biological studies evaluating pion dose equivalents for various CNS exposure scenarios of mice, a careful dosimetric validation study is required. Within our ultimate goal of evaluating the functional consequences of defined pion exposures to CNS functionality, we report in this article the detailed dosimetry of the PiMI pion beam line at the Paul Scherrer Institute, which was developed in support of radiobiological experiments. Beam profiles and contamination of the beam by protons, electrons, positrons and muons were characterized prior to the mice irradiations. The dose to the back and top of the mice was measured using thermoluminescent dosimeters (TLD) and optically simulated luminescence (OSL) to cross-validate the dosimetry results. Geant4 Monte Carlo simulations of radiation exposure of a mouse phantom in water by charged pions were also performed to quantify the difference between the absorbed dose from the OSL and TLD and the absorbed dose to water, using a simple model of the mouse brain. The absorbed dose measured by the OSL dosimeters and TLDs agreed within 5–10%. A 30% difference between the measured absorbed dose and the dose calculated by Geant4 in the dosimeters was obtained, probably due to the approximated Monte Carlo configuration compared to the experiment. A difference of 15–20% between the calculated absorbed dose to water at a 5 mm depth and in the passive dosimeters was obtained, suggesting the need for a correction factor of the measured dose to obtain the absorbed dose in the mouse brain. Finally, based on the comparison of the experimental data and the Monte Carlo calculations, we consider the dose measurement to be accurate to within 15–20%.
Recent data have highlighted some unexpected and potentially concerning effects of cosmic-radiation exposure on the brain (1, 2). Animal models, subjected to charged particles ranging from protons to the heavier ionized nuclei typically found in the galactic-cosmic ray (GCR) spectrum, have revealed that such exposures lead to neurocognitive complications involving a range of associated pathologies (3). Importantly, space relevant doses ≤50 cGy of nearly any individual or combination of ions have now been found to elicit persistent neuroinflammation, degradation of mature neurons throughout the brain, and impairments in learning and memory along with a host of mood disorders (4, 5). Even more concerning for space agencies is that most of these cosmic radiation-induced disruptions in the central nervous system (CNS) functionality do not appear to resolve with time (1), even over the course of one year following exposure (2, 6).
A relatively recent finding shows that astronauts engaged in deep space travel may receive up to <20% of their total absorbed dose from pions, including exposures to positive and negative pions and photons derived from neutral pion decay (7). Importantly, pion doses were found to increase with shielding thickness. While absorbed dose can be modeled and calculated, it informs little on biological dose equivalents, and points to the need to more fully understand organ specific dose equivalents and the potential risks associated with pion exposures. While potentially promising, the use of pions in the clinic never gained traction, likely due to difficulties in generating homogeneous beams and minimal improvements in the therapeutic index (8, 9). However, in the setting of a highly interconnected network such as the brain, capture reactions associated with negative pions may turn out to be particularly deleterious. To our knowledge, pion-induced neurocognitive deficits at space relevant total doses (≤50 cGy) have never been evaluated, and as such, represents a significant and problematic gap in knowledge for space agencies (3).
To investigate the biological effect of pion irradiations, we irradiated mice with 150 MeV charged pions from the PiM1 beam line at the Paul Scherrer Institute (PSI). Pion irradiation does not benefit from a well-established dosimetry. Only few pion dosimetry studies were published in the 1970s and 1980s when the potentiality of pion negative beam therapy were considered (10–13). As a prerequisite to biological studies, evaluating pion-induced absorbed dose to water for various mice exposure scenarios is required. For this purpose, beam profiles and beam contamination by protons, electrons, positrons and muons were characterized. The dose at the head and the tail of the mice were measured with thermoluminescent dosimeters (TLD) and optically simulated luminescence (OSL). The redundant use of two different types of dosimeters allowed cross-validating the dosimetry results. Geant4 Monte Carlo simulations were also performed to quantify the difference between the dose measured by OSL and TLD and the dose absorbed in a mouse brain modeled by water (14).
This article presents the characterization of the PiM1 beam and the dosimetry used during radiation exposures of mice at PSI.
MATERIALS AND METHODS
Description of the PIM1 Secondary Beam at Paul Scherrer Institute
Figure 1 shows the secondary PiM1 beam line at the Paul Scherrer Institute. Secondary pions, muons, electrons and protons are generated by the interaction of the high-intensity 590 MeV initial proton beam with a 2-mm thick graphite target and are guided by the PiM1 beam line to the experimental area (15, 16). The main proton beam and the secondary beam are pulsed with a 20 ns cycle. A couple of dipole magnets selects the momentum of the secondary charged particles while the quadrupole magnets are used for the focusing of the beam. The polarity of the magnets can be changed to select a positive or a negative charged beam. A 4 mm polymethylmethacrylat (PMMA) plate, located between the two dipoles, and a low-magnetic field on the second dipole ensure to select pions. This separation was crucial for our experiment as the contamination of protons is significant for 150 MeV pions and detrimental for our radiobiology experiment. Positively charged particles of a given momentum (here 254 MeV/c) are selected by the first magnet. Protons having higher stopping power than the other particles composing the secondary beam (charged pions, positrons, muons) lose more momentum in the PMMA plate placed between the dipole magnets and are, therefore, kept in the beam by selecting a lower current in the second magnet than for the other particles.
Description of the Experimental Setup
Mice were irradiated with 150 MeV positive and negative pions of the PiM1 beam line. The setup used for the calibration measurement and the mice irradiations is shown in Fig. 2. The beam was characterized by a plastic scintillator (PIL detector) with a 12.028 mm2 sensitive area and set on an X-Y motorized stage. We measured lateral beam profiles by scanning the detector across the beam. The stage was then used to place the mice at the beam center. We measured the contamination level of the beam with muons, electrons, and positrons using a time-of-flight analysis of the PIL detector signal relative to a pulsed signal with a 20 ns cycle, synchronized with the main proton beam. For the monitoring of the beam during irradiation of the mice, we used a second plastic scintillator (PL10) located at the exit of the beam off axis. Calibration factors of the counting rates of the PL10 detector as a function of the beam intensity measured by the PIL detector was determined during the calibration measurements.
The beam intensity was calibrated as measured by the PIL detector in term of absorbed dose to water for Co-60 for our radiobiological set-up. The dose induced by the pion beam was measured by thermoluminescent dosimeters (TLD) calibrated at the Institute of Radiation Physics’s calibration laboratory and optically stimulated luminescence (OSL) dosimeters calibrated at the PSI’s calibration laboratory, which are both traceable to the Swiss national metrology institute (METAS). The reported quantity is the absorbed dose to water (for Co-60). For TLDs, we used square TLD-100 LiF:Mg,Ti chips (Thermo Fisher) 3.2 mm × 3.2 mm × 0.9 mm (16). Each OSL dosimeter consisted of two BeO Thermalox 995™ 4.7 mm × 4.7 mm × 0.5 mm detector chips placed side by side and wrapped in a lightproof package. Before the exposures, the detectors were annealed at a temperature of 900°C for 15 min. The detectors were evaluated using the methodology described in the literature (17); after the OSL signal readout, a second identical readout of each detector was performed after a 20 s reference irradiation with a 90Sr/90Y 1.53 GBq beta source, equivalent to an absorbed dose to water (for Co-60) of 1,130 mGy. This second readout was used to individually calibrate the detectors, accounting for variations in their sensitivity and mass.
Calibration procedures used cylindrical water phantoms mimicking mice, which were located along the beam axis, with dosimeters at the entrance and exit of each phantom (see Figs. 3 and 4a) for the calibration of the positive pion beam. For the calibration of the negative pion beam, four mouse phantoms were irradiated simultaneously and positioned as shown in Fig. 4b.
As dosimetry based on passive dosimeters is time-consuming and cannot immediately provide information required for exploratory experiments typical of setup adjustments and optimization, we additionally used Gafchromic EBT3 film sheets (Ashland Inc., Wayne, NJ) for fast relative dosimetry. The dose was estimated measuring density changes in films using a hand-held densitometer. This simplified procedure made the optimization of beam parameters and set-ups more efficient.
After the lateral and longitudinal beam profiles was measured, a decision was made to irradiate mice in groups of 4. Consequently, we could irradiate 50 mice in two hours, which helped minimize the stay of mice in the experimental area. The mice were placed in the beam as shown in Fig. 5 by using cylindrical plastic containers to immobilize them, with their heads facing each other. The holder was centered on the beam axis, which located the beam lateral maximum on the line separating the adjacent mice. The cranial dose received by the mice was measured by OSL dosimeters and TLDs placed at the top (P1 positions in Fig. 5) of the mouse containers. OSL dosimeters were also placed at the opposite side of each container (positions P2 on Fig. 5) for each irradiated mouse while TLDs were placed at this position only for selected mice.
Dose Computation with Geant4 Monte Carlo Simulations
Geant4 Monte Carlo simulations were used to compute the absorbed dose deposited by the charged pion beams in water mouse phantoms (Fig. 6) (14). Four mice were modeled by water cylinders with 8 cm height and 3.5 cm diameter and were placed in the pion beam to model closely the irradiation geometry. OSL dosimeters and TLDs were placed in front and at the back of the water phantoms and were modeled as square chips, made, respectively, of beryllium oxide and lithium fluoride. Absorbed doses were computed in water at 5 mm depth from the surface of the phantom and in the OSL dosimeters and TLDs. The Geant4 shielding physics list was selected for all simulations (14). Simulations were performed with 150 MeV monoenergetic positive and negative pions, with no angular dispersion and with a Gaussian lateral profile. We fitted the full width half maximum (FWHM) obtained from the X and Y profile measurements (Table 1).
RESULTS AND DISCUSSIONS
Characterization of the Pion Beam
Figure 7 represents the intensity of the beam measured by the PIL detector as a function of the current of the second dipole used for the positively charged beam. The peaks at lower and higher magnet currents represent the proton and the pion beam, respectively. For mice irradiations, we used the corresponding value of the beam current of the second dipole that maximized the positive pion beam intensity.
Figure 8 represents the time-of-flight spectrum of the different particles present in the beam. The time of flight is presented as the time differential relative to a pulse signal with a 20 ns period (representing the period of the main proton beam). Different types of particles with identical momentum will take different times to travel along the beam line from the target to the detector PIL. Thus, particle discrimination can be detected as different peak signatures on the time-of-flight spectrum. With this set-up, expected time-of-flight differences between muons, electrons, and charged pions from the target to the PIL detector could be resolved, where specific particles could be assigned to a single peak in the distribution presented in Fig. 8. Results indicate that for positive and negative charged beams pions are dominant, with a contamination of electrons/positrons and muons lower than 5% for the positive beam and lower than 20% for the negative beam.
The X (horizontal) and Y (vertical) lateral profiles of the positive and negative pion beams as measured by the PIL detector are shown in Fig. 9. The FWHM values of the X profiles and Y profiles are given in Table 1. These values of the FWHM were taken as input parameter of the Gaussian beam profiles in the Geant4 Monte Carlo simulations.
Tables 2 and 3 show the dose measured by OSL dosimeters and TLDs for the irradiation of mice in the 2 × 2 configuration illustrated in Fig. 5 with the positive pion beam. The target doses were 200 mGy and 50 mGy, respectively. In general, the measured doses at position P1 were 15–20% lower than the expected target doses, indicating the need to measure real time doses during the irradiation of mice. One possible reason for this discrepancy is the fact that the calibration measurements were performed with two water phantoms placed in tandem at the center of the beam as shown in Fig. 4a, while the mice were irradiated in groups of four as shown in Fig. 5. In this second configuration, the mice heads are located ∼2 cm laterally from the beam axis, where the beam intensity decreased by roughly 20% (see beam profile on Fig. 9).
Tables 4 and 5 present the dose measured by the OSL dosimeters and TLDs for irradiation of the mice with the negative pion beam to a target dose of 160 mGy and 45 mGy, respectively. The measured doses at position P1 agree within 5–10% with the expected target dose. This value is in better agreement than the predictions for positive pions.
In all the reported measurements, OSL and TLD readings show a good agreement: the average relative bias is lower than 6%, with few values above 10%. This validates our dosimetric strategy based on two different chains of traceability. However, inherent differences between pion and photon beams cannot be dismissed, because OSL and TLD possess comparable physical detecting principles, which may exhibit similar biases in pion beam measurements.
The dose measured is given in terms of dose in water relative to the 60Co reference irradiation. In principle, this dose should be corrected for the quality of the beam (mainly charged pions at 150 MeV) used during our irradiation. To our knowledge, such a correction has never been investigated for a pion beam. Nevertheless, the relative efficiency of OSL and TLD dosimeters has been studied for different beam qualities showing a dependence of the relative efficiency as a function of LET. Typical variations with respect to the Co-60 reference of >10% for LET > 1 keV/μm have been reported (19–22). The charged pion beam used in this study (254 MeV/c momentum) has a LET much lower than 1 keV/μm, thus the relative efficiency of the dosimeters for the pion beam should not be underestimated by more than 10%. For the purposes of this proof-of-principle study, we consider this value as acceptable with the acknowledged uncertainties.
With the mouse placement configuration used during irradiations, we ensured a similar dose exposure to the brain for all mice. However, for different parts of the body, the dose exposure for mice 1 and 2 and mice 3 and 4 are no longer equivalent, due to the different orientations along the beam axis. From the comparison of measured dose at positions P2 we see that this difference is around 30% towards the rear of the mice, a difference that is clearly much smaller closer to the head. Thus, these differences are not expected to have a significant impact on the outcome of neurocognitive testing conducted over the course of 1 year after exposure.
Dose Computation with Monte Carlo Simulations
Table 6 presents the dose computed by Geant4 for the water phantom irradiations with the positive pion beam. The same fluence at the center of the beam was considered as in the experimental irradiation of the mice at high dose. The dose computed in the OSL and TLD is 30–35% lower than the measured one. A possible explanation of this discrepancy is that beam dispersion at the entrance of the irradiated targets was not considered, since a pure, 100% pion beam was used for water phantoms in the simulations. Since the dose in water in the phantom computed with the simulation is higher than the dose computed in the TLD by 15–20%, the data suggests the use of a correction factor for the dose measured by the OSL dosimeters and TLDs of 1.15–1.2. These results again point to the need to perform real-time dosimetry during irradiation, to decrease the uncertainties on the absorbed dose to water.
In this article we have detailed the calibration and dosimetry measurements for the irradiation of mice with positive and negative pions from the PiM1 beam at Paul Sherrer Institute, Switzerland. This work was performed to support future radiobiological experiments and presented real dosimetric challenges. First, the beam was characterized by a plastic scintillator combined with a time-of-flight analysis and the absorbed dose was measured by TLDs and OSL dosimeters. Beam parameters were set to suppress the contamination of protons to the positive pion beam. From the time-of-flight analysis we measured a contamination of electrons/positrons and muons lower than 5% for the positive beam and at the level of 20% for the negative beam. The beam size and its intensity were sufficient for the simultaneous irradiation of 4 mice, where a total absorbed dose of ∼160 mGy could be achieved within 5 min. To assess any difference between the doses measured at the top of the mouse head by TLDs and OSL dosimeters and the effective absorbed dose in the brain (5 mm depth) we performed Geant4 Monte Carlo simulations of irradiation of mouse water phantoms by the PiM1 beam.
Table 7 summarizes the average doses measured by the TLDs and OSL dosimeters for the different series of mice irradiation performed in this study. The dose in water at 5 mm depth computed by Monte Carlo simulations is also presented. While the computed dose in OSL and TLD (see Table 6) is generally 30% lower than the measured values, the absorbed dose computed in water is only 15% lower. Based on the comparison of experimental data and Monte Carlo computations and on the correction needed between the dose measured and the dose in water, we consider the measurement of the dose accurate within 20%.
The authors would like to thank P. Socha for his help in the setting of the experiment. This work was supported by the NASA Specialized Center of Research (NSCOR) grant NNX15AI22G (CLL).