The study, published in EJNMMI Physics, compares the performance of AlfaMC against the general-purpose codes MCNPX and Geant4. The primary objective was to determine microdosimetric quantities for uniform and non-uniform distributions of astatine-211 within thyroid models for humans, rats, and mice. The results demonstrate that AlfaMC produces results within 5% of MCNPX and 2% of Geant4, confirming its accuracy for cellular and subcellular dosimetry.
Astatine-211 is an alpha-particle emitter with a half-life of 7.2 hours, making it suitable for radionuclide therapy. It emits alpha particles with energies of 5.867 MeV and 7.450 MeV, which have a short range of up to 70 micrometers in water. This short range results in high linear energy transfer, causing complex DNA damage that is more effective at killing cancer cells than beta-particle emitters. However, because astatine is a halogen, free astatine-211 accumulates in the thyroid gland, which becomes a primary normal organ at risk during treatment.
Accurate estimation of radiation absorbed dose to the thyroid is critical for patient safety. Traditional mean absorbed dose calculations may be insufficient for alpha particles due to the stochastic nature of energy deposition. Microdosimetry addresses this by considering the variation in energy deposited in different neighboring cells. The study utilized previously published single and multiple thyroid follicle models for three species to simulate these conditions. The models included spherical follicle lumens, layers of follicle cells, and six spherical follicle cell nuclei distributed symmetrically.
The researchers simulated four different distributions of astatine-211 within the thyroid models. These included uniform distribution within the follicle lumen, within the follicle cells, within the follicle cell nuclei, and on the surface of concentric spheres. The concentric sphere distribution represented a heterogeneous distribution, potentially mimicking astatine bound to a pharmaceutical agent. The targets for all simulations were the six follicle cell nuclei. The simulations tracked the transport of alpha particles and the energy imparted to these targets.
AlfaMC is a Monte Carlo code dedicated exclusively to alpha-particle transport. It was developed at the University of Lisbon and customized for calculating microdosimetric quantities. The code uses the Continuous Slowing Down Approximation and considers energy and angular straggling. A key advantage of AlfaMC is its computational efficiency, with a typical simulation run tracking 100,000 alpha particles taking approximately two minutes. This speed is significantly faster than general-purpose codes like MCNPX and Geant4, which require more complex setup and longer calculation times.
MCNPX is a general-purpose Monte Carlo code developed at Los Alamos National Laboratory, while Geant4 is an open-source toolkit developed at CERN. Both codes are capable of simulating the transport of various particles in complex geometries. However, they require specialized expertise in physics processes and programming. Geant4 was extended with the Geant4-DNA project to allow calculations at the molecular level, which is useful for radiobiology. The study compared the microdosimetric outputs of these three codes to assess the reliability of AlfaMC for thyroid dosimetry.
The mean specific energy and single-hit mean specific energy were calculated for each species and source distribution. The mean specific energy represents the average energy per mass, including zero-hit events, and is used for S-value calculations. The single-hit mean specific energy provides information about the average energy deposited by single alpha particle hits. The intercomparison showed that the largest difference between AlfaMC and MCNPX was 4.6%, which was between the rat and human models. The standard deviation for all Monte Carlo results was within 0.4%, indicating high statistical precision.
For the single thyroid follicle models, the agreement between AlfaMC and Geant4 was even closer, with differences up to 1.7%. The single-hit specific energy distributions obtained by AlfaMC agreed well with those from MCNPX and Geant4. The probability distributions were normalized to have an area of one, allowing for direct comparison of the shapes. The maximum and minimum values of the probability density functions occurred at approximately the same specific energy for each species and source location.
In the multiple thyroid follicle models, the researchers examined the effect of source radius on the specific energy. The radius of the concentric spheres varied from 0 to 150 micrometers with 5 micrometer increments. The results showed a strong agreement between AlfaMC and MCNPX for the human, rat, and mouse models. The ratio between the two codes was near one for most data points, except for two points in the human model where the ratio was 2.2 and 1.2.
The differences between the codes are attributed to their distinct methods of describing alpha-particle transport and energy deposition. AlfaMC uses the NIST ASTAR database for stopping power, which has stated uncertainties of 1-4% for high-energy alpha particles. MCNPX uses a condensed history approach, grouping many collision processes into single steps. Geant4 uses tabulated or analytical cross sections and can simulate DNA physics explicitly. The activation of DNA physics in Geant4 did not substantially alter the results compared to standard low-energy physics.
The study highlights the importance of species-specific dosimetry. The size differences in thyroid follicle lumens, cell thickness, and nucleus diameter between rodents and humans lead to substantial differences in absorbed dose. Translation of absorbed doses from rodents to humans cannot be done simply due to the limited range of alpha particles. Realistic follicle models that account for the larger follicle size in humans are necessary for accurate dosimetric estimation. This is particularly important for the contribution from astatine in the follicle lumen.
The validation of AlfaMC suggests it is a suitable tool for evaluating different strategies to reduce thyroid dose. Its speed and customization for alpha particles make it preferable for comparing microdosimetric parameters between species. Future work could address the effect of extending DNA physics regions beyond the target. The findings support the use of microdosimetry in optimizing radionuclide therapy for non-thyroid cancers where the thyroid is a risk organ.
The thyroid gland is considered one of the normal organs with the highest risk for side effects in radionuclide therapy involving astatine-211 because free astatine accumulates there. Although astatine is a halogen like iodine, its biodistribution differs significantly from radioiodide in animal studies. In humans, relatively high uptake of free astatine was observed in patients with thyroid disorders as early as 1954. The physiological behavior of astatine remains limited in understanding, with major differences in biodistribution across various organs and tissues in mice and rats. While astatine is transported into thyroid follicle cells and into the follicle lumen, similar to iodide, the amount transported is lower. Furthermore, astatine transport is not solely via the sodium-iodide transporter, involving other mechanisms, and unlike iodide, astatine is not organified to thyroglobulin before entering the lumen.
The study utilized specific geometric parameters for the thyroid follicle models to reflect species differences. For the single follicle model, the diameter of the follicle lumen, the thickness of the follicle cell layer, and the diameter of the follicle cell nucleus were 150, 10, and 8 micrometers for humans; 70, 8, and 6 micrometers for rats; and 50, 6, and 4 micrometers for mice. Each component in these models was assigned a liquid water density of 1.0 g/cm3. The multiple follicle model simplified the scenario where a central follicle is surrounded by neighboring follicles, with surrounding lumen and cell layers matching the single model dimensions for each species. These geometric distinctions are critical because the limited range of alpha particles means that dose contributions from astatine in the follicle lumen vary substantially between species.
Simulations were conducted with specific computational parameters to ensure statistical accuracy. In AlfaMC, 10 to the power of 7 alpha particles were tracked per calculation, with a cut-off energy set to 1 keV. The code used stopping-power values from the NIST ASTAR database and applied a Gaussian, Vavilov, or Landau model for energy straggling. Each simulation was repeated twice with different random seeds to generate twelve independent values for statistical analysis. In Geant4, typically 5 to the power of 7 particles were simulated, with energy deposition collected event-by-event.
The validation process highlighted that intercomparison of independent theoretical data is important because experimental validation of these specific microdosimetric results is not possible. Agreement within 5% with published results is generally considered good for such comparisons. The study noted that deviations up to 9% between analytical models and Geant4-DNA results for spherical geometries with diameters of 1 to 10 micrometers were acceptable in previous contexts. The current findings support the use of AlfaMC for evaluating different strategies to reduce absorbed dose to the thyroid, such as the administration of thyroid blockers like stable iodide or perchlorate. This approach is relevant for radiation safety in patients treated with astatine-211-labelled pharmaceuticals for non-thyroid cancers, where the thyroid, along with bone marrow and kidneys, is a risk organ.
Subject of Research: Validation of Monte Carlo microdosimetry codes for astatine-211 in thyroid follicle models
Article Title: Monte Carlo microdosimetry of 211At in multi-species thyroid follicle models: AlfaMC versus MCNPX and Geant4
Article References: Romiani, A., Josefsson, A., Chakarova, R., & Forssell-Aronsson, E. (2026). Monte Carlo microdosimetry of 211At in multi-species thyroid follicle models: AlfaMC versus MCNPX and Geant4. EJNMMI Physics, 13(1), Article 83. https://doi.org/10.1186/s40658-026-00925-w
Image Credits: AI Generated
DOI: 10.1186/s40658-026-00925-w
Keywords: astatine-211, microdosimetry, Monte Carlo simulation, thyroid dosimetry, AlfaMC, MCNPX, Geant4, radionuclide therapy, alpha particles, radiobiology, Monte, Carlo
News Source: Denise Maddox. (November 10, 2026). Alpha Particle Dosimetry Validated Across Thyroid Models. Scienmag.
