Compensation for Cranial Spill-In Into theCerebellum Improves Quantitation of

Striatal Dopamine D2/3 Receptors in RatsWith Prolonged [18F]-DMFP Infusions

ERIK MILLE,1 PAUL CUMMING,1 AXEL ROMINGER,1 CHRISTIAN LA FOUGERE,1 KLAUS TATSCH,2

BJORN WANGLER,1,3 PETER BARTENSTEIN,1 AND GUIDO BONING1*1Department of Nuclear Medicine, University Hospital of the Ludwig-Maximilians University of Munich, Munich,

Germany2Department of Nuclear Medicine, Municipal Hospital Karlsruhe Inc., Karlsruhe, Germany

3Molecular Imaging and Radiochemistry, Institute of Clinical Radiology and Nuclear Medicine, University MedicalCenter Mannheim, Medical Faculty Mannheim, Heidelberg University, Germany

KEY WORDS bolus; infusion; steady-state; dopamine receptors; amphetamine;[18F]-DMFP

ABSTRACT The condition of steady-state receptor binding in positron emission to-mography (PET) studies is best obtained through the use of a bolus plus steady-infu-sion paradigm. This is a particularly important consideration in the context of in vivocompetition studies, where a pharmacological challenge can be administered duringthe interval of steady-state ligand binding, as in the case of [11C]-raclopride studieswith amphetamine challenge. However, the short half-life of 11C imposes limits on thepractical duration of constant infusions. Therefore, we chose to test [18F]-DMFP as atracer for dopamine D2/3 receptors in rat striatum in the paradigm. Using a conven-tional bolus injection, the [18F]-DMFP BPND was 3.8 in striatum of anesthetized rats.When followed by a constant infusion, we obtained quasi-stable BPND estimates of 4.5within an interval of 45 min. During infusions lasting up to 4 h, BPND declined pro-gressively. This seemed due to the progressive spill-in of radioactivity from the cra-nium to the cerebellum reference region, despite optimized iterative reconstruction ofthe images. Therefore, we propose a new concept of compensation for this spill-ineffect using pharmacokinetic considerations, without requiring high-resolution ana-tomical images. Challenge with amphetamine (1 and 4 mg/kg) evoked an �25% reduc-tion in BPND. There was no clear evidence of dose-dependence in the striatal-bindingchanges, despite the considerably greater physiological effect, as documented by ECG.Thus, the general applicability of the bolus plus infusion method with [18F]-DMFP forsmall animal studies is impeded by the substantial labeling of the cranium. The cra-nial uptake was linear, indicating first-order kinetics for the enzymatic defluorinationof the tracer. Based on this phenomenon, we developed an analytic method compensat-ing for the effects of progressive cranial labeling on the estimation of specific bindingin striatum. Synapse 66:705–713, 2012. VVC 2012 Wiley Periodicals, Inc.

INTRODUCTION

The availability of binding sites for the benzamide

class of dopamine D2/3 receptor antagonists in living

brain is influenced by competition from endogenous

dopamine in the interstitial fluid (Laruelle, 2000;

Rominger et al., 2010c). When the availability of

striatal-binding sites is measured first in a baseline

condition, and again after challenge with ampheta-

mine, or some other treatment altering dopamine

release, the change in receptor availability revealsaltered dopamine tonus. This in vivo competition par-adigm has furnished new insights into the presynap-

*Correspondence to: Guido Boning, Department of Nuclear Medicine, March-ioninistr. 15, 81377 Munich, Germany.E-mail: Guido.Boening@med.uni-muenchen.de

Received 16 December 2011; Accepted 21 March 2012

DOI 10.1002/syn.21558

Published online 27 March 2012 in Wiley Online Library (wileyonlinelibrary.com).

VVC 2012 WILEY PERIODICALS, INC.

SYNAPSE 66:705–713 (2012)

tic abnormalities of the dopamine system in schizo-phrenia (Nikisch et al., 2010; Patel et al., 2010) andsubstance abuse (Volkow et al., 2004, 2009). Manypositron emission tomography (PET) occupancy stud-ies use a bolus tracer method, in which the equilib-rium between bound and unbound tracer is onlytransient. This property of bolus methods limits theinterpretation of results from the competition model,which incorrectly assumes that pharmacologicallyevoked changes in dopamine release occur as a time-invariant step function. Consequently, the kinetics ofa pharmacodynamic effect at dopamine D2/3 receptorscannot be ascertained. As an alternate approach bet-ter satisfying the requirement for a steady-state invivo, methods with continuous infusion of the tracerhave been developed (Villemagne et al., 1999). Thisapproach has been established in human PET studieswith the dopamine D2/3 antagonist [11C]-raclopride(Carson et al., 1997; Watabe et al., 2000) and a dopa-mine agonist ligand (Seneca et al., 2008). In general,a small amount of tracer is first administered as abolus and followed by a continuous infusion, so as toobtain a steady-state for the specific binding, which iscalculated as the simple ratio between the radioactiv-ity concentration in the region of interest to that in areference region nearly devoid of specific binding.

Despite the several theoretical advantages of thebolus plus infusion paradigm, it has not found widedeployment, in part because of the requirement foran on-site cyclotron for provision of [11C]-labeled trac-ers and also because the short physical half-life of theisotope (20 min) imposes practical limits of the maxi-mum possible duration of infusions. In the particularcase of [11C]-raclopride PET studies, infusions haveextended for only 90 min, a duration that is clearlyinsufficient to monitor the temporal dynamics of am-phetamine effects on dopamine release. On the otherhand, the physical half-life of 18F (110 min) is suffi-cient for central production and distribution to dis-tant PET centers of benzamide ligands such as [18F]-fallypride and its lower affinity congener [18F]-DMFP(Mukherjee et al., 1996). The competition model hasbeen established for these tracers in mouse striatumusing conventional tracer bolus methods (Romingeret al., 2010a,c), but analysis of mouse brain imageswas complicated by substantial defluorination, whichresulted in [18F]-fluoride uptake in the cranium. Ifthe problems arising from bone labeling could besolved, [18F]-fallypride and [18F]-DMFP could poten-tially be used for bolus plus infusion studies lastingseveral hours. In the present study, we first estab-lished a bolus method for the assay by micro-PET of[18F]-DMFP specific binding in rat striatum. We thendeveloped a method for obtaining steady-state bindingin rat striatum, using a bolus injection followed by acontinuous infusion lasting for 4 h. To compensate forthe effects resulting from degradation of the tracer

and progressive bone-uptake, we analytically deriveda new computational method to estimate the magni-tude of the spill-in from the cranium into the refer-ence region based upon pharmacokinetic observationsand considerations. During steady-state infusions, wetested the vulnerability of [18F]-DMFP binding to aspecific pharmacological challenge, that is, adminis-tration of amphetamine (1 and 4 mg/kg, i.v.).

MATERIALS AND METHODSAnimal handling

Female Sprague–Dawley rats (Harlan Laboratories,NL) weighting �250 g were housed in pairs in indi-vidually ventilated cages (IVC, Tecniplast, IT) with a12-h day/night cycle. The animals had access to foodand water ad libitum. Anesthesia was induced with3.5% isoflurane in oxygen and maintained by theplacement of the nose in a cone administering 1.5%isoflurane in oxygen at a constant flow rate of 2 l/min. An intravenous line was then placed into a tailvein. All experiments were conducted in accordanceto the German laws for animal experiments and ani-mal protection.

Physiological monitoring

A small animal monitoring system (BioVet, m2mImaging Corp.) was used to record continuously thephysiological state during the entire course of thePET measurements. An ECG was derived from leadII using subcutaneously attached electrode pins. Ab-dominal respiratory motion was registered with apressure sensitive cushion and the core body temper-ature measured with a rectal temperature sensor.

PET measurements

[18F]-DMFP was synthesized according to a proce-dure published earlier (Rominger et al., 2010c); spe-cific activity was in the range 60–200 GBq/lmol. PETdata were acquired in list-mode with a dedicatedsmall animal tomograph (Inveon DPET, PreclinicalSolutions, Siemens Healthcare Molecular Imaging).The rats were positioned on a thermostatically con-trolled heating pad (388C) and immobilized by place-ment of pins from a custom-made head-holder intothe external auditory meatus. The dynamic emissionrecordings were started upon tracer injection. Aftercompletion of the entire emission recordings, brieftransmission scans were obtained with a built-in 57Cotransmission source.

Bolus injection experiments

We first measured the [18F]-DMFP BPND in stria-tum (N 5 12) using an established bolus injectionmethod (Constantinescu et al., 2011; Rominger et al.,2010c). Here, list mode emission recordings lasting 90

706 E. MILLE ET AL.

Synapse

min were initiated upon administration of [18F]-DMFP (47 6 4 MBq), delivered in 1 ml of saline as aslow bolus lasting 30 s, followed by flushing of theline with 0.3 ml saline.

Continuous infusion experiments

Approximately 100 MBq [18F]-DMFP (98.3 6 12MBq) mixed in 5 ml of physiological electrolyte solu-tion was administered using a syringe-pump (SP100i,World Precision Instruments, Sarasota, FL) connectedto the intravenous catheter. An empirically deter-mined bolus plus infusion protocol was intended toattain rapidly a constant [18F]-DMFP concentrationin brain. An initial 28.4 MBq [18F]-DMFP (1.42 ml)was infused during 1.7 min (50 ml/h), followed by pro-longed infusion of the remaining 71.6 MBq [18F]-DMFP (3.58 ml) during the subsequent 4 h (0.89 ml/h). The syringe was shielded in a tungsten cylinder(5-mm lead equivalent) to provide radiation protectionfor the personnel and also to minimize random detec-tion by the nearby PET. The temperature of theradiotracer was maintained at 388C by heating foilplaced around the shield. Pharmacological challengewas administered at 95 min after the start of theexperiment. The challenge consisted of manual bolusinjections of 0.5 ml saline (N 5 5), 1 mg/kg (N 5 5),or 4 mg/kg (N 5 5) d-amphetamine sulfate (Sigma-Aldrich, Germany), likewise in 0.5 ml saline.

Data processing and analysis

Analysis of physiological data

The recordings of physiological data consisted oftabular-separated values sampled each millisecond.Cardiac and respiratory motions were determined bysearching for peaks in the relevant signals. To thisend, we applied a centered moving average filter (2-min width) for simultaneously calculating the signalmeans and standard deviations. A heart beat (Rpeak) was registered when the peak amplitude of theECG signal exceeded the floating mean amplitude bymore than 1.3 standard deviations at a time at least100 ms (‘‘inhibit’’) after the previous R-peak. For re-spiratory data, the inhibit was set to 300 ms. Meancardiac and respiratory rates were calculated for eachPET frame, as defined below. Physiological parame-ters of each animal were calculated relative to thebaseline condition, defined as the mean prevailingduring 45–90 min after start of the PET recording(i.e., before the pharmacological challenge).

PET image reconstruction

PET list-mode data were spatially and temporallyrebinned into dynamic 3D sinograms. For the bolusstudies, recordings were divided into 28 frames (6 320, 6 3 30, 5 3 60, 2 3 150, 3 3 300, and 6 3 600 s),

whereas recordings for the continuous infusion experi-ments were divided into 32 frames of 7.5 min each.Dynamic image sequences were generated using fouriterations of the 3D-ordered subset expectation maximi-zation reconstruction (Hudson and Larkin, 1994; Qiet al., 1998), followed by 32 iterations of the 3D ordi-nary-Poisson maximum a priori (OP-MAP) reconstruc-tion (Mumcuoglu et al., 1994; Qi and Leahy, 2000;zoom 5 1.42, b 5 0.01), with attenuation and scattercorrections (Watson, 2000; Watson et al., 1996). Thisprocedure, based upon our earlier optimization of spa-tial resolution and image smoothness [(Rominger et al.,2010b) and unpublished], gave 3D frames consisting of128 3 128 3 159 voxels (0.51 mm 3 0.51 mm 3 0.80mm). The final reconstructed transaxial field-of-viewwas 64 mm in diameter and 127 mm in length. Recon-struction time for each 3D frame was 1 h using a high-throughput Linux cluster (Transtec GmbH, Germany)with a total of 64 CPU cores (AMD Opteron, 64 bit).

Spatial normalization of image data

The dynamic image sequences were converted fromthe original MicroPET format to the minc format(http://www.bic.mni.mcgill.ca/ServicesSoftware/MINC).Mean images of the first 10 min were manually nor-malized to a cryosection atlas of the rat brain (Peder-sen et al., 2007) using the program register, asdescribed previously (la Fougere et al., 2010a;Rominger et al., 2010c), and the transformation wasthen applied to the entire dynamic image sequence.Voxel-wise radioactivity concentrations were scaled to50 MBq [18F]-DMFP per 250 g body weight for thesimple bolus studies or 100 MBq per 250 g bodyweight for the continuous infusion studies.

Analysis of the bolus experiments

The reference tissue volume of interest (VOI) wasdefined as an ellipsoid (25.3 mm3) placed at the cen-ter of the cerebellum in atlas coordinates. The timeactivity curve (TAC) extracted from cerebellum wasused for the voxel-wise calculation of BPND in thenative space by the method of Logan (Logan et al.,1990) as implemented in PMOD (PMOD Technologies,Switzerland). The native space maps were then con-verted to minc format and resampled to the brainatlas space as described earlier. The mean voxel-wisevalues were then calculated within spheroidal VOIs(14.0 mm3) placed over the bilateral striatum. Anadditional VOI was placed within the lower occipitalbone for the calculation of a skull TAC.

Analysis of the continuous infusion experiments

The voxel-wise [18F]-DMFP-binding ratio, which isequal to BPND when steady-state prevails (Inniset al., 2007), was calculated as shown in Eq. 1, below.

707AMPHETAMINE EFFECTS ON [18F]-DMFP BINDING IN RATS

Synapse

Here, the mean voxel-wise activity concentration incerebellum VOI, Cref (voi, t), was first calculated foreach frame. Then, a denoised approximation Cref (t) ofthis cerebellum TAC was estimated by linear regres-sion and used for the subsequent computations:

BPNDðvoxel; tÞ ¼ Cðvoxel; tÞ � ~Cref ðtÞ~Cref ðtÞ

: ð1Þ

The VOI TACs and striatal-binding estimates weremeasured as described for analysis of bolus experi-ments mentioned earlier. For each rat, the baselineBPND was defined as the mean during the interval of45–90 min after initiation of the tracer infusion. Theindividual BPND curves were then normalized tothese baseline values. The postchallenge conditionwas defined as the mean normalized measurementsduring the interval 135–180 min. Finally, the relativecurves and postchallenge values were averaged foreach group.

Compensation for spill-in

In contrast to methods for explicit partial volumecorrection, which use high-resolution anatomicalimaging and segmentation (Fazio and Perani, 2000;Morris et al., 1999; Rousset et al., 1998, 2008; Shida-hara et al., 2009), we propose an algorithm to esti-mate and compensate for spill-in from the craniumbased upon pharmacokinetic considerations. To thisend, we distinguish between a radioactivity concen-tration as measured, Cm, and the corresponding con-centration with compensation for spill-in, Cc. If thetotal concentration measured in the reference regionas a function of time is the composite of the ‘‘true’’concentration and a spill-in component arising frombone, Cm

bone, we can then estimate the magnitude ofthat spill-in through a scaling factor, a,

Cmref ðvoi; tÞ ¼ Cc

ref ðvoi; tÞ þ a � Cmboneðvoi; tÞ: ð2Þ

Here, we exploit the fact that the geometric effectand the corresponding scaling factor a are time-invar-iant, depending only on the size and location of thereference and bone VOIs, relative to the spatial reso-lution of the tomograph. Hence, an estimate of thebinding potential, BPc

ND, based on a reference tissueconcentration being compensated for spill-in, can beobtained by combining Eqs. 1 and 2:

BPcNDðvoxel; tÞ

¼ Cmðvoxel; tÞ � Cmref ðvoi; tÞ � a � Cm

boneðvoi; tÞ� �

Cmref ðvoi; tÞ � a � Cm

boneðvoi; tÞ: ð3Þ

In the present continuous infusion studies, weobserve that bone-labeling increases as a linear func-tion of tracer infusion time (see Fig. 1). This empirical

finding, given nearly unchanging plasma inputimposed by the constant infusion, implies that themetabolic process underlying bone-labeling follows afirst-order pharmacokinetic model:

d

dtCboneðtÞ � CplasmaðtÞ: ð4Þ

This assumption can be confirmed graphically,because the slope of bone uptake is not constant dur-ing bolus experiments (Fig. 1), but declines with time.Furthermore, the implicit rate constants linkingCplasma and Cref can be ignored as soon as steady-state is attained during a constant infusion. Thefollowing approximation can then be retrieved fromEqs. 2 and 4,

Cmref ðvoi; tÞ � a � Cm

boneðvoi; tÞ ¼ b � ddt

Cmboneðvoi; tÞ þ g;

ð5Þ

where the coefficients (a, b, and g) can be determinedby linear regression from the measured TACs. Focus-ing on a, the coefficient relevant for compensating thespill-in, the regression analysis for all animals yieldeda median magnitude of 0.9%, which seems entirelyconsistent with the relative magnitudes of bone radio-activity and signal increases in cerebellum, asdepicted in Figure 2B:

Ccref ðvoi; tÞ ¼ Cm

ref ðvoi; tÞ � 0:009 � Cmboneðvoi; tÞ: ð6Þ

Noise reduction, as introduced above for the referenceVOI, Cref (t), can now be applied to produce adenoised and spill-in compensated reference tissueconcentration Cc

ref from Eq. 6, which reforms Eq. 3 to

BPcNDðvoxel; tÞ ¼

Cmðvoxel; tÞ � ~Ccref ðtÞ

~Ccref ðtÞ

: ð7Þ

Now, the compensated binding potential is calculateddirectly, without influence of spill-in from the craniuminto the reference VOI.

RESULTS

The mean striatal BPND for (N 5 12) bolus experi-ments (3.8 6 0.9) and the mean baseline BPND for (N5 15) constant infusion experiments (4.5 6 1.4) didnot differ significantly (P 5 0.15, two-tailed t-test). Inall [18F]-DMFP recordings, there was substantial andprogressive uptake of radioactivity throughout theskeleton; the ratio of radioactivity in cranium to thatin cerebellum reached 20:1 at 90 min. At the end of4-h infusions, this ratio attained 40:1 (Fig. 1). Thesemilog line profiles (Fig. 2) show that spill-in at theend of 4 h increases the measured signal by �10% atthe center of the cerebellum VOI and by 20% at theedges.

708 E. MILLE ET AL.

Synapse

Amphetamine treatment evoked a dose-dependentincrease in normalized heart rate (Fig. 3), whichincreased by �5% after challenge with 1 mg/kg (one-tailed t-test; P < 1025) and by 20% after 4 mg/kg dose

(P < 1026); this dose effect was statistically signifi-cant (P < 1026). The heart rate remained constant inthe control group. There were no significant treat-ment effects on respiratory rate (not shown), but the

Fig. 2. Location of the cerebellar VOI and line-profile (A) and line profiles on semilog scale at varioustime points (B) to illustrate the bone uptake in one representative animal. The VOI is represented by thegrey box.

Fig. 1. Mean normalized [18F]-DMFP TACs in striatum (circles),cerebellum (squares), and bone (triangles) from the conventionalbolus injection (A) and from continuous infusion protocols in whichrats were challenged with (B) saline, (C) 1 mg/kg amphetamine,

and (D) 4 mg/kg amphetamine. Each symbol represents the meanobservations in 12 rats (A) or five rats (B–D) and correspondingstandard deviations. Arrows represent the injection of the chal-lenge.

709AMPHETAMINE EFFECTS ON [18F]-DMFP BINDING IN RATS

Synapse

amphetamine treatments evoked a hyperthermia of�2–38C.

Relative to the baseline condition, mean uncompen-sated BPND declined continuously in the controlgroup, falling by 13% in the course of 3 h of continu-ous tracer infusion (Fig. 4A). Both amphetaminedoses evoked similar 30% declines in uncompensatedBPND (Table I). In contrast, the magnitude of BPC

ND

was stable in the control group throughout the 4-hinfusions (Fig. 4B) and tended to exceed the uncom-pensated BPND although this difference was not sig-nificant (P 5 0.30, two-tailed t-test). Amphetaminetreatment evoked 22–25% reductions in BPC

ND. Timeseries of mean BPC

ND are shown in Figure 5 at repre-sentative time points before and after the challenge.The mean BPC

ND maps before and after treatment

are summarized in Figure 6 as a superposition ontothe histological atlas for rat brain.

DISCUSSION

We found very close concordance between [18F]-DMFP BPND estimates obtained by Logan analysis ofconventional bolus data and by simple ratio analysisof bolus plus continuous infusion data. Present esti-mates of the magnitude of BPND in rat striatumsomewhat exceed those reported earlier for thisligand in mouse striatum (Rominger et al., 2010c)and in binding ratio studies of the human brain (laFougere et al., 2010b). The quantitation of [18F]-DMFP binding in rat striatum was considerably hin-dered by the progressive uptake of radioactivity inthe skeleton, which is almost certainly due to libera-tion of [18F]-fluoride (Tipre et al., 2006). We have ear-lier found that labeling of the mouse cranium waseven more extensive in PET studies with [18F]-DMFPthan for the case of its high-affinity congener [18F]-fallypride and have formally demonstrated the extentof spill-in relative to radioactivity measurements intissue samples (Rominger et al., 2010c). Especially, inthis study, which entailed very prolonged tracer infu-sions, the accumulation of radioactivity in bone altersthe signal apparently arising from the cerebellumand other brain regions, as illustrated in Figure 2.This figure also shows that progressively increasingspill-in from bone results in overestimation of thetracer concentration in the cerebellum; this phenom-enon propagates to the underestimation of the striatalBPND (see Fig. 4A), which is computed as the simple-ratio relative to Cref, the cerebellum reference region.Alternately, it might be supposed that accumulationof a brain-penetrating plasma metabolite could like-wise have resulted in the increasing magnitude of

Fig. 3. Relative heart rates in groups of (N 5 5) rats with(squares) saline treatment or (N 5 5) challenge with amphetamineat a dose of (circles) 1 or (triangles) 5 mg/kg. Time ranges for com-putation of mean values at baseline and postamphetamine are illus-trated as gray bands.

Fig. 4. Relative [18F]-DMFP BPND (A) and BPCND (B) in striatum during prolonged tracer infu-

sions in groups of (N 5 5) rats with (squares) saline treatment or (N 5 5) challenge with ampheta-mine at a dose of (circles) 1 or (triangles) 5 mg/kg. Time ranges for computation of mean values atbaseline and postamphetamine are illustrated as gray bands.

710 E. MILLE ET AL.

Synapse

Cref, as is the case for [18F]-altanserin (Pinborg et al.,2003). However, we found no evidence for progressiveaccumulation of [18F]-DMFP metabolites in brainTACs of control animals with continuous infusion (seeFig. 1).

An incidental goal of this study was to demonstratedose-dependence of the effect of amphetamine on [18F]-DMFP binding during constant infusion, as has beendemonstrated for the case of bolus [11C]-raclopride(Houston et al., 2004). In that study, there was a signif-icant dose effect between 1 and 4 mg/kg amphetamine.Indeed, our BPND and BPC

ND findings even suggest ashorter duration of the treatment effect on [18F]-DMFPbinding at the higher amphetamine dose (Fig. 4), thisdespite clear evidence for a pharmacodynamic dose de-pendence on heart rate (Fig. 3). However, the study in

groups of five animals was not specifically designed totest for dose-dependence, but rather to establish betterthe principle of using [18F]-DMFP in challenge para-digms, hitherto only reported in a very few studies(Rominger et al., 2010c). As such, we expect that pres-ent findings with [18F]-DMFP should be translatablefor clinical PET studies of psychostimulant action atcenters lacking a medical cyclotron.

Extensive bone-labeling in the course of rodentPET interferes with the analysis of brain uptake ofdiverse other [18F]-labeled tracers, including theVMAT-2 ligand [18F]-fluorolethyl-DTBZ (Jahan et al.,2011) the beta-amyloid ligand [18F]-florbetaben(unpublished), the serotonin 5HT-1A ligand [18F]-FCWAY (Tipre et al., 2006), and the tyrosine kinasesubstrate [18F]-FHBG (Alauddin et al., 2001). The

TABLE I. Mean uncompensated (BPND) and compensated (BPCND) binding potentials of [18F]-DMFP in rat striatum assessed during

continuous infusion experiment at baseline and after saline or amphetamine challenge

Condition NInjected dose

(MBq) Animal weight (g) BPND at baseline BPND postchallengeRelative

reduction (%)

Saline 5 99 6 11 251 6 22 4.7 6 1.8 (BPND) 4.1 6 1.8 135.4 6 2.3 (BPC

ND) 5.4 6 3.0 22Amphetamine 1 mg/kg 5 91.8 6 9.5 244 6 22 3.5 6 0.7 (BPND) 2.4 6 0.5 32

3.9 6 0.8 (BPCND) 2.9 6 0.6 25

Amphetamine 4 mg/kg 4 5 103 6 12 250 6 15 5.3 6 1.0 (BPND) 3.7 6 0.9 316.0 6 1.1 (BPC

ND) 4.6 6 1.0 22

Fig. 5. Time series of mean [18F]-DMFP-binding potential mapscompensated for spill-in into the cerebellum (BPC

ND) obtained ingroups of (N 5 5) rats treated with saline or amphetamine at a dose

of 1 or 4 mg/kg at 95 min into the infusion (yellow band). Paramet-ric maps in the coronal plane are projected onto the histologicalatlas for rat brain in common coordinates.

711AMPHETAMINE EFFECTS ON [18F]-DMFP BINDING IN RATS

Synapse

spatial resolution afforded by current small animalPET instrumentation is approaching the theoreticallimit imposed by positron travel before annihilation,such that spill-in from the cranium will necessarilyremain an issue for the quantitation of rodent brainscans with [18F]-fluorinated tracers. Our method isdesigned to compensate for progressive spill-in of cra-nial radioactivity based on pharmacokinetic consider-ations, rather than geometric or anatomic informa-tion. Although our approach is tailored to the particu-lar circumstances of a continuous tracer infusion, theconcept can be generalized to any circumstance inwhich the kinetics of tracer uptake in an interferingstructure is known.

Explicit approaches for partial volume and spill-over corrections would likewise reduce such spill-ineffects, albeit with the requirement for precise CT- orMRT-based delineation of the relevant structures(Fazio and Perani, 2000; Morris et al., 1999; Roussetet al., 1998, 2008; Shidahara et al., 2009) and gener-ally also based upon the assumption of spatially uni-form tracer distribution throughout relevant struc-tures. Because our present instrumentation did notprovide high-resolution anatomic images, we deriveda method specifically to compensate for the effect ofspill-in from the cranium to the cerebellum reference

region. The evident stabilization of BPCND achieved

in control animals with constant tracer infusion, de-spite the presence of substantial cranial uptakesupports the validity of our approach. This procedurecreated the appearance of lesser magnitude of theamphetamine-effect; whereas the uncorrected BPND

declined by 31–33% and the magnitude of BPCND

declined only by 22–25% after amphetamine chal-lenge. However, this seeming difference was due tothe projection of amphetamine effect on a spontane-ously declining BPND in the saline condition, whichwe show to be an artifact of the uncompensated spill-in from the cranium. The constant infusion paradigmpresents advantages with respect to the attainment ofprolonged steady-state binding of [18F]-DMFP atstriatal dopamine receptors, here for as long as 4 h,but with the requirement for compensation or spill-over correction in order to avoid progressive bias inthe binding estimates.

REFERENCES

Alauddin MM, Shahinian A, Gordon EM, Bading JR, Conti PS.2001. Preclinical evaluation of the penciclovir analog 9-(4-[18F]flu-oro-3-hydroxymethylbutyl)guanine for in vivo measurement of sui-cide gene expression with PET. J Nucl Med 42:1682–1690.

Carson RE, Breier A, de Bartolomeis A, Saunders RC, Su TP,Schmall B, Der MG, Pickar D, Eckelman WC. 1997. Quantifica-

Fig. 6. Mean [18F]-DMFP-binding potential maps compensatedfor spill-in into the cerebellum (BPC

ND) obtained in groups of (N 55) rats at baseline (A–C) and after treatment with saline (A0) or am-

phetamine at a dose of 1 (B0) or 4 mg/kg (C0) at 95 min into theinfusion. Triplanar projections of the parametric maps are projectedonto the histological atlas for rat brain in common coordinates.

712 E. MILLE ET AL.

Synapse

tion of amphetamine-induced changes in [11C]raclopride bindingwith continuous infusion. J Cereb Blood Flow Metab 17:437–447.

Constantinescu CC, Coleman RA, Pan ML, Mukherjee J. 2011.Striatal and extrastriatal microPET imaging of D2/D3 dopaminereceptors in rat brain with [1F]fallypride and [1F]desmethoxyfal-lypride. Synapse 65:778–787.

Fazio F, Perani D. 2000. Importance of partial-volume correction inbrain PET studies. J Nucl Med 41:1849–1850.

Hudson HM, Larkin RS. 1994. Accelerated image reconstructionusing ordered subsets of projection data. IEEE Trans Med Imag-ing 13:601–609.

Innis RB, Cunningham VJ, Delforge J, Fujita M, Gjedde A, GunnRN, Holden J, Houle S, Huang SC, Ichise M, Iida H, Ito H,Kimura Y, Koeppe RA, Knudsen GM, Knuuti J, Lammertsma AA,Laruelle M, Logan J, Maguire RP, Mintun MA, Morris ED, ParseyR, Price JC, Slifstein M, Sossi V, Suhara T, Votaw JR, Wong DF,Carson RE. 2007. Consensus nomenclature for in vivo imaging ofreversibly binding radioligands. J Cereb Blood Flow Metab27:1533–1539.

Jahan M, Eriksson O, Johnstrom P, Korsgren O, Sundin A, Johans-son L, Halldin C. 2011. Decreased defluorination using the novelbeta-cell imaging agent [18F]FE-DTBZ-d4 in pigs examined byPET. EJNMMI Res 1:33.

la Fougere C, Boning G, Bartmann H, Wangler B, Nowak S, Just T,Wagner E, Winter P, Rominger A, Forster S, Gildehaus FJ, Rosa-Neto P, Minuzzi L, Bartenstein P, Potschka H, Cumming P.2010a. Uptake and binding of the serotonin 5-HT1A antagonist[18F]-MPPF in brain of rats: Effects of the novel P-glycoprotein in-hibitor tariquidar. Neuroimage 49:1406–1415.

la Fougere C, Popperl G, Levin J, Wangler B, Boning G, Uebleis C,Cumming P, Bartenstein P, Botzel K, Tatsch K. 2010b. The valueof the dopamine D2/3 receptor ligand 18F-desmethoxyfallypride forthe differentiation of idiopathic and nonidiopathic parkinsoniansyndromes. J Nucl Med 51:581–587.

Laruelle M. 2000. Imaging synaptic neurotransmission with in vivobinding competition techniques: A critical review. J Cereb BloodFlow Metab 20:423–451.

Logan J, Fowler JS, Volkow ND, Wolf AP, Dewey SL, Schlyer DJ,MacGregor RR, Hitzemann R, Bendriem B, Gatley SJ, et al.1990. Graphical analysis of reversible radioligand binding fromtime-activity measurements applied to [N-11C-methyl]-(2)-cocainePET studies in human subjects. J Cereb Blood Flow Metab10:740–747.

Morris ED, Chefer SI, Lane MA, Muzic RF Jr, Wong DF, DannalsRF, Matochik JA, Bonab AA, Villemagne VL, Grant SJ, IngramDK, Roth GS, London ED. 1999. Loss of D2 receptor binding withage in rhesus monkeys: Importance of correction for differences instriatal size. J Cereb Blood Flow Metab 19:218–229.

Mukherjee J, Yang ZY, Brown T, Roemer J, Cooper M. 1996. 18F-desmethoxyfallypride: A fluorine-18 labeled radiotracer with prop-erties similar to carbon-11 raclopride for PET imaging studies ofdopamine D2 receptors. Life Sci 59:669–678.

Mumcuoglu EU, Leahy R, Cherry SR, Zhou Z. 1994. Fast gradient-based methods for Bayesian reconstruction of transmission andemission PET images. IEEE Trans Med Imaging 13:687–701.

Nikisch G, Baumann P, Kiessling B, Reinert M, Wiedemann G,Kehr J, Mathe AA, Piel M, Roesch F, Weisser H, Schneider P,Hertel A. 2010. Relationship between dopamine D2 receptor occu-pancy, clinical response, and drug and monoamine metaboliteslevels in plasma and cerebrospinal fluid. A pilot study in patientssuffering from first-episode schizophrenia treated with quetiapine.J Psychiatr Res 44:754–759.

Patel NH, Vyas NS, Puri BK, Nijran KS, Al-Nahhas A. 2010. Posi-tron emission tomography in schizophrenia: A new perspective. JNucl Med 51:511–520.

Pedersen K, Simonsen M, Østergaard SD, Munk OL, Rosa-Neto P,Olsen AK, Jensen SB, Møller A, Cumming P. 2007. Mapping theamphetamine-evoked changes in [11C]raclopride binding in living

rat using small animal PET: Modulation by MAO-inhibition. Neu-roimage 35:38–46.

Pinborg LH, Adams KH, Svarer C, Holm S, Hasselbalch SG,Haugbøl S, Madsen J, Knudsen GM. 2003. Quantification of 5-HT2A receptors in the human brain using [18F]altanserin-PETand the bolus/infusion approach. J Cereb Blood Flow Metab23:985–996.

Qi J, Leahy RM. 2000. Resolution and noise properties of MAPreconstruction for fully 3-D PET. IEEE Trans Med Imaging19:493–506.

Qi J, Leahy RM, Cherry SR, Chatziioannou A, Farquhar TH. 1998.High-resolution 3D Bayesian image reconstruction using themicroPET small-animal scanner. Phys Med Biol 43:1001–1013.

Rominger A, Mille E, Boning G, Wangler B, Gildehaus FJ, Arszol C,Bartenstein P, Cumming P. 2010a. a2-Adrenergic drugs modulatethe binding of [18F]fallypride to dopamine D2/3 receptors in stria-tum of living mouse. Synapse 64:654–657.

Rominger A, Mille E, Zhang S, Boning G, Forster S, Nowak S,Gildehaus FJ, Wangler B, Bartenstein P, Cumming P. 2010b.Validation of the octamouse for simultaneous 18F-fallypridesmall-animal PET recordings from 8 mice. J Nucl Med 51:1576–1583.

Rominger A, Wagner E, Mille E, Boning G, Esmaeilzadeh M, Wan-gler B, Gildehaus FJ, Nowak S, Bruche A, Tatsch K, BartensteinP, Cumming P. 2010c. Endogenous competition against binding of[18F]DMFP and [18F]fallypride to dopamine D2/3 receptors in brainof living mouse. Synapse 64:313–322.

Rousset OG, Ma Y, Evans AC. 1998. Correction for partial volumeeffects in PET: Principle and validation. J Nucl Med 39:904–911.

Rousset OG, Collins DL, Rahmim A, Wong DF. 2008. Design andimplementation of an automated partial volume correction inPET: Application to dopamine receptor quantification in the nor-mal human striatum. J Nucl Med 49:1097–1106.

Seneca N, Zoghbi SS, Skinbjerg M, Liow JS, Hong J, Sibley DR,Pike VW, Halldin C, Innis RB. 2008. Occupancy of dopamine D2/3

receptors in rat brain by endogenous dopamine measured withthe agonist positron emission tomography radioligand [11C]MNPA.Synapse 62:756–763.

Shidahara M, Tsoumpas C, Hammers A, Boussion N, Visvikis D,Suhara T, Kanno I, Turkheimer FE. 2009. Functional and struc-tural synergy for resolution recovery and partial volume correc-tion in brain PET. Neuroimage 44:340–348.

Tipre DN, Zoghbi SS, Liow JS, Green MV, Seidel J, Ichise M, InnisRB, Pike VW. 2006. PET imaging of brain 5-HT1A receptors inrat in vivo with 18F-FCWAY and improvement by successful inhi-bition of radioligand defluorination with miconazole. J Nucl Med47:345–353.

Villemagne VL, Wong DF, Yokoi F, Stephane M, Rice KC, MateckaD, Clough DJ, Dannals RF, Rothman RB. 1999. GBR12909attenuates amphetamine-induced striatal dopamine release asmeasured by [11C]raclopride continuous infusion PET scans. Syn-apse 33:268–273.

Volkow ND, Fowler JS, Wang GJ, Swanson JM. 2004. Dopamine indrug abuse and addiction: Results from imaging studies and treat-ment implications. Mol Psychiatry 9:557–569.

Volkow ND, Fowler JS, Wang GJ, Baler R, Telang F. 2009. Imagingdopamine’s role in drug abuse and addiction. Neuropharmacology56 (Suppl 1):3–8.

Watabe H, Endres CJ, Breier A, Schmall B, Eckelman WC, CarsonRE. 2000. Measurement of dopamine release with continuousinfusion of [11C]raclopride: Optimization and signal-to-noise con-siderations. J Nucl Med 41:522–530.

Watson CC. 2000. New, faster, image-based scatter correction for 3DPET. IEEE Trans Nucl Sci 47:1587–1594.

Watson CC, Newport D, Casey ME. 1996. A single scatter simula-tion technique for scatter correction in 3D pet. Comp Imag Vis4:255–268.

713AMPHETAMINE EFFECTS ON [18F]-DMFP BINDING IN RATS

Synapse