529

NEUROIMAGING CLINICS OF NORTH AMERICA

Neuroimag Clin N Am 16 (2006) 529–552 Opioid Imaging

Alexander Hammers, PhDa,b,c,*, Anne Lingford-Hughes, PhDd,e

- Derivation, release, peptide action, Changes in receptor availability in pain and metabolism and discomfort: between-group - Receptors and ligands comparisons Receptors Direct intrasubject comparisons of periods Species differences with pain and pain-free states Regional and layer-specific subtype - Opioid imaging in epilepsy distributions Focal epilepsies Ligands Idiopathic generalized epilepsy - Positron emission tomography imaging Summary of opioid receptors - Opioid imaging in other specialties Introduction PET imaging Movement disorders Quantification of images Dementia Available ligands and their quantification Cardiology - Opioid receptor imaging in healthy - Opioid imaging in psychiatry: addiction volunteers - Summary - Opioid imaging in pain-related studies - Acknowledgments - References

Opioids derive their name from the Greek o´pioy agonists provided evidence for the existence of mul- for poppy sap. Various preparations of the opium tiple receptors [4]. In the early 1980s, there was ev- poppy Papaver somniferum have been used for pain idence for the existence of at least three types of relief for centuries. Structure and stereochemistry opiate receptors: m, k, and d [5,6]. A fourth ‘‘orphan’’ are essential for the analgesic actions of morphine receptor (ORL1 or NOP1) displays a high degree and other opiates, leading to the hypothesis of the of structural homology with conventional opioid existence of specific receptors. Receptors were iden- receptors and was identified through homology tified simultaneously by three laboratories in 1973 with the d receptor [7], but the endogenous ligand, [1–3]. The different pharmacologic activity of orphanin FQ/nociceptin, does not interact directly

Dr. Hammers was funded by an MRC Clinician Scientist Fellowship (G108/585). a Department of Clinical Neuroscience, Division of Neuroscience and Mental Health, Imperial College London, Hammersmith Hospital, DuCane Rd., London W12 0NN, UK b Epilepsy Group, MRC Clinical Sciences Centre, Room 243, Cyclotron Building, Hammersmith Hospital, DuCane Rd., London W12 0NN, UK c Department of Clinical and Experimental Epilepsy, Institute of Neurology, University College London, Queen Square, London WC1N 3BG, UK d Academic Unit of Psychiatry, University of Bristol, Cotham House, Cotham Hill, Bristol BS6 6JL, UK e Imaging Department, Division of Clinical Sciences, Faculty of Medicine, Hammersmith Hospital, Imperial College London, DuCane Rd., London W12 0NN, UK * Corresponding author. Epilepsy Group, MRC Clinical Sciences Centre, Room 243, Cyclotron Building, Hammersmith Hospital, DuCane Rd., London W12 0NN, UK. E-mail address: [email protected] (A. Hammers).

1052-5149/06/$ – see front matter ª 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.nic.2006.06.004 neuroimaging.theclinics.com 530 Hammers & Lingford-Hughes

with classical opioid receptors [8] and is not dis- leads to receptor desensitization, and long-term ex- cussed further herein. Endogenous opioids have posure to agonists leads to receptor downregulation been identified [9]. These substances were origi- [17]. Desensitization is achieved through phos- nally distinguished from exogenous opiates, but phorylation of agonist-activated receptors and sub- the term opioids is now generally used for all ligands. sequent receptor endocytosis via clathrin-coated Besides their role as endogenous and exogenous pits. The ubiquitin/proteasome pathway seems to pain killers, opioids and their receptors are impli- be important in agonist-induced downregulation cated in reward, addiction, mood regulation and as well as basal turnover of opioid receptors. Mono- mood disorders, epilepsy, movement disorders, ubiquinated receptors are simply internalized, and dementia. whereas ubiquitin chains of more than three units target the receptor to the proteasome. Further down- stream, trafficking to lysosomes may have a role. Derivation, release, peptide action, Triggered release of endogenous opioids (eg, in and metabolism seizure models) leads to major changes [18]. In lab- All opioid peptides are derived from three different oratory animals, transmitter peptide levels generally gene products: proopiomelanocortin, proenkepha- decrease to about 40% to 50% of baseline in the lin, and prodynorphin. Proopiomelanocortin gives hours after release. Levels of transmitter peptide rise to b-endorphin (m- and d- preferring) and non- mRNA tend to increase temporarily. Depending opioid peptides. Proenkephalin contains four cop- on the model, increases can be relatively minor (ap- ies of different enkephalins (most d-preferring). proximately 150% of baseline) or major (300% to Prodynorphin gives rise to several dynorphin pep- 1400%). Limited data are available on opioid recep- tides and neoendorphin (all k-preferring) (Table 1) tor level changes following events like epileptic sei- [10]. Endogenous opioids consist of between zures. Acute release leads to decreased availability of four (endomorphin) and 31 amino acids (b- receptors as measured by [3H]diprenorphine bind- endorphin). ing [13] or [3H]U69,593 binding [14]. Over slightly Similar to other neuropeptides, opioids are stored longer time courses, both decreases, particularly for in large dense core vesicles [11], which are distin- d-opioid receptors [19], and increases (of the m re- guishable electron microscopically from small clear ceptor) [20] have been found. Receptor endocytosis vesicles containing the fast-acting transmitters such may be irreversible for d receptors, requiring de as GABA and glutamate. Opioid vesicle containing novo protein synthesis [17]. neurons also contain classical fast-acting transmit- Opioid receptor function changes over the hours ters; therefore, opioids act as co-transmitters to following neuronal events have received far less at- modulate the actions of the primary transmitter. tention. After 4 hours of exposure to the m-agonist Exocytosis of the dense core vesicles is calcium de- DAMGO to induce desensitization, some func- pendent [12] but, unlike small clear vesicles, exocy- tional resensitization of the m receptor occurred tosis is not limited to restricted zones of the plasma after 10 minutes, with 100% of control responses membrane. Exocytosis requires high-frequency stim- reached after 60 minutes [21]. In another study, ulation of the opioid-containing neurons [13,14]. m-receptor protein reached normal levels and func- In contrast to amino acid transmitters, opioid tion after 6 hours through recycling [22]. Basal peptides can affect the excitability of neurons at levels can be achieved through recycling in 30 the relatively large distance of 50 to 100 mm from [23] to 60 minutes, with some agonists inducing the site of release [15] owing to their diffusion receptor levels of 110% to 120% of control as mea- through the extracellular space and their high affin- sured with [3H]diprenorphine [24]. ities for their receptors, with nanomolar concentra- tions of the peptides being effective [16]. After Receptors and ligands release from nerve terminals, opioid peptides are rapidly degraded by a variety of peptidases [9]. Receptors The time courses of changes of transmitter pep- There are three main types of opioid receptors, m, d, tides and receptor protein are of great interest and k. For all, the existence of subtypes has been when interpreting positron emission tomography proposed [10], and one subtype each has been (PET) studies of opioid function in experiments de- cloned in several species including man. The recep- signed to show the role of neurotransmitters, usu- tor proteins consist of about 370 amino acids [17]. ally by employing two PET scans acquired at They belong to the G-protein coupled receptor fam- different delays from an experimental manipula- ily and have their characteristic structure with seven tion or spontaneously occurring event. hydrophobic transmembrane domains, connected As is true for most G-protein coupled receptors, by relatively short intracellular and extracellular short-term exposure of opioid receptors to agonists loops. The amino acid sequences are about 60% Opioid Imaging 531

Table 1: Opioid receptors and their ligands Receptor m (MOR) d (DOR) k (KOR) Endogenous b-endorphin > b-endorphin, Dynorphin A >> ligand potency dynorphin A > leu-enkephalin, b-endorphin > met-enkephalin, met-enkephalin > leu-enkephalin > leu-enkephalin dynorphin A met-enkephalin Agonists DAMGO DPDPE Enadoline (CI-977) DAGO DSLET U69,593 Dihydromorphine DADLE (d, m) Selective CTAP Naltrindole nor-BNI antagonists Nonselective Naloxone antagonists Naltrexone (orally active) PET ligands

Abbreviations: DAMGO, [D-Ala2,N-Me-Phe4,Gly-ol5]-enkephalin; DAGO, D-Ala2-MePhe4-Gly(ol)5 enkephalin: DPDPE, [D-Pen2,5]-enkephalin; DSLET, [D-Ser2, Leu5] enkephalin-Thr6; DADLE, D-Ala2-D-Leu5 enkephalin; CTAP, D-Phe-Cys- Tyr-D-Trp-Arg-Thr-Pen-Thr-NH2; nor-BNI, nor-binaltorphimine. 532 Hammers & Lingford-Hughes

identical between opioid receptor types and about (except medial occipital), as well as some midbrain 90% identical between the same receptor types and deeper nuclei [31]. m-opioid receptor binding is cloned from different species. There are marked present in all of these areas but is most dense in differences in the degree of sequence conservation layers I, IIIa, and IV [26,28,29], suggesting a somato- between domains. Most transmembrane segments dentritic expression. Functional coupling of m re- and the three intracellular loops are highly con- ceptors to intracellular signal transduction served [17], the latter probably because they medi- mechanisms, measured as DAMGO-stimulated ate the G-protein coupling. Extracellular loops are [35S]GTPgS-binding, was around 200% above basal more variable and are responsible for receptor se- levels and evenly distributed throughout human lectivity and affinity. The determinants of the opi- frontal gray matter [32,33]. In comparison, oid receptor binding pocket differ from ligand to mRNA, receptor protein, and functional coupling ligand for a given receptor. This difference means are all much lower in the human hippocampus, that there is not a single receptor binding site, as whereas m receptors contribute considerably to the is true for ionotropic receptors; rather, opioid recep- high density of opioid receptors in the amygdala. tors are capable of considerable plasticity regarding Delta receptors are generally less prominent in engagement of ligands and subsequent events lead- the human brain than in rodent brains. They are ing to receptor activation. most numerous in the striatum and throughout the neocortex, and occur notably in medial occipi- Species differences tal cortex [25,29]. d-opioid receptor binding is pres- There are well-known species differences. In gen- ent in all layers and shows peak densities in layers I eral, there is relatively less d binding in the human to IIIa [26,29]. Functional coupling of d receptors, compared with the rat brain and relatively more assessed as DPDPE-stimulated [35S]GTPgS-bind- k binding [25]. The latter finding has been corrobo- ing, was evenly distributed throughout human rated by quantitative autoradiography [26,27] and frontal gray matter [32]. There is comparatively lit- by the detection of a more widespread k-opioid re- tle d-opioid receptor binding in hippocampus and ceptor mRNA expression in humans compared with amygdala, very little in the thalamus (approxi- rodents [28]. Other examples include k-opioid re- mately 9% of total [25]), and none in the cerebel- ceptors found in deep cortical layers in human lum, making the cerebellum a possible reference but not in rat brain and the lack in humans of the region for d-opioid receptor ligands. typical patchy distribution of m receptors seen in ro- Kappa receptors are most numerous in the neo- dents [29]. It is prudent to be careful in extrapolat- cortex, amygdala, and hippocampus and notably ing results from rodent studies to humans. sparse in the striatum [29]. k-opioid receptor bind- ing is present in all layers but concentrated in layers Regional and layer-specific subtype Vand VI, matching the distribution of mRNA expres- distributions sion [26,29]. Enadoline-stimulated [35S]GTPgS- The receptor subtypes are unevenly distributed be- binding, indicating functional coupling of k tween regions. Outside the medial occipital lobe, receptors to intracellular signal transduction mecha- human neocortical binding consists of 40% to nisms, was present throughout human frontal gray 50% k, 20% to 40% d, and 15% to 40% m matter but twice as strong in lamina V-VI when com- [25,29]. In the thalamus, m binding predominates pared with lamina I-IV [32]. k-opioid receptor bind- at approximately two-thirds, whereas in mesial tem- ing in the human hippocampus is higher than m-or poral structures, approximately 60% of binding is d-receptor binding [25] and restricted to the pyrami- attributed to k receptors. Absolute levels are about dal cell layer [26]. About 60% of the high level of 120 to 145 fmol receptors/mg protein in most of opioid receptor binding in the amygdala represents the neocortex, compared with about 72 to 74 k-opioid receptor binding [25,29]. In contrast to fmol receptors/mg protein in the lateral and medial the rat, mRNA and k-binding sites are present in occipitotemporal gyri [26]. the human cerebellum [30]. Importantly for PET studies, there is evidence of At least for the m receptor, in some regions a sub- m-receptor binding and, at a lower level, k-receptor stantial influence of age and sex on the binding po- binding in the human cerebellum [30], indicating tential of the m-selective ligand [11 C]carfentanil as that, in contrast to many other transmitter systems, measured by PET has been found [34] in a total it cannot be used as a classical reference region for of 36 men and 30 women spanning several decades m-ork-receptor PET tracers; however, the medial and scanned in two different scanners. occipital lobe is a potential reference region for m-receptor PET tracers. Ligands Mu receptors are found abundantly in the human Selective agonists and antagonists are now available thalamus, amygdala, striatum, and neocortex (Table 1). Selectivity is never complete, but the Opioid Imaging 533 affinity for the subtype or subtypes indicated is usu- A measure of the brain availability of the parent ra- ally more than an order of magnitude higher than diotracer is desirable. This measurement can be for the other subtypes. PET ligands are discussed achieved with arterial blood sampling in which to- in detail in the following sections. tal radioactivity and radiolabeled metabolites in plasma are assayed in blood samples throughout Positron emission tomography imaging the scanning period. These data can be used to of opioid receptors produce an ‘‘input function’’ for the available unmetabolized ligand in arterial plasma. Mathe- Introduction PET imaging matical modeling techniques are then used to de- PET enables tomographic imaging of local concen- rive parameters of interest from the data. These trations of injected biologically active substances models make the assumption that the time-activity that have been radioactively labeled (radioligands curve in each region or voxel of interest represents or tracers) [35–37]. Subjects can be investigated in the convolution of the input function and the tissue the resting state or in relationship to an event response function. Compartmental models are typ- such as the injection of a substance, the induction ically used, usually with two tissue compartments, of pain, or the occurrence of a seizure. Positron- of which one represents free and nonspecific bind- emitting isotopes used for opioid receptor imaging ing and the other specific binding. Outcome param- have half-lives of approximately 20 minutes (11 C) eters are the volume-of-distribution (VD) or and 110 minutes (18F); therefore, the use of PET is sometimes individual rate constants of radioligand limited by the need for an on-site cyclotron to pro- exchange between different compartments. When duce 11 C, whereas 18F-labeled substances can be a brain region devoid or nearly devoid of specific transported off site. Only minimal or ‘‘tracer doses’’ binding is available, for example, in the medial oc- of neuroreceptor radiotracers (in the microgram cipital lobe for m receptors and the cerebellum for range) need to be injected to estimate receptor pa- d receptors, an image-based input function can be rameters; importantly for potent substances like derived by measuring the tissue kinetics in this opioid agonists, such small doses are usually with- ‘‘reference region’’ [38]. A simpler method is to use out clinical effect. Radiotracers for receptor imaging late summed images, subtract a reference region’s need to have certain physical properties (ie, a suit- integrated radioactivity concentration from that able lipophilicity [measured as logP octanol-water] of the target region, and divide by the reference for penetrating the blood brain barrier), and bio- region’s integrated radioactivity concentration (ratio logic properties (ie, they should not be substrates images). Radioligand properties, the radiation for multidrug transporters). The radiation dose de- dose to patients and particularly healthy volunteers, pends on the tracer used and the activity injected accuracy, bias, reproducibility, reliability, and the but is typically in the range of 2.5 to 5 mSv using sensitivity to clinically meaningful changes should contemporary tomographs and three-dimensional always be considered before choosing a particular scanning, which should always be used for brain method of analysis. In addition, in more clinically imaging. The raw data collected by the PET scanner orientated settings, it may not be possible to imple- is mathematically reconstructed to produce tomo- ment more complex acquisition techniques, such as graphic images of tissue radioactivity concentra- those involving arterial cannulation and metabolite tion. Requirements for absolute quantification are analysis. Simpler methods may often be less sensi- normalization for the current efficiency of detector tive to changes and require the scanning of larger blocks, correction for nonuniform efficiency along numbers of subjects, with additional cost and radi- the longitudinal (z) axis, and scatter and attenua- ation protection implications [39–41]. tion correction. Spectral analysis [42] is an alternative modeling technique to compartmental modeling that has Quantification of images the advantage of not requiring a predefined number The distribution of the radioligand changes over of compartments to produce estimates of VD and time, and scanning is subdivided into time frames. has been widely used in the analysis of opioid re- Shortly after injection, radioactivity distribution re- ceptor PET data. Spectral analysis convolves a multi- flects blood flow and radioligand delivery. Later, ra- exponential function with the parent tracer arterial dioactivity levels and their changes will gradually plasma input function to generate an optimum fit reflect the ligand-receptor interaction more than for the observed tissue data and can often be used blood flow and delivery. at the voxel level for the generation of parametric Radiotracer availability in the brain depends on maps. The number and components of the exponen- the blood concentration of the parent radioligand, tials represent the spectral contributions of the im- its partitioning between blood cells and plasma, pulse function. The integral of the impulse its binding to plasma proteins, and its metabolism. response function (IRF), extrapolated to infinity, 534 Hammers & Lingford-Hughes

corresponds to the total VD of the tracer; however, receptor subtypes [58] and does not have a reference for tracers with slow kinetics, VD estimates may region because binding in all putative reference re- be noisy owing to poor extrapolation of the late gions can be blocked with naloxone [59]. It is a par- time courses. In this situation, suitable restrictions tial agonist at d and k receptors and an antagonist at of the expected range of slow exponential decays the m receptor [60]. It has been labeled with 11 C can be used but will lead to underestimation of [61,62] and 18F [63], with the 18F derivative having high binding regions. Alternatives or extensions similar properties [63]. Ratio images derived from currently under study include derivations of spec- the use of an occipital reference region despite the tral analysis such as rank shaping [43], basis pursuit presence of d and k receptors [25,28,29] have re- [44], bootstrapped techniques to derive error esti- peatedly been shown to have less sensitivity to de- mates simultaneously [45,46], or the use of a spe- tect changes in patients than images quantified cific time point on the fitted time course of the with spectral analysis [40,41].[11 C]buprenorphine unit IRF (eg, IRF60 at 60 minutes), which is less de- is a partial agonist at m receptors and an antagonist pendent on the extrapolation to infinity than VD. at k and d receptors but has not found widespread Information derived from a reference region can use compared with [11 C]DPN. In rats, the signal is be used to correct VD for nonspecific binding, smaller than that of [11 C]DPN, with more nonspe- yielding binding potentials (BP). VD and BP are cific binding. proportional to the available receptor density The agonist [11 C]carfentanil (CFN) is used for im- (Bmax) over receptor affinity (Kd). More complex aging m receptors. Its affinity for m receptors (KI at experimental protocols usually requiring multiple 37C, 0.051 nM) is approximately 100-fold higher injections may be used to derive Bmax and Kd than for d receptors and more than 200-fold higher separately. Region-and voxel-based methods (the than for k receptors. Medial occipital cortex can be latter when tracer characteristics allow the genera- used as a reference region, and quantification has tion of parametric maps) are used for the analysis often been performed as the specific-nonspecific ra- of changes; both are complementary [47]. tio on late summed images (eg, 35–70 minutes) as Quantitative or semi-quantitative evaluation of (region of interest occipital)/occipital [64]. More datasets, typically in comparison with healthy con- recently, the use of graphic methods and the simpli- trols, is often more sensitive to abnormalities than fied reference tissue model [65] have been more ac- the qualitative assessment of PET images [47–50]. curate in quantifying [11 C]CFN binding [66]. Automatic methods for the co-registration of MR Similar 18F-labeled tracers are under development imaging and PET data are readily available and al- for use by centers without access to a cyclotron [67]. low the interpretation of functional data compared Delta receptors are visualized with the antagonist with higher resolution structural data [51]. [11 C]methylnaltrindole (MeNTI). Its affinity for d re- Partial volume effects arise owing to the restricted ceptors of 0.02 nM is approximately 700-fold higher spatial resolution PET provides. When structural than for m receptors and more than 3000-fold changes are present, that is, the anatomy differs be- higher than for k receptors [68]. Cerebellum has tween patients and controls, these effects are partic- been used as a reference region, and quantification ularly important because they can lead to spurious has been performed on late summed images differences. In these situations, they should be cor- (eg, 34–90 minutes [68] or 50–90 minutes [69]) rected for [52,53]. Region-based correction methods as (region of interest cerebellum)/cerebellum. [54–56] and voxel-based methods [57] are used. [11 C]MeNTI binding is essentially irreversible over Even in the absence of partial volume effects dif- the time course of a 90-minute PET scan [68,70]. fering between conditions or groups, interpretation It is unclear how far flow effects will enter these of binding estimates tends not to be unequivocal. binding estimates, and the ratio method was aban- For example, reductions in binding estimates can doned by the same researchers in a later systematic be due to reduced receptor concentration, reduced modeling study [70]. receptor affinity, increased occupancy by endoge- [18F]cyclofoxy (CFX) is an antagonist at m and k re- nous ligands, or opposite changes in the control ceptors. Affinities were reported as a Ki of 2.6 nM for group or control condition. In the following sec- m sites, 9.3 nM for k sites, and 89 nM for d sites [71]. tions, the interpretation thought to be most likely The medial occipital cortex may contain some k re- in the circumstances of the study has been ceptors [28] but has been successfully used as a refer- indicated. ence region following estimation of total VD with an arterial input function and kinetic modeling with Available ligands and their quantification a one tissue compartment model [72–74]. Available ligands are tabulated in Table 1. Dipre- [11 C]GR103545 is the active () enantiomer of norphine (DPN) binds with similar, approximately the racemate GR89696 and a selective k agonist, 1 nmol/L, affinities at all three major opioid with subnanomolar to low nanomolar affinities Opioid Imaging 535

[75]. It has shown promise as the first k-selective the insula and the parietal operculum, have high PET tracer in mice [76] and baboons [75] but will [18F]DPN BPs [78]. Genotype has an important require radiochemistry improvements to increase role in regional binding estimates [79]. There is specific activity and reduce co-injected unlabeled a common co-dominant valine/methionine ligand to become useful in humans. polymorphism in the gene encoding catechol- O-methyl-transferase (COMT), leading to higher activity of this enzyme that metabolizes catechol- Opioid receptor imaging in healthy amines like dopamine, adrenaline, and noradrena- volunteers line in the order of val/val > val/met > met/met The marked regional differences in overall opioid carrying individuals. The more met polymorphism, receptor concentrations and subtype distributions the less breakdown of catecholamines, presumably described previously are reproduced by in vivo im- leading to more continuous stimulation of the do- aging techniques. Regarding total opioid receptor paminergic system. In animal models, such chronic binding, Jones and coworkers [77] commented on activation has been shown to lead to reduced a marked difference, estimated from [11 C]DPN ratio enkephalin (Table 1) levels and compensatory images, between the ‘‘medial’’ or ‘‘affective’’ pain increases in regional m receptors. Although global system (which comprises the medial thalamus, [11 C]CFN BP, reflecting m receptor concentration, amygdala, caudate nucleus, insula, cingulate gyrus, was not influenced by the polymorphism in 18 sub- and orbitofrontal cortex) (Fig. 1) with high opioid jects of whom 11 were heterozygous, marked differ- receptor binding and the ‘‘lateral’’ or ‘‘discriminat- ences in the expected direction were seen regionally ing’’ pain system (projecting to the somatosensory in the thalamus [79]. Comparison of the various cortex) with much lower binding. This difference groups showed further effects of genotype on base- is also obvious on parametric maps of [11 C]DPN line [11 C]CFN BP in ventral pallidum and nucleus VD (Fig. 2). Other areas now thought to be part accumbens, in the order of about 30% (Fig. 3). of the lateral nociceptive system, such as parts of The finding of higher m-receptor binding in

Fig. 1. Localized clusters of negative correlations between [11C]diprenorphine (DPN) VD (n 5 14) and restless legs syndrome severity, measured on the International Restless Legs Scale, analyzed with SPM99 and displayed at P < .01 uncorrected threshold, cluster extent of 50 voxels. All clusters throughout the whole brain are demon- strated (top left) in the maximum intensity projection ‘‘glass brain.’’ The top right and bottom two panels show significant clusters overlain on a single subject brain [133]. Color bar represents Z-values. Many structures of the medial pain system show this negative correlation. (From von Spiczak S, et al. The role of opioids in restless legs syndrome: an [11C]diprenorphine PET study. Brain 2005;128(Pt 4):911; with permission.) 536 Hammers & Lingford-Hughes

Fig. 2. Parametric maps of [11C]diprenorphine (DPN) VD in a healthy male volunteer, acquired after injection of approximately 185 MBq of [11C]DPN on a Siemens/CTI HR11 scanner displayed alongside co-registered MR im- aging slices. (Top) Coronal slices at the level of the amygdala. (Bottom) Superior transverse slices. Note the het- erogeneity of binding in the coronal slices and the relative lack of binding centered on the postcentral gyrus in the transverse slices. Color bar represents the VD.

individuals with the met/met phenotype was con- [34]. Similarly, total VD and specific VD (regional firmed by a different group in postmortem autora- total VD minus occipital VD) of [18F]CFX, reflecting diographic studies in the caudate and accumbens m- as well as k-opioid receptor binding, were reduced nucleus and some thalamic nuclei; however, that in nine postmenopausal women compared with 15 study showed increased mRNA levels for pre-proen- men in the thalamus but not in several other regions kephalin, contrary to predictions [80]. of interest, including caudate and putamen [74]. Other factors affect receptor concentrations. Age The opioid system is involved in mood regulation. may regionally increase or decrease receptor con- m agonists tend to be rewarding, whereas k agonists centrations as seen in postmortem studies and us- induce dysphoria. In 14 healthy women scanned ing [11 C]CFN BP via a ratio method in subjects with a bolus-infusion paradigm [82], neutral and aged between 19 and 79 years [34]. Women had sad mood states were induced in a counterbalanced generally higher m receptor BP, thought to relate at order between 5 and 45 and 45 and 100 minutes, re- least partly to hormonal status [34]. A high estradiol spectively, after the start of the radiotracer infusion. state induced with patches led to higher m-receptor Sadness was induced by instructing subjects to con- availability in eight women ranging between 15% centrate on an autobiographical event that, before and 32% [81]. These effects are likely to be region scanning, had been established to be associated specific, and some regions may show interactions with profound sadness. Logan plot derived VD ratios between age and sex, adding to difficulties in inter- (over occipital lobe) for the periods of 10 to 45 and pretation. In the thalamus, m receptor BP was higher 50 to 100 minutes were compared using statistical in premenopausal women when compared with parametric mapping. Increased availability of m re- their male contemporaries, whereas the opposite ceptors (ie, a ‘‘deactivation’’ of the opioid system) relationship was seen in postmenopausal women was seen in the perigenual anterior cingulate gyrus, Opioid Imaging 537

Fig. 3. Catechol-O-methyl-transferase (COMT) polymorphism with substitution of valine (val) by methionine (met) at codon 158 (val158met). Effect of genotypes and associated COMT activity on [11C]carfentanil (CFN) vol- ume-of-distribution ratio (DVR) binding potential (BP), reflecting m-opioid receptor binding and m-opioid recep- tor system activation in response to sustained pain. (Upper left) Correlations between COMT val158met polymorphisms and baseline m-opioid receptor BP. Significant correlations were obtained in the anterior thala- mus (1), the nucleus accumbens (2), and adjacent ventral pallidum (3) ipsilateral to pain. (Upper right) Mean standard deviation of nucleus accumbens (ipsilateral) BP values. (Lower left) Correlations between COMT activity and m-opioid system activation in the ipsilateral posterior thalamus (1), nucleus accumbens (2), and ventral pal- lidum (3), bilaterally, and the contralateral amygdala and adjacent temporal cortex. (Lower right) Mean stan- dard deviation values for the percent change in the magnitude of m-opioid receptor system activation in the nucleus accumbens. Color bar represents the Z-scores. Threshold for display of results was set at P < .01 for all analyses. (From Zubieta JK, Heitzeg MM, Smith YR, et al. COMT val158met genotype affects m-opioid neuro- transmitter responses to a pain stressor. Science 2003;299(5610):1241; with permission. Copyright 2003 AAAS.) left inferior temporal cortex, and amygdala and ven- Opioid imaging in the wider sense also includes tral pallidum bilaterally. Most of these changes were indirect measures of opioid neurotransmission, more clearly seen after global normalization of VD that is, studies investigating cerebral blood flow in ratios. The magnitude of deactivations correlated relation to opioid receptor agonists. Although these with the increase in negative affect ratings or the de- studies are not discussed herein, a recent detailed crease in positive affect ratings in the perigenual cin- review is available elsewhere [84]. gulate gyrus, ventral pallidum bilaterally, and on the left in the amygdala, insula, and hypothalamus. The Opioid imaging in pain-related studies study provides direct in vivo evidence for a role of the m-opioid receptor system in affect regulation in Changes in receptor availability in humans. In another study in 12 healthy men, higher pain and discomfort: between-group baseline Logan plot VD ratios were weakly associ- comparisons ated with smaller blood flow responses to aversive Four patients with central neuropathic pain visual stimuli in the left anterior middle temporal (mainly resulting from strokes) when compared gyrus [83]. with an undisclosed number of healthy controls 538 Hammers & Lingford-Hughes

had reduced [11 C]DPN VD centered on medial pain volunteers by applying a heat stimulus (approxi- system structures (cingulate gyrus, insula, and thal- mately 44C) from 60 to 70 minutes after injection amus), with some reductions in the lateral pain sys- of [18F]DPN. A VD from data 80 to 120 minutes tem (inferior parietal cortex) [85]. The effect sizes post injection was calculated using an invasive Lo- were around 15% to 30%. In this particular study, gan plot and compared with the findings in another it was thought to be unlikely that the reductions eight healthy volunteers of similar age and sex were due to occupancy of receptors by endogenous distribution scanned with the same protocol but opioids, and the interpretation of a disease-related without pain induction. In the group with pain loss of binding sites with subsequent failure of the stimulation, there were decreases in [18F]DPN VD opioid system to suppress pain was favored, partly in the ipsilateral nucleus accumbens and amygdala supported by the lack of clinical change after pro- and bilaterally in the middle frontal gyri, the ante- longed infusions of naloxone in two patients. A dif- rior insulae, the thalami, and the perigenual ante- ferent group reported a similar finding of regionally rior cingulate gyrus. reduced [11 C]DPN ratios in medial (contralateral thalamus, insula, cingulate gyrus, midbrain) and Direct intrasubject comparisons of periods lateral (parietal, somatosensory, and lateral frontal with pain and pain-free states cortices) pain system structures in five patients Such studies are more complex to perform, because with poststroke pain compared with 12 controls us- paired studies need to be obtained for patients in ing the occipital lobe as a reference region [86], ex- two different states and ideally compared with panding on an earlier case report [87]. In contrast paired control scans. Nevertheless, they allow the to the previous study [85], all images had been pre- attribution of changes in a neurotransmitter system processed to show the anatomic lesion on the same to changes in clinical or experimental stage and pro- side, and decreases showed lateralization to the side vide more powerful evidence than studies compar- contralateral to the perceived pain. ing groups of patients and controls. When compared with healthy controls, 15 pa- Four patients with rheumatoid arthritis were ex- tients with meticulously evaluated primary restless amined twice with [11 C]DPN [91]. Global increases legs syndrome had no global or regional differences in opioid receptor availability and additional re- in [11 C]DPN VD [40]; however, the more severe the gional increases in the frontal and temporal lobe restless legs syndrome, the less [11 C]DPN VD in and cingulate gyrus were seen in the reduced pain much of the medial pain system (Fig. 1), suggesting state, consistent with the idea that occupancy by en- enhanced release of endogenous opioids. Similarly, dogenous ligands, triggered by pain, may lead to re- ratings on the McGill Pain Questionnaire correlated duced availability. The same group obtained similar inversely with [11 C]DPN VD in the orbitofrontal results in six patients undergoing thermocoagula- cortex and anterior cingulate gyrus. tion for refractory trigeminal neuralgia. Regional in- Overall, these studies are remarkable for delineat- creases of [11 C]DPN VD after surgical pain relief were ing the medial pain system irrespectively of the type seen in frontal, insular, perigenual, midcingulate, of pain or anatomic localization of lesions. They of- and inferior parietal cortices, basal ganglia, and thal- fer no explanation of how such systematic changes amus bilaterally [92]. Although the preceding stud- could be mediated. ies suffer from the lack of a control group scanned Six men with left-sided cluster headache when twice, they have the merit of being the only ones at- compared with eight healthy controls had de- tempting intrasubject comparisons in patients so far. 11 creased signal of [ C]DPN (quantified as IRF60) More work has been done in healthy controls in in the pineal gland, and binding in hypothalamus experimental paradigms. When pain was induced and cingulate gyrus correlated negatively with dis- through capsaicin application to the hand, eight ease duration [88]. The validity of quantifying healthy volunteers showed reductions in contralat- 11 11 [ C]DPN with IRF60 for the pineal gland, which eral thalamic [ C]CFN 34- to 82-minute tissue ra- is situated outside the blood-brain barrier, was tios over the occipital lobe when compared with not explicitly examined. Although the changes are baseline; in addition, the magnitude of the decrease in brain regions previously implicated in the path- of m-receptor binding correlated directly with pain ophysiology of cluster headache [89], this makes intensity [93]. them harder to interpret. All studies discussed hereafter used healthy con- Ultimately, patients should be the beneficiaries trols in a paradigm whereby sustained deep tissue and subjects in imaging research; however, experi- pain was induced by infusion of 5% hypertonic sa- mental paradigms offer the possibility of tightly line into the masseter muscle, with nonpainful in- controlling several variables that would be impossi- fusion of 0.9% (isotonic) saline in the control ble to standardize in real patients. In one such ex- condition into the other masseter muscle. By con- periment [90], pain was induced in eight healthy trolling the rate of infusion, a similar level of pain Opioid Imaging 539 was maintained, around 50 or slightly below on differences in the same direction were replicated a visual analogue scale where 0 denotes no pain later for the nucleus accumbens and hypothalamus and 100 the most intense pain imaginable. Both (but not the amygdala) [81]. scans were acquired on the same day, separated by To further characterize the role estradiol has in at least six isotope half-lives. Intramasseter infu- women’s m-opioid transmission in relation to sions were administered from 20 to 40 minutes pain, eight women were subjected to two paired after radiotracer administration [79,81,94,95] or studies—one pair in the low-estrogen, low-proges- from 40 to 60 minutes [96]. A bolus-infusion radio- terone early follicular menstrual cycle phase and tracer administration protocol was used. Modified one pair after transdermal delivery of estradiol for Logan graphical analysis with occipital cortex as 7 to 9 days, which increased estradiol plasma levels a reference region on data from 20 to 70 about fivefold [81]. In the high estradiol state, base- [79,81,94,95] or 40 to 90 minutes [96] post radio- line DVRs of [11 C]CFN increased an average of 15% tracer injection was used to quantify m-receptor to 32% in the thalamus, amygdala, nucleus accum- binding as DVR, a VD ratio (ie, the slopes of the bens, and hypothalamus. These increases were cor- Logan plot for voxels of interest divided by the related with estradiol plasma levels in the latter two. reference region slope). The high estradiol state also led to 12% to 19% de- Using this paradigm, the Ann Arbor group has creases of pain-induced [11 C]CFN DVRs (ie, activa- initially shown in 20 healthy volunteers that sus- tion of m receptors through endogenous opioid tained pain can induce a release of endogenous opi- receptors) in the hypothalamus/nucleus accumbens oids, measurable as a decrease in [11 C]CFN DVR, and the amygdala ipsilateral to pain. Even more indicating interaction of endogenous opioids with strikingly than in the earlier study [95], women in m-opioid receptors in specific brain regions. m-opioid the low estradiol state did not show any activation receptor mediated activation is associated with of m receptors but, on the contrary, ‘‘deactivations’’ reductions in pain ratings in the sensory domain in these same regions and the medial thalamus; (ipsilateral nucleus accumbens, thalamus, and these effects ranged from 11% to 16%. amygdala) and affective domain (bilaterally in the Such increases are difficult to interpret, because cingulate gyrus and thalamus, and ipsilaterally in opioid release generally requires high frequency fir- the nucleus accumbens) [94]. Having established ing [97], and the system overall does not seem ton- this principle, the researchers went on to demon- ically active, however, the authors discuss [81,95] strate sex differences in 14 men and 14 women some evidence that there may be a tonic release of aged between 20 and 30 years, with the women endogenous opioids at baseline, particularly in scanned during the low-estrogen, low-progesterone the nucleus accumbens, but possibly also in the early follicular menstrual cycle phase [95]. The in- brainstem, amygdala, and other regions. tensity-controlled paradigm did not show any dif- The group went on to demonstrate an effect of ferences in subjective positive and negative affect genotype on m-receptor neurotransmission and scores or pain questionnaire ratings. Nevertheless, used the common COMT val/met polymorphism men activated the m-opioid receptor system contral- described above in the section on opioid imaging aterally to pain infliction in the anterior thalamus, in healthy volunteers. In addition to the substantial ventral pallidum/substantia innominata, and ante- influence on baseline m-receptor binding, with indi- rior insula, as well as ipsilaterally in the amygdala viduals with the met/met phenotype showing higher and ventral basal ganglia, whereas the latter was levels of m-opioid receptors in the thalamus, ventral the only area with significant DVR decreases in pallidum, and nucleus accumbens, these individuals women who, in addition, showed DVR increases also had substantially smaller activations of the following pain in the ipsilateral nucleus accumbens. m-receptor system (with a gene ‘‘dose’’ effect and Across all subjects, greater activation of m-opioid met/val heterozygotes intermediate between homo- receptors in ipsilateral nucleus accumbens and zygotes) and, in parallel, higher negative ratings on amygdala was correlated with less pain perception, affect scores and pain questionnaires. a correlation in the expected direction. Interestingly, the same group provided evidence Baseline [11 C]CFN DVR indicated higher binding that the placebo response is mediated via m-opioid in women only in one amygdala, in contrast to the receptors. Fourteen healthy male volunteers under- group’s earlier study showing more widespread and went a three-scan paradigm consisting of a baseline impressive increases in women compared with men scan and two scans with pain induction. During [34]. Characteristics of this earlier study which one of these, a placebo was administered with the might explain some of these differences include expectation of pain relief. Placebo did relieve pain a wider age range, a ratio method of quantification, and also led to release of endogenous opioids in no correction for multiple comparisons, and a less the middle frontal gyrus and nucleus accumbens ip- objective method of region definition. Sex silateral to the pain and in the perigenual anterior 540 Hammers & Lingford-Hughes

cingulate gyrus and insula contralaterally when receptors overall, increases of m-receptor binding ra- compared with the pain only condition [96]. tios were seen on the side of the epileptogenic focus In summary, work performed thus far has firmly in lateral midtemporal and posterior temporal neo- established that activation of the endogenous opi- cortex, whereas binding in the amygdala was de- oid transmitter system in response to chronic or creased. The latter finding might be due to partial acute and prolonged pain can reliably be imaged volume effects that could be corrected for in con- using [11 C]CFN and [11 C]DPN in groups of healthy temporary studies [103]; however, the finding of lat- volunteers and patients. In addition, reductions of eral neocortical increases confirms an earlier study opioid receptor binding are seen as group effects [104]. These increases of [11 C]CFN binding coin- when comparing pain patients with healthy con- cided with areas of hypometabolism seen on trols using [11 C]DPN or [18F]DPN, with a remark- [18F]FDG PET in the earlier [104] but not the later ably similar spatial pattern corresponding largely study [103]. Temporal neocortical binding increases to the medial pain system despite different etiolo- were also seen in ten patients using the d-receptor gies and localizations of the primary pathology. In subtype selective antagonist [11 C]MeNTI. Region paired studies, many of the effects could be seen definition was similar to that in the earlier study in individual subjects. Having established some [100] but performed on three slices, and binding major determinants of baseline binding and opioid was quantified via ratios on 50- to 90-minute con- release, the next steps for the field will be to start in- centration images as (region cerebellum)/cere- vestigating single patients—the ultimate beneficia- bellum [69]. Because of [11 C]MeNTI’s essentially ries of medical research—and establish clinical irreversible binding, this quantification method is links that help in management and prognosis. likely suboptimal and was later abandoned [70]. Ratio increases for d receptors were seen over all Opioid imaging in epilepsy of the temporal neocortex ipsilateral to the seizure focus and included the amygdala. Significance was Epilepsy is the most common serious neurologic reached for the superior temporal pole, reaching disorder, affecting 0.5% to 0.8% of the population. more posteriorly in the most inferior of three PET Epileptic seizures can arise from one part of the slices available [69]. The increases in the inferior brain (focal epilepsies) or can appear generalized most slice coincided with increases of [11 C]CFN ra- from the outset (generalized epilepsies). Among tios derived from the same patients. In contrast to the focal epilepsies, temporal lobe epilepsy is the the earlier study, no increases were seen in temporal most common, and hippocampal sclerosis is the neocortex at the level of the amygdala, although the most frequent underlying cause. There is consider- amygdala [11 C]CFN ratio decrease was replicated. able interest in opioidergic mechanisms in epilepsy. One speculative explanation is that an increase in Two recent reviews covering opioid imaging in epi- m receptors may be a manifestation of a tonic anti- lepsy are available [98,99]. epileptic system in the temporal neocortex to limit the spread of epileptic activity, while the amygdalar Focal epilepsies decrease of [11 C]CFN ratios could be related to me- Interictal studies sial temporal hyperexcitability. Opposite changes of Overall opioid receptor binding has been studied [11 C]CFN and [11 C]MeNTI ratios in the amygdala, with [11 C]DPN in two studies. Using [11 C]DPN ra- implying increases of m receptors but decreases of tios (region occipital/occipital) for four regions d receptors, may explain why no asymmetries were (amygdala, anterior temporal/midtemporal/poste- detected with the nonselective tracer [11 C]DPN. rior temporal, ie, not hippocampus) derived from Using [18F]CFX, the m and k antagonist, quanti- standard-sized rectangular regions of interest in fied with a kinetic model as a VD, there was no over- a single PET plane in 11patients with a video electro- all asymmetry in a group of 14 patients with encephalogram confirmed temporal lobe epilepsy, temporal lobe epilepsy, but, in some individual pa- no differences between the focus and nonfocus sides tients, binding was increased in the ipsilateral tem- were seen [100]. Eleven controls were scanned but poral lobe [72]. Taken together with the temporal no results reported. A later study confirmed the ab- neocortical increases of m receptors described previ- sence of asymmetries of [11 C]DPN binding in two ously, this finding would be consistent with a de- patients [101]. In contrast, work in progress has crease in the affinity or number of k receptors, or shown abnormalities of [11 C]DPN binding in 80% decreased availability of k receptors through occu- of patients with focal epilepsies due to malforma- pation by an endogenous ligand. tions of cortical development [102]. [11 C]CFN was used in the same 11 patients stud- Ictal studies ied with [11 C]DPN [100] and quantified in the same Ictal studies have been performed in five patients way in the same regions. In contrast to opioid with reading epilepsy compared with six controls Opioid Imaging 541 with [11 C]DPN and a two-scan (rest-activation) epilepsy. Currently, PET opioid studies are confined paradigm [105]. Quantification was performed to research in a few centers and are not used using a metabolite-corrected arterial plasma input clinically. function and spectral analysis. This approach al- lowed the computation of parametric maps of the Idiopathic generalized epilepsy 11 IRF60 and use of statistical parametric mapping. [ C]DPN VD was used to image all opioid receptor Reading epilepsy provides a model for focal epilepsy subtypes in eight patients with childhood and juve- and has the advantage that seizures can be provoked nile absence epilepsy compared with eight controls easily through reading yet do not lead to significant [107]. No significant differences were found, sug- head movement. Comparison of the resting condi- gesting no overall interictal abnormalities of the tion in patients with that in controls revealed no sig- opioid system in idiopathic generalized epilepsy. nificant differences. Reading-induced seizures were Nevertheless, changes were seen in a dynamic 11 11 associated with reduced [ C]DPN IRF60 in the left [ C]DPN study investigating eight patients with parietotemporo-occipital cortex and to a lesser ex- primary generalized epilepsy and eight controls. tent in the left middle temporal gyrus and posterior In patients, absence seizures were induced by hy- parieto-occipital junction (Fig. 4), implying release perventilation 30 to 40 minutes after tracer injec- of endogenous opioids at the time of seizures. These tion, and hyperventilation maintained for 10 areas are known to be involved in reading, visual minutes led to generalized spike-wave discharges processing, and recognition of words, and the re- for 10% to 51% of this period. After this provoca- searchers have since provided fMR imaging mea- tion of serial absence seizures, a faster elimination sures of increased cerebral blood flow during of [11 C]DPN from association cortices but not reading overlapping with these regions (Fig. 4). thalamus, basal ganglia, or cerebellum was seen, Further evidence for the release of endogenous suggested by simulations and a two-compartment opioids at the time of seizures comes from a fortu- model to correspond to an estimated 15% to 41% itous ictal [18F]CFX PET scan from the National In- decrease in the specific tracer uptake rate constant stitutes of Health laboratory reported as a personal (k3) [108], compatible with the release of endoge- communication [106]. Frequent intermittent right nous opioids in association cortices at the time of medial temporal discharges developed about 6 seizures. minutes after injection, and the time-activity curve for the right medial temporal lobe remained con- Summary stantly below that of the contralateral side for the Opioid receptor alterations seem to be involved in remaining 60 minutes. temporal lobe epilepsy and other epilepsy syn- Taken together, these findings suggest an involve- dromes. Some of the earlier studies could be repli- ment of the opioid system in several forms of focal cated to benefit from the rapid progress in the

Fig. 4. Diprenorphine binding during reading-induced seizures compared with control conditions and cerebral blood flow during reading. (Left) Statistical parametric map on a normalized MR brain image. Areas with sig- 11 nificantly decreased [ C]diprenorphine (DPN) impulse response function at 60 minutes (IRF60) during seizures are shown in yellow (Z 5 5.22; P < .05 corrected) in the left parietotemporo-occipital cortex and red (uncorrected P < .001; not significant when corrected for multiple comparisons) in the right inferior parietal lobe (Z 5 4.12), left superior temporal gyrus (Z 5 4.02), left fusiform gyrus (Z 5 4.00), left middle temporal gyrus (Z 5 3.51), and left temporoparietal junction (Z 5 3.37). (From Koepp MJ, Richardson MP, Brooks DJ, et al. Focal cortical release of endogenous opioids during reading-induced seizures. Lancet 1998;352:954; with permission.) (Right)fMR imaging results in healthy volunteers reading sentences versus pseudofont [134]. The activated areas overlap with those in which endogenous opioid release is seen. 542 Hammers & Lingford-Hughes

fields of MR imaging and PET in the last decade. No [11 C]DPN radioactivity concentration between 30 detailed studies of extratemporal epilepsies and and 90 minutes postinjection, dividing caudate, malformations of cortical development have been putamen, and thalamus standard-sized region of reported. In the absence of suitable k-selective ra- interest values by occipital values, and in addition dioligands, the question of the interictal and ictal used spectral analysis to generate parametric images role of the k-opioid system in the epilepsies remains of IRF60 values. Again, no differences between non- open. dyskinetic Parkinson’s disease patients and controls There is good evidence for ictal release of endog- were seen. Dyskinetic patients showed reduced up- enous opioids in several syndromes. Because many take ratios when compared with controls and their of the opioid receptor ligands seem to be displace- non-dyskinetic counterparts in the putamen and able by endogenous transmitters, further systematic thalamus. A smaller difference was found between comparisons of the ictal state with the interictal dyskinetic patients and controls in the caudate nu- state will be highly relevant. Future studies aiming cleus; this finding did not differentiate the dyski- to elucidate the pathophysiology of the epilepsies netic from the non-dyskinetic group. There was may benefit from the use of subtype-specific opioid a nonsignificant (P < .1) but large (11 years on aver- ligands. age) age difference between the patient groups; however, the same group had seen no age effect 11 Opioid imaging in other specialties on striatal [ C]DPN ratios in the earlier study [110]. The findings have some support in a decreas- Movement disorders ing putaminal uptake ratio with increased dyskine- The high concentration of opioid peptides and re- sia severity, defined as the summed scores (1–4) of ceptors in the basal ganglia has prompted investiga- the Unified Parkinson’s Disease Rating Scale tions in movement disorders. Parkinson’s disease (UPDRS) items 32 and 33 (duration and disability), is the most common akinetic rigid movement dis- which just reached significance using a Pearson order and is due to degeneration of nigral dop- correlation, and the (nonsignificant) greater reduc- aminergic neurons projecting to the striatum; tion of contralateral uptake ratios in two patients consequently, major differences have been shown with unilateral dyskinesias. This study was one of in the dopaminergic system [109]. Several atypical the first to harness the power of statistical paramet- parkinsonian syndromes, including multiple sys- ric mapping, developed in the same institution orig- tem atrophy and Steele-Richardson-Olszewski syn- inally for blood flow studies, for the investigation of drome (SRO) (supranuclear palsy), typically do receptor binding studies. The statistical parametric not respond to levodopa replacement therapy. mapping investigation of IRF60 not only confirmed There were no differences in standard-sized puta- the differences between dyskinetic Parkinson’s dis- minal and caudate region of interest [11 C]DPN ease patients and controls but also showed addi- ratios over occipital cortex in eight normal controls tional decreases in the pallidum and anterior and eight patients with a clinical diagnosis of Par- cingulate gyrus, as well as increases in the middle kinson’s disease [110]. In contrast, the same study and superior frontal gyri. Based on experimental showed that mean putaminal ratios were reduced evidence, the authors favored the interpretation in seven patients with multiple system atrophy that the striatal and thalamic reductions in opioid (striatonigral degeneration subtype), with three of receptor binding were due to increased occupancy seven patients showing significant abnormalities due to increased endogenous opioid transmission. defined as binding greater than 2.5 standard devia- In view of no differences in dopamine D1 or D2 tions below the control mean. Six patients with receptors between dyskinetic and non-dyskinetic SRO syndrome (supranuclear palsy) showed de- Parkinson’s disease patients and increased blood creased ratios in caudate and putamen as a group, flow in basal ganglia and frontal cortex during levo- with abnormal caudate binding in four of six pa- dopa-induced dyskinesias, this study overall sug- tients and abnormal putaminal binding in all six. gests that dyskinesias are instead associated with In the SRO group, mean binding was also signifi- alterations of opioid transmission in the basal gan- cantly decreased when compared with that in glia, with resulting overactivity of basal ganglia– Parkinson’s disease patients. frontal connections [112]. Another study by the same group investigated six Another condition in which opioid neurotrans- Parkinson’s disease patients with dyskinesias (in- mission has been investigated is Huntington’s voluntary movements that are a complication of disease, an autosomal dominantly transmitted long-term treatment with levodopa therapy) com- neurodegeneration predominantly affecting the pared with seven Parkinson’s disease patients with- caudate nucleus which leads to psychiatric abnor- out dyskinesias and ten controls [111]. They used malities, involuntary movements, and, ultimately, a region of interest approach on ratio images of death in early middle age. In five patients with Opioid Imaging 543

Huntington’s disease compared with nine controls, Dementia 11 decreases of [ C]DPN binding measures were seen Opioid receptor alterations have been described in with a variety of approaches. Importantly, IRF60 im- the most frequent dementia, Alzheimer’s disease. ages outperformed ratio images (striatal regions A quantitative autoradiographic study investigated over occipital lobe for 30- to 90-minute radioactiv- 11 brains of patients with Alzheimer’s disease com- ity images), and statistical parametric mapping out- pared with 10 brains of age-matched controls who performed regional approaches in the sense that had had normal neurological status in life. There peak changes were more marked, and additional was decreased m-receptor binding in the hippocam- nonhypothesized decreases were found in the pus (48%) and subiculum (46%), decreased mid-to-anterior cingulate gyrus as well as nonhypo- d-receptor binding in the amygdala (51%) and thesized increases in thalami and pregenual cingu- ventral putamen (30%), decreased k-receptor late gyrus/superior frontal gyrus [41], arguing binding in the hippocampus (39%), and in- strongly for the use of kinetic modeling techniques. creased k-receptor binding in the dorsal (157%) Huntington’s disease is accompanied by major and ventral (193%) putamen and cerebellar cortex atrophy, particularly of the caudate nucleus. The (1155%, but on the background of relatively low study did not correct for partial volume effects, absolute binding) [27]. Quantifying receptor bind- and it is likely that this factor accounts at least for ing decreases in neurodegenerative diseases in vivo some of the observed decreases but not for the tha- requires partial volume effect correction, which can lamic and cortical increases. be difficult. Nevertheless, the effect sizes shown in The findings in dyskinetic patients with Parkin- this postmortem study are remarkable, and the large son’s disease and patients with Huntington’s increases in k-receptor binding could potentially be disease prompted investigation of carriers of a muta- useful for early diagnostic and therapeutic studies. tion in the DYT1 gene manifesting with primary Early reports of m-opioid receptor decrease in the torsion dystonia. There were no regional differences amygdala in Alzheimer’s disease measured by 11 in [ C]DPN VD in seven patients compared with [11 C]CFN have not been pursued [115]. Work with 15 healthy controls and no correlation of opioid re- [18C]CFX in 12 patients with Alzheimer’s disease ceptor binding with disease severity, arguing against and 12 age-matched controls using specific VD cal- a major role of opioid neurotransmission across all culated as total VD minus medial occipital VD and dyskinesias [113]. standard-sized circular regions of interest in both In the movement disorder restless legs syndrome, groups revealed decreased specific VD in the Alz- no differences between 15 patients and 12 age- heimer patients in a global fashion but accentuated matched healthy controls were seen using in parietal, frontal, and limbic areas [73].Itis 11 11 [ C]DPN and either ratio methods or [ C]DPN unclear how partial volume effects affect kinetic 11 VD (n 5 14 patients) [40].[ C]DPN VD was neg- modeling of time-activity curves derived from atively correlated with the severity of restless legs standard-sized regions of interest, and the global syndrome in structures belonging to the medial nature of changes as well as their smaller magnitude pain system, and negatively correlated with the af- compared with cerebral blood flow changes make fect score of the McGill Pain Questionnaire in orbi- clinical usefulness of [18C]CFX in Alzheimer’s dis- tofrontal cortex and anterior cingulate gyrus; both ease unlikely, particularly in view of the recent avail- findings were interpreted as occupancy by endoge- ability of in vivo markers of Alzheimer’s pathology nous opioids. Similar changes were seen using ratio like [11 C]PIB or [18F]FDDNP [116]. images, but effect sizes and spatial extents were much smaller, underlining the value of quantitative Cardiology kinetic analysis. There are other areas in which opioid imaging could Tourette’s syndrome is a tic disorder with motor potentially be useful. For example, a recent prelim- and vocal tics starting before the age of 18 years. inary study in five subjects showed BPs in myocar- In five patients with this syndrome, basal ganglia to- dium of about 4 with [11 C]CFN and [11 C]MeNTI. tal opioid receptor binding was not different from High doses of naloxone, however, when given as that in an undisclosed number of controls. Spectral a single pre-PET bolus without subsequent infu- analysis and statistical parametric mapping local- sion, only reduced BPs by about 20% [117]. None- ized decreased binding in the cingulate gyrus, as theless, this study shows the potential of opioid well as increased binding in the insula bilaterally, imaging techniques used outside the brain. the left premotor cortex and perigenual cingulate cortex [114]. These findings, although etiologically Opioid imaging in psychiatry: addiction unclear, are interesting in the context of case reports of therapeutic utility of opioid neurotransmission In contrast to the widespread use of PET/single- modifying drugs (both antagonists and agonists). photon emission computed tomography (SPECT) 544 Hammers & Lingford-Hughes

imaging of dopaminergic systems in psychiatric dis- individuals could be due to an increase in receptor orders, opioid imaging has not been widely ap- number or reduced endogenous opioid levels. The plied. The exception is in the addiction field. The preclinical literature has reported that both may opioid system is the target for opioid drugs of abuse occur. and has a key role in modulating the dopaminergic The opioid system has a key role in mediating the system, mediating, for example, the pleasurable and pleasurable effects of alcohol, and several lines of reinforcing effects of alcohol [118]. evidence suggest that altered function in the opioid The first study to explore the opioid system in system may be involved in vulnerability to alcohol addiction used [11 C]CFN to investigate the m-opioid dependence in humans [121]. In addition, the opi- receptor in cocaine addiction. Increased levels of oid antagonist naltrexone is an effective medication the m-opioid receptor were found in ten cocaine in treating alcohol misuse [122]. To test the hypoth- addicts who were 1 to 4 days abstinent when com- esis that increased opiate receptor levels may also pared with six controls [119]. Increased [11 C]CFN mediate craving for alcohol, Heinz and coworkers binding was seen throughout the brain in the ante- [123] used [11 C]CFN PET to assess m-opioid recep- rior cingulate, frontal and temporal cortices, cau- tor levels in alcohol dependent individuals who date, and thalamus. The increased levels in the had been abstinent for 1 to 3 weeks. Using regional amygdala, anterior cingulate, and frontal cortex pos- analysis and statistical parametric mapping to itively correlated with the severity of cocaine craving compare [11 C]CFN levels in patients with that in reported. When rescanned at approximately 4 weeks controls, increased binding was seen in the ven- of abstinence, in some but not all of the individuals, tral striatum, including the nucleus accumbens the increased levels of [11 C]CFN binding were re- (Fig. 5). This area is involved in mediating re- duced to levels comparable with those in healthy ward-based learning, and its dopaminergic func- controls. A later study in 17 cocaine addicts com- tion is modulated by the opioid system. No pared with 16 controls also reported increased differences were seen in other areas, including the m-opioid receptor levels as measured with cortex. Using statistical parametric mapping, a posi- [11 C]CFN in the anterior cingulate and frontal and tive correlation between [11 C]CFN binding levels in lateral temporal cortices in cocaine addicts after 1 the ventral striatum and self-reported craving was day of abstinence from cocaine [120]. Similar to seen (Fig. 5). This finding is similar to observations the earlier study, the increased m-opioid receptor in cocaine dependence. In addition, a positive cor- levels correlated with self-reported craving in several relation was seen between m-opiate receptor levels brain regions (eg, the anterior cingulate, dorsolateral in the frontal cortex and craving, as measured by frontal cortex) but not ratings of depression or anx- the obsessive-compulsive drinking scale. This area iety. Interestingly, the increased levels of [11 C]CFN of the cortex is involved in executive functioning; binding in some frontal cortical regions (eg, dorso- therefore, a dysfunctional m-opioid system may lateral frontal cortex) positively correlated with the contribute to the poor decision making often seen amount of cocaine recently used. This finding sug- in dependent patients. Several other clinical vari- gests that the increased levels occur in response to ables were studied, including a family history of cocaine use. Cocaine addicts were then scanned at alcoholism, age, early versus late alcoholism (estab- 1 week and 12 weeks after monitored abstinence. lished dependency before and after the age of The levels of [11 C]CFN binding and changes over 25 years), and smoking, but none of these were time varied, depending on the brain region. For in- significantly associated with m-opioid receptor stance, m-opioid receptors remained elevated in the availability. anterior cingulate cortex after 1 week but in the Some individuals maintained their sobriety and dorsolateral cortex had reduced to control levels. were rescanned 5 weeks later. Unlike in cocaine de- Over the 12-week period, binding of [11 C]CFN de- pendence, the higher [11 C]CFN binding levels per- creased in all brain regions with the exception of sisted. It was not possible to determine whether the anterior cingulate cortex. Craving was still pos- the increased binding was due to reduced endoge- itively correlated with increased m-opioid receptors nous opioid levels or increased receptor levels; at 1 week; however, after 12 weeks, craving was however, the researchers speculated that if negligible, so correlational analysis was not receptor levels are increased, naltrexone may work performed. through normalizing this increase and reducing The results of these two studies support a role for the rewarding effects of alcohol. the opioid system in cocaine addiction. The level of In opiate addiction, a similar increase in opioid m-opioid receptor availability seems to be related to receptor availability in early abstinence has been cocaine use and craving. [11 C]CFN is sensitive to en- reported. A preliminary study of three heroin de- dogenous opioid levels, and the reported increase pendent patients used [11 C]CFN PET to measure in early abstinence from cocaine in dependent m-opioid receptor availability during and after Opioid Imaging 545

00 Fig. 5. Brain imaging of central m-opiate receptor availability (V3 ). Region of interest analysis: (A) definition, (C) 00 interindividually averaged time-activity curves, and (D)V3 of all participants, obtained from Logan plot. (B) 00 Voxelwise averaged V3 parametric images of detoxified alcoholic patients, abstinent for 1 to 3 weeks (top 00 left) compared with healthy volunteers (top right). The highest V3 in alcoholic patients and the largest differ- ence between alcoholic patients and controls were found in the ventral striatum and the adjacent putamen. [11C]carfentanil indicates 11C-labeled carfentanil. In part (D), the open circles represent the mean; the box sur- rounding the circles, the standard error of the mean; and the bars, the standard deviation. (From Heinz A, Reimold M, Wrase J, et al. Correlation of stable elevations in striatal mu-opioid receptor availability in detoxified alcoholic patients with alcohol craving: a positron emission tomography study using carbon 11-labeled carfen- tanil. Arch Gen Psychiatry 2005;62(1):59; with permission.) a brief 6-week ascending and descending regimen binding was seen compared with that in controls of the substitute drug, buprenorphine [124]. Bu- in the inferofrontal and anterior cingulate cortices. prenorphine is a partial agonist at the m-opiate re- The small number of patients studied likely con- ceptors and a k antagonist. After withdrawal from tributed to the restricted areas showing a significant buprenorphine, a significant increase in [11 C]CFN increase. Using [11 C]DPN PET, higher binding 546 Hammers & Lingford-Hughes

levels throughout the brain have been reported in reduced [18C]CFX binding (19% to 32%) was recently abstinent methadone-maintained heroin seen in these methadone-maintained opiate addicts dependent patients compared with control sub- when compared with controls. The scans were per- jects [125]. Unlike for cocaine or alcohol depen- formed 22 hours after the last dose of methadone, dence, no significant correlation with craving was almost at a time when their next daily dose was found. due. Plasma levels of methadone only weakly corre- Together these studies suggest that early absti- lated with [18C]CFX binding in the caudate and pu- nence from dependence is associated with in- tamen and not significantly in any other region. It creased availability of m-opioid receptors. Because was speculated that only a small number of opioid this increase has been shown for alcohol, cocaine, receptors were occupied by clinically efficacious and opioid dependence, it suggests that the opioid doses of methadone (30–90 mg/day), leaving a sig- system may have a fundamental role in addiction nificant number of opioid receptors available for and that changes occur regardless of the substance other functions such as pain relief. of abuse. Although an increase in receptor availabil- Another study using [11 C]DPN PET has also ity is associated with craving in alcohol and cocaine explored the dose-occupancy relationship for meth- dependence, this is not true in opioid dependence. adone [128]. In this study, no significant relation- Perhaps the direct action of the opioid abuse on the ship between the dose of methadone and [11 C]DPN opioid receptors obscures any more subtle effects binding was evident, and complementary preclini- on the opiate system involved in reward and addic- cal work revealed that [11 C]DPN binding was not tion. An important issue that needs to be fully altered by an acute dose of methadone despite explored is the contribution of the effects of having profound behavioral effects. The conclusion substances of abuse on the opioid system, the ef- was that methadone needs to occupy a small per- fects of abstinence, or underlying vulnerability to centage of receptors to be clinical efficacious and at the raised opioid receptor levels seen. a level that is undetectable using [11 C]DPN PET. PET imaging has been widely used to explore the These findings were slightly at odds with a previous dose-occupancy relationship for a number of drugs study from the same group that had shown blockade in psychiatry, in particular D2 occupancy by the an- of opioid agonist effects in a similar group of meth- tipsychotic drugs [126]. In clinical practice, it be- adone-maintained patients [129], suggesting that came clear that high doses of these drugs were few opioid receptors were available. unnecessary because most D2 receptors were al- Although both studies suggest that methadone ready blocked, and the risk of adverse side effects does not need to occupy a large percentage of opi- outweighed any small potential increase in antipsy- oid receptors for efficacy, neither can help with ra- chotic efficacy. In the management of opioid addic- tional prescribing, that is, how much methadone tion, patients are generally put on a substitute drug to give to reach a maximal clinical effect. The stud- such as methadone or buprenorphine. In clinical ies suggest that the tolerance seen to opiates in such practice, the dose of the substitute drug given is gen- patients is due to altered opioid receptor function erally based on converting how much ‘‘street’’ opi- other than removal of the opioid receptor binding oid drugs such as heroin the patient is using, site, because this site is clearly still available for together with clinical observation of withdrawal the radiotracers to bind to. It is likely that mecha- or intoxication once on the substitute drug. There nisms involved in transmembrane signaling under- can be a wide variation in the dose of substitute lie this phenomenon. Evidence from preclinical or drug, particularly methadone, that the patient even- human postmortem studies is also not consistent tually is maintained on (eg, tens of milligrams to as to what happens to the opioid receptor in opioid over 100 mg). In addition, patients can request dependence, with increases, reductions, or no more substitute drug when it is not clear to the cli- change reported [130,131]. It is notable that opioid nician that they are in withdrawal or need more. withdrawal can be ameliorated with non-opioid PET imaging can assess the dose-occupancy rela- drugs, suggesting that this system is not critical in tionship for these two substitute drugs at the opioid mediating all of the effects of abused opioid drugs receptor in an attempt to determine how much [132]. drug to give to a patient. By contrast, a clear dose-occupancy relationship Methadone, a m agonist, is the most widely pre- is apparent with buprenorphine, a partial agonist scribed substitute drug. A group of opioid depen- at the m receptor and k antagonist. In a preliminary dent patients who had been maintained on study of three heroin dependent patients using methadone for many years (2–27 years) and were [11 C]CFN PET, 2 mg of buprenorphine, a relatively stable underwent a [18C]CFX PET scan [127]. In re- low clinical dose, occupied 36% to 50% of opioid gions of interest such as the anterior cingulate cor- receptors [124]. A relatively large clinical dose, 16 tex, thalamus, caudate nucleus, and amygdala, mg, resulted in 79% to 95% occupancy; however, Opioid Imaging 547 there was a large variation in the percentage occu- Etorphine to rat-brain homogenate. Proc Natl pancy seen. In a follow-up study using more pa- Acad Sci USA 1973;70:1947–9. tients, buprenorphine resulted in significant [3] Terenius L. Stereospecific interaction between reduction of [11 C]CFN binding by 41% with 2 mg, narcotic analgesics and a synaptic plasma 80% with 16 mg, and 84% with 32 mg [132]. membrane fraction of rat cerebral cortex. Acta Pharmacol Toxicol (Copenh) 1973;32:317–20. This degree of occupancy was seen throughout the [4] Martin WR, Eades CG, Thompson JA, et al. The brain, including areas of interest such as the ante- effects of morphine- and nalorphine-like drugs rior cingulate gyrus, amygdala, and frontal cortical in the nondependent and morphine-dependent regions, and was correlated with plasma buprenor- chronic spinal dog. J Pharmacol Exp Ther 1976; phine levels. In addition, blockade of opioid 197:517–32. agonist effects correlated with buprenorphine occu- [5] Lord JA, Waterfield AA, Hughes J, et al. Endoge- pancy. PET opioid receptor imaging has been infor- nous opioid peptides: multiple agonists and re- mative, showing the amount of buprenorphine ceptors. Nature 1977;267:495–9. required to result in almost complete occupancy [6] Kosterlitz HW, Paterson SJ, Robson LE. Charac- of opioid receptors. terization of the kappa-subtype of the opiate re- ceptor in the guinea-pig brain. Br J Pharmacol 1981;73:939–49. Summary [7] Darland T, Heinricher MM, Grandy DK. Orpha- nin FQ/nococeptin: a role in pain and analge- The excitement of the 1970s, when opioid receptors sia, but so much more. Trends Neurosci 1998; and the various subtypes were described, has been 21:215–21. matched in the 1990s and early years of this century [8] Mogil JS, Pasternak GW. The molecular and be- by the application of ever higher resolution whole- havioral pharmacology of the orphanin FQ/no- brain PET scanners, the availability of several radio- ciceptin peptide and receptor family. Pharmacol ligands, the combination of PET with advanced Rev 2001;53:381–415. structural imaging, advances in modeling of macro- [9] Pleuvry BJ. Opioid receptors and their ligands: natural and unnatural. Br J Anaesth 1991;66: parameters of PET ligand binding, and efforts in 370–80. large-scale statistical analysis of imaging datasets. [10] Alexander S, Peters J, Mead A, et al. TiPS recep- Suitable SPECT tracers are lacking, but with the cur- tor and ion channel nomenclature supplement rent increase in the number of available PET (or 1999. Trends Pharmacol Sci 1999;19(Suppl 1): PET/CT) cameras and cyclotrons owing to the clin- 62–3. ical success of PET in oncology, PET may become [11] Simmons ML, Chavkin C. Endogenous opioid widespread enough to overcome this. regulation of hippocampal function. Int Rev Many breakthrough scientific discoveries have Neurobiol 1996;39:145–96. been made using opioid imaging. It is to be hoped [12] Bayon A, Rossier J, Mauss A, et al. In vitro re- that, in the coming decade, there will be a more lease of [5-methionine] enkephalin and [5-leu- cine] enkephalin from the rat globus pallidus. widespread application of the available techniques Proc Natl Acad Sci USA 1978;75:3505–6. to patients and an impact in clinical medicine. [13] Neumaier JF, Chavkin C. Release of endogenous opioid peptides displaces [3H] diprenorphine Acknowledgments binding in rat hippocampal slices. Brain Res 1989;493:292–302. The authors would like to thank their colleagues at [14] Wagner JJ, Evans CJ, Chavkin C. Focal stimula- the MRC Clinical Sciences Centre, Hammersmith tion of the mossy fibers releases endogenous Hospital, and the Department of Clinical and dynorphins that bind kappa 1-opioid receptors Experimental Epilepsy, Institute of Neurology, in guinea pig hippocampus. J Neurochem 1991; 57(1):333–43. particularly Dr. Matthias Koepp and Professor [15] Drake CT, Terman GW, Simmons ML, et al. Dy- John Duncan for helpful discussions, Dr. Federico norphin opioids present in dentate granule Turkheimer for proofreading parts of this review, cells may function as retrograde inhibitory neu- and Dr. Matthias Glaser for the drawings of the rotransmitters. J Neurosci 1994;14:3736–50. chemical structures. [16] Corbett A, Paterson S, McKnight A, et al. Dynor- phin (1–8) and dynorphin (1–9) are ligands References for the kappa-subtype of opiate receptor. Na- ture 1982;299:79–81. [1] Pert CB, Snyder SH. Properties of opiate-recep- [17] Chaturvedi K, Christoffers KH, Singh K, et al. tor binding in rat brain. Proc Natl Acad Sci Structure and regulation of opioid receptors. USA 1973;70:2243–7. Biopolymers 2000;55:334–46. [2] Simon EJ, Hiller JM, Edelman I. Stereospecific [18] Simonato M, Romualdi P. Dynorphin and epi- binding of the potent narcotic analgesic (3H) lepsy. Prog Neurobiol 1996;50:557–83. 548 Hammers & Lingford-Hughes

[19] Crain BJ, Chang KJ, McNamara JO. An in vitro [33] Rodrı´guez-Puertas R, Gonza´lez-Maeso J, autoradiographic analysis of mu and delta opi- Meana JJ, et al. Autoradiography of receptor- oid binding in the hippocampal formation of activated G-proteins in post mortem human kindled rats. Brain Res 1987;412:305–10. brain. Neuroscience 2000;96:169–80. [20] Perry DC, Grimes LM. Administration of kainic [34] Zubieta J-K, Dannals RF, Frost JJ. Gender and acid and colchicine alters mu and lambda opi- age influences on human brain mu-opioid re- ate binding in rat hippocampus. Brain Res ceptor binding measured by PET. Am J Psychia- 1989;477:100–8. try 1999;156:842–8. [21] Pfeiffer M, Kirscht S, Stumm R, et al. Heterodime- [35] Cherry SR, Phelps ME. Imaging brain function rization of substance P and mu-opioid receptors with positron emission tomography. In: regulates receptor trafficking and resensitization. Toga AW, Mazziotta JC, editors. Brain mapping: J Biol Chem 2003;278(51):51630–7. the methods. San Diego: Academic Press; 1996. [22] Minnis JG, Patierno S, Kohlmeier SE, et al. Li- p. 191–221. gand-induced mu opioid receptor endocytosis [36] Wienhard K, Wagner R, Heiss W-D. PET. Grund- and recycling in enteric neurons. Neuroscience lagen und Anwendungen der Positronen-Emis- 2003;119(1):33–42. sions-Tomographie. Berlin, Heidelberg, New [23] Lecoq I, Marie N, Jauzac P, et al. Different regu- York: Springer-Verlag; 1989. lation of human delta-opioid receptors by [37] Valk PE, Bailey DL, Townsend DW, et al, edi- SNC-80 [(1)-4-[(alphaR)-alpha-((2S,5R)-4-allyl- tors. Positron emission tomography: basic sci- 2,5-dimethyl-1-piperazinyl)-3-meth oxyben- ence and clinical practice. London, Berlin, zyl]-N,N-diethylbenzamide] and endogenous Heidelberg: Springer-Verlag; 2003. p. 884. enkephalins. J Pharmacol Exp Ther 2004;310(2): [38] Gunn RN, Lammertsma AA, Hume SP, et al. 666–77. Parametric imaging of ligand-receptor binding [24] Chen LE, Gao C, Chen J, et al. Internalization in PET using a simplified reference region and recycling of human mu opioid receptors model. NeuroImage 1997;6(4):279–87. expressed in Sf9 insect cells. Life Sci 2003; [39] Hammers A, Asselin MC, Osman S, et al. To- 73(1):115–28. wards improved test-retest reliability in quanti- [25] Pfeiffer A, Pasi A, Mehraein P, et al. Opiate re- tative ligand PET: [11 C]diprenorphine as an ceptor binding sites in human brain. Brain example. [abstract]. J Cereb Blood Flow Metab Res 1982;248:87–96. 2005;25:S665. [26] Hiller JM, Fan LQ. Laminar distribution of the [40] von Spiczak S, Whone AL, Hammers A, et al. multiple opioid receptors in the human cere- The role of opioids in restless legs syndrome: bral cortex. Neurochem Res 1996;21:1333–45. an [11C]diprenorphine PET study. Brain 2005; [27] Mathieu-Kia AM, Fan LQ, Kreek MJ, et al. Mu-, 128(Pt 4):906–17. delta- and kappa-opioid receptor populations [41] Weeks RA, Cunningham VJ, Piccini P, et al. 11 C- are differentially altered in distinct areas of diprenorphine binding in Huntington’s disease: postmortem brains of Alzheimer’s disease pa- a comparison of region of interest analysis with tients. Brain Res 2001;893(1–2):121–34. statistical parametric mapping. J Cereb Blood [28] Peckys D, Landwehrmeyer GB. Expression of Flow Metab 1997;17:943–9. mu, kappa, and delta opioid receptor messen- [42] Tadokoro M, Jones AKP, Cunningham VJ, et al. ger RNA in the human CNS: a 33P in situ hy- Parametric images of 11 C-diprenorphine bind- bridization study. Neuroscience 1999;88: ing using spectral analysis of dynamic PET 1093–135. images acquired in 3D. In: Uemura K, et al, [29] Pilapil C, Welner S, Magnan J, et al. Autoradio- editors. Quantification of brain function: graphic distribution of multiple classes of opi- tracer kinetics and image analysis in brain oid receptor binding sites in human forebrain. PET. Amsterdam: Excerpta Medica; 1993. p. Brain Res Bull 1987;19:611–5. 289–94. [30] Schadrack J, Willoch F, Platzer S, et al. Opioid re- [43] Turkheimer FE, Hinz R, Gunn RN, et al. Rank- ceptors in the human cerebellum: evidence from shaping regularization of exponential spectral [11C]diprenorphine PET, mRNA expression and analysis for application to functional paramet- autoradiography. Neuroreport 1999;10(3): ric mapping. Phys Med Biol 2003;48(23): 619–24. 3819–41. [31] Kuhar MJ, Pert CB, Snyder SH. Regional distri- [44] Gunn RN, Gunn SR, Turkheimer FE, et al. Pos- bution of opiate receptor binding in monkey itron emission tomography compartmental and human brain. Nature 1973;245:447–50. models: a basis pursuit strategy for kinetic mod- [32] Platzer S, Winkler A, Schadrack J, et al. Autora- eling. J Cereb Blood Flow Metab 2002;22(12): diographic distribution of mu-, delta- and 1425–39. kappa 1-opioid stimulated [35S]guanylyl-50-O- [45] Turkheimer F, Sokoloff L, Bertoldo A, et al. Esti- (gamma-thio)-triphosphate binding in human mation of component and parameter distribu- frontal cortex and cerebellum. Neurosci Lett tions in spectral analysis. J Cereb Blood Flow 2000;283:213–6. Metab 1998;18(11):1211–22. Opioid Imaging 549

[46] Kukreja SL, Gunn RN. Bootstrapped DEPICT [11 C]diprenorphine studies [abstract]. J Cereb for error estimation in PET functional imaging. Blood Flow Metab 2003;23(Suppl):679. Neuroimage 2004;21(3):1096–104. [60] Lee KO, Akil H, Woods JH, et al. Differential [47] Hammers A, Koepp MJ, Labbe´ C, et al. Neocor- binding properties of oripavines at cloned tical abnormalities of [11 C]-flumazenil PET in mu- and delta-opioid receptors. Eur J Pharma- mesial temporal lobe epilepsy. Neurology 2001; col 1999;378(3):323–30. 56:897–906. [61] Frost JJ, Wagner HN Jr, Dannals RF, et al. Imag- [48] Theodore WH, Sato S, Kufta C, et al. Temporal ing opiate receptors in the human brain by pos- lobectomy for uncontrolled seizures: the role of itron tomography. J Comput Assist Tomogr positron emission tomography. Ann Neurol 1985;9:231–6. 1992;32:789–94. [62] Jones AKP, Luthra SK, Maziere B, et al. Re- [49] Hajek M, Antonini A, Leenders KL, et al. Me- gional cerebral opioid receptor studies with siobasal versus lateral temporal lobe epilepsy: [11 C]diprenorphine in normal volunteers. metabolic differences shown in the temporal J Neurosci Methods 1988;23:121–9. lobe shown by interictal 18F-FDG positron [63] Wester HJ, Willoch F, Tolle TR, et al. 6-O-(2- emission tomography. Neurology 1993;43: [18F]fluoroethyl)-6-O-desmethyldiprenorphine 79–86. ([18F]DPN): synthesis, biologic evaluation, and [50] Henry TR, Mazziotta JC, Engel JJ, et al. Quanti- comparison with [11 C]DPN in humans. J Nucl fying interictal metabolic activity in human Med 2000;41(7):1279–86. temporal lobe epilepsy. J Cereb Blood Flow [64] Frost JJ, Douglass KH, Mayberg HS, et al. Metab 1990;10:748–57. Multicompartmental analysis of [11 C]-carfenta- [51] Kiebel SJ, Ashburner J, Poline JB, et al. MRI and nil binding to opiate receptors in humans PET coregistration: a cross validation of statisti- measured by positron emission tomography. cal parametric mapping and automated image J Cereb Blood Flow Metab 1989;9(3): registration. Neuroimage 1997;5:271–9. 398–409. [52] Koepp MJ, Richardson MP, Labbe´ C, et al. 11 C- [65] Lammertsma AA, Hume SP. Simplified refer- flumazenil PET, volumetric MRI, and quantita- ence tissue model for PET receptor studies. tive pathology in mesial temporal lobe epilepsy. Neuroimage 1996;4:153–8. Neurology 1997;49(3):764–73. [66] Endres CJ, Bencherif B, Hilton J, et al. Quantifi- [53] Knowlton RC, Laxer KD, Klein G, et al. In vivo cation of brain mu-opioid receptors with hippocampal glucose metabolism in mesial [11 C]carfentanil: reference-tissue methods. temporal lobe epilepsy. Neurology 2001;57: Nucl Med Biol 2003;30(2):177–86. 1184–90. [67] Henriksen G, Platzer S, Marton J, et al. Syn- [54] Labbe´ C, Koepp MJ, Ashburner J, et al. Abso- theses, biological evaluation, and molecular lute PET quantification with correction for modeling of 18F-labeled 4-anilidopiperidines partial volume effects within cerebral struc- as mu-opioid receptor imaging agents. J Med tures. In: Carson C, Daube-Witherspoon M, Chem 2005;48(24):7720–32. Herscovitch P, editors. Quantitative functional [68] Madar I, Lever JR, Kinter CM, et al. Imaging of brain imaging with positron emission tomog- delta opioid receptors in human brain by N10- raphy. San Diego: Academic Press; 1998. p. ([11 C]methyl)naltrindole and PET. Synapse 59–66. 1996;24(1):19–28. [55] Rousset OG, Ma Y, Evans AC. Correction for [69] Madar I, Lesser RP, Krauss G, et al. Imaging of ð- partial volume effects in PET: principle and val- and m-opioid receptors in temporal lobe epi- idation. J Nucl Med 1998;39:904–11. lepsy by positron emission tomography. Ann [56] Aston JAD, Cunningham VJ, Asselin MC, et al. Neurol 1997;41:358–67. Positron emission tomography partial volume [70] Smith JS, Zubieta JK, Price JC, et al. Quantifica- correction: estimation and algorithms. J Cereb tion of delta-opioid receptors in human brain Blood Flow Metab 2002;22:1019–34. with N10-([11 C]methyl) naltrindole and posi- [57] Richardson MP, Friston KJ, Sisodiya SM, et al. tron emission tomography. J Cereb Blood Cortical grey matter and benzodiazepine recep- Flow Metab 1999;19(9):956–66. tors in malformations of cortical development: [71] Rothman RB, Bykov V, Reid A, et al. A brief a voxel-based comparison of structural and study of the selectivity of norbinaltorphimine, functional imaging data. Brain 1997;120(Pt 11): (-)-cyclofoxy, and (1)-cyclofoxy among opioid 1961–73. receptor subtypes in vitro. Neuropeptides [58] Richards ML, Sadee W. In vivo opiate receptor 1988;12(3):181–7. binding of oripavines to mu, delta and kappa [72] Theodore WH, Carson RE, Andreasen P, et al. sites in rat brain as determined by an ex vivo la- PET imaging of opiate receptor binding in hu- beling method. Eur J Pharmacol 1985;114(3): man epilepsy using [18F]cyclofoxy. Epilepsy 343–53. Res 1992;13:129–39. [59] Asselin M-C, Hammers A, Sethi FN, et al. [73] Cohen RM, Andreason PJ, Doudet DJ, et al. Definition of a reference region for PET/ Opiate receptor avidity and cerebral blood 550 Hammers & Lingford-Hughes

flow in Alzheimer’s disease. J Neurol Sci 1997; changes in opioid receptor binding: a PET study 148(2):171–80. with 11 C-diprenorphine. AJNR Am J Neurora- [74] Cohen RM, Carson RE, Sunderland T. Opiate re- diol 1999;20(4):686–90. ceptor avidity in the thalamus is sexually dimor- [88] Sprenger T, Willoch F, Miederer M, et al. Opioi- phic in the elderly. Synapse 2000;38(2):226–9. dergic changes in the pineal gland and hypo- [75] Talbot PS, Narendran R, Butelman ER, et al. thalamus in cluster headache: a ligand PET 11 C–GR103545, a radiotracer for imaging study. Neurology 2006;66(7):1108–10. kappa-opioid receptors in vivo with PET: syn- [89] May A, Ashburner J, Buchel C, et al. Correlation thesis and evaluation in baboons. J Nucl Med between structural and functional changes in 2005;46(3):484–94. brain in an idiopathic headache syndrome. [76] Ravert HT, Scheffel U, Mathews WB, et al. Nat Med 1999;5:836–8. [(11)C]-GR89696, a potent kappa opiate recep- [90] Sprenger T, Valet M, Boecker H, et al. Opioider- tor radioligand; in vivo binding of the R and S gic activation in the medial pain system after enantiomers. Nucl Med Biol 2002;29(1): heat pain. Pain 2006;122(1–2):63–7. 47–53. [91] Jones AK, Cunningham VJ, Ha-Kawa S, et al. [77] Jones AK, Qi LY, Fujirawa T, et al. In vivo distribu- Changes in central opioid receptor binding in tion of opioid receptors in man in relation to the relation to inflammation and pain in patients cortical projections of the medial and lateral with rheumatoid arthritis. Br J Rheumatol pain systems measured with positron emission 1994;33(10):909–16. tomography. Neurosci Lett 1991;126(1):25–8. [92] Jones AK, Kitchen ND, Watabe H, et al. Mea- [78] Baumgartner U, Buchholz HG, Bellosevich A, surement of changes in opioid receptor binding et al. High opiate receptor binding potential in vivo during trigeminal neuralgic pain using in the human lateral pain system. Neuroimage [11 C] diprenorphine and positron emission to- 2006;30(3):692–9. mography. J Cereb Blood Flow Metab 1999; [79] Zubieta JK, Heitzeg MM, Smith YR, et al. COMT 19(7):803–8. val158met genotype affects mu-opioid neuro- [93] Bencherif B, Fuchs PN, Sheth R, et al. Pain acti- transmitter responses to a pain stressor. Science vation of human supraspinal opioid pathways 2003;299(5610):1240–3. as demonstrated by [11 C]-carfentanil and posi- [80] Berthele A, Platzer S, Jochim B, et al. COMT tron emission tomography (PET). Pain 2002; Val108/158Met genotype affects the mu-opioid 99(3):589–98. receptor system in the human brain: evidence [94] Zubieta JK, Smith YR, Bueller JA, et al. Regional from ligand-binding, G-protein activation and mu opioid receptor regulation of sensory and preproenkephalin mRNA expression. Neuro- affective dimensions of pain. Science 2001; image 2005;28(1):185–93. 293(5528):311–5. [81] Smith YR, Stohler CS, Nichols TE, et al. Prono- [95] Zubieta JK, Smith YR, Bueller JA, et al. m-opioid ciceptive and antinociceptive effects of estradiol receptor-mediated antinociceptive responses through endogenous opioid neurotransmission differ in men and women. J Neurosci 2002; in women. J Neurosci 2006;26(21):5777–85. 22(12):5100–7. [82] Zubieta JK, Ketter TA, Bueller JA, et al. Regulation [96] Zubieta JK, Bueller JA, Jackson LR, et al. Placebo of human affective responses by anterior cingu- effects mediated by endogenous opioid activity late and limbic mu-opioid neurotransmission. on mu-opioid receptors. J Neurosci 2005; Arch Gen Psychiatry 2003;60(11):1145–53. 25(34):7754–62. [83] Liberzon I, Zubieta JK, Fig LM, et al. mu-Opioid [97] Tortella FC. Endogenous opioid peptides and receptors and limbic responses to aversive emo- epilepsy: quieting the seizing brain? Trends tional stimuli. Proc Natl Acad Sci USA 2002; Pharmacol Sci 1988;9:366–72. 99(10):7084–9. [98] Koepp MJ, Duncan JS. PET: opiate neurorecep- [84] Apkarian AV, Bushnell MC, Treede RD, et al. tor mapping. Adv Neurol 2000;83:145–56. Human brain mechanisms of pain perception [99] Hammers A. Flumazenil PET and other ligands and regulation in health and disease. Eur J for functional imaging. Neuroimaging Clin N Pain 2005;9(4):463–84. Am 2004;14(3):537–51. [85] Jones AK, Watabe H, Cunningham VJ, et al. [100] Mayberg HS, Sadzot B, Meltzer CC, et al. Quan- Cerebral decreases in opioid receptor binding tification of mu and non-mu receptors in tem- in patients with central neuropathic pain mea- poral lobe epilepsy using positron emission sured by [11 C]diprenorphine binding and PET. tomography. Ann Neurol 1991;30:3–11. Eur J Pain 2004;8(5):479–85. [101] Bartenstein PA, Prevett MC, Duncan JS, et al. [86] Willoch F, Tolle TR, Wester HJ, et al. Central Quantification of opiate receptors in two pa- poststroke pain and reduced opioid receptor tients with mesiobasal temporal lobe epilepsy, binding within pain processing circuitries: before and after selective amygdalohippocam- a[11 C]diprenorphine PET study. Pain 2004; pectomy, using positron emission tomography. 108(3):213–20. Epilepsy Res 1994;18(2):119–25. [87] Willoch F, Schindler F, Wester HJ, et al. Central [102] Hammers A, Koepp MJ, Richardson MP, et al. pain after pontine infarction is associated with [11 C]-diprenorphine PET in malformations of Opioid Imaging 551

cortical development (MCD) [abstract]. Epilep- [118] Kreek MJ, LaForge KS, Butelman E. Pharmaco- sia 2001;42(S7):100. therapy of addictions. Nat Rev Drug Discov [103] Meltzer CC, Zubieta JK, Links JM, et al. MR- 2002;1(9):710–26. based correction for brain PET measurements [119] Zubieta JK, Gorelick DA, Stauffer R, et al. In- for heterogeneous gray matter radioactivity dis- creased mu opioid receptor binding detected tribution. J Cereb Blood Flow Metab 1996;16: by PET in cocaine-dependent men is associated 650–8. with cocaine craving. Nat Med 1996;2(11): [104] Frost JJ, Mayberg HS, Fisher RS, et al. Mu-opiate 1225–9. receptors measured by positron emission to- [120] Gorelick DA, Kim YK, Bencherif B, et al. Imag- mography are increased in temporal lobe epi- ing brain mu-opioid receptors in abstinent co- lepsy. Ann Neurol 1988;23:231–7. caine users: time course and relation to [105] Koepp MJ, Richardson MP, Brooks DJ, et al. Fo- cocaine craving. Biol Psychiatry 2005;57(12): cal cortical release of endogenous opioids dur- 1573–82. ing reading-induced seizures. Lancet 1998;352: [121] Lingford-Hughes A, Nutt D. Neurobiology of 952–5. addiction and implications for treatment. Br J [106] Koepp MJ, Duncan JS. PET: opiate neurorecep- Psychiatry 2003;182:97–100. tor mapping. In: Henry TR, Duncan JS, [122] Lingford-Hughes AR, Welch S, Nutt DJ. Evi- Berkovic SF, editors. Functional imaging in the dence-based guidelines for the pharmacologi- epilepsies. Philadelphia: Lippincott Williams cal management of substance misuse, & Wilkins; 2000. p. 145–55. addiction and comorbidity: recommendations [107] Prevett MC, Cunningham VJ, Brooks DJ, et al. from the British Association for Psychophar- Opiate receptors in idiopathic generalised epi- macology. J Psychopharmacol 2004;18(3): lepsy measured with 11 C-diprenorphine and 293–335. PET. Epilepsy Res 1994;19:71–7. [123] Heinz A, Reimold M, Wrase J, et al. Correlation [108] Bartenstein PA, Duncan JS, Prevett MC, et al. In- of stable elevations in striatal mu-opioid recep- vestigation of the opioid system in absence tor availability in detoxified alcoholic patients seizures with positron emission tomography. with alcohol craving: a positron emission to- J Neurol Neurosurg Psychiatry 1993;56: mography study using carbon 11-labeled car- 1295–302. fentanil. Arch Gen Psychiatry 2005;62(1): [109] Brooks DJ. Neuroimaging in Parkinson’s dis- 57–64. ease. NeuroRx 2004;1(2):243–54. [124] Zubieta J, Greenwald MK, Lombardi U, et al. [110] Burn DJ, Rinne JO, Quinn NP, et al. Striatal opi- Buprenorphine-induced changes in mu-opioid oid receptor binding in Parkinson’s disease, receptor availability in male heroin-dependent striatonigral degeneration and Steele-Richard- volunteers: a preliminary study. Neuropsycho- son-Olszewski syndrome, a [11 C]diprenorphine pharmacology 2000;23(3):326–34. PET study. Brain 1995;118(Pt 4):951–8. [125] Williams TM, Daglish MRC, Lingford-Hughes [111] Piccini P, Weeks RA, Brooks DJ. Alterations in AR, et al. Increased availability of opioid recep- opioid receptor binding in Parkinson’s disease tors in early abstinence from opioid depen- patients with levodopa-induced dyskinesias. dence: an [11 C]diprenorphine PET study. Ann Neurol 1997;42(5):720–6. Submitted. [112] Brooks DJ, Piccini P, Turjanski N, et al. Neuro- [126] Pilowsky LS. Probing targets for antipsychotic imaging of dyskinesia. Ann Neurol 2000; drug action with PET and SPET receptor im- 47(4, Suppl 1):S154–8 [discussion: S158–9]. aging. Nucl Med Commun 2001;22(7): [113] Whone AL, Von Spiczak S, Edwards M, et al. 829–33. Opioid binding in DYT1 primary torsion dysto- [127] Kling MA, Carson RE, Borg L, et al. Opioid nia: an 11 C-diprenorphine PET study. Mov Dis- receptor imaging with positron emission ord 2004;19(12):1498–503. tomography and [(18)F]cyclofoxy in long- [114] Weeks RA, Turjanski N, Brooks DJ. Tourette’s term, methadone-treated former heroin ad- syndrome: a disorder of cingulate and orbito- dicts. J Pharmacol Exp Ther 2000;295(3): frontal function? QJM 1996;89(6):401–8. 1070–6. [115] Mueller-Gaertner HW, Mayberg HS, Ravert HT, [128] Melichar JK, Hume SP, Williams TM, et al. et al. Mu opiate receptor binding in amygdala Using [11 C]diprenorphine to image opioid in Alzheimer’s disease: in vivo quantification receptor occupancy by methadone in opioid by 11 C carfentanil and PET [abstract]. J Cereb addiction: clinical and preclinical studies. Blood Flow Metab 1991;11(Suppl 2):S20. J Pharmacol Exp Ther 2005;312(1):309–15. [116] Nordberg A. PET imaging of amyloid in Alz- [129] Melichar JK, Myles JS, Eap CB, et al. Using sac- heimer’s disease. Lancet Neurol 2004;3(9): cadic eye movements as objective measures of 519–27. tolerance in methadone dependent individuals [117] Villemagne PS, Dannals RF, Ravert HT, et al. during the hydromorphone challenge test. Ad- PET imaging of human cardiac opioid recep- dict Biol 2003;8(1):59–66. tors. Eur J Nucl Med Mol Imaging 2002; [130] Goodman CB, Emilien B, Becketts K, et al. 29(10):1385–8. Downregulation of mu-opioid binding sites 552 Hammers & Lingford-Hughes

following chronic administration of neuropep- concentrations, and antagonist blockade in tide FF (NPFF) and morphine. Peptides 1996; heroin-dependent volunteers. Neuropsycho- 17(3):389–97. pharmacology 2003;28(11):2000–9. [131] Zadina JE, Kastin AJ, Harrison LM, et al. Opiate [133] Holmes CJ, Hoge R, Collins L, et al. Enhance- receptor changes after chronic exposure to ago- ment of MR images using registration for signal nists and antagonists. Ann N Y Acad Sci 1995; averaging. J Comput Assist Tomogr 1998;18: 757:353–61. 192–205. [132] Greenwald MK, Johanson CE, Moody DE, et al. [134] Noppeney U, Price CJ. An FMRI study of Effects of buprenorphine maintenance dose syntactic adaptation. J Cogn Neurosci 2004; on mu-opioid receptor availability, plasma 16(4):702–13.