Background
Radioiodine has been used clinically for about 70 years, both for diagnosis (
123I,
125I and
131I) and therapy (
125I and
131I) of various diseases. Radioiodine is to a high extent accumulated in the thyroid gland and metabolically built into the thyroid hormones, and the biokinetics of radioiodine is used for measurement of thyroid function [[
1]].
131I, as iodide, is therefore routinely used for diagnosis and treatment of various thyroid disorders such as thyrotoxicosis (hyperthyroidism) and thyroid cancer [[
1]-[
3]]. Lately,
131I has sometimes been replaced by
123I as iodide for diagnostic purposes to avoid the thyroid stunning phenomenon [[
4],[
5]]. Radioiodine bound to specific vector molecules is also widely used. For example,
131I- and
123I-MIBG are used for scintigraphy and
131I- and
125I-MIBG for treatment of various neuroendocrine tumours [[
6],[
7]],
123I-receptor ligands for brain scintigraphy [[
8]],
125I methylene blue for sentinel node localization [[
9]] and
131I-labelled monoclonal antibodies for treatment of lymphoma [[
10]]. When, radioiodine labelled pharmaceuticals are administered to the body, radioiodide might be released into the circulation, e.g. by enzymatic reactions by dehalogenases in tissues, and taken up by the thyroid gland [[
11]]. For radiopharmaceuticals or tracers labelled with radioiodine, the thyroid is thus an organ at risk.
Furthermore,
131I is a radionuclide of importance in nuclear accidents, where it is a rest product from the nuclear fission process in nuclear energy plants. After the Chernobyl accident in 1986, contamination with
131I (and other short-lived isotopes such as
132I and
133I) led to an increased incidence of differentiated thyroid cancers in children but not in adults, with a higher incidence with lower age [[
12]–[
15]].
There is thus a need for accurate dosimetric calculations of the absorbed dose for both patients examined or treated with radioiodine, personnel handling radioiodine and for personnel and the general population in case of accidental exposure to radioiodine.
Dosimetric estimations using the MIRD formalism is mostly utilised due to its simplicity, and the use of mean absorbed dose is of interest if the radionuclides and the energy deposited are homogeneously distributed within each organ/tissue. This assumption is adequate as long as the range of the emitted particles is long compared to the size of the cells. For radionuclides emitting particles with shorter range, e.g. Auger and internal conversion electrons, non-uniform distribution within an organ/tissue will give heterogeneous absorbed dose distribution, and more detailed dosimetric approaches are clearly needed.
The physical properties differ between these radioiodine isotopes. For
131I, emitting relatively high-energy β particles with a range up to 2 mm in tissue [[
16]], the energy distribution will be relatively homogeneous within the thyroid, less dependent of the radionuclide distribution, while for
125I, emitting cascades of Auger electrons with ranges from a few nanometres up to around 23 μm in tissue, the energy deposition within the thyroid gland will be more heterogeneous, dependent of the distribution of the radionuclide, electrons emitted, half-life, and the amount of photons emitted [[
17],[
18]].
To be able to determine more detailed dosimetric parameters for the thyroid cells from heterogeneously distributed radioiodine isotopes in the thyroid tissue, thyroid tissue models are needed. We have recently published thyroid models for man, but also for mouse and rat, and performed microdosimetric studies of the α particle emitting radiohalogen
211At, demonstrating the importance of detailed dosimetry for the thyroid [[
19]]. A few thyroid models for radioiodine dosimetry have been previously published for normal and thyrotoxic thyroid follicles [[
20]-[
24]]. To our knowledge, few dosimetric studies have been published demonstrating the dosimetric properties of these radioiodine nuclides in these models.
The aim of this study was to compare dosimetric calculations for
123I,
125I and
131I using the general purpose Monte Carlo radiation transport code MCNPX 2.6.0 [[
25]] with nuclear decay data from ICRP 107 [[
26]] and the recently developed thyroid models for man, rat and mouse [[
19]].
Discussion
In general, the
S values determined in the present study were in good agreement with the few corresponding data available in the literature. Our results for
123I,
125I and
131I homogeneously distributed within the follicle cell nuclei in the species-specific models were in excellent agreement with published cellular S values calculated with an analytical method [[
31]] based on the experimental range-energy relationship for electrons by Cole [[
32]].
The mean absorbed dose to the follicle cells for
131I homogeneously distributed within the follicle lumens was determined for a similar human thyroid follicle model as ours [[
24]]. Their model included interstitial tissue (51% of the total tissue volume), and 12 surrounding follicles layers, compared with no interstitial tissue and 8 follicle layers in our human model (since we found that the contribution from surrounding layers beyond the 8th was <1%). The contribution from the central follicle was 7% in our model compared with 17% in their model, and the contribution from the surrounding follicles in our model was slightly higher, e.g. 32% for the first surrounding layer compared with 29%. These differences could mainly be explained by the difference in consideration of interstitial tissue. Despite these model differences we found in both studies that the follicle cells received 0.98 Gy when the mean absorbed dose was 1 Gy to thyroid tissue.
There are some simplifications and assumptions made in the calculations, both for the mathematical models and for physical data. The use of unit density water (1.0 g/cm
3) in the models instead of the density of the thyroid gland, which according to ICRP publication 23 is 1.05 g/cm
3 [[
33]]. Dosimetric calculations performed with a 3% mass concentration of
127I (stable iodine) homogenously distributed within the follicle lumens only showed a small difference for
211At [[
19]], and after normalisation of the mass for the follicle cell nuclei this difference would be even less. Other assumptions regarding the thyroid models have previously been discussed [[
19]]. In general, we estimate that the assumptions in the dosimetrical calculations, such as limitations in nuclear decay data and transport physics used by the Monte Carlo code only contribute to a minor extent to the results. The conventional AE spectrum from ICRP 107 was used in the calculations [[
26]] and have been regarded adequate when calculating absorbed doses to regions with diameters larger than 1 μm [[
34]], and AE with initial kinetic energy lower than 1 keV (the lowest cutoff energy for electrons in MCNPX 2.6.0) were assumed to be fully absorbed within the source volume. This assumption is realistic since experiments have shown that electrons with a kinetic energy of 1 keV have a range of about 61 nm in unit density matter [[
32]], verified by calculations: absorbed fraction very close to unity for unit density water spheres with radius of 2 μm for monoenergetic 1 keV electrons [[
35]]. Bremsstrahlung generated by the electrons was not accounted for in the Monte Carlo calculations, but the contribution was low in this application (for 1 MeV electrons, only about 0.7% of the kinetic energy is transferred to bremsstrahlung in liquid water, and this fraction is even less for lower kinetic energies [[
36]]). The contribution from
131mXe was not considered in the dosimetric calculations for
131I, which would result in an underestimation of the
S value to the cell nucleus of about 3.6% for
131I homogenously distributed in an 8-μm cell nucleus and about 1.5% for a homogenous distribution in a 150-μm diameter lumen (unpublished data). Furthermore, the contribution from the xenon daughter is probably less significant due to the short retention in the thyroid gland because of its gaseous state [[
23]].
For
123I, the contributions from the
123Te and
123mTe daughters were not included in the dosimetric calculations due to a very long half-life and low yield, respectively. Furthermore, the effects of the charge of the tellurium atoms (average charge of about +9 [[
37]] due to multiple ionisation when
123I and
125I emit cascades of AE) were not considered. Otherwise, such charged atoms may produce ionizations and excitations in the immediate vicinity of the decay site [[
37]], which could enhance the biological effect when covalently bound to the DNA [[
38]].
With
123I and
125I homogeneously distributed within the follicle cell nucleus, the
S value was 2.3 times higher for
125I than that for
123I in the mouse, rat and human models, a result in accordance with similar calculations for a 10-μm diameter tissue sphere, excluding charge neutralisation [[
37]].
Biodistribution studies performed on mice, rats and guinea pigs have shown that the highest uptake of radioiodide was in the thyroid gland, with the highest concentration occurring around 18 to 24 h after injection [[
39]-[
41]], while the maximal concentration is obtained after approximately 1 to 2 days in normal humans [[
42]]. Preclinical studies have shown that radioiodide is rapidly transported through the follicle cell cytoplasm. At early time-points, radioiodine appears as rings peripherally in the follicle lumen close to the apical cell surface [[
43]-[
46]], and thereafter the radioiodine is more homogeneously distributed in the follicle lumen [[
45],[
46]]. However peripheral rings have been observed as long as 99 days after injection [[
46]]. The specific activity was initially highest in the smallest follicles but became independent of follicle size with time [[
45]]. In the human model, the
S value for
123I,
125I and
131I distributed on the apical follicle cell surface was 2.2, 5.9 and 1.5 times higher than for a homogeneous distribution within the follicle lumen, respectively. Due to the much shorter half-life of
123I (13 h), the fraction of decays in the follicle cells and at the apical surface would be highest for
123I, indicating a possible higher absorbed dose when biokinetic data are considered.
The MIRD formalism assumes a homogeneous distribution of the radionuclide within the source compartment when determining the mean absorbed dose. Compared with our results, the mean absorbed dose calculated according to MIRD formalism (Equation
1) and nuclear decay data was lower, with the largest difference of 18% for
125I, and the smallest of 4% for
131I. This comparison together with the dosimetric data obtained for inhomogeneous distribution shows the importance of taking the range of the emitted particles into account. For
125I, the emitted low-energy AE and CE, with a range of up to 23 μm in water [[
36]], could contribute to a heterogeneous absorbed dose distribution, and about 90% of the absorbed dose originates from the follicle itself. The high-energy β particles emitted by
131I, with a range of up to 2.1 mm in water [[
24]], contribute to a cross-fire effect with contributions from eight surrounding layers of follicles, which results in a more homogeneous absorbed dose distribution, and only about 11% of the mean absorbed dose originates from the follicle itself. For
123I, the emitted low-energetic AE and somewhat higher-energetic CE, the mean absorbed dose was about 9% higher than that according to MIRD formalism. The absorbed dose is then a combination between heterogeneous absorbed dose distribution from the very short-ranged AE and the more homogeneous absorbed dose distribution from the more long-ranged CE, and about 38% of the mean absorbed dose originates from the follicle itself. Furthermore, in the peripheral regions of the thyroid gland where fewer surrounding follicles contribute, the absorbed dose may be lower and more heterogeneous. This effect would be largest for
131I with eight surrounding follicle layers contributing and could lead to a reduced mean absorbed dose by up to about 45%.
For radioiodine homogeneously distributed only within the follicle lumens, the mean absorbed dose was 0.80, 0.11 and 0.98 Gy, respectively, for 123I, 125I and 131I, compared with 1 Gy calculated with MIRD formalism and nuclear decay data for radioiodine homogenously distributed within both follicle cells and lumens. Thus, the MIRD formalism overestimates the mean absorbed dose for 123I and 125I.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AJ designed the study, performed the Monte Carlo simulations and drafted the manuscript. Both authors contributed to the scientific and intellectual discussion and interpretation of the data and revision of the manuscript. Both authors read and approved the final manuscript.