Radium deposition in human brain tissue: A Geant4-DNA Monte Carlo toolkit study

Abstract

NASA has encouraged studies on ²²⁶Ra deposition in the human brain to investigate the effects of exposure to alpha particles with high linear energy transfer, which could mimic some of the exposure astronauts face during space travel. However, this approach was criticized, noting that radium is a bone-seeker and accumulates in the skull, which means that the radiation dose from alpha particles emitted by ²²⁶Ra would be heavily concentrated in areas close to cranial bones rather than uniformly distributed throughout the brain. In the high background radiation areas of Ramsar, Iran, extremely high levels of ²²⁶Ra in soil contribute to a large proportion of the inhabitants' radiation exposure. A prospective study on Ramsar residents with a calcium-rich diet was conducted to improve the dose uniformity due to ²²⁶Ra throughout the cerebral and cerebellar parenchyma. The study found that exposure of the human brain to alpha particles did not significantly affect working memory but was significantly associated with increased reaction times. This finding is crucial because astronauts on deep space missions may face similar cognitive impairments due to exposure to high charge and energy particles. The current study was aimed to evaluate the validity of the terrestrial model using the Geant4 Monte Carlo toolkit to simulate the interactions of alpha particles and representative cosmic ray particles, acknowledging that these radiation types are only a subset of the complete space radiation environment.

1. Introduction

NASA has encouraged studies on ²²⁶Ra deposition in the human brain because the organization is interested in investigating the biological, medical, and cognitive effects of exposure of human brain tissue to high linear energy transfer (LET) alpha particles [1]. To explore the reason behind this interest, Dr. John Boice, Past-President of NCRP, and his colleagues have explained that alpha particles, at least to some extent, mimic astronauts’ exposure to high charge and energy (HZE) particles during anticipated missions to the Moon, Mars and similar space exploration missions. In other words, they believe that a better surrogate for the cognitive effects of long-term exposure of the human brain to high-LET radiation during space travel, actual human brain tissue exposed to alpha particles on Earth for years, may be more relevant than mouse brain tissue exposed to heavy ions for a short time (∼few minutes). Our collaboration has criticized this terrestrial model [2], [3], since that despite the attractiveness of this model, neither Boice nor NASA paid sufficient attention to the fact that ²²⁶Ra, as an alkaline earth element, is a bone-seeker and accumulates in the skull. Thus, the alpha particles emitted from ²²⁶Ra with energies of about 5 MeV have a very short range (maximum range can be about tens of µm), and the radiation dose due to these particles would be heavily concentrated in areas in close proximity to cranial bones rather than being uniformly distributed throughout the cerebral and cerebellar parenchyma. Because of this, even with higher total uptakes of ²²⁶Ra, the dose to various neuroanatomical compartments of the human brain would not represent the space radiation environment.

Our previous studies indicated that in the high background radiation areas of Ramsar, Iran, the extremely high levels of ²²⁶Ra in the soil contribute a very large proportion of the inhabitants’ radiation exposure [2], [3], [4]. It is worth noting that people in these areas mostly consume foods locally grown in this radium-rich soil. As we were aware of the Achilles heel of radium ingestion studies for space research, we designed a prospective study on Ramsar residents with a calcium-rich diet to improve the uniformity of dose due to ²²⁶Ra throughout the cerebral and cerebellar parenchyma. Our study showed that exposure of the human brain to alpha particles did not significantly affect working memory. However, a significant correlation was found between radium ingestion level and reaction times. Residents with higher levels of ingestion of ²²⁶Ra had longer reaction times. This finding was crucial because the same situation might occur in future deep space missions during which the astronauts’ brain cells will be exposed to HZE particles. As astronauts should be able to rapidly respond to the wide variety of new situations they face, increased reaction times, along with other potential cognitive impairments, may endanger the success of the mission.

While acute exposure to low-LET radiation on inbred lab animals at terrestrial experiments is cost-effective and straightforward, they may not reflect the actual radiobiological hazards of manned space missions. A more predictive model would involve studying humans chronically exposed to high-LET radiation similar to that which may be experienced in space. Regions on Earth with high background radiation, such as Ramsar, Iran, offer an opportunity to examine the health of residents and gain insight that could be relevant to space missions.

Our present study attempts to evaluate the validity of the terrestrial model using the Geant4 Monte Carlo (MC) toolkit [5] to simulate the interactions of the 5.59 MeV alpha particles with water as a biological material. By simulating a mammalian cell model, a comparison of DNA damage induced by GCR as well as alpha particles (5.59 MeV) has been investigated. It should be noted that the complete space radiation environment, including Solar Particle Events (SPE) and Galactic Cosmic Radiation (GCR), spans a considerable range of energies and charged particles. Therefore, we used the reference energy spectrum models available in the literature.

2. Materials and methods

2.1. MC simulation toolkit

Geant4-DNA [6], an extension of the Geant4 MC toolkit (version 11.0), was used to simulate the interactions of 5.59 MeV alpha particles and HZE particles, as well as the interactions of the secondary electrons with water. It is an appropriate tool to model biological damage induced by ionizing radiation at the scale of the DNA structure. The “G4EmDNAPhysics_option2” physics list, which is recommended for cellular and sub-cellular scale simulations, was implemented in this study [7], [8]. It includes several physics models covering the interactions needed for particle transport in water. For alpha particles, the implemented physics models were:

• G4DNAIonElasticModel (0.0001–1 MeV),

• G4DNAMillerGreenExcitationModel (0.001–400 MeV),

• G4DNARuddIonisationModel (0–400 MeV), G4DNADingfelderChargeDecreaseModel (0.001–400 MeV),

for nuclear scattering, electronic excitation, ionization, and electron capture, respectively.

For ions, the implemented physics model was:

G4DNARuddIonisationExtendedModel (0.5–10⁶ MeV/n) for ionization. It should be mentioned that the transport of all ions is not supported by the Geant4-DNA toolkit. However, the dominant ions in the cosmic spectrum (i.e., ⁷Li, ⁹Be, ¹¹B, ¹²C, ¹⁴N, ¹⁶O, ²⁸Si, ⁵⁶Fe) can be transported.

For electrons, the implemented physics models were:

• G4DNABornExcitationModel (9 eV–1 MeV),

• G4DNAChampionElasticModel (7.4 eV–1 MeV),

• G4DNABornIonisationModel (11 eV–1 MeV),

• G4DNAMeltonAttachmentModel (4–13 eV),

• G4DNASancheExcitationModel (2–100 eV),

for electronic excitation, elastic scattering, ionization, molecular attachment, and vibrational excitation, respectively.

The default energy cutoff for electrons is 7.4 eV, below which the transport of electrons stops, and their remaining energy is deposited locally. Other physics models are also available [7], [8]. All these interactions occur in the physical stage, i.e., the first stage of the Geant4-DNA simulation process. The Geant4-DNA has the capability of simulating chemical interactions through the pre-chemical and chemical stages [9], [10]. For a comprehensive list of available chemical interactions and the reaction rates of chemical species production, which are implemented in Geant4-DNA, refer to [10], [11]. The chemical stage was accelerated by using the independent reaction times (IRT) method [12] (i.e., the “G4EmDNAChemistry_option3” class) instead of the dynamic step-by-step approach [10]. The number of generated chemical species ($O{H}^{\bullet }$, $H^{\bullet }$, $H_2^{\bullet }$, $e_{\mathit{aq}}^{-}$, $H_2{O}^{+}$, $H_2{O}_2$), following the water radiolysis in the nucleus, was obtained up to 2.5 ns after the beginning of the interactions of the 5.59 MeV alpha particles and a GCR model. Among the above-mentioned reactive oxygen species (ROS), the number of hydroxyl radicals ($O{H}^{\bullet }$ radical) is highly important due to their high risk of interactions with the DNA molecules [13].

2.2. Geometry and source simulation

A combination of two Geant4 examples was used in this study: the “clustering” [14] and the “wholeNuclearDNA” example [15]. The structure of a fibroblast cell nucleus based on the “wholeNuclearDNA” example includes different geometric levels: nucleosome (which is defined as a histone protein core wrapped by a 200 bp length DNA double helix), chromatin fiber (contains 90 nucleosomes), flower-shaped loops, chromosome territories and an ellipsoidal nucleus with 23.64×17.04×6 µm long axes, which is seen in Fig. 1. The histone protein is defined as a cylinder with a radius of 3.75 nm and a height of 5.7 nm. The fibroblast cell nucleus model comprised 46 chromosome territories, 11875 flower-shaped loops, 332500 chromatin fibers, and 29925000 nucleosomes [15]. The total genome length is nearly 6 giga base pairs (Gbps). The nucleus was placed in an ellipsoid with 26×20×10 μm long axes as the cytoplasm. The material of all geometric levels is water with a density of 1 g/cm³. The primary particles were emitted from an ellipsoidal surface with the exact dimensions as the cytoplasm from the left side towards the simulated geometry.

Figure 1.

The geometric levels of forming a fibroblast cell nucleus from left to right: nucleosome (histone protein + DNA double helix), chromatin fiber (contains 90 nucleosomes), flower-shaped loops, chromosome territories (green cubes) and an ellipsoid nucleus with 23.64×17.04×6 µm long axes.

Many attempts have been made to model the GCR spectrum [16], [17], [18]. Although it is impossible to implement the full GCR spectrum accurately, a reasonable estimate of the GCR spectrum can be simulated with the help of the data available in the literature. We followed reference [18], which presented the developed version of the Badhwar‐O'Neill model, to define the GCR spectrum. Fig. 2 shows the relative abundance of ions with an atomic number (Z) of 1 to 28 in the GCR spectrum.

Figure 2.

The relative abundance of ions with an atomic number (Z) of 1 to 28 in the GCR spectrum [18].

2.3. Damage yield calculation

To simulate the direct yields of the DNA damage, the positions and the energy depositions of all particles were obtained in the target volume (i.e., sugar-phosphate backbone). The frequencies of single- and double-strand breaks (SSBs and DSBs) were calculated based on the clustering example of Geant4. In this example, a damage probability with a linear relationship between two energy thresholds of 5 and 37.5 eV was implemented. If the energy deposited in the target volume was less than 5 eV, the probability of DNA break (i.e., an SSB) was zero, and if it was more than 37.5 eV, the break probability was 1. The break probability increases linearly between these two thresholds with increasing the deposited energy. If the distance between two subsequent SSBs in opposite strands was less than the threshold of 3.4 nm, one DSB would be counted. Furthermore, we developed the clustering example to consider indirect DNA damages. To estimate the indirect damage yields, we only consider $O{H}^{\bullet }$ radical as the most reactive chemical species. The number and positions of all $O{H}^{\bullet }$ radicals produced in the target volume were recorded. In Geant4-DNA, the total simulation time for the chemistry stage is up to 1 µs. However, for DNA damage calculations, a time of up to 2.5 ns is considered. Limiting the time of the chemistry stage leads to a simple modeling of the scavenging process of free radicals that occurs in reality and reduces the possibility of indirect damage [19]. Note that not all $O{H}^{\bullet }$ radicals lead to strand breaks. A probability of 40% for the interaction of the $O{H}^{\bullet }$ radical with the DNA molecule was assumed following [20]. According to Charlton et al. [21], a damage yield of any kind is obtained by eq. (1):

$$\mathit{Damage}yield\left(G{y}^{-1}{\mathit{Gbp}}^{-1}\right)=\frac{N_{\mathit{break}}}{D\times N_{\mathit{bp}}}$$ (1)

in which N_break is the number of strand breaks, D is the total absorbed dose to the cell (Gy), and N_bp (Gbp) is the total number of bps. Note that the DSBs may be induced by direct (physical stage), indirect (chemical stage), and hybrid effects. The classification of DNA damage is shown in Fig. 3. With 5000 primary particles, all statistical uncertainties were less than 1%. Approximately, it took 50 h and 55 h in a PC of a 60-core 3.8 GHZ processor, 100 GB RAM for 5.59 MeV alpha particles and GCR simulations, respectively.

Figure 3.

DNA damage classification. (a) Direct and indirect single-strand breaks (SSBs). (b) A direct double-strand break (DSB) is formed by two direct SSBs with a distance of less than 3.4 nm. (c) An indirect DSB is formed by two indirect SSBs with a distance of less than 3.4 nm or 10 bp. (d) A hybrid DSB.

3. Results

To validate the simulation, we defined a radiation source similar to the work of Friedland et al. [22] and compared the results. A circular source was simulated, where alpha particles and carbon ions of different energies were emitted from its surface towards the cell. Friedland et al. obtained direct and indirect damage yields induced by several light ions using the PARTRAC MC code [23]. Fig. 4 compares the total (direct, indirect, and hybrid) DSBs caused by helium- and carbon-ion beams as a function of LET. The error bars in Fig. 4 are 5% of the extracted PARTRAC results. As can be seen, there is a good agreement between the results.

Figure 4.

A comparison of the total yield of double-strand breaks between the Geant4-DNA (in our study) and the PARTRAC code [22] for incident alpha particles (A) and C-12 ions (B).

Fig. 5 illustrates the normalized yield of SSBs (direct and indirect) and DSBs (direct, indirect, and hybrid) obtained by eq. (1). The total dose delivered to the cell nucleus per incident particle, the ratio of direct/indirect DNA damage, and the ratio of total SSB/DSB DNA damage are tabulated in Table 1 for each radiation field.

Figure 5.

The normalized DNA damage yields of single-strand breaks (A) and double-strand breaks (B).

Table 1.

The total absorbed dose to the cell nucleus per particle, the ratio of direct/indirect DNA damage, and the ratio of SSB/DSB DNA damages for alpha particles and the GCR model.

Direct/Indirect	SSB/DSB	Dose (Gy/particle)	Radiation
0.296	22.74	0.0118	alpha (5.59 MeV)
0.324	23.03	0.0127	GCR spectrum

Fig. 6 shows the G-values of several water radiolysis products as a function of time, i.e., the number of produced ROS ($O{H}^{\bullet }$, $H^{\bullet }$, $H{2}^{\bullet }$, $e_{\mathit{aq}}^{-}$, $H_2{O}^{+}$, $H_2{O}_2$) per 100 eV energy deposition, up to 1 μs following the physical stage of the simulation. The error bars are not shown for clarity. It should be noted that no time limit has been applied in the calculation of G-values so that chemical species are considered until the end of their diffusion range. The mean number of common chemical species produced in the nucleus within 2.5 ns after the physical stage is displayed in Fig. 7. The yield of chemical species is normalized to one primary particle. The error bars in Fig. 7 indicate a statistical uncertainty of 5%.

Figure 6.

The G-values of typical chemical species of water radiolysis as a function of time for 5.59 MeV alpha particles (A) and the GCR spectrum model (B).

Figure 7.

The normalized chemical species yields in the cell nucleus after 2.5 ns following the physical stage for 5.59 MeV alpha particles and GCR radiation.

4. Discussion

The use of acute exposures of low-LET radiation on inbred lab animals for terrestrial experiments is simple and inexpensive but does not give all the needed information about the actual radiobiological hazards of manned space missions. A more predictive model could involve studies of humans who have chronically been exposed to high LET radiation similar to what could be experienced in space. The ongoing high LET radiation exposure in Ramsar, Iran, and other high background radiation regions on Earth provides an opportunity to examine the health of residents and gain insight that could be relevant to space missions [24]. Our previous studies [2], [3], [4] have shown that exposure of the human brain to alpha particles does not significantly affect working memory, but it influences the reaction times. Residents with higher levels of ingestion of ²²⁶Ra have longer reaction times. This finding is important because the same situation might occur in future deep space missions during which the astronauts’ brain cells will be exposed to HZE particles, which might induce oxidative stress and neuroinflammation that can lead to neuronal damage and cognitive deficits, including memory loss and reduced reactions times. As astronauts should be able to rapidly respond to the wide variety of new situations they face, increased reaction times, along with other potential cognitive impairments, may endanger the success of the mission.

The Geant4-DNA MC toolkit can simulate the production of radicals and DNA damage, but it cannot simulate the correlation of radiation-induced oxidative stress, neuroinflammation, and cognitive deficits. However, our present study using Geant4-DNA is an attempt to better understand how much absorbed dose the cell nucleus will get per primary particle from exposure to 5.59 MeV alpha particles (representing residents with high levels of ingestion of ²²⁶Ra) and a model of GCR radiation (representing astronauts being exposed to GCR), as well as from the secondary electrons, in water. Calculations using the Geant4-DNA toolkit show that the total direct DNA damage (normalized to the DNA length and the total absorbed dose in the nucleus) induced in the nucleus is 53.9 and 49.1 Gy^-1Gbp^-1 irradiated by 5.59 MeV alpha particles and GCR, respectively. The total indirect DNA damage is about 167.4 and 166.8 Gy^-1Gbp^-1 for alpha particles and GCR, respectively. As seen in Fig. 5, the DNA damage yield caused by cell irradiation from alpha particles and GCR is quantitatively close. The normalized absorbed dose to the cell nucleus is 0.0118 and 0.0127 Gy per primary alpha particles and GCR, respectively. Although there are different particles in the GCR spectrum, the dose deposited per primary particle in the cell is almost the same for 5.59 MeV alpha particles and the GCR model. This indicates that these two models may have similar effects on a mammalian cell. According to Figure 6, Figure 7, although the difference in the G-values obtained for the two radiation fields reaches up to 25%, the normalized chemical species produced in the cell (per particle) are almost the same. It has been shown that indirect DNA damage induced by high-LET radiation is lower than that caused by low-LET radiation [25]. However, the impact of secondary electrons is vital and can significantly increase the indirect effect. This is what is shown as the ratio of Direct/Indirect DNA damage in Table 1. Considering that the GCR flux of alpha and HZE particles is of the order of 10^-5-10² particles cm^-2 (MeV/n)^-1 day^-1, for each particle, depending on the particle, its energy, and on the solar cycle, an 8-10 months one-way journey to Mars might cause a significant damage to the brain.

5. Conclusions

In this study, a comparison of induced DNA damage in a fibroblast cell nucleus exposed to 5.59 MeV alpha particles, and a model of GCR, was performed using the Geant4-DNA MC toolkit. The results indicate that the induced DNA damage in a human cell caused by these two radiation fields is nearly identical. This would imply that the astronauts could develop the same cognitive damage as Ramsar inhabitants. This is an important result that should be investigated in future space studies. Note that one of the limitations of the Geant4-DNA MC toolkit is that the maximum possible energy for ion transport is up to 1 MeV/n, while the maximum kinetic energy of HZE particles in the implemented GCR model is up to 1 GeV/n. However, the resulting information, even with limited cross-sections, will contribute to a better understanding of our Ramsar-based terrestrial model of how ²²⁶Ra alpha particle radiation might cause neuronal damage and cognitive deficits. Given the importance of these results to the success of future space missions, a thorough evaluation of the space radiation environment is required. This evaluation should include the full spectrum of particle types and their energies as encountered in the SPE and GCR source terms. Nevertheless, the proposed model characterizes the extent of DNA damage induced by the representative source term.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.