A revised compartmental model for biokinetics and dosimetry of 2-[18F]FDG

The revised compartmental model presented in this work describes the distribution of 2-[18F]FDG in a more physiologically realistic way compared to the descriptive biokinetic model previously reported in ICRP Publication 128 and includes recent measurements of activity in blood and urinary bladder. Important features in the revised model are the presence of blood as a central compartment that, after an intravenous injection, transfers 2-[18F]FDG to other body organs and tissues, and the explicit inclusion of pancreas and spleen as source regions, which were not considered in ICRP Publication 128. The source regions in the revised model were selected based on the available biodistribution data for 2-[18F]FDG. New experimental data obtained with, e.g. whole-body PET might help to evaluate whether the selection of source regions was adequate enough or further regions could be modelled as distinct sources. No indication of organs and tissues, besides those considered in this study, showing an uptake of 2-[18F]FDG notably higher than a general level of 18F activity in the body, was found in the currently published works though. The derived very quick transfer from plasma to erythrocytes and back shows that, instead of plasma, the whole blood can be considered as central exchange compartment in biokinetic studies of 2-[18F]FDG, as also mentioned by other authors [4]. Note that the incorporation of second sub-compartments for source regions “Heart wall” and “Other” was justified by the presence of a long-term retention of 2-[18F]FDG in these regions and a need to obtain accurate TACs and does not have an obvious physiological rationale.

The proposed biokinetic model showed a good description of the experimental data. The highest difference in TIACs compared to ICRP Publication 128 was observed for lungs and liver (− 90% and − 44%, respectively). It should be considered that lungs and liver are highly vascularised organs. In the revised model presented here, the contribution of 2-[18F]FDG activity in the blood within these organs was modelled separately from the 2-[18F]FDG activity in the parenchyma. The reported values represent thus TIACs in the parenchyma only, without the contribution due to the activity in the regional blood volumes of these organs (the same holds here for pancreas and spleen). On the contrary, in ICRP Publication 128 blood was not considered as an explicit source region and the provided TIACs include 2-[18F]FDG activity in both organ parenchyma and blood. This explains the observed differences in the TIACs for lungs and liver. (Pancreas and spleen were not present in the previous model structure.) The separate modelling of the contribution of blood to the activity measured in a given organ enabled to estimate more realistically the TIACs in the organ parenchyma. Moreover, considering blood as a distinct source region overcomes a possible underestimation of organ-absorbed doses for non-source organs with substantial mass fraction of blood content [29].

The radiopharmaceutical 2-[18F]FDG is to a large extent eliminated via the urine. The assumption of voiding intervals affects the predicted TIAC in the urinary bladder content. Another improvement of the revised model was thus the usage of the dynamic bladder model that allows more realistic simulations of filling and emptying the urinary bladder and, thus, of the urinary excretion of 2-[18F]FDG and the corresponding absorbed dose to urinary bladder wall. Generally, any arbitrary voiding interval or fixed voiding scheme can be assumed to calculate the TIACs in the urinary bladder content. The ICRP Publication 128 assumes a constant voiding interval for adults of 3.5 h. However, for 2-[18F]FDG scans the patients are often hydrated and the first voiding occurs just before the image acquisition (45–60 min after administration). Therefore, the first voiding at 45 min was assumed in the calculations of this study. In spite of the early first voiding, the TIACs with the revised model are about 20–25% higher than the one calculated in ICRP Publication 128. This can be explained by the fact that the MIRD model (the basis of those calculations) notably underestimated the excretion data (Fig. 3). Urinary bladder activity measured at 2 h p.i. by Jones et al. [30] is consistent with the data by Mejia et al. [6] and well described by the model of this study. The cumulative bladder excretion reported in [31] lies between the MIRD and the revised models and amounts to 5.8–7.1%IA at 1 h p.i. (compared to ca. 3.6 and 9.7%IA predicted by the MIRD and the revised models, respectively). The urinary excretion of 2-[18F]FDG as predicted by the model showed a good agreement with the 18F activity in urinary bladder as obtained in the SUS studies (SUS data, Fig. 3); while the volumes of urinary bladder contents at 1 h p.i. (SUS data, Additional file 2: Table S2) were for most patients notably higher than those predicted by the dynamic bladder model under the assumption of a reference urinary production rate [22] and a residual urine volume of 10 ml. This could be due to (1) only partial emptying the bladder by the patients at the first voiding and, consequently, larger residual urine volumes and (2) higher urinary production rate because of hydration before the 2-[18F]FDG diagnostic scan. Underestimation of urine volumes can lead to an overestimation of the absorbed dose to bladder wall, since SAFs for bladder content as source and bladder wall as target regions decrease with increasing the volume of bladder content. Thus, the reported absorbed dose coefficient for bladder wall is possibly a conservative estimate.

The organ dose coefficients were calculated according to the latest dosimetric framework of the ICRP, in particular making use of the reference anthropomorphic voxel phantoms instead of the stylized mathematical phantoms employed in ICRP Publication 128. The revised absorbed dose coefficients for uterus and ovaries were notably higher compared to those reported in ICRP Publication 128 partly due to the higher revised TIAC for urinary bladder contents and also because of different (generally more realistic) anatomical description of the geometry in the voxel phantoms used in the calculations of the cross-fire energy deposition.

The dose to the urinary bladder wall is notably lower (− 47%) than in the previous calculations although the TIAC in the bladder content is higher. In this case, no direct comparison can be made due to several reasons. First, the dynamic bladder model employed here was used in a combination with SAFs that vary with varying bladder volume (and thickness of the bladder wall). In addition, the new dosimetric approach is based on a realistic assessment of the energy deposition in the organ wall also for beta-particles, and not on the previously used simplified assumption that half of the energy of beta-particles is absorbed in the wall. To evaluate the impact of these features, some additional calculations were made. First of all, the contributions to the absorbed dose to urinary bladder wall from the source region urinary bladder contents were computed also for the static bladder model, which assumes fixed volume of the bladder and thickness of the bladder wall. Using the bladder TIAC from Table 3, values of 5.7E−02 mGy/MBq and 4.3E−02 mGy/MBq were obtained for photons and 9.6E−03 mGy/MBq and 9.1E−03 mGy/MBq for beta-particles for the dynamic and the static bladder models, respectively. The use of smaller bladder volumes for emptied or not completely filled bladder in the dynamic model would thus lead to a higher dose to the bladder. However, the more realistic treatment of the beta-particles energy deposition in the bladder wall causes a drastic reduction of the beta contribution to the bladder dose of nearly an order of magnitude. (It was equal to 8.9E−02 mGy/MBq according to the previous assumptions.) The present calculations with more realistic biokinetic and dosimetric models thus show that, from one hand, the static bladder model underestimated the dose to bladder wall (especially from the photon component) and, from the other hand, that the previous simplistic assumptions on beta-particles dosimetry led to an unjustified overestimation of the absorbed dose to the bladder wall in spite of the fact that the excretion curve of 2-[18F]FDG was underestimated by the MIRD model, as shown in Fig. 3.

Finally, the influence of the selected initial settings for the dynamic bladder model on the absorbed dose to the urinary bladder wall was also analysed. Firstly, assuming an initial bladder volume of 10 ml (this corresponds to having emptied the bladder just before the time of 2-[18F]FDG injection) instead of a half-full bladder, as in our reference calculations, leads to an increase of the urinary bladder wall dose by approximately 10% despite the same TIAC value. This increase is due to the slightly larger SAFs for smaller bladder volumes. Secondly, if no forced voiding at 45 min p.i. is assumed and the voiding pattern befalls according to the dynamic bladder model (every 3,75 h for adult male and every 4 h for adult female), the urinary bladder wall dose would increase by 25%, due to the higher TIAC values. This shows the positive effect on dose (and not only on imaging) of forcing a bladder voiding before the first scan. Similarly, hydration helps to decrease the absorbed dose to the bladder wall by diluting the activity in a larger urine volume and inducing a higher voiding frequency. The dose reduction is, however, not substantial and amounts to about 5% in the case considered here.

Recently, Hu et al. [31] investigated an impact of TACs measured with a total-body high-sensitive PET/CT scanner up to 8 h p.i. on dosimetry of 2-[18F]FDG. The TIAC for brain in [31] was notably higher compared to the ones reported here and by other authors [2, 4, 6, 7, 32]. Brain-absorbed dose does not notably influence the effective dose though, since brain has a tissue weighting factor of only 0.01 [18]. Hu et al. obtained the effective dose coefficient of 1.4E−02 mSv/MBq compared to 1.7E−02 mSv/MBq in this work.

The calculation of the effective dose coefficients according to the current recommendations of the ICRP [18] is affected in a complicated and partially opposite way by the simultaneous combination of the different methodological improvements implemented in this work: revised and physiologically more realistic biokinetic model, improved and realistic description of the urinary excretion, anatomically realistic reference voxel phantoms, improved assumptions for the calculations of the energy deposition in the target organs, especially for beta-particles, revised values of radiation and tissue weighting factors. As a result, the revised effective dose coefficient is about 10% lower than the one in ICRP Publication 128 (1.7E−02 vs 1.9E−02 mSv/MBq).

留言 (0)

沒有登入
gif