In this study, we evaluated the absorbed doses from different OAR and the effective dose for a range of scandium and gallium radionuclides. More specifically, this work investigates the dosimetry of DOTATATE labelled with variable radionuclides in view of their clinical application. In previous work, scandium-43 and scandium-44 were described as favourable radioisotopes for peptide-based PET imaging due to their physical characteristics [9, 11]. We also reported previously that the EGSnrc and MCNP6.1 MC codes have good agreement among the results when used for dosimetric purposes [21]. If both codes independently produce similar results, it increases confidence in the accuracy of the computed dosimetry. The only concern to the authors was the possible radiation exposure from these radioisotopes to the patient.

In DOTATATE peptide receptor radionuclide therapy (PRRT), the spleen, RBM, liver, and kidneys are known to retain the majority of the administered therapeutic activity and can potentially be associated to the development of radio-induced toxicity [22]. However, published dosimetry data are available only for [68Ga]Ga-DOTATATE [23,24,25].

In this study, the dosimetry data for the previously-mentioned diagnostic scandium radioisotopes were assessed and quantified. Comparable to the increased half-life, these radioisotopes translated into longer organ residence times. Furthermore, the scandium radioisotopes, scandium-43/44/44m, have similar characteristics to that of lanthanide therapeutic nuclides such as lutetium-177 and terbium-161. However, the dose quantification of scandium radioisotopes labelled to DOTATATE has not been available before this study.

Table 4 shows the main results for AD/IA for the OAR calculated in this work, including the dosimetry for the 111In-Octreotide, a somatostatin receptor also used for NETs [24]. Although somatostatin receptor imaging using PET has replaced scintigraphy imaging, [111In]In-Octreotide is still performed when PET is not available [26]. In these findings, the dosimetry for the scandium radioisotopes (scandium-43 and scandium-44) presented higher absorbed dose per injected activity compared to gallium-68 for the main OARs. The results are in accordance due to the half-life of these radioisotopes being considerably longer than gallium-68.

One of the physical characteristics of indium-111 is its long half-life (t1/2 = 2.8 days), quite similar compared to scandium-44m. Analysing the results, one can observe the same order of magnitude between them. On the other hand, the results obtained for the other scandium radioisotopes show good concordance when compared to [111In]In-Octreotide. In this case, the dose results for the [43Sc]Sc- and [44Sc]Sc-DOTATATE present delivering similar dose amount to the OARs in a short time compared to the well-established Octreotide.

Supplementary Table 8 presents the results of this work for [68Ga]Ga-DOTATATE. The highest organ absorbed doses were observed in the spleen, kidneys, liver, and gallbladder, in descending order, for both male and female phantoms. The male phantom mean spleen dose result from this work presented differences of 51%, 81%, and 79% when compared to the studies of Sandstrom et al. [23], Walker et al. [24], and Josefsson et al. [25], respectively. For the female phantom, the difference was 45% when comparing the spleen dose result of this work with the result from Sandstrom et al. [23]. A 79% and 76% difference for the spleen in the female phantom was found for Walker et al. [24] and Josefsson et al. [25], respectively. The absorbed dose to kidneys for the male phantom was within 62% of Walker et al. [24] and Sandstrom et al. [23] results, presenting a 75% difference regarding the Josefsson et al. [25] result. Liver dose differences ranged between 12 and 63%. Regarding effective dose coefficients, the results presented in this work (3.6E−3 and 4.5E−3 mSv/MBq for male and female phantoms, respectively) are quite different from the literature (> 65%). The studies involving patients report an effective dose coefficient of roughly 2.3E−2 mSv/MBq [23,24,25]. These results were summarised in the Supplementary Table 9. The authors believe that this difference is due to the different biodistribution of this radiopeptide. The previous studies calculated the absorbed dose in organs based on the biodistribution data of [68Ga]Ga-DOTATATE measured in patients who received this radiopeptide. In this work, the [68Ga]Ga-DOTATATE residence times were scaled from the ones obtained from [177Lu]Lu-DOTATATE [17]. This allowed the comparisons done in this study to be solely based on the physical properties of the different radionuclides evaluated and further reduce uncertainties associated with pharmacokinetics [27]. Nevertheless, these residence times scaling is likely to explain the differences observed in effective dose. Recently, Wong et al. [28] investigated the differences in tumour-to-normal organ standard uptake value (SUV) ratios measured with [68Ga]Ga-DOTATATE compared with [177Lu]Lu-DOTATATE in patients with NET. They reported that there is evidence of kinetic differences in DOTATATE uptake and internalization [28].

Other publications have identified the potential of using scandium radioisotopes for radiotheranostic application [29, 30]. Benabdallah et al. [30] have published preclinical and dosimetric estimation models for the scandium-44 in different scenarios, where they tried to observe and analyse the excretion of this radioisotope and the radiation impact in research and clinical use. Their dosimetry shows that the effective dose extrapolated from mice data to human, for male and female adults, were 0.146 mSv/MBq and 0.179 mSv/MBq, respectively. Compared to our findings, their values presented relative difference (Δ%) of 89% higher for both results. Moreover, they pointed out the significant absorbed dose in some organs for the scandium-44 due to the extended activity residence time of the radiopeptide in the body [30]. Analysing the AD/IA for the OARs, some their results are comparable to our calculations. Additionally, they reported [30], the heart, gall bladder, and stomach wall exhibited notably high absorbed doses. However, their biokinetic investigation was based on in vivo experiments with mice and different chemical ligand, which can explain the differences to our methodology. The chemical properties of the scandium radioisotopes allow their use for radiolabelling of other pharmaceuticals, for example to obtain [44Sc]Sc-PSMA-617 [31]. Despite the same radionuclide and different targeting agents, the biodistribution of the PSMA-617 presents the same regions of interest as the DOTATATE. Khawar et al. reported dosimetry for individual patients and the kidneys presented highest mean absorbed dose, while in our findings, the critical organ is the spleen followed by the kidneys [32].

Nevertheless, Singh et al. [3] published an imaging study for scandium-44 and gallium-68 labelled with DOTATATE. For one of the imaging scenarios presented, it was pointed out that there was a 9-month delay in detecting lesion using [68Ga]Ga-DOTATATE, and 24 h after [44Sc]Sc-DOTATATE administration the lesions were detected [3]. The suitability of scandium-44 compared to gallium-68 may guarantee the quality of the imaging ensuring that the regions of interest would have time enough to be seen [31]. Meanwhile, the dosimetry shows that the use of scandium radioisotopes would result in a higher absorbed dose compared to the same tumour-targeting agent labelled with the gallium, but reasonable differences with results found in literature for other radiopeptides with same aim.

Figure 4 shows the differences in effective dose between scandium-43, scandium-44 (including a 15% impurity of scandium-44m) and a mixture containing various ratios of both radioisotopes. The impurity of 15% was used to show the worst-case scenario, but as presented in the supplementary material, the presence of the scandium-44m impurity is almost negligible with a maximum of 8.1% increase in the effective dose. Regarding the differences found between the effective dose for scandium-43, scandium-44 and a possible mixture of scandium-43/scandium-44, the largest difference in effective dose was 6.8E−3 mSv/MBq for a sample containing only scandium-44 and 8.1E−3 mSv/MBq for a scandium-44 with 15% scandium-44m impurity. For current practices, these levels of variation in the absorbed dose are likely to be within dosimetry uncertainties of molecular radiotherapy dosimetry [33] and should not impact on the choice of radioisotope used. The choice will be made based on availability of the radionuclide and PET device technology, as some devices will not be able to quantitatively overcome the presence of the high gamma emission of scandium-44 and, thus, favour the use of scadium-43 [8, 9, 11]. Additionally, the use of a mixture of scandium-43/scandium-44 can overcome the cost associated with scandium-43 production by the use of same target material as for the production of scandium-44 irradiated at higher proton energies and benefit from additional information from the scandium-44 emissions for possible multi-isotope PET imaging [34]. In relation to overall effective dose, for an injected activity of 100 MBq (which was used previously [3]) a patient undergoing this examination would receive 0.92, 1.60, and 1.74 mSv, respectively for scandium-43, scandium-44, and scandium-44 and 44m mixture.

Some considerations about the source data used in dosimetry extrapolations performed in this work must also be discussed. The source region TIAC for lutetium-177 used to derive TIAC for the others PET tracers of concern in this study were reported by Marin et al. [17]. They applied bi-exponential fitting to derive TAC in the considered source regions. In their work, the acquired time-data points were obtained at 4, 24, and 168 h post therapeutic activity administration (p.a.). The relatively early first acquisition at 4 h p.a., with a second acquisition at 24 h p.a., possibly enabled a sufficiently reliable description of the first “fast” wash-out phase. In parenchymal organs, it is known that the TAC is mainly driven by a relatively “slow” radio-pharmacokinetics with effective half-lives, also reported by Marin et al., in the range of 50–80 h. For the case of [177Lu]Lu-DOTATATE, it has been reported that a loss of radio-pharmacokinetics information in the early hours after administration impacts the absorbed dose estimates in parenchymal organs of only a few percent (typically less than 5%) [22]. Possible TAC bias induced by the different PR-molecular probe conjugation, when compared to the [177Lu]Lu-DOTATATE from which the source TIAC used in this work were obtained, was not investigated in this work, but could possibly result in absorbed dose bias.