In this study, we evaluated human radiation dosimetry for the 11C-labeled 5-HT2AR agonist PET radioligand, Cimbi-36, using two different 11C-labeling positions. Effective dose for 11C-Cimbi-36, the labeling position of choice for neuroimaging , was 5.5 μSv/MBq, resulting in a radiation dose of 3.3 mSv for a PET scan following injection of 600 MBq and allows for injection of 1.8 GBq per study in order to stay below 10 mSv, which is the recommended limit for studies involving healthy volunteers, that provide intermediate to moderate benefits to society .
11C-Cimbi-36 is a radioligand developed for brain imaging, and thus magnetic resonance imaging is often used for anatomical purpose, yielding no additional radioactive exposure. However, a low-dose brain-only CT for attenuation correction can be performed at a fraction of the PET effective dose.
Because of the short half-life of carbon-11 (20.4 min), organs with high perfusion tend to get the highest absorbed doses of radiation from 11C-labeled radioligands, but also excretory organs such as the liver, kidney, gallbladder, and urinary bladder get high exposure . Indeed, this is also true for 11C-Cimbi-36, as the liver, kidneys, and urinary bladder are among the five organs with highest absorbed doses. This reflects the relatively fast metabolism seen for Cimbi-36 [1, 15].
The effective dose for the alternative labeling position, 11C-Cimbi-36_5, was only slightly lower; 5.3 μSv/MBq (equivalent to a radiation dose of 3.2 mSv if injecting 600 MBq), with the difference being most pronounced for the urinary bladder, i.e., a fivefold reduction in absorbed dose. Despite the small sample size, the generally lower uptake for 11C-Cimbi-36_5, including in the urinary bladder, can be attributed to the higher level of radioactivity in the form of small diffusible substances (11C-formaldehyde, 11C-formic acid), and possibly also volatile (11C-CO2) substances, comprised in the M1 radiometabolite fraction (Fig. 1) .
The difference in effective dose between the two 11C-labelings of Cimbi-36 is within the limitations of the method and inter-individual variability. Thus, even the relatively fast metabolism of parent tracers (resulting in radiometabolites with different physical properties) does not affect dosimetry outcomes noticeably. This attests to the notion that dosimetry of 11C-labeled radioligands is mostly dependent on the initial blood perfusion phase.
The effective doses of both radioligands are in line with other 11C-labeled PET tracers; range 3.0–7.8 μSv/MBq, with the exception of one (11C-WAY-100635; 14.1 μSv/MBq) [13, 23]. The estimated effective dose for 11C-Cimbi-36 in this study is also in accordance with preclinical studies; effective dose was found to be 4.9 μSv/MBq and 7.7 μSv/MBq, when extrapolating from pig and rat dosimetry, respectively . In the case of 11C-Cimbi-36, studies in pigs thus proved to have a better translational value compared with rats, but differences across species cannot be predicted . The decision of whether to undertake human radiation dosimetry studies of a new radioligand, when 11C-labeled tracers show this limited variability, should therefore be considered. Dosimetry studies are both costly and time-consuming, and expose healthy individuals to radiation, with (perhaps) no added benefit compared with a conservative estimate based on the highest reported effective dose, as suggested in one study . A generic model for 11C-labeled substances for brain imaging predicts an effective dose of 4.5 μSv/MBq, based on the assumption that radioactivity is rapidly and uniformly distributed throughout all tissue . Yet, in the absence of preclinical data or model that can predict with sufficient certainty if one organ receives an excessive absorbed dose, it can be justified to conduct a small number of whole-body scans to take this possible scenario into account.
Despite similar radiation dosimetry for the two tracers, 11C-Cimbi-36 continues to be the preferred radioligand for 5-HT2AR neuroimaging studies, as it has a better signal-to-noise ratio in the brain .
Biodistribution, pharmacology, and metabolism
Biodistribution can help shed light on pharmacokinetics , but interpretation of the SUV curves for the internal organs should be done with caution, as they represent both the parent tracers and their radiolabeled metabolites. This leaves the following possible interpretations of radioactivity uptake: 1) specific binding of parent tracer; 2) enzymes or other sites of metabolism of parent tracer or labeled metabolites; 3) excretion of parent tracer or radiometabolites; 4) non-specific binding of parent tracer or radiometabolites. As biodistribution of Cimbi-36 is particularly interesting because of its use as a recreational drug (alias 25B-NBOMe) with many case reports of toxicity and even fatalities , we will discuss 1–3 in more detail in the following.
Specific binding: Cimbi-36 not only binds to the 5-HT2AR, but also has affinity for, e.g., the 5-HT2C receptor and Sigma 2 receptors; however, the affinity for these two receptors are 15-  and 120-fold  lower, respectively. As no blocking agent was used in the present study, potential 5-HT2A/CR binding in extracerebral tissue cannot be determined with complete certainty from our data. The primary focus of 5-HT is on its effects in the central nervous system, but peripheral 5-HT is implicated in many bodily functions, such as energy metabolism through actions in the gut, pancreas, and fatty tissues [28,29,30,31] but also immune response, inflammation , and pain stimuli . High uptake in, e.g., the small intestines or the pancreas might thus represent specific 5-HT2A/2CR binding. 5-HT2ARs are also found on human alpha, beta, and delta cells in the pancreas . In type 2 diabetic patients, increased expression of the 5-HT2AR was found in unspecified pancreatic islet cells relative to healthy controls , while 5-HT2CR inhibits insulin secretion by beta cells in a diabetic mouse model . These findings may in part explain the hyperglycemia seen in several cases of NBOMe intoxication .
Surprisingly, we found high uptake in what we believe to be brown adipose tissue (BAT). A newly conducted study links administration of 25B-NBOMe in rats to hyperthermia and thermogenesis of BAT . These effects are thought to be mediated through peripheral 5-HT2ARs as central adrenergic and serotonergic neurons were selectively destroyed by neurotoxins in these animals. BAT stimulation by 5-HT has been shown to inhibit beta-adrenergic signaling and BAT thermogenesis  through the 5-HT3R [30, 37], while 5-HT2AR stimulation increases BAT thermogenesis [38, 39]. Thus, peripheral 5-HT might have bidirectional physiological effects on BAT thermogenesis, depending on 5-HT receptor expression pattern, which in turn could reflect differences in the physiological state (active vs. inactivated BAT). Interestingly, hyperthermia is a common complication in NBOMe intoxications, with and without seizures , and peripheral 5-HT2A/2CR effects on BAT might be a contributing factor.
The severe toxicity of 25B-NBOMe and other 5-HT2R selective agonists is curious in light of the relatively low toxicity of classical non-selective 5-HT2AR hallucinogens such as psilocybin, LSD, and mescaline [40, 41]. Considering the different effects on BAT thermogenesis mediated by different 5-HT receptors, toxicity might then arise because of, rather than despite, high selectivity. Other factors, such as inter-individual differences in metabolism might also contribute, as discussed below.
Metabolism: Not surprisingly, the liver showed high uptake beyond the initial perfusion phase reflecting the extensive metabolism of the parent tracers. In both pig and human, the metabolic route is through O-demethylation (phase I reaction, Fig. 1), primarily at the 5’-position, followed by glucuronide conjugation (phase II reaction) . 11C-labeling in the 2-methoxybenzyl-position (11C-Cimbi-36) gives rise to two radiometabolite fractions: M1, comprising small polar radiometabolites, which is likely a mixture of 11C-formaldehyde, 11C-formic acid, and 11C-CO2/bicarbonate; and M2, which was identified as a 11C-glucuronide conjugate . Changing the 11C-labeling to the 5′-methoxy-4-bromophenethylamine position (11C-Cimbi-36_5) eliminates the radiolabeled form of the glucuronide conjugate (M2), leaving only the M1 fraction . Caspar et al.  found that the O-demethylation is catalyzed by CYP2C19 (cytochrome P-450 enzyme) and CYP2D6, and the relative contribution to hepatic clearance was estimated to be 69% for CYP2D6, with the remaining clearance attributed to CYP2C19 and CYP3A4 (catalyzing N-dealkylation and/or hydroxylation). These CYP isoforms are also found in the small intestines , which we hypothesize account for part of the high uptake of 11C-Cimbi-36 in the proximal small intestine throughout the scan. The other part possibly representing specific binding to 5-HT2A/2CRs. 11C-Cimbi-36 also shows higher uptake than 11C-Cimbi-36_5 in the liver, which we speculate is because the resulting phase I metabolite of the former goes on to being glucuronidated by glucuronosyltransferases (UGT) . The labeled metabolites of 11C-Cimbi-36_5 are small polar, likely volatile, substances  and therefore not likely retained in the liver to the same degree. The same pattern is seen in the kidneys, in which several UGTs are present , although transporters in the renal tubule cells responsible for the secretion of the 11C-labeled glucuronide conjugate might better explain the difference in uptake seen between the tracers. This is further substantiated as the difference in uptake does not emerge until approximately 20 min into the scan, when most of the activity is in the form of the glucuronide conjugate .
Inter-individual differences due to genetic variability in CYP isoforms or induction/inhibition by foods or drugs may contribute to the toxicity of NBOMes. We note that the red and purple lines are near identical in whole blood (Fig. 3) and follow the same shape in the liver, kidney, and bladder. The case is similar for the blue and green lines, which follow a different shape and have higher uptake in the liver, while the orange is somewhat in-between. The liver transit time may thus reflect differences in CYP profiles with regard to metabolic rate or route.
Excretion: The two organs with the most pronounced difference in uptake between the two labeling positions were the urinary bladder and the gallbladder, pointing to an unambiguous difference in excretory pathways of the 11C-labeled metabolites of the two tracers. For the gallbladder, this difference is not reflected in the dosimetry results, as the onset of the difference happens after three half-lives of the radioactive label. It is important to keep in mind that, unlike most of the organs in the human body, the urinary bladder and the gallbladder do not have fixed volumes but expand and contract on physiological demand, and thus the concentrations depend on two independent variables. Emptying by the gallbladder into the small intestines might also contribute to the high uptake in the proximal small intestines, but then again, these SUVs are relatively stable during the course of the scan. For 11C-Cimbi-36, the urinary bladder SUV curves (Fig. 2) vary immensely between the subjects as was the case for the absolute amount of excreted substance (Fig. 3d). As the cold doses in our study (ranging 0.25–0.61 μg) are well below pharmacologically active doses of NBOMes (usually 0.5–1 mg ), we cannot exclude the possibility that metabolic route and excretory rate differ in these settings.
To further explore the biodistribution of 5-HT2A/2C receptors and metabolism, whole-body scans after pre-treatment with blocking agents such as Ketanserin (5-HT2A receptor antagonist) or CYP inhibitors could be performed, and CYP isoform profiling of participants could be correlated.