Utility of Preclinical Dosimetry for First-in-Human Studies with PET Drugs Labeled with Common Radionuclides: A Systematic Review of Radiation Dose Estimates

Visual Abstract

Abstract

First-in-human studies of radiopharmaceuticals include an estimation of human radiation doses often extrapolated from preclinical biodistribution data. We reviewed radiation dose estimates derived from preclinical and clinical studies to determine the utility of preclinical dosimetry studies for estimating the human radiation dose of PET radiopharmaceuticals. Methods: Literature searches were performed for preclinical and clinical studies published between 1990 and 2021 that provided radiation dose estimates for common PET radiopharmaceuticals. Whole-body effective dose and maximum organ absorbed dose coefficients derived from animal and human studies were compared to evaluate their agreement. Ratios of animal-derived to human-derived dose estimates were calculated. Dosimetry data were further stratified by pharmacophore, animal species, and animal-to-human extrapolation method. Results: The SD in radiation dose estimates from clinical studies was generally lower compared with the SD in dose estimates extrapolated from animals. An overlap in the distribution of radiation dose estimates derived from animal and human studies was observed for short-lived radiopharmaceuticals (⁸²Rb-, ¹⁵O-, ¹³N-, ¹¹C-, ⁶⁸Ga-, ¹⁸F-, and ⁶⁴Cu-labeled, n = 394). The higher dose estimates and larger variability (spread) in preclinical and clinical radiation dose estimates observed for long-lived (⁸⁹Zr- and ¹²⁴I-labeled, n = 48) compared with short-lived radiopharmaceuticals highlight important differences in the radiation profile between the 2 groups. Conclusion: Our assessment of the agreement between animal and human radiation dosimetry data for PET drugs indicated a difference between short-and long-lived radiopharmaceuticals in terms of agreement, magnitude, and variability in dose estimates. A closer agreement between animal and human data, relatively lower doses, and lower variability in measured ED values in studies of short-lived PET radiopharmaceuticals suggest that approaches to forego preclinical dosimetry can be developed to facilitate clinical trials of new PET drugs for certain short-lived radionuclides. For first-in-human studies of long-lived PET radiopharmaceuticals or radiopharmaceuticals with long biological turnover, which are associated with higher radiation doses, animal-derived human dose estimates may provide necessary radiation safety information (e.g., identify unexpected high uptake in a particular organ) even though the radiation dose may be underestimated.

First-in-human (FIH) studies of diagnostic radiopharmaceuticals are typically supported by preclinical biodistribution and dosimetry studies to estimate radiation absorbed dose (AD) to various human organs. Federal regulation requires investigational new drug applications (INDs) for radiopharmaceuticals to provide data from animal and human studies estimating radiation AD to the whole body and critical organs in humans (1). Animal studies are conducted to evaluate biodistribution, stability, and metabolism and to inform the safety evaluation of a FIH study of a radiopharmaceutical. Thus, original IND submissions include human dose estimates that are often extrapolated from preclinical dosimetry studies. The extrapolation methods make assumptions about the differences in metabolism, anatomy, and biodistribution between animals and humans. These assumptions contribute to uncertainties in predicting the radiation dose to human organs (2).

The Food and Drug Administration has not defined organ AD or whole-body effective dose (ED) thresholds for diagnostic radiopharmaceuticals studied under an IND. However, the Code of Federal Regulations Title 21 §361.1(b)(3)(i) outlines the upper limits for radiation doses to individual organs and the whole body for radioactive drugs studied under an institutional Radioactive Drug Research Committee (RDRC) for human subjects. This regulation limits the whole-body ED and organ AD from a single or cumulative number of studies conducted within 1 y to 30 mSv for the whole body (or 30 mGy for radiosensitive organs, such as blood-forming organs, lens of the eye, and gonads), and to 50 mGy for all other organs, respectively (3).

To our knowledge, Sparks and Aydogan have provided the most comprehensive assessment of uncertainties and prediction bias in animal-to-human dosimetry extrapolations (2). They evaluated the performance of commonly used extrapolation methods to predict time-integrated activities (or residence time) for human organs using animal data and found that human organ time-integrated activity coefficients tend to be underestimated if derived from animal studies (2). Since the AD in a particular organ is proportional to the time-integrated activity coefficients within that organ, this underestimation is concerning, especially when time–activity measurements are used to estimate dosimetry of therapeutic radiopharmaceuticals.

Zanotti-Fregonara et al. proposed pathways to assess radiation safety for FIH studies with new ¹¹C- and ¹⁸F-labeled radiopharmaceuticals in the absence of animal-derived dosimetry data (4–6). Other studies have assessed the strengths and weaknesses of various methods and factors for animal-to-human dosimetry extrapolations (7–9).

We aimed to assess the degree of uncertainty in human radiation dosimetry derived from animal data and to evaluate the utility of preclinical dosimetry studies to derive human dose estimates for FIH studies. We systematically reviewed dosimetry estimates of PET radiopharmaceuticals derived from preclinical and clinical studies, focusing on ⁸²Rb-, ¹⁵O-, ¹³N-, ¹¹C-, ⁶⁸Ga-, ¹⁸F-, ⁶⁴Cu-, ⁸⁹Zr-, or ¹²⁴I-labeled radiopharmaceuticals.

MATERIALS AND METHODS

Literature searches were performed between January to June 2021 using PubMed, Embase, SciFinder, Web of Science, and Google Scholar to compile available radiation dosimetry data from full-text articles in English published between 1990 and 2021 for ⁸²Rb-, ¹⁵O-, ¹³N-, ¹¹C-, ⁶⁸Ga-, ¹⁸F-, ⁶⁴Cu-, ⁸⁹Zr-, and ¹²⁴I-labeled radiopharmaceuticals. Details of the full search strategy are provided in the supplemental materials, available at http://jnm.snmjournals.org.

Articles that reported human organ radiation dose estimates from animal and human studies (whole-body ED and human organ AD per unit of administered activity) were included in this review. The recommended administered activity and radiation dosimetry data from the Dosage and Administration section in the prescribing information of approved PET radiopharmaceuticals was also collected. Publications lacking whole-body ED and organ AD data were excluded.

Human radiation dose coefficients derived from preclinical dosimetry studies and calculated using data from clinical studies were compared and are presented in increasing physical half-life order throughout. Human radiation doses for each clinical study were calculated by multiplying the reported ED and maximum AD coefficient by the average reported administered activity to all participants. To compare the human radiation doses reported in the literature to the RDRC organ dose limits, the proportion of studies with ED and AD values above the RDRC organ dose thresholds was calculated.

Estimates based on clinical data are referred to as human ED and AD or human-derived whole-body ED and human organ AD. Human radiation dose estimates derived from animals are referred to as animal ED and AD or animal-derived human whole-body ED and human organ AD. The mean, median, and range of animal and human ED and maximum AD coefficients from each study were plotted and tabulated for each radionuclide.

For radiopharmaceuticals with ED and AD data available from both animal and human studies, animal-to-human ratios were calculated and are referred to as ED and AD ratios. The dose delivered to each target organ was tabulated, and the animal-derived doses were normalized to human-derived doses to obtain AD ratios for each target organ. AD ratios were grouped and plotted as histograms to display ratio frequency distributions. The geometric mean, SD, variance, and median for each AD ratio distribution were calculated. AD ratio calculations involved analyses of all target organ AD data, whereas all other discussion of AD estimates pertains to the maximum organ AD.

NOTEWORTHY

• This review compiles PET radiopharmaceutical dosimetry data from 442 studies covering 366 unique radiopharmaceuticals, including 157 preclinical and 327 clinical dose estimates, across 9 radionuclides.

• Greater animal-to-human dosimetry agreement, lower radiation dose, and lower variability in radiation dose were observed across short-lived PET radiopharmaceuticals labeled with ⁸²Rb, ¹⁵O, ¹³N, ¹¹C, ⁶⁸Ga, ¹⁸F, and ⁶⁴Cu compared with long-lived PET radiopharmaceuticals (⁶⁴Cu-labeled radiopharmaceuticals with long biological turnover or radiopharmaceuticals labeled with ⁸⁹Zr or¹²⁴I).

• Approaches could be developed to forego preclinical dosimetry studies for certain short-lived PET radiopharmaceuticals in FIH studies.

Additional information was collected regarding the pharmacophore, animal species, and animal-to-human dosimetry extrapolation methods to compare the study methodology of the selected publications and to analyze the degree of differences between methodologies.

To investigate possible differences in dosimetry estimates between radiopharmaceuticals labeled with small molecules, peptides, and antibodies, the animal and human ED values were plotted by radiopharmaceuticals labeled with these pharmacophores.

To assess differences between animal species, the mean ED and AD ratios were stratified by preclinical data from small animals (mice and rats) and large animals (nonhuman primates and pigs).

To compare our findings with those of Sparks and Aydogan (2), frequency distributions of organ AD ratios were plotted and assessed for each animal-to-human dosimetry extrapolation method used in the studies included.

RESULTS

Review of Radiation Dose Estimates

Our search identified 442 studies that contained ED or AD data from preclinical or clinical studies and were retained for analysis (Fig. 1). We identified 366 unique radiopharmaceuticals (i.e., unique chemical or biological entities) consisting mostly of molecular targets for neurologic and oncologic indications and targets for cardiovascular, metabolic, inflammatory, and infectious disease indications (Table 1). In total, 157 preclinical ED estimates, 154 preclinical AD estimates, 327 clinical ED estimates, and 320 clinical AD estimates were identified (Tables 2 and 3). Some studies reported ED and AD estimates for more than 1 species, reported ED and AD estimates from both animals and humans for 1 radiopharmaceutical, or reported dosimetry data for more than 1 radiopharmaceutical in a single publication. No data from preclinical studies were available for ⁸²Rb-, ¹⁵O-, and ¹³N-labeled radiopharmaceuticals.

FIGURE 1.

Preferred reporting items for systematic reviews and meta-analyses (PRISMA) flowchart of the study selection process.

TABLE 1.

Studies Selected for Systematic Review

		No. studies		No. PET radiopharmaceuticals, by subject
Radionuclide	Half-life	Identified	Included	Animal	Human	Animal and Human	Total	FDA-approved PET drug
82Rb	1.25 min	13	5	—	1	—	1	[82Rb]rubidium chloride
15O	2.03 min	14	5	—	1	—	1
13N	9.97 min	13	1	—	1	—	1	[13N]ammonia
11C	0.014 d	186	100	35	74	22	87	[11C]choline
68Ga	0.047 d	97	68	18	46	5	59	[68Ga]DOTATATE, [68Ga]edotreotide [68Ga]gozetotide
18F	0.076 d	219	175	42	121	22	141	[18F]FDG, [18F]sodium fluoride, [18F]fluciclovine, [18F]florbetapir, [18F]flutemetamol, [18F]florbetaben, [18F]fluorodopa, [18F]fluoroestradiol, [18F]flortaucipir, [18F]piflufolastat
64Cu	0.53 d	133	40	25	9	2	32	[64Cu]DOTATATE
89Zr	3.3 d	35	28	12	14	1	25
124I	4.2 d	101	20	8	13	2	19

TABLE 2.

Reported Animal- and Human-Derived Whole-Body ED Coefficients and Administered Activities

Variable	82Rb	15O	13N	11C	68Ga	18F	64Cu	89Zr	124I
Animal derived
n*	—	—	—	39	21	47	28	13	9
Mean ± SD (mSv/MBq)	—	—	—	0.006 ± 0.002	0.020 ± 0.006	0.023 ± 0.020	0.039 ± 0.029	0.468 ± 0.205	0.252 ± 0.157
Median (mSv/MBq)	—	—	—	0.005 (0.003–0.014)	0.017 (0.011–0.032)	0.019 (0.0001–0.123)	0.033 (0.005–0.140)	0.540 (0.047–0.880)	0.260 (0.011–0.550)
CV (%)	—	—	—	44	31	85	76	44	62
Human derived
n*	5	5	1	78	54	143	12	16	13
Mean ± SD (mSv/MBq)	0.0011 ± 0.0002	0.002 ± 0.002	0.007	0.006 ± 0.002	0.025 ± 0.012	0.021 ± 0.007	0.037 ± 0.014	0.544 ± 0.146	0.355 ± 0.142
Median (mSv/MBq)	0.001	0.001	0.007	0.005	0.022	0.020	0.035	0.533	0.368
Mean AA (MBq) Mean AA (mCi)	1129 (536–1502) 31 (15–41)	1185 (944–1427) 32 (26–39)	736 20	471 (106–1190) 12.7 (2.9–32.2)	175 (101–263) 4.7 (2.7–7.1)	259 (90–606) 7.0 (2.4–16.4)	226 (83–441) 6.1 (2.2–11.9)	91 (32–191) 2.4 (0.9–5.2)	115 (58–201) 3.1 (1.6–5.4)
% difference in mean ED†	—	—	—	0	−20	10	5	−14	−29
CV (%)	15	91	—	39	49	35	37	27	40

^*Number of effective dose estimates.

^†(Animal ED − human ED)/human ED × 100%.

CV = coefficient of variation; AA = administered activity.

Data in parentheses are ranges.

TABLE 3.

Maximum Reported Animal- and Human-Derived Organ AD Coefficients for Radionuclides

Variable	82Rb	15O	13N	11C	68Ga	18F	64Cu	89Zr	124I
Animal derived
n*	—	—	—	39	18	47	28	13	9
Mean ± SD (mGy/MBq)	—	—	—	0.030 ± 0.021	0.361 ± 0.417	0.186 ± 0.294	0.221 ± 0.309	1.405 ± 0.633	2.811 ± 2.093
Median (mGy/MBq)	—	—	—	0.025 (0.005–0.104)	0.193 (0.014–1.850)	0.081 (0.0008–1.810)	0.116 (0.005–1.480)	1.289 (0.603–2.900)	3.200 (0.078–6.450)
CV (%)	—	—	—	69	116	158	140	45	74
Organ with maximum AD	—	—	—	Small intestine	Kidneys	Gallbladder	Kidneys	Heart	Kidneys
Human derived
n	5	3	1	77	54	141	12	16	11
Mean ± SD (mGy/MBq)	0.006 ± 0.002	0.002 ± 0.0004	0.0071	0.035 ± 0.033	0.262 ± 0.181	0.160 ± 0.122	0.297 ± 0.115	2.375 ± 2.441	1.634 ± 0.922
Median (mGy/MBq)	0.005 (0.005–0.009)	0.002 (0.002–0.003)	0.0071	0.026 (0.010–0.194)	0.224 (0.048–1.090)	0.134 (0.014–0.797)	0.315 (0.076–0.460)	1.740 (1.180–11.400)	1.580 (0.680–3.760)
Organ with maximum AD	Kidneys	Heart	Heart	Gallbladder	Bladder	Gallbladder	Upper large intestine	Liver	Kidneys
% difference in mean maximum AD†	—	—	—	−6	44	16	−26	−41	72
CV (%)	30	19	—	95	69	76	39	103	56

^*Number of AD estimates.

^†(Animal ED − human ED)/human ED × 100%.

CV = coefficient of variation.

Data in parentheses are ranges.

The mean human whole-body ED and maximum organ AD coefficients for each radionuclide reflect sex-averaged values for each study. In general, minimal differences were observed between sexes. All dosimetry data and radiopharmaceutical targets are compiled in Supplemental Tables 1–9. The percent difference in mean ED coefficients between animal- and human-derived data for short-lived radionuclides was generally small (Fig. 2; Table 2). Higher percent differences were observed overall between animal- and human-derived maximum organ AD coefficients compared with ED coefficients (Fig. 3; Table 3).

FIGURE 2.

Human whole-body ED coefficients from published data.

FIGURE 3.

Maximum human organ AD coefficients from published data.

For ¹²⁴I- and ⁸⁹Zr-labeled radiopharmaceuticals, the large range observed in mean ED values was attributable to higher dose values. The mean ED coefficients from human studies were approximately 10- and 15-fold higher than ⁶⁴Cu-labeled radiopharmaceuticals, respectively (Fig. 2; Table 2), highlighting important differences in the radiation profile and risk for short- versus long-lived radionuclides.

The relative differences in ED values (coefficient of variation; Table 2) were generally smaller for human studies, with the exception of ⁶⁸Ga-labeled radiopharmaceuticals. For clinical studies, intragroup variability in ED was generally not significantly higher for longer-lived nuclides; however, considerable intragroup variability in maximum AD estimates was observed for short-lived nuclides, although this is likely a reflection of the number of studies and radiopharmaceutical targets within each nuclide group. The intragroup variability for animal-derived ED and AD estimates was generally larger than that observed in clinical studies.

ED values for all studies with ⁸²Rb-, ¹⁵O-, ¹³N-, ¹¹C-, ⁶⁸Ga-, ¹⁸F-, and ⁶⁴Cu-labeled radiopharmaceuticals were below the 30 mSv threshold (Table 4). The proportion of studies with maximum AD values above 50 mGy ranged between 0% and 50% of the studies with ⁸²Rb-, ¹⁵O-, ¹³N-, ¹¹C-, ⁶⁸Ga-, ¹⁸F-, and ⁶⁴Cu-labeled radiopharmaceuticals, with studies of ⁶⁴Cu-labeled radiopharmaceuticals having the highest percentage. For ⁸⁹Zr- and ¹²⁴I-labeled radiopharmaceuticals, the maximum AD values for all studies exceeded the dose threshold of 50 mGy.

TABLE 4.

Studies with Human Whole-Body ED Exceeding 30 mSv and Maximum ADs Exceeding 30 and 50 mGy

		Maximum organ AD
Radionuclide	ED > 30 mSv	>50 mGy	>30 mGy
82Rb	0/4 (0)	0/4 (0)	0/4 (0)
15O	0/2 (0)	0/2 (0)	0/2 (0)
13N	0/1 (0)	0/1 (0)	0/1 (0)
11C	0/77 (0)	1/76 (1)	5/76 (7)
68Ga	0/54 (0)	13/54 (24)	33/54 (61)
18F	0/141 (0)	37/140 (26)	71/140 (51)
64Cu	0/12 (0)	6/12 (50)	10/12 (83)
89Zr	11/15 (73)	15/15 (100)	15/15 (100)
124I	9/13 (69)	11/11 (100)	11/11 (100)
Total	20/319 (6)	83/315 (26)	145/315 (46)

Data are number of studies (numerator) over total studies (denominator), followed by percentage in parentheses.

These group-level differences in mean and maximum radiation dose values between animal and human studies for various PET radiopharmaceuticals do not account for the varying types of radiopharmaceuticals, kinetics, or tissues involved in animal and human studies. Supplemental Table 10 shows the percent differences between animal- and human-derived ED and AD coefficients, respectively, for a subset of cases when the same radiopharmaceutical was the subject of both animal and human studies. Animal and human data for the same radiopharmaceutical were primarily available for short-lived radionuclides only. The study-level percent differences showed both underestimation and overestimation of animal-derived compared with human-derived estimates in varying degrees, similar to the group differences observed (Tables 2 and 3).

Radiation Dose Stratified by Pharmacophore

Except for ⁶⁸Ga (Fig. 4A), major differences in ED coefficients were not observed among different pharmacophores for short-lived radionuclides. Compared with ⁶⁸Ga- and ¹⁸F-labeled radiopharmaceuticals, ¹¹C-labeled radiopharmaceuticals had a relatively low radiation profile (mean human ED was approximately 4-fold smaller) and less variability in both animal and human ED, independent of pharmacophore (absolute range, 0.012 versus 0.071 mSv/MBq; coefficient of variation, ∼40% versus 60% for ¹¹C-labeled vs. ⁶⁸Ga- and ¹⁸F-labeled radiopharmaceuticals, respectively). These results were expected, given the short physical half-life of ¹¹C-labeled radiopharmaceuticals and the different effective half-lives of the radiolabeled molecule types. For the long-lived radionuclides, the mean ED value was more than 2-fold higher (⁶⁴Cu-labeled radiopharmaceuticals) and more than 13-fold higher (¹²⁴I-labeled radiopharmaceuticals) than the mean ED for ¹⁸F-labeled radiopharmaceuticals (Fig. 4B). Major differences in ED values for ⁶⁴Cu-and ¹²⁴I-labeled radiopharmaceuticals were not observed across different pharmacophores, although the disparity of available data may explain these observations. All except 1 ⁸⁹Zr-labeled radiopharmaceutical were identified as antibody-based, and a pharmacophore-type stratification was not conducted for ⁸⁹Zr-labeled radiopharmaceuticals.

FIGURE 4.

Human whole-body ED coefficients stratified by pharmacophore.

Radiation Dose Stratified by Animal-to-Human Extrapolation Method

The stratification of preclinical studies included in this review by animal-to-human extrapolation method consisted of 3 methodologies: no extrapolation, relative organ mass extrapolation, and combined relative organ mass and physiologic time extrapolation methods.

The distributions of AD ratios showed that most animal-based calculations used the relative organ mass extrapolation method (Figs. 5A–5D). For all radiopharmaceuticals and extrapolation methods combined, the geometric mean and median of the AD ratios (0.73 and 0.94, respectively) highlight the underestimation of human doses derived from animal biodistribution studies. Across all extrapolation methods, the highest organ AD ratios ranged from 0.6 (testes) to 34 (lower large intestine wall) (Fig. 5A).

FIGURE 5.

Distribution of human organ AD ratios.

DISCUSSION

Radiation Dose

To our knowledge, this review provides the largest compilation of radiation dose estimates from preclinical and clinical studies thus far for 366 PET radiopharmaceuticals containing ⁸²Rb-, ¹⁵O-, ¹³N-, ¹¹C-, ⁶⁸Ga-, ¹⁸F-, ⁶⁴Cu-, ⁸⁹Zr-, or ¹²⁴I-labeled radiopharmaceuticals and extends the previous reviews focused on ¹¹C- and ¹⁸F-labeled radiopharmaceuticals (4–6). This review provides a comprehensive assessment of the level of agreement between animal- and human-derived dosimetry of PET radiopharmaceuticals and establishes a basis for evaluating whether available dosimetry data are sufficient to support new approaches for assessing the safety of FIH studies with investigational PET radiopharmaceuticals.

The predictive value of animal studies for human organ radiation dosimetry has been analyzed in smaller studies (2,9–11). Overall, we have found agreement of mean dosimetry estimates between animal- and human-derived data, particularly with short-lived radionuclides.

These results show that ⁸²Rb-, ¹⁵O-, ¹³N-, ¹¹C-, ⁶⁸Ga-, ¹⁸F-, and ⁶⁴Cu-labeled radiopharmaceuticals have a safe radiation profile relative to the RDRC radiation dose thresholds. The whole-body ED estimates for these radiopharmaceuticals were below the whole-body radiation dose threshold of 30 mSv and only a small proportion of studies showed an organ AD exceeding the 50-mGy threshold.

Greater variability was observed between the animal and human mean AD values compared with the mean ED values for ¹¹C-, ⁶⁸Ga-, ¹⁸F-, ⁶⁴Cu-, ⁸⁹Zr-, and ¹²⁴I-labeled radiopharmaceuticals. However, there are limitations when using mean values to evaluate agreement between animal- and human-derived dose estimates. Therefore, it is valuable to consider animal-to-human dosimetry agreement, particularly for radiopharmaceuticals with maximum organ AD values above the RDRC radiation dose thresholds.

Other general considerations for the variability in the mean dose estimates observed across preclinical and clinical studies include factors such as radiopharmaceutical pharmacokinetic profile, biological half-life, small sample size for dosimetry studies, and intersubject variability across various diseases. Although we did not examine systematic differences between the type of radiopharmaceuticals used in animal versus human studies, we expect that considerations of the factors listed above would contribute to such systematic differences in our comparison between animal- and human-derived radiation dose estimates.

Level of Agreement

Molecular entities (e.g., antibodies, small molecules, peptides) affect the biodistribution and dosimetry profile of radiopharmaceuticals. Compared with small molecules and large proteins, peptides have unique pharmacokinetic characteristics because of their susceptibility to proteolytic degradation, short plasma half-lives, and rapid renal metabolism and excretion (12). This is a particularly important consideration for radiopharmaceuticals of long-lived PET radionuclides (i.e., ⁶⁴Cu-, ⁸⁹Zr-, and ¹²⁴I-labeled radiopharmaceuticals) because longer physical and biological half-lives would result in a higher organ AD.

Animal biodistribution data both underestimated and overestimated human ED and specific human organ doses. ED and AD derived from larger animals provided less overestimation of dosimetry compared with those from smaller animals; however, data on long-lived radiopharmaceuticals in larger animals are lacking (Table 5). Data from rats showed the greatest overestimation of ED and AD estimates over other species. Despite some overestimation for AD, data from mice provided the most accurate ED estimates.

TABLE 5.

Mean Animal-Derived to Human-Measured Human Whole-Body ED and Human Organ AD Ratios

Variable	n	11C	68Ga	18F	64Cu	89Zr	124I	Overall
Mean ED ratio*		1.00 ± 0.38	1.00 ± 0.22	1.22 ± 1.06	1.03 ± 0.11	1.75	0.64	1.10 ± 0.74
No. ratios per species		26	6	26	3	1	1	63
By species
Mouse	27	0.94	0.64	1.07	0.97	1.75	0.64	1.00
Rat	9	1.29	1.06	3.46	1.14	—	—	1.74
Monkey	23	1.07	1.13	1.28	—	—	—	1.16
Pig	4	0.95	1.06	0.62	—	—	—	0.88
By species size
Small	36	1.12	0.85	2.27	1.06	1.75	0.64	1.28
Large	27	1.01	1.10	0.95	—	—	—	1.02
Mean AD ratio†		1.04 ± 0.22	1.02 ± 0.77	1.12 ± 0.48	1.50 ± 0.81	2.64	0.91	1.37 ± 1.01
No. organs‡		25	24	26	24	24	24	147
By species
Mouse	27	1.00	0.58	1.03	1.50	2.64	0.91	1.28
Rat	9	1.34	1.49	2.18	1.48	—	—	1.62
Monkey	23	1.07	1.56	1.34	—	—	—	1.32
Pig	4	1.18	0.88	0.87	—	—	—	0.98
By species size
Small	36	1.17	1.04	1.61	1.49	2.64	0.91	1.48
Large	27	1.13	1.22	1.11	—	—	—	1.15

^*One animal-to-human ED ratio was calculated per radiopharmaceutical and species if more than 1 species was studied with the same radiopharmaceutical.

^†Animal-to-human AD ratios for each target organ were calculated per radiopharmaceutical and species if more than 1 species was studied with the same radiopharmaceutical. Ratios were then averaged across radiopharmaceuticals for each target organ.

^‡Includes all available target organs except whole body.

Sparks and Aydogan analyzed human organ residence times derived from animal dosimetry data for ³H-, ¹⁸F-, ⁷⁵Se-, ^99mTc-, ¹¹¹In-, ¹²³I-, ¹²⁵I-, ¹³¹I-, ^117mSn-, ¹⁸⁶Re-, and ²⁰¹Tl-labeled radiopharmaceuticals (2). Their analyses showed an underestimation of radiation exposure to a specific organ, regardless of the extrapolation method used. Our analyses showed that the AD ratio distributions using either the mass-extrapolation method or the no-extrapolation method were similar (geometric means of 0.74 and 0.73, respectively) and correspond to a 1.3-fold underestimation of animal-derived human dose estimates compared with a 2.2-fold underestimation in the same distributions plotted by Sparks and Aydogan (geometric means of 0.43 and 0.45) (Figs. 5B and 5C). Additionally, our review addressed certain limitations in the Sparks and Aydogan analysis (e.g., consideration of specific organs and species because of insufficient data).

Comparisons of the geometric mean, geometric SD, and median of methods with and without relative organ mass extrapolation showed that the 2 distributions are quite similar. Combined time and mass extrapolation resulted in the most underestimation of AD from animal-to-human compared with the other 2 extrapolation methods. However, there are limitations to this comparison because of the insufficient sample size.

For short-lived radiopharmaceuticals, which comprised most of the available data, the degree of concordance between animal and human AD estimates for the same radiopharmaceutical varied widely across radionuclides, pharmacophores, and organs with maximum AD, indicating both underestimation and overestimation and large variability in extrapolating the human radiation dose from animal data. Such specific cases of radiopharmaceuticals where both animal and human data are available could be considered for future analysis. The overall goal of this systematic review, which included a large body of data, was to investigate and present overarching trends related to estimating human radiation dose estimates from animal and human studies.

Several preclinical dosimetry studies lacked a description of the extrapolation methods used. Moreover, the comparison of dosimetry estimates for each method of extrapolation used does not necessarily reflect the other assumptions used in dose calculations. For example, the software used for dose calculation and the incorporated models may have differed among studies.

Many studies lack information on the dose calculation software or versions used (e.g., OLINDA/EXM, MIRDOSE), as well as other assumptions in the calculations (e.g., use of the bladder model and specification of the bladder-voiding interval), which limits interpretation of the data. The choice of the bladder-voiding interval affects the AD to the bladder, other organs, and the whole-body ED. Although MIRD Pamphlet No. 14 provides the framework for the calculation of radiation-absorbed dose to the urinary bladder wall, it does not define a standard voiding schedule to follow using the dynamic bladder model (13). The void times used across studies varied greatly (between 0.5 and 6 h), and many studies did not report dose calculations using bladder models or lack description about the use of one. The variability of the bladder-voiding models, voiding dynamics, and time–activity curve for radiopharmaceutical bladder activity contribute to challenges of standardizing the voiding parameters for calculating dosimetry data from preclinical and clinical studies (13,14).

CONCLUSION

This comprehensive review of PET radiopharmaceutical dosimetry assessed the agreement between animal and human radiation dosimetry. The SD and range of animal-derived human radiation dose estimates were generally higher than those from clinical data. Nevertheless, the distribution of radiation dose estimates from animal and human studies demonstrated a reasonable agreement between animal- and human-derived dosimetry with a relatively safe radiation profile, particularly with short-lived PET radiopharmaceuticals. These observations indicate that approaches to forego preclinical dosimetry studies for certain short-lived PET radiopharmaceuticals can be developed to facilitate the study of new investigational PET drugs. For long-lived PET radionuclides and PET radiopharmaceuticals with long biological turnover, animal-derived human dose estimates remain necessary to inform radiation safety before beginning an FIH trial.

KEY POINTS

QUESTION: Are there clinical data that could justify foregoing nonclinical dosimetry studies generally used to predict the radiation risk of PET drugs in FIH studies?

PERTINENT FINDINGS: Our assessment of the agreement between animal- and human-derived radiation dosimetry data for PET drugs indicated a difference between short-and long-lived radiopharmaceuticals in terms of agreement, magnitude, and variability in dose estimates. A closer agreement between animal and human data, relatively lower whole-body ED and organ AD values, and lower variability in measured dosimetry estimates were observed in studies of short-lived PET radiopharmaceuticals labeled with ⁸²Rb, ¹⁵O, ¹³N, ¹¹C, ⁶⁸Ga, ¹⁸F, and ⁶⁴Cu compared with long-lived radiopharmaceuticals (⁶⁴Cu-labeled radiopharmaceuticals with long biological turnover or radiopharmaceuticals labeled with ⁸⁹Zr or ¹²⁴I).

IMPLICATIONS FOR PATIENT CARE: These findings inform the radiation dosimetry and safety of PET drugs and suggest that approaches to forego preclinical dosimetry studies for certain short-lived PET radiopharmaceuticals can be developed. These approaches are expected to facilitate the study of investigational PET drugs and expedite the development of new diagnostic and theranostic options for patient care.

DISCLOSURE

No potential conflict of interest relevant to this article was reported. This article reflects the views of its authors and should not be construed to represent FDA’s views or policies.