EANM guidance document: dosimetry for first-in-human studies and early phase clinical trials

Relevant types of emitters, including examples of currently common radionuclides, are listed in Table 1. Possible imaging methods, obtained from previous studies with focus on quantification and optimisation, are also provided. Imaging approaches are relatively straightforward for many nuclides, such as fluorine-18, technetium-99 m, and lutetium-177. However, some nuclides are more challenging. Imaging of alpha-emitters is often difficult due to the lower amounts of activity commonly administered, and pure beta-emitters such as yttrium-90 lack the primary photon emission to isolate for energy discrimination. The biodistribution, amount of activity, and equipment performance (e.g. energy range, sensitivity) may then determine whether imaging approaches are feasible or should be replaced or supplemented by other methods. Potential radioactive daughters, metastable states (e.g. lutetium-177 m), and other production impurity isotopes (e.g. rhenium-188 impurities in rhenium-186) should also be considered. While these are not included explicitly in Table 1, some of the photons listed are emitted by daughters. When imaging daughter emissions, care should be taken to ensure the imaging adequately indicates the position of the parent radionuclide rather than the biodistribution of different radionuclides in the decay chain. In addition, depending on the half-life and probability for relocalisation, separate dosimetry of the other radionuclides may need independent investigation.

A range of potential carriers exist, including antibodies and related structures (such as antibody fragments), peptides and proteins, cells, vehicles for non-targeted transport, and various small molecules. Radionuclides can also be injected in salt forms (e.g. [¹³¹I]NaI or [⁸⁹Sr]SrCl₂). In Table 2, examples of different carrier molecules are given, together with a few key pharmacokinetic features. While some overall features are demonstrated, such as prolonged circulation of antibodies compared to peptides or small molecules, the behaviour of individual compounds may vary greatly, and the normal tissues of interest and expected kinetics should be evaluated for each case. It should also be noted that the parameters in Table 2 are derived from studies of radiopharmaceuticals; therefore, pure carrier molecules, or carrier molecules labelled with other radionuclides, might not display identical behaviour. Non-targeted compounds are often special in the sense of being injected loco-regionally, and there is limited transference value for similar compounds unless they also share administration site.

Administrations can be intravenous, oral, inhalation-based, or performed by loco-regional injection (e.g. intra-arterial, intra-tumoural, or in cavities) or surface application (skin, on cavity surfaces). For oral, inhalation-based, or loco-regional injections and surface applications, stomach and gastrointestinal systems, lungs, and administration site should be included as potential source organs (see section “Biodistribution and time-integrated activity estimation” for definition of source organs).

Dosimetry investigations are among the preclinical safety data recommended in the EMA document “ICH guideline M3(R2) on non-clinical safety studies for the conduct of human clinical trials and marketing authorisation for pharmaceuticals” to support exploratory clinical trials [48]. The EMA “Guideline on the non-clinical requirements for radiopharmaceuticals” defines the requirements to be met by a new radiopharmaceutical in order to be translated clinically, and according to the 2018 version a biodistribution study in animals should be conducted, with dosimetry performed [49]. Consequently, the most recent interpretation of the EU regulatory framework for radiopharmaceutical testing and marketing approval considers non-clinical dosimetry data as essential information to be included in the investigational medicinal product dossier (IMPD) of a new compound [50, 51]. Some of the similar considerations apply to the United States legislation, where FDA expects that an IND contains “sufficient data from animal or human studies to allow a reasonable calculation of radiation-absorbed dose to the whole body and critical organs upon administration to a human subject” (title 21 CFR 312.23(a)(10)(ii) [3]). For diagnostics, it is also specified that preclinical models should be included; “the radiation safety assessment must establish the radiation dose of a diagnostic radiopharmaceutical by radiation dosimetry evaluations in humans and appropriate animal models” (title 21 CFR 315.6 (d), [3]). However, the 2014 EANM guidance document on how to prepare an IMPD for a radiopharmaceutical to be used in a clinical trial does not cover internal dosimetry [52]. A recent review published following an international meeting on “Preclinical testing of radiopharmaceuticals” supported by the International Atomic Energy Agency (IAEA) recognizes the relevance of preclinical dosimetry but, at the same time, acknowledges its lack of standardization [53]. In summary, in vivo preclinical pharmacokinetic studies of new radiolabelled compounds are almost invariably performed, whereas an estimation of human dosimetry based on biodistribution experiments on laboratory animals may not necessarily be available at the time a first-in-human trial is initiated.

In many cases, there will be similar radiopharmaceuticals already investigated in humans, such as when therapeutic compounds are developed following diagnostic tracers. Compassionate use cases of therapeutic compounds may also have been performed. This allows for initial estimates of source organs and measurement timing, after corrections for physical half-live differences are performed. It should also be noted that generic absorbed dose estimates of some classes of radiopharmaceuticals, for example, diagnostic brain receptor substances or amino acids, are given by the ICRP (e.g. ICRP 128) [44]. If highly similar tracers with the same radionuclide already exist, this information should be used as a starting point and evaluated during the first-in-human study.

It is important to mention that the well-established physiologically based pharmacokinetic-pharmacodynamic (PBPK/PD) modelling approaches for conventional drug development have only rarely been applied in radiopharmaceutical development. Both the EMA and FDA have guidelines on the reporting of PBPK modelling [54, 55]. A recent review focused on this aspect and came to the conclusion that the available literature would support the notion that these modelling concepts can result in a better understanding of PK and whole-body distribution of radiopharmaceuticals in general and could contribute to the evolving research of radiopharmaceuticals [56].

In the development of new radiopharmaceuticals, the translation from in vitro cell line uptake studies to small animal studies and finally into first in human investigations is a common path to define the pharmacokinetics (PK) and prediction of absorbed doses to humans [57]. This bench-to-bedside approach requires at several steps caution and harmonisations to prevent erroneous assumptions.

In preclinical biodistribution studies, a set of animals is administrated the radiopharmaceutical and the organ activities measured at several time points. This can be performed by sacrificing the animals and measuring tissues in well counters or by imaging the animal at multiple time points [58]. Investigation of PK and behaviour of new radiopharmaceuticals can also be done by using theranostic pairs to visualize the biodistribution by means of small animal imaging [59]. The imaging approaches require fewer animals. However, quantification may be challenging for smaller volumes and the procedures requires anaesthesia, which may alter the pharmacokinetics of the studied compound. This holds especially true for radiopharmaceuticals where the biodistribution depends on the level of specific hormones or neurotransmitters [60, 61]. In contrast, anaesthetics have virtually no impact on tumour targeting. With regard to predictive models of human pharmacokinetics, an initiative some ten years ago published some proposals how to proceed and summarized the potential pitfalls [62‐66]. Besides anaesthesia, other important confounding factors in this context include fasting state, food-drug interactions, and circadian differences in metabolism [67]. All these aspects can result in significant impact on the extrapolation of the acquired results into the human situation, and as mentioned above, it is important to harmonize and to disentangle the potential impact.

Several methods have been described to extrapolate human absorbed doses from time-activity information obtained experimentally in small animals [68, 69]. The most straightforward approach often utilized is the direct application of the source organ time-integrated activity coefficient (TIACs) to human organ masses in order to calculate absorbed doses. However, owing to the differences in organ sizes and metabolic rates of physiological functions between species, power-law equations relating the variable of interest as a dependent function of the body mass, namely, allometric equations, have been developed in the pharmaceutical field [70, 71]. A popular approach, which is supported by some national guidelines such as the Swiss federal guideline [72], considers the application of a relative mass scaling factor to the animal TIAC to obtain the human TIAC [68, 73]. More complex allometric approaches have been proposed which consider time scaling and/or organ-specific metabolic scaling [7, 74]. These methods may provide different results depending on the characteristics of the radiopharmaceutical, and none has been universally adopted as the reference method for prediction of human absorbed doses, so far. Due to the various challenges, as briefly described above, it is important to realize that any animal model is at its best only an approximation of the human situation. It should be emphasized that any extrapolation should only be used as a starting point to inform the first-in-human study design, and cannot replace the study completely.

Blood samples are typically drawn at multiple time points and measured in a gamma counter. The counter should be calibrated and characterised over the theoretical range of activity values. If the blood samples could contain contributions from multiple radionuclides of a decay chain, consideration should be given to using different energy channels or delayed measurements to separate the individual radionuclides based on half-life [89].

Further consideration should also be given as to whether whole-blood measurements are sufficient or if plasma or serum should be measured independently. Some compounds will bind to components in the blood (such as unbound lead to red blood cells [77]), and separating the compartments may then provide valuable information for establishing proper kinetics. Blood (or urine) samples can also be used to evaluate metabolic stability of the compounds by chromatographic techniques [83], preferably at several time points as the stability can change over time. This evaluation is recommended when the measured activity in fluid samples is used directly for dosimetric calculations, and the in vivo human stability unknown.

The purpose of collecting and measuring activity within excrement is to indirectly estimate the retention within the body over time and/or for use as input for pharmacokinetic modelling or special dosimetric models. When knowledge of the total activity excreted is desired, this can be achieved by urine or faecal collection or spot samples in combination with diuresis measurements or weighing. Spot samples alone are more relevant for radiation protection considerations or verification of quantitative imaging approaches. The main route of excretion is also useful when ascertaining radiation protection requirements; an example of this is given in the [²²³Ra]RaCl₂ example where faecal excretion was demonstrated to dominate over urinary excretion (see example section in supplementary information; “[²²³Ra]RaCl₂”).

Choice of probe should be selected based on photon flux, for which a simple ionisation detector or a more sensitive scintillator-based probe may be appropriate. The suitability and properties of the probe for the intended radiopharmaceutical and the relevant time/activity span should be characterised before inclusion of subjects. Measurements at different fixed distances is advised, especially if the predicted biokinetics are uncertain or the expected input close to the operating range of the probe. In most cases, the first measurement of a subject post-administration serves as calibration for the whole-body activity at later time points, and this measurement should be performed before micturition.

Depending on how well the performance characteristics of the imaging equipment for the given radionuclide are established, it may be necessary to first explore the potential for quantitative imaging by phantom studies [28]. For single photon emitters, the use of planar scintigraphy is generally suboptimal compared to SPECT imaging, in particular for overlapping organs [90], but can be useful for early phase dynamic studies [91]. Example approaches are suggested for the nuclides in Table 1. However, the radiopharmaceutical particulars may render the stated approach too uncertain for novel applications. Renewed calibration and imaging quality studies are then needed to determine the optimal imaging protocol. Investigations for dead time, noise equivalent count rate, and quantification uncertainties should be performed for all nuclides if the amount of activity to be imaged exceeds or is significantly less than the equipment’s previously established operating range for the specific radionuclide. This also applies if the structures to be investigated (i.e. source organs) is likely to be different in size or shape than previously charted structures, or the background activity or attenuating properties (e.g. anatomical size of the body part) may differ significantly. Typically, phantoms containing spheres of different sizes are used for such evaluations, but more anthropomorphic phantoms may also be considered [92]. Quantitative tomographic emission imaging enables the simultaneous collection of dosimetry-relevant information from multiple organs in a patient. However, different organs can present differing kinetic phases and/or different effective half-lives. Hence, the choice of the acquisition time points may represent a trade-off that should prioritize a more accurate TAC estimation of the most irradiated organs and/or organs at risk. It is recommendable to make sure patient positioning and imaging field of view are consistent between imaging time points.

Accurate estimation of individual tissue masses are in most cases essential for absorbed dose calculations for therapies. Anatomical imaging of diagnostic quality should then be available for all organs considered source volumes, acquired for at least one time point. Outlines drawn on CT or MRI in combination with tabular densities from, e.g. ICRP 89 or ICRU 46 [93, 94] may be used to derive organ masses. While stand-alone diagnostic scans can be used for this purpose, co-registered anatomical and nuclear medicine images provided by hybrid imaging systems (such as PET/CT or SPECT/CT) are beneficial to guide the delineation of activity. If delineated masks of organs based on anatomical imaging (or masks made from one of the nuclear medicine imaging time points) are used to derive the activity values, it must be considered that significant changes in patient positioning may take place between different image acquisition time points. This plays a particularly important role for small organs such as salivary glands, areas of high positional variability like the head and neck, or hollow organs such as bladder, stomach, and intestines.

Source organs, i.e. organs characterized by significant transit of radioactivity due to the radiopharmaceutical uptake followed by physical decay and possible biological washout, should be initially identified based on pre-clinical data, and existing clinical information collected from in-human studies involving radiolabelled compounds with expected similar biodistribution (Fig. 3).

As a rule of thumb, major parenchymal organs such as the liver, lungs, kidneys, and spleen should always be considered as potential source organs in a first-in-human study. Similarly, the blood pool should be included, as well as organs involved in the excretion pathways such as the various parts of the gastro-intestinal tract and the urinary system including the respective contents. Due to the comparatively high radiosensitivity of red bone marrow, it is generally recommended to consider this as a potential source organ, even if specific uptake is not initially identified. For loco-regionally delivered radiopharmaceuticals (see section “Administration route”), systemic leakage should not be disregarded. For example, using microspheres for selective internal radiotherapy of disease in the liver, lung shunt has been found a potential limiting factor [20].

Depending on the specific biodistribution, other tissues should also be considered source organs. A non-complete list includes adrenal glands, brain, cortical bone, eyes, gallbladder contents, heart wall, ovaries, pancreas, prostate, red bone marrow, salivary glands, testes, thymus, thyroid, trabecular bone, and uterus. These tissues are commonly available as source organs for input of TIA in most organ-based dosimetry software, for example, Olinda EXM, IDAC, OpenDose, and MIRDcalc [95‐99]. Many of the tissues are also among the listed target organs in the output. (Note that “target organs” here refers to organs for which absorbed doses are calculated in the context of safety, not to be confused with, e.g. targets for a treatment.) If tissues of interest are not included in standard organ-based dosimetry software, dedicated absorbed dose calculations should be performed. An example approach, used for the choroid plexuses for [⁶⁸ Ga]Ga-NODAGA-RGDyK [100, 101], is described in the example section (supplementary information).

The activity (or the activity concentration) present in source organs at different time points is required for the dosimetry workflow, as the absorbed dose calculations consider the energy deposited in target volumes based on the radiations emitted from source organs. As described in “Measurement protocol”, a sufficient number of measurements have to be available to appropriately characterize the kinetics in source organs and derive TACs. TAC fitting is a crucial step for computing accurate source-organ TIACs. Numerical and/or analytical time integration of source organ TACs provides the TIA. TIACs are then obtained by dividing the TIA by the total administered patient activity. In general, three models may be used for mathematically describing the biodistribution of a radiopharmaceutical over time, analytical models using a sum of exponential functions, empirical models, or compartmental models [102]. Mono-exponential analytical fitting is suited for tissues presenting TACs which can be described by a single clearance phase with a prompt initial uptake and also for whole-body (WB) retention fitting. In general, biological processes are expected to follow first-order kinetics, and multi-exponential TAC fitting can then describe the biodistribution and pharmacokinetics in source tissues characterized by multiple phases (e.g. fast uptake, fast and slow clearance). Model selection can for example be performed with the corrected Akaike information criterion (AICc) [103]. Linear interpolation followed by trapezoidal integration is a common empirical model which should be avoided if possible. However, it can offer a valuable solution when the TAC (or parts of the TAC) does not present any exponential behaviour. A disadvantage is the need to extrapolate information for the period after the last measurement (also before the first measurement if the initial conditions are undefined). Two methods are generally adopted to evaluate the late TIA contribution. The first method is a most often conservative approach that extends the TAC to infinity by considering only physical decay. The second method considers the TAC tail beyond the latest acquired time point as prolongation of the late clearance phase (represented by the effective half-life from the last measurements) to infinity. In this case, one must verify that the late effective half-life does not exceed the physical radionuclide half-life. Only in absence of a realistic late effective half-life should the “conservative” physical decay approach be used. In either case, the fraction of the TAC that is extrapolated should be reported. Compartmental models use differential equations to describe the transfer of radiopharmaceuticals between compartments of a system, often at organ or sub-organ level. While in silico models in theory could replace first-in-human studies, data size can be a challenge for the development of models for most first-in-human studies. However, in some cases, a priori knowledge can be integrated. For transfer rates, this requires previous studies of similar carriers in humans. A highly relevant scenario is the presence of unbound radionuclides, in which biokinetics of the radionuclides can be derived from ICRP publications as described [75‐78].

Regardless of the fitting approach, a larger number of acquisitions generally improve the accuracy in estimating the actual tissue TAC and hence the absorbed dose. Different publications have demonstrated the importance of well-selected time points, especially considering late data acquisitions, in providing accurate TIA estimates [17, 87, 104, 105]. By the nature of first-in-human studies, the initial temporal sampling may not represent the optimal schedule. If one includes an initial cohort of subjects (Fig. 3), the evaluations performed for investigated fit models and selection criteria for this population should be reported. However, other measurement time points, resulting in other best fit models, may be used for the remaining subjects. The final TIA values should therefore ideally be established first after evaluations of the entire cohort are performed, to ensure consistency.

For therapeutic radiopharmaceuticals, the mean absorbed dose to the organs or sub-organs that demonstrate activity accumulation should be calculated and reported for individual patients. The absorbed dose is the primary parameter responsible for the biological effects observed, both efficacy and toxicity. Available evidence indicates sigmoidal correlations for the absorbed doses required to induce biological effects in both tumours and healthy tissues. Beyond certain thresholds, an increase in activity may therefore be accompanied by additional toxicities without providing any major therapeutic benefits.

Absorbed dose calculations should be performed using patient-specific kinetic data and patient-specific anatomy. Mass scaled, model-based S-values may be used when direct radiation transport simulations using patient-specific anatomy are unavailable. If organ mass is determined empirically, the methodology should be clearly stated and justified. Lack of published or reference S-values for an organ or target should not negate the requirement to calculate the absorbed dose, and an appropriate modelling method should be used as an alternative. Unless modelling the cross dose to the target organ, reference masses of non-source organs should be used for model-based approaches. Mean absorbed doses to the total body and to typical organs at risk, such as bone marrow or kidney, should also be reported irrespective of active uptake. Alongside mean organ absorbed doses, other dosimetric parameters such as percentage of injected activity (%IA), mean TIAC, and parameters describing the kinetics, such as mean effective half-life, should be reported. In addition to mean organ absorbed doses, dose calculations at the voxel level might also be given, provided that robust methods for image registration and resolution modelling are used to minimise fitting- or resolution-related errors [111]. Examples of, e.g. the time-activity curves at the voxel level should also be reported to demonstrate the validity of the fits. The penetration range of the emissions should be considered with respect to whether Monte Carlo–based voxel-level dosimetry is needed or if kernel or multi-kernel methods can be used.

Evaluation of the stochastic risk from a new radiological test is often made in comparison to that of similar exams or approaches. For diagnostic applications, the ICRP system of protection is based on the use of standard reference anatomical and physiological models. The absorbed doses to individuals therefore have less relevance and the absorbed doses calculated using absorbed fractions derived from the standard reference phantoms should either be used or reported in conjunction with patient-specific doses. The effective dose is designed by ICRP to be calculated using the absorbed fractions derived from the standard phantoms (ICRP 133) [112], and should not be calculated using individual anatomical data. These S-values are also available in software such as Olinda/EXM 2.0, IDAC and MIRDCalc.

The effective dose is used to estimate the radiation detriment to a population, averaged over the full age distribution and for an equal number of both sexes. Therefore, in a biodistribution study, a sufficiently large sample of both male and female patients across the weight class of the ICRP reference phantoms should be included. The age distribution should also be considered.

When determining the effective dose, the TIACs assigned to the male and female reference phantom should ideally be based on a pharmacokinetic model using decay data averaged from the study cohort. TIAC values can also be evaluated using the mean organ concentration of the population and total organ activity scaled to the mass of the phantom model. If taking this approach, care should be taken to ensure that 100% of the activity is accounted for, which may require using a “remainder of body” compartment such that the summed activity in all source organs equates to that measured in the total body. Effective dose is conventionally reported using the sex-averaged equivalent doses calculated using the respective phantoms.

Some tissues may require special considerations or models. A non-complete list include various hollow organs, bone marrow, lesions and other small structures, or sub-compartments of organs, for which some examples are given in the following.

In general, while the absorbed dose to the contents of a hollow organ is not of interest, potential filling and excretion schedules may influence the doses to surrounding tissues. Examples of hollow organs include the urinary bladder [113], the gastro-intestinal tract [114‐117], and the peritoneal cavity [118]. For beta- or photon contributions, models incorporated within OLINDA/EXM 2.0 may be used for absorbed dose estimations. Assumptions related to the models, e.g. voiding schedule, should be stated in the dosimetry report.

Special considerations for red bone marrow and dosimetric approaches have been previously described in an EANM guideline [81]. In brief, blood-based approaches can be utilised as long as there is no specific uptake in the region; if so, quantitative imaging needs to be performed. The underestimation that can occur using an incomplete model has been demonstrated for several radiopharmaceuticals, including one of the provided examples: [¹⁷⁷Lu]Lu-lilotomab satetraxetan (see Supplementary information) [119]. The dose estimates will also depend on the fraction of red marrow [120], and the approach to estimating cellularity should be reported.

For sphere-like structures, e.g. lesions or small organs such as lacrimal glands, the absorbed self-dose can be estimated by spherical models of various emitters and sphere sizes [121].

In general, differences in TIAs or assumed radiosensitivities for sub-compartments of tissues, such as the renal cortex and medulla [122, 123], may necessitate separate small- or micro-scale investigations to provide clinical relevant information. Since such approaches are likely to be based on extrapolations, the assumptions should be clearly stated in the report. Small- or micro-scale investigations will also need to be evaluated in relation to the range of the radiations and is hence contextualised by the recent increase in alpha-emitter based treatments (Fig. 1).

When reporting absorbed dose values, it is important to know the uncertainty associated with the calculated result. This uncertainty aims to characterize the range of values within which the true value lies, specified with a stated level of confidence. The EANM practical guidance document on uncertainty analysis [4] recommends an uncertainty schema that addresses many of the systematic and random sources of error in a dose calculation. Important aspects include the uncertainty of each TAC value and the appropriateness of the fit function [124, 125] and other fundamental aspects such as the uncertainty associated with system calibration, image registration, and volume-of-interest (VOI) delineation. Reporting of uncertainties in first-in-human dosimetry studies is currently often lacking. However, given the potential impact these data have, both from a legislative, clinical, and radiation protection perspective, it is arguably most important in this regard to have an indication of the reliability of the results.

A source of uncertainty that can easily be reported and potentially reduced is that associated with the timing choice of activity measurements. The goodness-of-fit is a straightforward method allowing an uncertainty estimate of the fitting parameters and TIA to be reported. This is potentially one of the most important aspects for first-in-human dosimetry, where the pharmacokinetics of the new radiopharmaceutical and hence optimal timing of the dosimetry measurements are largely undefined. Thus, an interim analysis of uncertainty and the influence of individual measurement time points can be used to optimise and potentially reduce the measurements regimen for dosimetry in subsequent patients or cycles. In the diagnostic setting when determining biokinetics of a study cohort, uncertainty can either be determined by performing the fitting analysis on all participant data points simultaneously or by using the average of the cohort for each time point (assuming consistent timings across the population). In the former approach, the uncertainty in fitting can be determined using an ordinary least squares fit (assuming equal weighting of all data points). In the latter approach, an uncertainty for each time point can be determined from the population distribution and a generalised least squares approach used to propagate this into uncertainty in the fitting parameters [126].

The accuracy and traceability of activity concentration in the body are significant factors for dosimetry applications. For SPECT/CT-based dosimetry, the accuracy of the activity measurement is largely dependent on the accuracy of the radionuclide calibrator used to cross-calibrate the imaging system. Therefore, this should ideally be traceable to a national metrology institute and quoted as part of the uncertainty report. In the context of multi-centre, first-in-human dosimetry, harmonization of activity measurements is a key aspect to ensure that the dosage information obtained at an individual site can later be used for implementation and/or improvement of the therapy at other sites.

A summary of parameters to be reported is provided in Table 4. Methodological descriptions and qualitative considerations that should accompany the report are described in the following. Summary statistics including mean, standard deviation, and range of absorbed doses should be provided, even though the population sizes are normally too small (< 100) for what is commonly assumed reliable standard deviations. The values should be provided for each investigated tissue, in addition to WB. The method used to select source volumes should be described, and if multiple methods have been investigated, the results should be compared taking the uncertainties of each approach into account. For FDA or EMA approval, safety estimations for a “complete” set of human organs are traditionally required. For example, the list of target organs in the ICRP 128 [44] includes adrenals, bone surfaces, breast, brain, gallbladder wall, gastrointestinal tract, heart wall, kidneys, liver, lungs, oesophagus (thymus), ovaries, pancreas, red marrow, skin, spleen, testes, thymus, thyroid, urinary bladder wall, uterus, and other tissues, with the possible addition of lachrymal glands, salivary glands, and spinal cord. For diagnostic tracers, equivalent dose and effective dose should also be reported using the most recent radiation and tissue weighting factors. The sets of weighting factors used should be stated. Considerations should be given to including reports of effective dose using older factors for comparison purposes. For treatments with alpha-emitters, relative biological effectiveness (RBE) weighted absorbed doses can be calculated [110], although this is currently not recommended as the primary parameter to be reported due to uncertainty of the RBE-values. The choice of RBE value should be stated indicating its biological end-point and ideally also justified if this is included in addition to the absorbed doses. The separate contributions from all nuclides and radiations to the total absorbed dose should be listed if applicable. Reporting of biological effective dose (BED) is encouraged for therapies. However, the choice of radiobiological formula(s) and parameters should be stated. Ideally, a range of relevant radiobiological parameters should be explored if the exact values are unknown. Similar considerations for normal tissues apply to tumours or other regions of disease, which should also be included in the report if feasible. Absorbed doses for sub-structures or voxel-dose-maps can be reported if providing relevant clinical information. Volume cut-offs such as D90% or D98% can also be reported as summary statistics for relevant tissues. In accordance with principles of findability, accessibility, interoperability, and reusability (FAIR), all dose parameters to individual participants, as well as all activity measurements for each time point, TAC parameters (such as effective half-life), and potentially the volumes should be made available. Uncertainty estimates should be included for the sources investigated, and potential sources not investigated described. In addition, information about all underlying assumptions, details of acquisition or measurement settings, factors related to the radiopharmaceutical, as well as details of the dose calculations themselves [5] should be reported. How and why the final dosimetry study protocol was adapted based on the initial subject cohort should be detailed, if relevant. In cases where dosimetry planning is performed prior to a therapy, both the predictive and actual dosimetry results should be reported. This could insights of the validity of the planning approach, but also about the assumptions made increasing the information gained within the trial. When available, the results from the first-in-human dosimetry study should preferably also be compared to any available preclinical data of similar radiopharmaceuticals, contributing to a foundation for improving translational knowledge regarding biodistribution and dosimetry for radiopharmaceuticals.