Quantification of μ-opioid receptors with [11C]carfentanil PET
Carfentanil (4-carbomethoxy-fentanyl, R-31833, Wildnil) is an agonist of the opioid receptor. Carfentanil is thousands of times more potent analgesic than morphine.
[11C]carfentanil ([11C]CFN, [11C]CAF) is a selective agonist for μ1 opioid receptors, with low binding to μ2 receptors (Eriksson and Antoni, 2015), showing excellent test-retest reliability (Hirvonen et al., 2009), and possesses favourable dosimetric properties (Newberg et al., 2009). Due to the high specific activity and small masses that are used in human PET studies [11C]carfentanil does not cause marked pharmacological effects (Scott et al., 2007; Wand et al., 2011).
[11C]carfentanil binding is affected by polymorphism of the gene coding μOR, with carriers of the Asn40Asp (A118G) variant having lower cerebral binding potential (Weerts et al., 2013). Also age and gender have some effect on the cerebral distribution of the binding. Both μOR density and concentration of the endogenous ligands have an effect on cerebral [11C]carfentanil binding.
Endogenous opioid peptide (EOP) release in the brain may be detected as reduction of [11C]carfentanil binding potential (Zubieta et al., 2001), as demonstrated with amphetamine administration (Colasanti et al., 2012; Mick et al., 2014), although this could not be replicated in the study by Guterstam et al. (2013). Reduced μOR availability could be at least partially caused by agonist induced receptor internalization (Quelch et al., 2014). EOP has been demonstrated using [11C]carfentanil PET for example after high-intensity interval training (Saanijoki et al., 2017).
Published analysis methods
In brain PET studies the occipital cortex, especially lateral prt, can be considered as a reference region, with only negligible μOR density (Frost et al. 1989; Hirvonen et al., 2009). Cerebellum is suitable reference region in rodents (Quelch et al., 2014), but not in human studies.
In brain studies with arterial input function, four-compartmental model was found to best fit the regional data, with (Endres et al., 2003) or without (Frost et al., 1988 and 1989) blood volume correction; three-tissue model (parameters K1, k2, k5, and k6) was first fitted to the data from occipital cortex; k5 and k6 were then used as fixed parameters when fitting the data from other regions to estimate Bmax/KD as k3/k3 (Frost et al., 1988 and 1989). Parameters k5 and k6 in occipital cortex were not found to change with increasing plasma metabolite fractions, suggesting that requirement for the second tissue compartment in occipital cortex is not due to uptake of label-carrying metabolites.
Tissue ratio method (SUVR) has been shown to provide binding estimates that are highly correlated with specific binding estimates obtained using blood-input (Frost et al., 1988 and 1989; Endres et al., 2003); simulation suggested that SUVR is also almost independent on changes in blood flow or BBB permeability (Frost et al., 1988 and 1989; Zubieta et al., 1999). For SPM analysis, “μ-opioid receptor binding” images have been calculated by the same group with the ratio method: time frames 34-82 min have been summed and the sum image is divided by the value in occipital cortex, and 1 is subtracted from the ratio image (Bencherif et al. 2004a and 2004b).
Logan plot analysis with occipital cortex input has also been validated as an analysis method (Endres et al. 2003). However, choosing the start time for linear fit was found to be problematic. A later time value (30 min) gave less bias than an earlier time (10 min), but increased the noise level (Endres et al. 2003). This may be acceptable for regional analysis, but not for calculation of parametric images. Problems with Logan plot with occipital cortex input can be partly avoided by correcting for the washout rate from the occipital cortex (k2’); Heinz et al. (2005), Weerts et al. (2008), and Wand et al. (2011) used this method, assuming k2‘=0.1 (or 0.104) min-1, based on the study by Endres et al. (2003). Hirvonen et al. (2009) estimated similar value (0.1237 min-1) for k2’. Measurement noise introduces additional bias in reference region input Logan plots, which can be reduced for example by PCA (Joshi et al., 2008).
A bolus + infusion protocol has been frequently used in [11C]carfentanil studies to achieve steady-state tracer levels. It is possible that this approach solves the nonlinearity problem with reference input Logan plot analysis. Bmax/Kd (DVR-1) images were produced with this technique and used in SPM analysis (Zubieta et al. 2001; Liberzon et al. 2002; Burghardt et al., 2015, Hiura et al., 2017).
Logan plot analysis with plasma input has been used in regional analysis of myocardium, skeletal muscle and lung (Villemagne et al., 2002). Skeletal muscle was used as a reference tissue to achieve an estimate of binding potential as
Usage of simplified reference tissue model (SRTM) is an alternative analysis method that provides BPND estimates without the non-linearity issue of Logan plot. Investigation of the performance of SRTM in [11C]carfentanil bolus studies was already proposed by Endres et al. (2003), and it has since then been used successfully in for example in μOR occupancy studies (Rabiner et al., 2011). Based on simulations, Endres et al. (2003) concluded that the effect of blood flow changes on BPND from SRTM analysis would be minimal, even lower than when using SUVR or Logan plot. Modification of the SRTM method with fixed reference tissue k2 (k2’) may work even better in basal conditions, but if μOR occupancy changes during the PET scan, an additional time-dependent term is needed (Johansson et al., 2019).
Partial volume correction increases μOR estimates especially in atrophic brain (Meltzer et al., 1990). It can also reduce the variance and reveals an age-dependent increase in binding (Bencherif et al. 2004b).
Suggested analysis method for Turku
BPND images are calculated with SRTM using occipital cortex as reference tissue (Hirvonen et al., 2009; Hagelberg et al., 2012; Tuominen et al., 2012; Karlsson et al., 2015; Nummenmaa et al., 2015; Tuominen et al., 2015; Karjalainen et al., 2016; Karlsson et al., 2016; Nummenmaa et al., 2016; Lamusuo et al., 2017 Majuri et al., 2017 and 2018,). Use software that applies basis function method in calculation of BPND maps, either PMOD, or imgbfbp (Ingman et al., 2005).
Plasma metabolite correction
Carfentanil is mainly metabolized by CYP enzymes in the liver. N-O-dealkylation is the major metabolism pathway of fentanyl derivatives. Metabolites do not have affinity to opioid receptors. Most of the radioactivity is cleared away through kidneys and urinary system as label-carrying metabolites. Frost et al. (1989) report that no significant amounts of volatile metabolites from demethylation ([11C]methanol, [11C]formaldehyde, or [11C]CO2) could be found in the blood. Major metabolite is more polar than [11C]carfentanil, but minor metabolite is more lipophilic (Endres et al., 2003).
Either Hill-type function
(Endres et al., 2003) or
Power function can be fitted to the parent tracer
fractions of individual study subjects; both functions provide essentially the same results.
Initial parent fraction should be constrained to 1.0 by using option
Figure 1. Hill-type function fitted to the parent fractions of all PET studies in the test-retest study (Hirvonen et al., 2009).
Carfentanil is one of the fentanyl analogues. Fentanyl and alfentanil diffuse freely into red blood cells (RBCs). Plasma protein binding is generally high for all opiates. Fentanyl, but not alfentanil, is bound to erythrocyte proteins, leading to RBC/plasma partition coefficients of 1.01 and 0.14, respectively (Bower and Hull, 1982). In rat studies with [3H]cyclofoxy the blood-to-plasma ratio ranged from 0.8 to 1.2 after bolus injection, and was 1.30±0.08 in in vitro (Sawada et al., 1991); metabolite corrected blood TAC was used as input function.
Figure 2. “Hill type” functions fitted to plasma-to-blood and RBC-to-plasma ratio data from eight [11C]carfentanil studies (unpublished data from TPC).
[11C]carfentanil and its label-carrying metabolite(s) seem to be instantly equilibrated between plasma and RBCs, with the unchanged tracer showing higher affinity to plasma proteins. The ratios from individual subjects are highly variable, even when blood hematocrit is accounted for (RBC-to-plasma ratio); therefore input sampling methods that would require blood-to-plasma conversion, such as automatic blood sampling system or image-derived input are not recommended.
Arterial input function
Compartmental model parameters
Simulation of the brain tissue curves in regions with specific binding can be based on the parameter values published by Endres et al. (2003): K1=0.166 mL/(min*mL), k3=0.242 min-1, k4=0.115 min-1, Vb=0.04 mL/mL, and K1/k2=1.59 mL/mL, when the 2-tissue compartment model fitting results from the reference tissue (occipital cortex) are converted to one-tissue compartment model parameters. Reference tissue curves were simulated using one-tissue compartment model (although two-tissue compartment model was used in fitting), assuming the same K1, Vb, and K1/k2 as for the specific binding regions (Endres et al., 2003). The effect of perfusion on binding parameters was simulated by varying K1 in the range 0.05-0.50 mL/(min*mL). Different binding potentials were simulated by giving k3 values 0.15, 0.25, and 0.35 min-1 (Endres et al., 2003).
Alternatively, we could simulate the tissue data using three-tissue compartment model. Frost et al. (1989) reported two-tissue compartment model fitting results for the occipital cortex: K1=0.094 mL/(min*mL), k2=0.173 min-1, k5=0.029 min-1, and k6=0.036 min-1. Vb was assumed to be 0, possibly leading to overestimation of K1. K1/k2 and k5/k6 from 5 subjects can be calculated to be on average 0.55±0.27 mL/mL and 0.82±0.12, respectively. In frontal cortex and thalamus, when k5 and k6 were fixed to the individual values retrieved from occipital cortex, the estimates of K1/k2 were similar than in occipital cortex. Estimates of k3 were 0.382±0.240 and 0.201±0.070 min-1, and k3/k4 were 1.78±0.39 and 3.38±0.93, respectively (Frost et al., 1989).
Endres CJ, Bencherif B, Hilton J, Madar I, Frost JJ. Quantification of brain μ-opioid receptors with [11C]carfentanil: reference-tissue methods. Nucl Med Biol. 2003; 30: 177-186. doi: 10.1016/S0969-8051(02)00411-0.
Frost JJ, Wagner HN Jr, Dannals RF, Ravert HT, Links JM, Wilson AA, Burns HD, Wong DF, McPherson RW, Rosenbaum AE, Kuhar MJ, Snyder SH. Imaging opiate receptors in the human brain by positron emission tomography. J Comput Assist Tomogr. 1985; 9(2): 231-236. PMID: 2982931.
Frost JJ, Mayberg HS, Fisher RS, Douglass KH, Dannals RF, Links JM, Wilson AA, Ravert HT, Rosenbaum AE, Snyder SH, Wagner HN Jr. Mu-opiate receptors measured by positron emission tomography are increased in temporal lobe epilepsy. Ann Neurol. 1988; 23: 231-237. doi: 10.1002/ana.410230304.
Frost JJ, Douglass KH, Mayberg HS, Dannals RF, Links JM, Wilson AA, Ravert HT, Crozier WC, Wagner HN Jr. Multicompartmental analysis of [11C]-carfentanil binding to opiate receptors in humans measured by positron emission tomography. J Cereb Blood Flow Metab. 1989; 9: 398-409. doi: 10.1038/jcbfm.1989.59.
Hirvonen J, Aalto S, Hagelberg N, Maksimow A, Ingman K, Oikonen V, Virkkala J, Någren K, Scheinin H. Measurement of central μ-opioid receptor binding in vivo with PET and [11C]carfentanil: a test-retest study in healthy subjects. Eur J Nucl Med Mol Imaging 2009; 36: 275-286. doi: 10.1007/s00259-008-0935-6.
Ingman K, Hagelberg N, Aalto S, Någren K, Juhakoski A, Karhuvaara S, Kallio A, Oikonen V, Hietala J, Scheinin H. Prolonged central μ-opioid receptor occupancy after single and repeated nalmefene dosing. Neuropsychopharmacology 2005; 30(12): 2245-2253. doi: 10.1038/sj.npp.1300790.
National Center for Biotechnology Information. PubChem Compound Database; CID=62156.
Sawada Y, Kawai R, McManaway M, Otsuki H, Rice KC, Patlak CS, Blasberg RG. Kinetic analysis of transport and opioid receptor binding of [3H](-)-cyclofoxy in rat brain in vivo: implications for human studies. J Cereb Blood Flow Metab. 1991; 11: 183-203. doi: 10.1038/jcbfm.1991.51.
Tuominen L, Nummenmaa L, Keltikangas-Järvinen L, Raitakari O, Hietala J. Mapping neurotransmitter networks with PET: an example on serotonin and opioid systems. Hum Brain Mapp. 2014; 35(5): 1875-1884. doi: 10.1002/hbm.22298.
Villemagne PSR, Dannals RF, Ravert HT, Frost JJ. PET imaging of human cardiac opioid receptors. Eur J Nucl Med. 2002; 29: 1385-1388. doi: 10.1007/s00259-002-0897-z.
Zubieta J-K, Dannals RF, Frost JJ. Gender and age influences on human brain mu-opioid receptor binding measured by PET. Am J Psychiatry 1999; 156: 842-848. doi: 10.1176/ajp.156.6.842.
Zubieta J-K, Smith YR, Bueller JA, Xu Y, Kilbourn MR, Jewett DM, Meyer CR, Koeppe RA, Stohler CS. Regional mu opioid receptor regulation of sensory and affective dimensions of pain. Science 2001; 293:311-315. doi: 10.1126/science.1060952.
Updated at: 2019-07-22
Created at: 2007-04-17
Written by: Vesa Oikonen