Analysis of [18F]FMPEP-d2 PET data

[18F]FMPEP-d2 is an inverse agonist radioligand for CB1 receptors of the endocannabinoid system. The tracer has high affinity and selectivity for the CB1R, and since it also can pass the blood-brain barrier, and is not a substrate for P-gp, it is suitable for quantifying the CB1 receptors in the human brain (Donohue et al., 2008; Terry et al., 2010b). Longitudinal study in mouse model of AD suggests that changes in [18F]FMPEP-d2 binding represents the level of endocannabinoids, as the CB1R expression was not changed (Takkinen et al., 2018).

A brain region devoid of CB1 receptors does not exist, or is not identifiable with PET scanners, preventing the application of reference region input methods and calculation of receptor binding potentials. Instead, receptor density must be measured as the distribution volume, VT, using metabolite-corrected arterial plasma curve as input function. The specific binding is about 80-90% of the VT, based on receptor blocking studies in monkeys (Terry et al., 2010b).

In a dosimetry study (Terry et al., 2010a), the liver had the highest [18F]FMPEP-d2 uptake, followed by the lungs. In addition, the brain, heart, gallbladder, spleen, kidneys, lumbar vertebrae, intestine, and urinary bladder could be seen in the uptake images. Radioactivity is excreted through both bile and urine (Terry et al., 2010a), and thus [18F]FMPEP-d2 cannot be used to precisely quantitate CB1Rs in the liver, intestinal tract, or urinary bladder.

CB1 receptors were found in rats in brown adipose tissue (BAT) but not in white adipose tissue (WAT), suggesting that [18F]FMPEP-d2 could be used to locate BAT in PET images (Eriksson et al., 2015). In human subjects, [18F]FMPEP-2 imaging has shown upregulation of CB1Rs in acute activation of BAT (Lahesmaa et al., 2018).


Plasma input function

[18F]FMPEP-d2 has several 18F-carrying metabolites in plasma, and the parent tracer constitutes 11% of the total plasma radioactivity at 60 min after injection (Terry et al., 2010b). The two deuterium atoms in the ligand reduce defluorination, but marked uptake of [18F]F- in the bone is still visible in late PET scans. In mice, two radiometabolites are found in the plasma, both being polar (Takkinen et al., 2018).

The free fraction in plasma (fp) is about 0.6% (Terry et al., 2010b).

In rats, both the measured plasma-to-blood ratios and parent tracer fractions in plasma can be fitted using monoexponential function (Eriksson et al., 2015). Parent tracer does not seem to pass red blood cell membranes (probably due to high plasma protein binding), but one or more of the radioactive metabolites readily do, leading to a stable plasma-to-blood ratio curve about 30 min after injection (unpublished finding in human subjects). In human studies the blood-to-plasma conversion and metabolite correction is more complex, probably because several 18F-carrying metabolites are present (Terry et al., 2010b).

Population-based input function, scaled individually with two arterial samples, was found to provide reliable VT values in healthy controls and in an alcoholism study, but erroneous result in a study with cannabis smokers (Zanotti-Fregonara et al., 2013). Notice that gender differences in plasma curves and metabolite fractions have been observed with [11C]OMAR, another CB1R ligand (Normandin et al., 2015).

Arterial blood curve can be derived from PET image, if a large arterial blood pool such as LV cavity or aortic arch is visible in the image. This is not possible in brain studies because of the limited field-of-view, and blood sampling is still required for the metabolite analysis. Arterialized venous blood sampling can produce comparable results to image-derived blood data (unpublished results), providing that arterialization has succeeded and venous samples can be taken at sufficient rate.

Recommended procedure for human studies in TPC:

1. Arterial blood TAC, either image-derived or collected with ABSS, is converted to plasma TAC using program b2plasma, for example with command:

b2plasma FMPEP-D2 z123_lvcavity.tac 0.42 z123_lvcavity_ap.tac

The plasma-to-blood conversion function is based on unpublished human data from TPC.

2. Plasma TAC from manual blood sampling is combined with the previously made plasma TAC using program dftcat.

If ABSS was used, then data from manual samples are usually preferred when they overlap with ABSS data, for example with command:

dftcat -second z123_bloodpump_ap.tac z123ap.kbq z123ap_combined.tac

But if initial phase of blood data is based on LV cavity ROI, then that should be preferred over manual samples, especially in case of arterialized venous sampling; myocardial uptake of [18F]FMPEP-d2 and its metabolites is low, and no spill-over or recovery correction for LV cavity TAC is necessary. To add data from manual sampling only after the end of LV cavity data, use command:

dftcat -first z123_lvcavity_ap.tac z123ap.kbq z123ap_combined.tac

3. For metabolite correction, the parent fractions in plasma are saved in a file, and then a power function is fitted to the fractions using program fit_ppf, for example with command

fit_ppf -model=pf -d=1 -b=1.5 -e=0.5 -svg=z123apratfit.svg z123ap.rat

to fit this power function with two free parameters:

, where 0 < a ≤ 1, and c > 0.

Both power function and Hill function fit the parent fraction data well, but power function seems to provide slightly better extrapolation of the parent fractions.

The combined plasma TAC can then be corrected for metabolites using program metabcor, for example with command:

metabcor -fnpure=z123apc.tac z123ap_combined.tac

, giving metabolite corrected plasma TAC in file z123apc.tac in the example.

The metabolite corrected plasma TAC can be used as input function in the quantitation of tissue uptake. Delay time correction between the input function and tissue TACs may still be necessary.

CB1R binding


In humans, two-tissue compartmental model has been fitted to 120-min PET data with arterial metabolite-corrected plasma input (Terry et al., 2010b; Hirvonen et al., 2012 and 2013). VT is then calculated from the rate constants of the compartmental model. One-tissue compartmental model does not fit the data adequately (Terry et al., 2010b). In the two-tissue compartmental model, constraining the K1/k2 to a value determined from all brain regions except cerebellum did not fit the data as well as the unconstrained model (Terry et al., 2010b). Optimal scanning time was determined to be 120 min, after which VT started to increase due to the accumulation of tracer metabolites in the brain (Terry et al., 2010b). In mice, one polar radiometabolite is found in the brain cortex, accounting for 14% of the activity 60-180 min after administration (Takkinen et al., 2018).

VT of [18F]FMPEP-d2 in the brain correlates negatively with BMI, and therefore individual VT values may need to be adjusted for the BMI (Hirvonen et al., 2012). VT is markedly higher (∼41%) in men than in women; difference is highest in the posterior limbic cortex (Laurikainen et al., 2019).

Other organs

In rats, one-tissue compartmental model fitted regional data from WAT and BAT reasonably well. VT was two times higher in BAT than in WAT (Eriksson et al., 2015). Human BAT and WAT can also be analyzed using one-tissue compartment model, with K1, k2, and VB as fitted parameters, and VT calculated as K1/k2; FUR can be used to analyze late-scans in these regions with relatively low blood flow (Lahesmaa et al., 2018).

At late time points, most of the radioactivity in plasma and blood is due to [18F]F-, and it is also present in the tissues. In soft tissues the [18F]F- concentration is usually lower than in plasma, but nevertheless, substantial part of radioactivity concentration in peripheral tissues is expected to be derived by [18F]F- and other radioactive metabolites instead of intact receptor-bound tracer.

See also:


Donohue SR, Krushinski JH, Pike VW, Chernet E, Phebus L, Chesterfield AK, Felder CC, Halldin C, Schaus JM. Synthesis, ex vivo evaluation, and radiolabeling of potent 1,5-diphenylpyrrolidin-2-one cannabinoid subtype-1 receptor ligands as candidates for in vivo imaging. J Med Chem. 2008; 51(18): 5833-5842.

Eriksson O, Mikkola K, Espes D, Tuominen L, Virtanen K, Forsbäck S, Haaparanta-Solin M, Hietala J, Solin O, Nuutila P. The cannabinoid receptor-1 is an imaging biomarker of brown adipose tissue. J Nucl Med. 2015; 56: 1937-1941.

Hirvonen J, Terry GE, Halldin C, Pike VW, Innis RB. Approaches to quantify radioligands that wash out slowly from target organs. Eur J Nucl Med Mol Imaging 2010; 37(5): 917-919.

Hirvonen J, Goodwin RS, Li CT, Terry GE, Zoghbi SS, Morse C, Pike VW, Volkow ND, Huestis MA, Innis RB. Reversible and regionally selective downregulation of brain cannabinoid CB1 receptors in chronic daily cannabis smokers. Mol Psychiatry 2012; 17(6): 642-649. doi: 10.1038/mp.2011.82.

Hirvonen J, Zanotti-Fregonara P, Umhau JC, George DT, Rallis-Frutos D, Lyoo CH, Li CT, Hines CS, Sun H, Terry GE, Morse C, Zoghbi SS, Pike VW, Innis RB, Heilig M. Reduced cannabinoid CB1 receptor binding in alcohol dependence measured with positron emission tomography. Mol Psychiatry 2013; 18(8): 916-921. doi: 10.1038/mp.2012.100.

Hirvonen J. In vivo imaging of the cannabinoid CB1 receptor with positron emission tomography. Clin Pharmacol Ther. 2015; 97(6): 565-567. doi: 10.1002/cpt.116.

Horti AG, Raymont V, Terry GE. PET imaging of endocannabinoid system. In: Dierckx RAJO et al. (eds.) PET and SPECT of Neurobiological Systems. Springer, 2014.

Lahesmaa M, Eriksson O, Gnad T, Oikonen V, Bucci M, Hirvonen J, Koskensalo K, Teuho J, Niemi T, Taittonen M, Lahdenpohja S, U Din M, Haaparanta-Solin M, Pfeifer A, Virtanen KA, Nuutila P. Cannabinoid type 1 receptors are upregulated during acute activation of brown adipose tissue. Diabetes 2018; 67(7): 1226-1236. doi: 10.2337/db17-1366.

Leung K, Donohue S. Molecular Imaging and Contrast Agent Database (MICAD).

Normandin MD, Zheng M-Q, Lin K-S, Mason NS, Lin S-F, Ropchan J, Labaree D, Henry S, Williams WA, Carson RE, Neumeister A, Huang Y. Imaging the cannabinoid CB1 receptor in humans with [11C]OMAR: assessment of kinetic analysis methods, test-retest reproducibility, and gender differences. J Cereb Blood Flow Metab. 2015; 35: 1313-1322.

Suter TM, Chesterfield AK, Bao C, Schaus JM, Krushinski JH, Statnick MA, Felder CC. Pharmacological characterization of the cannabinoid CB1 receptor PET ligand ortholog, [3H]MePPEP. Eur J Pharmacol. 2010; 649(1-3): 44-50.

Terry G. In vivo imaging of the cannabinoid CB1 receptor using positron emission tomography. Ph.D. thesis, Karolinska Institutet, Stockholm, 2009.

Terry GE, Hirvonen J, Liow JS, Seneca N, Tauscher JT, Schaus JM, Phebus L, Felder CC, Morse CL, Pike VW, Halldin C, Innis RB. Biodistribution and dosimetry in humans of two inverse agonists to image cannabinoid CB1 receptors using positron emission tomography. Eur J Nucl Med Mol Imaging 2010a; 37(8): 1499-1506.

Terry GE, Hirvonen J, Liow JS, Zoghbi SS, Gladding R, Tauscher JT, Schaus JM, Phebus L, Felder CC, Morse CL, Donohue SR, Pike VW, Halldin C, Innis RB. Imaging and quantitation of cannabinoid CB1 receptors in human and monkey brains using 18F-labeled inverse agonist radioligands. J Nucl Med. 2010b; 51(1): 112-120.

Zanotti-Fregonara P, Hirvonen J, Lyoo CH, Zoghbi SS, Rallis-Frutos D, Huestis MA, Morse C, Pike VW, Innis RB. Population-based input function modeling for [18F]FMPEP-d2, an inverse agonist radioligand for cannabinoid CB1 receptors: validation in clinical studies. PLoS One 2013; 8(4): e60231.

Tags: ,

Created at: 2015-09-23
Updated at: 2018-10-31
Written by: Vesa Oikonen