Abstract
DAILY VARIATIONS IN CEREBRAL BLOOD VOLUME AND CONSEQUENCES FOR QUANTITATIVE PET STUDIES
Mark Lubberink, Alie Schuitemaker, Saskia P.A. Wolfensberger, Gert Luurtsema, Bart N.M. van Berckel, Adriaan A. Lammertsma
Department of Nuclear Medicine and PET Research, VU University Medical Centre, Amsterdam, The Netherlands
Background and aim
Although fractional cerebral blood volume (Vb) can easily be included as a separate parameter in full compartmental analysis of PET studies, it is more difficult to account for in simplified analyses. In a previous study a significant change in distribution volume of (R)-[11C]verapamil from morning to afternoon scans was found using Logan analysis, but compartmental analysis revealed that this difference could be attributed solely to differences in Vb [1]. The aim of the present study was to verify whether there is indeed significant daily variation in Vb.
Methods
Whole brain grey matter Vb of 73 scans (51 (R)-[11C]PK11195, 12 (R)-[11C]verapamil and 10 [11C]R116301 scans) was related to scan start time retrospectively. (R)-[11C]verapamil and [11C]R116301 scans were part of a test-retest protocol, with the same volunteer being scanned twice on the same day at 10:30 and 15:00 h. All 60 min dynamic scans were acquired in 3D mode following administration of 370 MBq of tracer. A metabolite corrected arterial plasma input function was obtained using continuous arterial sampling together with discrete manual samples. A co-registered segmented T1-weighted MRI scan was used to define whole brain grey matter regions of interest. Data were analysed using single ((R)-[11C]verapamil) or two ((R)-[11C]PK11195, [11C]R116301) tissue compartment models, including Vb as a fit parameter. Differences in Vb between morning and afternoon scans were assessed using two-tailed t-tests.
Results
In all scans, whole brain grey matter Vb could be determined with a standard error of less than 10%. A shown in table 1, a significant decrease in blood volume between morning and afternoon scans in the same subject was found for both (R)-[11C]verapamil and [11C]R116301. In addition, a significant difference in Vb between morning and afternoon (R)-[11C]PK11195 scans was found.
Conclusion
For all three tracers, a decrease of ~10% in average Vb was found from morning to afternoon scans. This decrease could not be attributed to blood sampling in the morning scan itself, as the effect was also seen in (R)-[11C]PK11195 scans where subjects underwent only a single scan. As demonstrated previously, a decrease in Vb during the day may lead to erroneous Results: when it is not included as a fit parameter [1]. This effect should be considered especially in challenge studies where successive scans may be acquired during the day.
DIFFERENCES IN DISTRIBUTION VOLUME OF THE NICOTINERGIC ACETYLCHOLINE RECEPTOR-LIGAND 2-[F-18]F-A-85380 IN NON-SMOKERS AND SMOKERS AS A FUNCTION OF SCAN TIME
Hans Herzog1, Martina Minnerop2, Wolfgang Eschner3, Joern Schmaljohann4, Holger Brockmann4, Daniela Guendisch5, Ullrich Wuellner2
1Institute of Medicine, Forschungszentrum Juelich, 2Department of Neurology, Rheinische Friedrich-Wilhelms-University, Bonn, 3Department of Nuclear Medicine, University of Cologne, Cologne, 4Department of Nuclear Medicine, 5Department of Pharmaceutical Cemistry, Rheinische Friedrich-Wilhelms-University, Bonn, Germany
Papers on PET-studies of the cerebral dynamics of the nicotinergic acetylcholine receptor ligand 2-[F-18]F-A-85380 report scan times of up to 7 h. In a recent study we evaluated the uptake and cerebral kinetics of 2-[F-18]F-A-85380 in smokers compared to non-smokers with a scan duration of 6 h. To prepare future examinations we asked whether such a long scan time is necessary for determining the distribution volume and for finding relevant differences between the two groups. For this purpose the calculation of the distribution volume was based on scan times differing between 1.5 and 6 h. Possible differences in the distribution volume (DV) between non-smokers (NS) and smokers (S) were analyzed in regard to the scan time.
13 normal volunteers ? seven NS and six age-matched S ? were scanned with a PET-Scanner ECAT Exact after injection of 320 ± 40 MBq 2-[F-18]F-A-85380. The smokers stopped smoking two days before the measurement. Each acquisition consisted of three parts of 90 min separated by breaks of 45 min. Before the injection and the second and third scan, transmission scans of 10 min each were performed. The first transmission scan was used for attenuation correction of all data and possible head movements between the three scan periods were taken into account. Intra-and interscan movements were corrected with the MPItool. Using the dynamic PET images, the arterial plasma input function and the Logan-plot model implemented in the analysis and modeling software PMOD, parametric DV images were calculated for four different periods: from 0 min to 90 min, from 0 min 165 min, from 0 min to 225 min, and finally from 0min to 6 h. The DV images were analyzed with regions of interest placed over the thalamus, the cortex, and the cerebellum.
Looking at the cortical and the cerebellar DV values only minor differences between the four time periods were detected. This finding held for each group. In contrast, the calculated thalamic DV values increased with the longer scan time by 34% for NS and by 18% for S. If the cortical and the cerebellar DV values of NS and S were compared, significant differences for all scan times were received with p < 0.036. A significant (p < 0.009) intergroup difference of the thalamic DV was found only for the shortest scan time lasting to 90 min. For longer scan times with a greater variance present in the data there was only a trend (p > 0.05) of a different thalamic DV.
The influence of the maximal scan duration on the DV calculation nicotinergic acetylcholine receptor ligand 2-[F-18]F-A-85380 does not show a unique behavior. For a comparative analysis between two different groups it has to be assessed which cerebral regions are of specific interest. If regions outside the thalamus with a low density of the nicotinergic acetylcholine receptors are of main interest, a short scan time may be sufficient. In contrast, stable thalamus DV values are obtained only when a longer save time of e.g. 3 hours is evaluated.
LINEAR STATISTICAL ANALYSIS OF NEUROIMAGES USING VOXELWISE COVARIATES: APPLICATION TO THE STUDY OF DIFFUSION ABNORMALITIES IN SCHIZOPHRENIA
Babak A. Ardekani1,2, Arthika Bappal1, Matthew J. Hoptman1,3
1Center for Advanced Brain Imaging, Nathan Kline Institute for Psychiatric Research, Orangeburg, NY, USA, 2Department of Psychiatry, New York University School of Medicine, New York, NY, USA, 3Clinical Research Division, Nathan Kline Institute for Psychiatric Research, Orangeburg, NY, USA
Background and aims: Linear statistical models have been applied extensively to voxelwise analysis of PET and MRI data. Most analyses fit a predefined linear model with fixed covariates to every voxel within the dataset. Recently, two groups (Casanova, Ryali, et al., 2006; Oakes, Fox, et al., 2007) independently introduced Methods: that allow modeling voxelwise covariates. These covariates can change from voxel to voxel, in effect, fitting different linear models at different brain locations. We have also independently developed such capability and applied it to the analysis of diffusion tensor imaging (DTI) data (Ashtari, Kumra, et al., 2005). Here we demonstrate the application of our software, voxelwise analysis of covariance (VANCOVA), to the study of white matter abnormalities in schizophrenia.
Methods: Thirty-seven patients with DSM-IV diagnosis of schizophrenia and 29 age-matched controls underwent DTI and 2D and 3D anatomical imaging using a Siemens Vision 1.5T scanner. The DTI data consisted of seven averages of 8 diffusion-weighted (b=1000 s/mm2) volumes with diffusion sensitizing gradients along 8 different directions, and a b=0 volume. From the DTI data, fractional anisotropy (FA) and mean diffusivity (MD) images were computed. The high-resolution anatomical images were used to correct the spatial distortion in the DTI images as well as for spatial normalization of the FA and MD images to a standard space. The normalized images were smoothed with a 3 mm FWHM Gaussian filter. Using VANCOVA, we performed voxelwise statistical analyses of the FA images to determine regions were patients and controls had significantly different FA values (p<0.001 two-tailed uncorrected; cluster size > 100), with and without the option of controlling for MD.
Results: We found several regions in both white and gray matter where patients had reduced FA relative to controls. Some of the significantly different FA clusters are shown in the Figure with (bottom row) and without (top row) controlling for MD. When MD was used as a covariate, the number of significant clusters reduced in regions prone to partial volume effects (PVEs), while the differences were enhanced for some of the remaining clusters.
Conclusions: We have developed a software package that enables integrating voxelwise covariates in linear statistical models of neuroimaging data. This method can be applied to study of FA differences in patient populations, where it is important to control for MD as a confounding covariate in order to account for PVEs.
IN VIVO QUANTIFICATION OF 5-HT1A-[18F]MPPF INTERACTIONS IN RATS USING THE YAP-(S)PET AND THE BETA MICROPROBE
Philippe Millet1, M. Moulin-Sallanon2, A. Bartoli3, A. Del Guerra3, L. Lemoucheux4, N. Ginovart5,1, V. Ibanez1
1Psychiatric Neuroimaging Unit, Division of Neuropsychiatry, Department of Psychiatry, University Hospital of Geneva, Geneva, Switzerland, 2INSERM, Unit E340, La Tronche, France, 3Department of Physics, University of Pisa, Pisa, Italy, 4Advanced Accelerator Application, Saint Genix Pouilly, France, 5Department of Psychiatry, University of Geneva, Geneva, Switzerland
Introduction: Several studies have provided evidence that PET can be used to measure fluctuations in synaptic concentration of neurotransmitters. In this context, [18F]MPPF, a 5-HT1A radioligand with an affinity close to that of serotonin has been proposed as a tool to evaluate 5-HT levels. However, discordant Results: have been reported regarding the sensitivity of this radioligand to endogenous 5-HT. In this study, a full quantitative analysis of the pharmacokinetic properties of MPPF binding was performed using the β-microprobe (βP) (1) and the YAP-(S)PET (YAP) (2) to better understand the 5-HT1A-[18F]MPPF interactions. Methods: Sixteen Wistar rats were used for βP (n=5) and YAP (n=5) acquisitions, and for metabolite studies (n=6). β-microprobe: four probes were implanted to measure simultaneously time-concentration curves in the hippocampus, raphe dorsalis, frontal cortex and cerebellum. YAP-(S)PET : a 3D data acquisition mode and an expectation maximization algorithm with 50 iterations were used for image reconstruction (voxel size = 0.5 × 0.5 × 2 mm3). We used a multi-injection protocol which consists of a [18F]MPPF injection (T0=0 min), two successive co-injections of [18F]MPPF and unlabeled MPPF at T0 + 60 min and at T0 + 120 min. During acquisitions, about 95 arterial blood samples were withdrawn, then corrected for metabolites to obtain the model input function. A three tissue-compartment seven-parameter model (3T-7k) including non-specific binding, and a two tissue-compartment five parameter model (2T-5k), were used to analyze βP and YAP time-concentration curves. The simplified reference tissue model (SRTM) was used to estimate the binding potential (BPSRTM) values using the data obtained with the first injection and the cerebellum as the reference region.
Results: The multi-injection approach allowed to quantify all binding parameters with correct standard errors in all regions. Overall, the 3T-7k model provided better fits than the 3T-5k model as evaluated with the F-test in all experiments. The non-specific binding estimated with the 3T-7k model was similar across regions. The rank order for B'max values was: hippocampus> raphe ? frontal cortex > cerebellum, a result consistent with previously reported in vitro data. We found non-negligeable specific binding in cerebellum (B'max(βP) = 1.5 ± 0.9 pmol/ml). Significant correlations (p < 0.001) were found between B'max and BPSRTM values for βP (r=0.695) and YAP (r=0.695). The YAP-(S)PET underestimated [18F]MPPF binding levels in brain due to the limited resolution of the system, but lead to similar Conclusion.
Conclusion: The multi-injection approach has allowed to highlight non-specific binding in all brain regions and non-negligeable specific binding in cerebellum. Therefore, the use of cerebellum as a reference region could explain the fact that [18F]MPPF cannot detect changes in endogenous 5-HT levels when using semi-quantitative measures such as the BP.
Grant support: Swiss National Science Foundation N° 3100A0-104185
QUANTITATIVE ANALYSIS OF AMYLOID DEPOSITION IN HUMAN BRAIN USING PET AND A NEW IMAGING PROBE [11C]BF-227
Manabu Tashiro1, K. Kumagai1, N. Okamura3, S. Furumoto2, K. Furukawa2, Y. Funaki1, M. Maruyama4, Y. Kimura5, M. Itoh1, R. Iwata1, Y. Kudo3, H. Arai4, K. Yanai3
1Cyclotron and Radioisotope Center, Tohoku University, Sendai, Miyagi, Japan, 2Tohoku University Biomedical Engineering Research Organization (TUBERO), Sendai, Miyagi, Japan, 3Department of Pharmacology, Tohoku University Graduate School of Medicine, Sendai, Miyagi, Japan, 4Center for Asian Traditional Medicine, Department of Geriatrics and Gerontology, Tohoku University School of Medicine, Sendai, Miyagi, Japan, 5Positron Medical Center, Tokyo Metropolitan Institute of Gerontology, Tokyo, Japan
Introduction: In vivo detection of amyloid deposits would be useful for early diagnosis of Alzheimer's disease (AD). A new imaging probe, 2-[2-(2-Dimethylaminothiazol-5-yl)ethenyl]-6-[2-(fluoro)ethoxy]benzoxazole (BF-227), that was picked up out of hundreds of candidate compounds through systematic evaluation using animals, was [11C]-labeled for clinical evaluation. In vivo amyloid imaging using the [11C]BF-227 was successfully conducted in AD patients at Tohoku University (Ref.). The purpose of the present study is to examine the method for quantitative analysis of amyloid deposition in human brain using PET and [11C]BF-227.
Methods: Eleven healthy controls (M:F=8:3; mean age: 56.3 +/− 15.0 y.o.) and ten AD patients (M:F=4:6; 73.3 +/− 7.7 y.o.) were studied. Dynamic PET images were obtained for 60 min (23 sequential scans). Blood samples were obtained through arterialized veins as well as through radial arteries in only one case for each subgroup. The all cases were studied with PET and FDG as well. Regions of interest were placed on various cortical and subcortical regions based on the coregistered MRI images. The whole data were analyzed using tissue time-activity curves, corrected for partial volume effects due to white matter and cerebrospinal space, and metabolite-corrected plasma time-activity curves. Ratio of distribution volume (DVR: each region/cerebellum) were calculated using Logan graphical method, and additionally using full kinetic analysis based on three-compartment model in the cases with arterial sampling data. Results: In AD patients, [11C]BF-227 displayed significantly higher DVR values than in controls in various cortical regions such as the cingulate, frontal, temporal, parietal and occipital regions (i.e. temporal; control vs. AD: 1.16 +/− 0.05 vs. 1.29 +/− 0.06, respectively)(p< 0.001). Further analysis revealed that the highest retention of [11C]BF-227 was observed in the temporoparietal association area, the predilected site for dense amyloid plaque depositions, which was corresponding to the hypometabolic area observed in the FDG study.
Conclusion: These findings demonstrated that [11C]BF-227 was a promising PET probe and that Logan graphical method can be used for evaluation of dense amyloid deposits in AD patients.
Figure: Tissue time-activity curves of the cerebellum and lateral temporal cortex in controls and AD patients
ON THE VALIDITY OF REFERENCE TISSUE MODELS FOR ANALYSING (R)-[11C]PK11195 STUDIES IN TRAUMATIC BRAIN INJURY
Ronald Boellaard1, Hedy Folkersma2, Marc A. Kropholler1, Gert Luurtsema1, Pieter W. VanderTop2, Adriaan A. Lammertsma1, Bart N.N. van Berckel1
1Department of Nuclear Medicine and PET Research, 2Department of Neurosurgery, VU University Medical Center, Amsterdam, The Netherlands
Introduction
Activated microglia in the brain can be imaged in vivo using (R)-[11C]PK11195 and PET. In theory, this activation can be quantified using a reference tissue approach, provided that an appropriate reference tissue is available. The purpose of the present study was to assess whether a reference tissue approach is valid for analysing (R)-[11C]-PK11195 studies in traumatic brain injury (TBI).
Methods
Five TBI patients were included. About 10 days after TBI, subjects underwent a 60 minutes 3D dynamic (R)-[11C]PK11195 brain scan, including on-line and manual arterial sampling. Volumes of interest (VOI) were drawn on a co-registered T1-weighted MRI scan for frontal, occipital, parietal, temporal and cingulate cortex, and for thalamus and pons. In addition, VOI were defined in and near (penumbra) traumatic brain areas. A cerebellum VOI was used as reference tissue. Binding potential (BP-PI), volume of distribution (Vd), distribution volume ratio (DVR; Vd normalised to Vd cerebellum), blood volume fraction (Vb) and K1/k2 were derived for all VOI using a plasma input model. In addition, binding potential (BP-SRTM) was obtained using the simplified reference tissue model. Finally, simulations were performed to assess effects of varying Vb and K1/k2, the latter simulating blood-brain barrier (BBB) disruption, between reference and target tissues.
Results
Plasma input analysis revealed a strong correlation (R2=0.81, slope=3.1) between Vd and K1/k2 over all regions and subjects, whilst no correlation was observed between BPPI and K1/k2. Furthermore, BP-PI did not significantly differ from those seen in healthy subjects. These findings indicate that, in TBI, variation in Vd is almost exclusively determined by variation of the K1/k2 ratio. In addition, both DVR and BP-SRTM correlated well (R2=0.74) with variation of K1/k2 between target and reference tissues. DVR and BP-SRTM therefore primarily reflect changes K1/k2 rather then specific binding. Areas in and near TBI (penumbra) showed large increases in DVR and BP-SRTM on parametric images, but this (false) positive signal could simply be explained by a corresponding large increase in K1/k2 compared with that of the reference region. On the other hand BP-PI showed no evidence for increased binding. Simulations confirmed that DVR and BP-SRTM may provide false positive Results: in case of changes (increases) in (bidirectional) BBB transport. For BP-SRTM this was also the case for increased Vb.
Conclusion
Increases in (R)-[11C]PK11195 DVR and BP-SRTM, supposedly representing specific binding, in and near traumatic areas in TBI patients can be explained by differences in BBB transport and, in case of BP-SRTM, by changes in Vb. The large variation of K1/k2 across the brain of TBI patients prohibits the definition of an appropriate reference tissue. Therefore, in TBI patients, (R)-[11C]PK11195 studies should only be analysed using plasma input models.
POSSIBILITY OF SCAN TIME REDUCTION FOR COUNT-BASED OEF ELEVATION IN O-15 GAS PET
Masato Kobayashi1, Hidehiko Okazawa1, Tetsuya Tsujikawa1, Tatsuro Tsuchida2, Makoto Isozaki3, Yoshikazu Arai3, Keiichi Kawai1, Yasuhisa Fujibayashi1
1Biomedical Imaging Research Center, 2Department of Radiology, 3Department of Neurosurgery, University of Fukui, Fukui, Japan
Background and Aim
O-15 gas PET is useful to evaluate hemodynamic status in patients with ischemic cerebrovascular disease. We reported the simplified count-based method for diagnosis of misery perfusion; however the method needed long scan time to reach steady-state in continuous 15O2 inhalation. To reduce the exposure time and examination time, we assessed the accuracy of the count-based method with shorter scan time after 15O2-gas inhalation.
Methods
Twenty-four patients (64±11y) with unilateral cerebrovascular stenoocclusive disease were studied for measurement of asymmetric OEF elevation. Dynamic PET scans started with 1 min/frame at 2 min after the beginning of 15O2-gas inhalation with a constant flow rate (740 MBq/min). Each subject also underwent 15O-labeled water PET scan using the bolus injection method. To evaluate effects of scan time during 15O2 inhalation, dynamic PET data were summed for different scan duration started from each acquisition start time minute-by-minute. Count-based ratio images (cbOEF) were calculated pixel-by-pixel by 15O2/H215O PET images. Asymmetric indexes (cbOEF-AI) were obtained using the regions of interest drawn on the bilateral cerebral cortices. These AI were compared with the AI of quantitative OEF (qOEF-AI).
Results
Table showed the slope and correlation coefficients between qOEF-AI and cbOEF-AI with different duration and acquisition start time after 15O2-gas inhalation. The slopes of the regression lines were close to 1.00 with the correlation coefficients greater than 0.58. The cbOEF-AI using longer scan time was closer to the qOEF-AI than that using shorter scan time. Longer scan time provided better correlation between cbOEF-AI and qOEF-AI regardless of scanning start time. The cbOEF with 15O2 images summated from 2 min to about 5 min after 15O2-gas inhalation did not also correlate well with qOEF-AI. These Results: suggested that cbOEF-AI using scans with image acquisition started from 3 min after the 15O2 inhalation for 4 min or longer would estimate qOEF-AI correctly.
Conclusion
The cbOEF using H215O and continuous 15O2 inhalation method would need scanning from 3 min after 15O2-gas inhalation. Although longer duration time for 15O2 PET provided better correlation between qOEF-AI and cbOEF-AI, 4 min scan of 15O2-gas inhalation would be enough. The count-based PET with less invasive procedure in this study will be able to reduce examination time and stress to the patients in clinical studies for diagnosis of misery perfusion.
SMALL EFFECT OF DOPAMINE RELEASE AND NO EFFECT OF DOPAMINE DEPLETION ON [F-18]FALLYPRIDE BINDING IN HEALTHY HUMANS
Masahiro Fujita1, Vanessa Cropley1, Pradeep Nathan2, Amira Brown1, Janet Sangare1, Alicja Lerner1, Yong Ryu1, Kathleen Sprague1, Karen Berman3, Victor Pike1, Robert Innis1
1Molecular Imaging Branch, National Institute of Mental Health, Bethesda, MD, USA, 2Behavioural Neuroscience Laboratory, Department of Physiology, Monash Center for Brain and Behaviour, Monash University, Melbourne, VIC, Australia, 3Integrative Neuroimaging Unit, Clinical Brain Disorders Branch, National Institute of Mental Health, Bethesda, MD, USA
Background and aims: D-amphetamine induced dopamine release displaces [F-18]fallypride at dopamine D2 receptors in extrastriatal areas as well as in the striatum (1). The objectives of the current study were to measure the test-retest reproducibility particularly in low density regions and the influences of dopamine depletion in addition to dopamine release on [F-18]fallypride binding by performing four PET scans in each healthy human subject: two baselines, one with amphetamine, and one with α-methyl-para-tyrosine (AMPT) administration.
Methods: Fourteen subjects participated in the study. Three participated in a pilot study with one baseline and one amphetamine scans. Eight participated in all four scans, and three had two baseline and one amphetamine scans. D-amphetamine (0.5 mg/kg) was administered orally 3 h before [F-18]fallypride injection. AMPT (43 mg/kg/day) was administered orally for 2 days. Equilibrium ratios of specific-to-nondisplaceable were measured using the Lammertsma's reference tissue model in caudate, putamen, thalamus, medial portion of orbitofrontal cortex, anterior cingulate, lateral and medial temporal cortices, substantia nigra, and colliculi. Except the initial three subjects, correlation was studied between amphetamine or AMPT-induced changes in [F-18]fallypride binding and performance in cognitive tests.
Results: Test-retest variability and intraclass correlation coefficient (ICC) was 3.7 ? 7.7% and > 0.90, respectively, in all regions except anterior cingulate and colliculi which showed more than 10% variability and < 0.90 ICC. These two regions were excluded from further analyses. D-amphetamine displaced [F-18]fallypride significantly across the seven regions (p = 0.006) and in the five extrastriatal regions (p = 0.015) (Table). However, changes in individual regions reached to significant levels only in putamen and substantia nigra. AMPT did not cause significant changes in [F-18]fallypride binding, and there was no correlation between D-amphetamine and AMPT induced changes. Among cognitive tests, total number of words generated in Controlled Oral Word Association Test showed significant negative correlation with D-amphetamine induced decrease in [F-18]fallypride binding in thalamus and substantia nigra.
Conclusions: The measurement of [F-18]fallypride binding showed excellent reproducibility even in extrastriatal regions. D-amphetamine induced dopamine release displaced [F-18]fallypride in both striatum and extrastriatal regions. However, the effects were weak showing correlation with cognitive measures only in thalamus, which is a relay center of information processing and substantia nigra, which may reflect overall dopamine release. There was no correlation in other regions, which may play more specific roles in cognitive function. These subjects who showed weak D-amphetamine effects did not show AMPT induced changes in [F-18]fallypride binding.
DEVELOPMENT OF REFERENCE TISSUE METHOD FOR MULTIPLE INJECTIONS OF [C-11]RACLOPRIDE
Hiroshi Watabe1, Youichiro Ohta1, Noboru Teramoto1, Yoshinori Miyake2, Maki Kurokawa1, Akihide Yamamoto1, Yasuyuki Ose1, Yoko Ikoma3, Takuya Hayashi1, Hidehiro Iida1
1Department of Investigative Radiology, National Cardiovascular Center Research Institute, Suita, Osaka, Japan, 2National Cardiovascular Center Hospital, Suita, Osaka, Japan, 3Department of Biophysics, Molecular Imaging Center, National Institute of Radiological Sciences, Inage, Chiba, Japan
Introduction: PET with [C-11]raclopride is often used to investigate D2-dopamine receptor system in brain. Temporal changes of the binding potential (BP) of [C-11]raclopride due to the internal synaptic neurotransmitter release has been attempted to measure using bolus-followed by infusion (B/I) method or simplified reference tissue (SRT) method. B/I method is attractive, as its simple mathematical formulation, but requires relatively long study period, particularly when one intends to apply it at multiple activated conditions. SRT method can be completed within a relatively short period, but still needs to be carried out independently at each of multiple [C-11]raclopride PET scans. The aim of this study is to develop reference tissue method to estimate multiple BP values from a single session of PET scanning in conjunction with multiple injections of [C-11]raclopride.
Methods: The second [C-11]raclopride will be injected followed by the first injection of [C-11]raclopride after certain period. In order to estimate BP after the second injection, SRT method is extended to consider the residual radioactivity from the first injection of [C-11]raclopride.
In order to validate the method, we performed rodent studies (n=4, Wister rat male, 8-10 weeks old, 203-334 g) with PET and [C-11]raclopride. Rats were anesthetize by propofol and 37MBq of [C-11]raclopride (specific radioactivity of 8.2–34. GBq/umol) was injected and simultaneously PET acquisition was started. PET camera was MicroPET focus 120(Siemens).
After 20 minutes of the injection of [C-11]raclopride, 37 MBq of [C-11]raclopride (specific radioactivity of 1.1–6..0 GBq/umol) diluted by cold raclopride was injected.
PET data was acquired as list mode data and histogramed to the multiple sinogram data with 2 minutes time interval. All sinogram data were reconstructed by 2D filtered backprojection after Fourier rebinning technique. Regions of interest were placed on regions of basal ganglia and cerebellum and time activity curves were generated. Computer simulation was also performed to evaluate the present method.
Results and Discussion: Time activities curves were well fitted to the proposed model. BP values were significantly different between the 1st and 2nd injection of [C-11]raclopride attributed to the difference of the specific activity. Figure shows the relationship between mass of injected raclopride and BP value measured by the present method. Computer simulation (solid line in the figure) could well reproduce the Results: of the experiments, which suggests the feasibility of the proposed method. This technique can also easily be extended to more than three multiple jections of tracer.
Conclusion: The proposed method has potential to shorten study period of several neurotransmitter competition studies.
PERFORMANCE EVALUATION OF VARIOUS REFERENCE TISSUE INPUT PARAMETRIC METHODS: FOR [18F]FDDNP
Maqsood Yaqub1, Nelleke Tolboom1,2, Bart N.M. Van Berckel1, Anke A. Dijkstra1, Albert D. Windhorst1, Gert Luurtsema1, Philip Scheltens2, Adriann A. Lammertsma1, Roland Boellaard1
1Department of Nuclear Medicine and PET Research, 2Neurology and Alzheimer Centre, VU University Medical Centre, Amsterdam, The Netherlands
Introduction: [18F]FDDNP is a PET ligand that has been introduced for imaging neurofibrillary tangles and beta-amyloid fibrils in the brain. Recently, Kepe et al.[1] studied [18F]FDDNP binding in patients with Alzheimer Disease (AD) using reference Logan analysis[2]. The purpose of the present study was to study performance of several reference tissue parametric Methods: for measuring [18F]FDDNP binding.
Methods: The following parametric Methods: were evaluated: reference Logan[2], two basis function Methods: (RPM1 & RPM2[3],[4]), and various multi-linear Methods: (MRTMo, MRTM, MRTM2[5]) and two newly developed multi-linear Methods: based on MRTMo (MRTM3 and MRTM4). RPM2 and MRTM2,3,4 all include fixing the reference tissue clearance rate (k2′; using RPM1, MRTM2, MRTMo and MRTM1 respectively). Both simulations and clinical data were used to determine the effects of flow, fractional blood volume (Vb) and binding potential (BP) on accuracy and precision of parametric BP. Simulations were done by varying only one parameter and fixing other parameters to default values (R1=0.9, k2=0.07 & BP=0.2). Various simulated [18F]FDDNP time activity curves (TAC) were generated at a 15% noise level using plasma input. Clinical [18F]FDDNP data were obtained from 3 controls, 3 MCI and 3 AD subjects. TAC and/or regional average parametric data were derived for 8 regions of interest (ROI) across the frontal cortex. Grey matter cerebellum was used as reference tissue. ROI were projected on parametric images for comparison with BP obtained using SRTM.
Results: Simulations over a range of BP values (0.1–0..4) showed a maximum BP bias of 7% (compared with SRTM) for all Methods:, which decreased with increasing BP. Lowest BP bias was found for MRTM2 (<0.7±18%) and reference Logan (< 1.7±23%). MRTM showed very poor precision of 9–216% for BP = 0.4 to 0.1. BP bias did not exceed 4% for all Methods and for a range of simulated flows (R1=0.6 to 0.9). Precisions equaled ~10% except for MRTM1 (10–65%). MRTM2 yielded lowest bias (11±10%) over a range of Vb (0.025–0..075).
For most Methods: clinical parametric BP data showed correlations of R2=~0.8 with BP-SRTM. The only exception was MRTM (R2=0.49). Lowest biases were found for MRTM2, RPM2 and MRTM4 (1, 3 and 3%, respectively). For all other Methods: bias exceeded 10% compared with SRTM. Although parametric data showed a trend towards increased BP in AD subjects, none showed a significant difference with healthy subjects, in line with the findings from SRTM.
Visual inspection of parametric images showed artifacts for voxels near and in CSF and skull for most multi-linear Methods:. RPM2 provided ‘best’ BP images without these artifacts and these images appeared less noisy.
Conclusion: Most parametric Methods: evaluated showed similar performance compared with SRTM for brain tissue voxels. Based on observed accuracies, precisions and image quality, RPM2 is recommended for generating parametric BP (and R1) images.
REGULATION OF 14C-ACETATE UPTAKE IN CULTURED RAT ASTROCYTES
Rie Hosoi1, Yasuyo Matsuyama1, Yutaka Koyama2, Toshio Matsuda2, Antony Gee3, Osamu Inoue1
1Course of Allied Health Sciences, Graduate School of Medicine, 2Laboratory of Medicinal Pharmacology, Graduate School of Pharmaceutical Sciences, Osaka University, Suita, Osaka, Japan, 3Clinical Research Unit, GlaxoSmithkline and ACCI, Addenbrookes Hospital, Cambridge, UK
Introduction: Exogenous acetate is preferentially taken into astrocytes by monocarboxylate transporter-1 (MCT1) mediated process. Previously, we found that 14C-acetate uptake in the rat brain appears to occur in parallel with glial energy metabolism and reflects glial conditions (J Cereb Blood Flow Metab. 2004;24(2):188-90). Also we have recently reported the brain uptake of 14C-acetate is very sensitive to brain ischemia (J Stroke Cerebrovasc Dis, in press). Immediately after the 3 minutes-MCAO and reperfusion, 14C-acetate uptake showed a significant (about 50%) and reversible reduction in the rat striatum (ischemic core), which means even such a short-term MCAO caused depression of glial metabolism. It is important to examine regulation mechanism of acetate uptake in glia. In this study, we examined the effects of glutamate, extracellular ion concentrations and oxidative radicals on 14C-acetate uptake in cultured rat astrocytes to characterize 14C-acetate uptake in glia.
Methods: Astrocytes were prepared from cerebral cortices of 1-day-old Sprague-Dawley rats. In brief, the cells were cultured in Eagle's minimal essential medium (MEM) containing 10% FCS and 2 mM l-glutamate at 37 degree. Then the secondary cultures were grown in complete medium for 14 days. To measure 14C-acetate uptake, astrocytes were incubated at 37 degree in 24 well culture plates with 0.0185 mM 1-14C-acetate (2.0 GBq/mmol; Perkin Elmer Life Sciences Inc. Boston, MA, USA) in glucose containing HEPES-buffered HBSS. The uptake process was terminated by rapid removal of the radioactive medium. Cells were harvested into 0.5 ml of 0.5 N NaOH with 0.05% lauryl sulfate over night, separate aliquots were taken for protein determination and for measurement of radioactivity.
Results and Discussion: The uptake of 14C-acetate into cultured rat astrocytes increased with time of incubation at 37 degree. The uptake was completely blocked by 1 mM pCMBS (a selective inhibitor of MCT-1), where no inhibition was observed by 5 mM α-cyano-3-hydroxycinnamate (a selective inhibitor of MCT-2). Acetate (0.0185 to 20 mM)- or H+-dependent 14C-acetate uptake into astrocytes appeared to be mediated by a carrier with properties similar to that of monocarboxylate transport, same as intact animal brain. The inhibitory action of fluorocitrate was also observed. A significant reduction (more than 50%) in 14C-acetate uptake was observed with 0.5 mM fluorocitrate treatment. Fluorocitrate has been used as a selective inhibitor of aconitase in the glial TCA cycle. Therefore, 14C-acetate uptake into astrocytes reflects not only the transport process but also the metabolic process. Neither glutamate (0.5 mM) nor NMDA (0.1 mM) caused significant alteration in 14C-acetate uptake, which shows glutamate had no effect by itself on 14C-acetate uptake by astrocytes in such a short period (30 minutes). The Ca2+ ionophore A23187 and the Na+ ionophore monensin caused a significant reduction in 14C-acetate uptake, whereas high K+ concentration did not affect the 14C-acetate uptake. The inhibitory effect of hydrogen peroxide on 14C-acetate uptake was also observed. These Results: indicate that Ca2+ and Na+ concentration and reactive oxygen species might have important roles in regulation of 14C-acetate uptake in astrocytes.
INCREASED BINDING POTENTIAL OF [11C]PK11195 IN ALZHEIMER DISEASE USING A NEW SRTM ACCOUNTING FOR VASCULAR BINDING
Giampaolo Tomasi1, Alesandra Bertoldo1, Paul Edison2,3, Federico Roncaroli4, Nicola Pavese2, Alexander Hammers2,3, Yen Foung Tai2,3, Claudio Cobelli1, David J. Brooks2,3, Alexander Gerhard3, Federico Turkheimer2,3
1Department of Information Engineering, University of Padova, Padova, Italy, 2Department of Clinical Neuroscience, Division of Neuroscience, Imperial College London, 3MRC Clinical Sciences Centre, Hammersmith Hospital, 4Department of Neuropathology, Division of Neuroscience and Mental Health, Imperial College London, London, UK
INTRODUCTION: [11C]PK11195 PET is a marker of activated microglia. Currently, the simplified reference tissue model (SRTM) is the most used approach to estimate binding potential (BP) of this tracer.
Standard SRTM can be modified by accounting for cerebral vasculature binding both in reference and target tissues (SRTMV) [1]. Here we analyze the use of SRTMV in comparison with SRTM in controls and patients suffering from Alzheimer's disease(AD).
MATERIAL AND METHODS: 18 healthy subjects and 10 patients suffering from AD were considered. Each study was performed on an ECAT EXACT 3D scanner and each summed image was coregistered to the corresponding T1-weighted MRI scan and normalized to the stereotaxic space using SPM5 software. The time-activity-curves of several ROIs were extracted from the normalized image using the Hammersmith maximum probability atlas. BPs were first estimated using a classical SRTM with the reference tissue time course (CR) computed with the supervised selection algorithm described in [2]. A new version of the SRTM (SRTMV) was subsequently used which accounts for binding of [11C]PK11195 to vasculature in both target (Vb) and reference (VbR) regions. A fast algorithm which selected the pixels with the highest peak in the early frames was used to extract the vasculature tracer activity (Cb) directly from the images. CR and Cb were then used as input functions to SRTMV, VbR fixed to 5%, and RI, k2, BP and Vb estimated.
Results: Figure 1 shows mean (±SD) BP estimates obtained in the six considered ROIs by using a SRTM (blue bars) and SRTMV (red bars) in controls(left panel) and patients(right panel). The inclusion of a vasculature component in SRTM increases BPs by about 10% in controls and 17% in AD patients. P-values obtained thorugh STRMV, indicators of higher BPs in AD patients, were significantly lower than SRTM p-values. In addition average Vb values derived in the 6 ROIs using SRTMV are 4% for controls but only 2.8% in patients.
CONCLUSION: This study shows that accounting for vasculature binding (SRTMV) leads to an increase of BP in comparison with classical SRTM. The extent of BP increase is different in controls with respect to AD patients. This fact, which requires further elucidation, has a diagnostic impact since it allows a clearer differentiation between controls and AD patients. Further investigations are needed to assess the potential role of vasculature changes in the pathophysiology of AD.
NORMAL DATABASE OF THE SEROTONERGIC NEUROTRANSMISSION SYSTEM IN LIVING HUMAN BRAIN AS MEASURED BY PET
Harumasa Takano1, Hiroshi Ito1, Hidehiko Takahashi1, Ryosuke Arakawa1, Ryohei Matsumoto1, Tetsuya Suhara2
1Clinical Neuroimaging Team, Molecular Neuroimaging Group, 2Molecular Neuroimaging Group, Molecular Imaging Center, National Institute of Radiological Sciences, Chiba, Japan
Background and aims: The central serotonergic (5-hydroxytryptamine; 5-HT) system is known to be closely involved in the pathophysiology of depression, anxiety disorders, and other neuropsychiatric disorders. The postsynaptic 5-HT receptors are known to exhibit great heterogeneity and can be classified into 7 major classes with more than 16 subtypes. Recent advances in PET techniques have facilitated visualization of diverse neurotransmitter systems in vivo. However, with regard to the serotonergic system, suitable radioligands have only been available for the 5-HT1A and 5-HT2A receptors and the 5-HT transporter (5-HTT). Moreover, no integrated database of both pre- and postsynaptic serotonergic functions that include the 5-HT receptors and the 5-HTT have been reported. In the present study, we have generated normal databases of the pre- and postsynaptic serotonergic functions in the living human brain by using PET.
Methods: PET scans were performed on young, healthy men after intravenous injections of [C-11]WAY-100635, [C-11]NMSP, or [C-11]DASB to measure the binding of 5-HT1A and 5-HT2A receptors or 5-HTT, respectively (9 men for WAY; 5 men for NMSP; and 9 men for DASB). With the cerebellum as a reference region, the binding potential (BP) was calculated by using the reference tissue model method on a voxel-by-voxel basis. All the PET images were anatomically standardized using the SPM2 software, and databases were generated for each radiotracer.
Results: These databases facilitate comparison of the regional distributions of the 5-HT1A and 5-HT2A receptor binding and 5-HTT binding. The BP values of each region-of-interest are presented in Table 1 (mean ± SD). With regard to [C-11]WAY-100635, the highest bindings were observed in the hippocampal region. Moderate bindings were observed in the raphe nucleus, anterior cingulate and neocortical regions that include the temporal and frontal cortices. In contrast, the relatively low bindings were observed in the striatum and thalamus. With regard to [C-11]NMSP, moderate bindings were observed in the thalamus, neocortical and limbic regions such as the anterior cingulate and hippocampal region. A high accumulation of [C-11]NMSP in the striatum reflects binding to the dopamine D2 receptors. Further, with regard to [C-11]DASB, relatively high bindings were observed in the thalamus, striatum, and raphe nucleus, while low bindings were observed in the neocortical and limbic regions. These distributions were in agreement with those observed in human post-mortem studies.
Conclusions: Although the limited spatial resolution of the PET scanner might hamper the precise observation of pre- and postsynaptic functions of 5-HT, such databases should facilitate the understanding of neurotransmission physiology in the living human brain. These databases can also be used to investigate the regional alterations of serotonergic transmission in various neuropsychiatric disorders.
CONSTRUCTION AND EVALATION OF F-18 FDG PET PROBABILISTIC MAP FOR VOXEL BASED ANALYSIS OF THE RAT BRAIN
Ki Chun Im1, Jae Seung Kim1, Young Shin Ra2, Dae Hyuk Moon1
1Department of Nuclear Medicine, 2Department of Neurosurgery, Asan Medical Center, Seoul, South Korea
Introduction
Automated voxel-based analysis of rodent small-animal PET data is necessary for optimal use of information because the number of available resolution elements is limited. Parametric maps may be statistically analyzed at the voxel level, given suitable techniques for both the spatial normalization of image data into a standard space and the application of appropriate statistical tests. The purpose of this study was to develop F-18 FDG PET and MRI template for normal rat brain. Also, feasibility of SPM in detailed regional analysis of molecular changes in the rat brain was explored for F-18 FDG PET imaging of a model of traumatic brain injury (TBI).
Method
Ten adult Sprague-Dawley normal rats were scanned with PET and MRI. The PET images were acquired with 3D mode using microPET focus 120 scanner after injection of 37 MBq F-18 FDG. T2-weighted MR images were acquired using Bruker Biospec 4.7T MRI system. A MRI-based spatial normalization was used. The normal rat PET images were coregistered to T2-weighted MR images. Maximum mutual information (MMI) registrations and affine spatial normalizations were performed using SPM2. The spatial normalization of the MRI to the standard MRI was applied to the integral images. The normalized PET images were averaged voxel wise to create rat PET template. F-18 FDG PET images were count normalized to the whole-brain uptake. Eight TBI model rats were subjected to a moderate lateral fluid percussion injury (FPI). At two days, one week, three weeks and five weeks post FPI, PET images of 8 TBI rats were acquired four times. Data voxel brain size was scaled by a factor of 10 to approximately fit the human brain size. Using SPM2 software, TBI PET images were realigned, spatially normalized to a created PET-template and smoothed (8 mm FWHM). To assess the extent and severity of significant hypometabolic lesions in TBI compared to normal controls were obtained by a two-sided t-test of SPM (uncorrected p < 0.001, extent threshold>50 voxel) in all PET images of TBI.
Result
Visually acceptable F-18 FDG rat PET and MRI templates were created. Registration errors were negligible when MMI procedure was used to register a translated or rotated image volume. Thirty-two PET studies of 8 TBI model subjects were obtained. SPM analysis showed injured distribution of decrease F-18 FDG uptake in TBI rats compared with normal rats. In SPM analysis, the extent and severity of significant hypometabolic lesions were decreased according to a significant effect of time. At five weeks injured animals showed F-18 FDG uptake recovery using SPM analysis.
Discussion & Conclusion
These Results: indicate that voxel-based Methods: will be useful for future larger longitudinal studies of rat brain. Small-animal PET with advanced image processing is likely to play a useful role in detailed in vivo molecular imaging of the rat brain. MRI-based small-animal PET templates facilitate accurate assessment and spatial localization of rat brain function using voxel-based analysis.
EPITOOL - A MULTIMODALITY FRAMEWORK FOR INTEGRATIVE ANALYSIS OF FUNCTIONAL, ANATOMICAL AND ELECTROPHYSIOLOGICAL DATA IN EPILEPSY
Otto Muzik1,2, Jing Hua3, Darshan Pai3, Guangyou Zou3, Malek Makki1,2, Diane Chugani1,2, Eishi Asano2, Harry Chugani1,2
1Carman and Ann Adams Department of Pediatrics, Children's Hospital of Michigan, 2Division of Pediatric Neurology, 3Department of Computer Science, Wayne State University, Detroit, MI, USA
Background and aims: In order to obtain a more detailed understanding of processes involved in normal and abnormal brain function, integrated analysis of multimodality neuroimaging data across patients is necessary. The present work describes a computationally efficient framework for integrative and quantitative analyses of data obtained from positron emission tomography (PET), diffusion tensor imaging (DTI) and electroencephalography (EEG) based on statistical conformal geometry.
Methods: Landmark-constrained conformal mapping of the brain surface preserves angular relationships between cortical landmarks and allows a unique and accurate transformation of each subject's brain surface to a canonical spherical domain (the Statistical Conformal Brain Model ? SCBM) where finite elements can be defined geometrically on the sphere. Following matching of surface landmarks in the spherical domain, these finite elements can be subsequently reversely mapped into native space where they represent homotopic finite elements defined on the brain surface of individual subjects. By extending these finite elements into the brain using inverse gradient fusion, finite volume elements (FVEs) of the cortex can be created. Subsequently, all data sampling and analysis is performed in the subject's native space. As the FVEs correspond spatially between different brains, the characteristics of a particular FVE in native space (function, connectivity, electrophysiology) is tied to the same index in the SCBM and can be assessed statistically. The PET tracer concentration in these FVEs can be then analyzed with respect to a normative tracer distribution, yielding the location of objectively defined functional abnormalities. Moreover, PET abnormalities can be assessed with respect to data obtained from intracranial EEG or with respect to the pattern of cortico-cortical and cortico-subcortical fiber tracts derived from DTI. We present here an integrated WinXP software application which includes advanced display features for PET, DTI and EEG data in order to allow the review of all multi-modality information in one easily distributed software application. The proposed method is largely automated, relying solely on surface landmarks and objective parameters calculated from the data. No arbitrary thresholds or manual brain region definitions are required from the user.
Results: To test the spatial accuracy of the conformal mapping technique, we applied our method to a group of 6 normal adult controls. Three landmarks not used for conformal matching were defined and following conformal mapping the agreement between FVEs which overlayed a given landmark in the different brains was calculated. We obtained an agreement of >90% for all three landmarks.
Conclusion: The integration of complementary modalities is likely to result in powerful paradigms that take advantage of qualitatively different information inherent in diverse datasets. The presented software tool allows accurate integration of multi-modality data sets in epilepsy and is likely to provide new insights regarding pathophysiological mechanisms underlying this, and probably other, neurological disorders.
EFFECT OF SCAN DURATION ON THE LOGAN ANALYSIS OF [11C]PIB PET
Noriko Tanaka, Koichi Sato, Kiyoshi Fukushi, HItoshi Shinotoh, Hitoshi Shimada, Tsuneyoshi Ota, Tetsuya Suhara, Toshiaki Irie
Molecular Imaging Center, National Institute of Radiological Sciences, Chiba, Japan
Background and aims: [11C]PIB has been successfully used for measurement of amyloid deposition in Alzheimer's disease and related disorders by PET. A previous study showed the Logan graphical analysis with arterial input function and 60 min of emission data were associated with higher test-retest variability by 33% on average than those that used 90 min of emission data in 8 subjects1). We study the effect of scan duration on the Logan graphical analysis of [11C]PIB PET data in our institute.
Methods: The subjects consisted of 11 healthy subjects (5 men and 6 women) ranging in age from 48 to 91 years with a mean age of 64 years. [11C]PIB -PET acquisition was a dynamic sequence of 19 PET scans (3 × 20s, 3 × 40s, 1 × 60 s, 2 × 180s, 5 × 360s, 5 × 600s) for 90 min. Twenty seven timed-arterial samples were collected over 85 min. The TLC analysis was performed to determine the unmetabolized fraction of PIB in plasma. ROIs were defined manually in the subcortical white matter, 16 cortical regions, striatum and cerebellum. The Logan graphical analysis were applied over the 35 to 60 min (3 points), 35 to 70 min (4 points), 35 to 80 min (5 points), and 35 to 90 mins PET scan intervals (6 points) to calculate distribution volume (DV). The cerebellum was used as the reference tissue to calculate DV ratio (DVR). The performance of these analyses was assessed based on the magnitude of coefficient of variation (COV) and bias from the DVR in the analysis with 90 min of emission data.
Results: The COV of DVR ranged from 10 to 25% in each ROI in the analysis. The COV of DVR did not change with the four scan intervals in each ROI in this study. The bias of DVR in each ROI raged from ?3 to +2% in the analysis with 70 and 80 min of emission data. The bias of DVR in each ROI ranged from ?5 to 6% in the analysis with 60 min of emission data.
Discussion and Conclusions: The bias of DVR in each ROI was relatively large in the analysis with 60 min of emission data probably because linear regression was performed on only three points when the steady-state of the PIB kinetics might not be reached in some subjects. Since the COV were similar in the analysis with the four scan intervals, the scan duration of [11C]PIB PET could be shortened to 60 min without loss of diagnostic sensitivity of Alzheimer's disease as the previous study suggested 1).
ACCURATE DELAY CORRECTION FOR QUANTITATIVE MEASUREMENT OF CEREBRAL BLOOD FLOW AND ARTERIAL-TO-CAPILLARY BLOOD VOLUME (V0)
Hidehiko Okazawa1, Tetsuya Tsujikawa1, Masato Kobayash1, Makoto Isozaki2, Yoshikazu Arai2
1Biomedical Imaging Research Center, 2Department of Neurosurgery, Faculty of Medical Sciences, University of Fukui, Eiheiji-Cho, Fukui, Japan
Background and aims: Evaluation of regional cerebral blood flow (rCBF) and arterial-to-capillary blood volume (V0) is a good indicator of critical hemodynamic status in the severely affected areas of cerebrovascular stenoocclusive disease. In calculation of these parameters, arterial input function is usually determined by arterial blood sampling, and delay of the input is corrected for the whole brain or slice-by-slice. However, delay of the tracer is assumed to be different between the hemispheres in patients with unilateral occlusive lesions. To calculate accurate quantitative values of rCBF and V0, delay of the tracer arrival time was estimated pixel-by-pixel, and applied the delay time for calculation of parameters.
Methods: Thirty-five patients (mean age=65±10y) with unilateral major cerebral arterial stenoocclusive disease underwent O-15 water and O-15 gas PET scans. All patients had occlusion or stenosis (>70%) in the internal carotid or middle cerebral arteries (MCA). For estimation of input function, arterial blood radioactivity was measured continuously using an automatic counter with a constant flow from the brachial artery. The delay of the input was estimated pixel-by-pixel and delay map of tracer arrival time was created. The 3-weighted-integral method was employed for calculation of rCBF and V0 using the delay map for pixel-by-pixel delay correction. Cerebral blood volume, oxygen extraction fraction (OEF) and cerebral metabolic rate of oxygen were also calculated from the O-15 gas PET scans.
Results: Three of 35 patients studied showed misery perfusion with elevation of OEF ipsilateral to the stenoocclusive lesion. All of them had significant decrease in rCBF and increase in delay of tracer arrival. Figure shows representative parametric images of misery perfusion in right MCA territory. The remaining patients had OEF in normal range and no differences between the cerebral hemispheres in delay images. V0 showed a slight decrease in the impaired hemisphere of patients with misery perfusion.
Conclusion: Pixel-by-pixel estimation of delay of the input function provided a delay map for regional differences and accurate parametric values for evaluation of hemodynamic status in the ischemic cerebrovascular disease. The delay image, correlated well with OEF change, may indicate OEF elevation in impaired circulation when combined with accurate rCBF and V0 images.
Figure: Parametric images of patients with right MCA occlusion
DISTRIBUTION VOLUMES AS AN ALTERNATIVE TO BINDING POTENTIALS FOR SIGMA1 IMAGING
Yuichi Kimura1, Mika Naganawa1,2, Muneyuki Sakata3, Masatomo Ishikawa4, Masahiro Mishina5, Keiichi Oda1, Kenji Ishii1, Kiichi Ishiwata1
1Positron Medical Center, Tokyo Metropolitan Institute of Gerontology, 2Japanese Society for The Promotion of Science, Tokyo, 3Graduate School of Information Science, Nara Institute of Science and Technology, Nara, 4Department of Psychiatry, Chiba University, Chiba, Japan, 5Neurological Institute, Nippon Medical School Chiba, Hokusoh Hospital, Chiba, Japan
Introduction
Visualization of neuroreceptor density is important, but the algorithms for receptor imaging assume the existence of reference regions. If the target receptor distributes globally in the brain, only total distribution volume (DVt) is obtainable using the Logan plot. In this study, the applicability of DVt is investigated as an alternative to binding potentials (Bp) for the receptor that have no reference regions such as the sigma1 receptors in the case of [11C]SA4503 using wide range of physiological data.
Method
Investigated data were originated from 54 cases including 46 normals, 5 Alzheimer type dementia cases, and 3 schizophrenia patients ranging from 21 to 70 yrs. 19 of the normals were participated drug loading experiments using haloperidol, fluvoxamine, and paroxetine that has high, moderate, and negligible affinity to the sigma1 receptors, respectively. Totally 444 ROIs were placed, and Bp and DVt were estimated using nonlinear model estimation proposed in [1].
Results and Duscussion: The good linear relationship was observed. Since SA4503 has height affinity to the sigma1 receptors, and about only 5% of DVt is occupied as a free compartment, the change of DVt is dominantly governed by Bp. We can conclude that DVt can be used as the alternative to Bp.
ACUTE AND CHRONIC EFFECTS OF CITALOPRAM ON [18F]MPPF BINDING TO 5-HT1A RECEPTORS AND RECEPTOR-G PROTEIN INTERACTIONS
Marcelle Moulin-Sallanon1, Yves Charnay2, Nathalie Ginovart3,4, Antonietta Bartoli5, Alberto Del Guerra5, Laurent Lemoucheux6, Vicente Ibanez3, Daniel Fagret1, Philippe Millet3
1INSERM, Unit E340, La Tronche, France, 2Morphology Unit, Division of Neuropsychiatry, Department of Psychiatry, University Hospital of Geneva, Geneva, Switzerland, 3Psychiatric Neuroimaging Unit, Division of Neuropsychiatry, Department of Psychiatry, University Hospital of Geneva (HUG), Geneva, Switzerland, 4Department of Psychiatry, University of Geneva, Geneva, Switzerland, 5Department of Physics, University of Pisa, Pisa, Italy, 6Advanced Accelerator Application, Saint Genix Pouilly, France
Introduction: Selective serotonin re-uptake inhibitors (SSRIs) are potent antidepressants that produce their maximal therapeutic effects with a 3 to 4 week delay. This delay has been mainly attributed to gradual desensitisation processes at somatodendritic serotonin 5-HT1A autoreceptors. In this study, we evaluated potential adaptative changes of 5-HT1A receptors after acute and chronic citalopram (CIT) challenges. Methods: CIT treatments were applied on male Wistar rats either acutely (0.5mg/kg) through intraperitoneal injections, or chronically (10mg/kg/day; 21 days) through osmotic minipumps implanted subcutaneously. (1) Small animal PET procedures were carried out on 14 acutely-treated rats, under urethane anaesthesia. Sixty minutes after treatment, 1.5 mCi [18F]MPPF was given intravenously. PET acquisitions were performed on a YAP(S)PET (*). The simplified reference tissue model (SRTM) was used to estimate binding potential (BPSTRM) values in frontal cortex (FCx), hippocampus and dorsalis raphe nucleus (DR) with the cerebellum as the reference region. (2) Quantitative autoradiography was performed on 6 acutely- (2a) and 12 chronically-treated (2b) rats. After 60 min (2a) and 21 days (2b) of treatment, respectively, acutely- and chronically-treated rats were intravenously injected with [18F]MPPF (1.5 mCi) and sacrificed by decapitation 30 minutes later. The brain was removed and quickly frozen. Four identical slide series of brain coronal sections (20µm of thickness) were prepared in the region of FCx and cingulate cortex (CCx), hippocampus, DR and median raphe nuclei (MR). One slide series was exposed for 20 min onto autoradiographic films which were then scanned for quantitative analyses. (3) Three series of slides were kept at −80oC until 8-OH-DPAT (5-HT1A receptors agonist) stimulated [35S]GTPγS binding assay as described by Hensler (**). Results: An acute treatment with 0.5mg/kg of CIT as well as a 21 day chronic exposure to CIT (10mk/kg; sc) did not significantly modify [18F]MPPF binding in serotoninergic cell body regions of DR and MR and in serotonin terminal regions of FCx, CCx and hippocampus. [18F]MPPF has previously been described as a poor tool to detect changes in extracellular 5-HT concentration but may detect alteration in 5-HT1A receptors number. We thus suggest that no evident 5-HT1A receptors number decrease occur during acute or chronic treatment with CIT. We did not observed any modification of the 5-HT1A receptor agonist capacity to activate G-protein in the DR or MR after either acute or chronic CIT treatment. However, chronic CIT treatment slightly increased 8-OH-DPAT stimulated GTPγS binding in two serotoninergic terminal regions, the FCx and hippocampus. Conclusion: Our study showed no significant alteration in 5-HT1A receptor densities in brain following an acute or a chronic CIT treatment. The delayed therapeutic efficacy of CIT thus do not appear to be linked to either a regulation of 5-HT1A autoreceptor number or to a 5-HT1A receptor-G protein decoupling process in 5-HT cell body regions.
Grant support: Swiss National Science Foundation N° 3100A0-104185
PARAMETRIC MAPPING OF CEREBRAL BLOOD FLOW WITH PARTIAL VOLUME EFFECT CORRECTION USING O-15 WATER PET AND CLUSTER ANALYSIS
Kyeong Min Kim1, Hiroshi Watabe2, Takuya Hayashi2, Gi Jeong Cheon1, Hidehiro Iida2
1Laboratory of Nuclear Medicine, Korea Institute of Radiological and Medical Sciences, Seoul, South Korea, 2Department of Investigative Radiology, National Cardiovascular Center Research Institute, Suita, Osaka, Japan
Introduction: Partial volume effect (PVE) due to limited spatial resolution of PET image is one of factor resulting in significant underestimation of cerebral blood flow (CBF) in O-15 water PET study. There have been many approaches for PVE correction of PET image. Among the approaches, Iida et al [1] suggested a method for regional CBF estimation with PVE correction based on kinetic modeling of O-15 water. In this study, we show an approach to generate parametric images of CBF with PVE effect, in addition to the images of cerebral vascular volume (V0) and perfusable tissue fraction (PTF), based on the method proposed by Iida et al.
Methods: Two-compartment model of O-15 water proposed by Iida et al [1],[2] and Ohta et al [3] was used in this approach. When applied to parametric mapping, cluster analysis [4] was employed to reduce image generation time and noise on parametric images. In the estimation, the value of partition coefficient of tissue was fixed to 0.9 [1]. Immediately after a bolus injection of O-15 water, dynamic PET scan (ECAT EXACT47 Siemens-CTI, TN) was performed on 7 patients with cerebrovascular disease. During PET scan, arterial input function was acquired and corrected for delay and dispersion. A global region of interest was placed on MCA area with higher perfusion
Results: The parametric images of CBF with PVE correction, PTF and V0 could be generated by means of this approach. Clustering analysis could shorten pixel-by-pixel estimation time and provide improved quality of parametric images (Figure). The values of CBF with and without PVE correction, PTF, and V0 were 0.47 ± 0.09 ml/min/g, 0.36 ± 0.06 ml/min/g, 0.70 ± 0.07 g/ml, and 0.02 ± 0.01 ml/ml, respectively.
Conclusion: PVE corrected CBF image with improved image quality could be obtained by means of model-based approach using two-compartment model of O-15 water and cluster analysis. This Results: show that the presented approach has a potential to contribute to improved diagnosis of cerebrovascular disease and activation using O-15 water PET.
MODELING SEROTONIN 1A RECEPTORS IN BABOON USING THE AGONIST TRACER [11C]CUNBID-101 AND POSITRON EMISSION TOMOGRAPHY
Ramin V. Parsey1,4, Matthew S. Milak1,4, Alin J. Severance4, R. Todd Ogden1,3,4, J.S. Dileep Kumar1,4, Vattoly J. Majo1, Jaya Prabhakaran4, Pratap Mali4, J. John Mann1,2,4
1Department of Psychiatry, 2Department of Radiology, Columbia University of Physicians and Surgeons, 3Department of Biostatistics, Mailman School of Public Health, Columbia University, 4Department of Neuroscience, New York State Psychiatric Institute, New York, NY, USA
Background: Several lines of evidence implicate serotonin 1A (5-HT1A) receptors in the pathophysiology of major neuropsychiatric disorders. In vivo quantification of 5-HT1A receptors to date has relied on various 5-HT1A antagonists. We have recently reported [11C]-[O-Methyl-11C] 2- {4- [4- (7- methoxynaphthalen-1-yl) piperazin-1-yl] -butyl} −4-methyl-2H- [1,2,4] triazine-3,5-dione ([11C]MPT) as a promising 5-HT1A receptor agonist radiotracer in baboons. Although [11C]MPT (5-HT1A Ki = 1.4 nM, Emax = 95% and EC50 = 0.05 nM) demonstrated specific binding the rate of decay of radioactivity in all brain regions was relatively slow in baboons making quantification difficult. Structure activity relationship studies suggest that phenyl analogues of MPT are excellent 5-HT1A agonists. The 2-methoxyphenyl analogue MMT possesses a 5-HT1A Ki = 0.7 nM, EC50 = 0.3 nM, and Emax = 95%. However [11C]MMT did not show specific binding in baboons. We recently published the synthesis and initial evaluation of [O-methyl-11C] 2- (4- (4- (2-methoxyphenyl) piperazin-1-yl) butyl) −4-methyl-1,2,4-triazine-3,5 (2H,4H) dione ([11C]MMP) as a superior agonist PET ligand than [11C]MPT. Here we report the evaluation of in vivo modeling options for [11C]MMP in anesthetized baboons under its new name, [11C]CUNBID-101.
Methods: A series of PET scans were performed in two male baboons (Papio anubis) with an ECAT EXACT HR+ scanner. 4.88 ± 0.52 mCi (SA 1.84 ± 0.39 Ci/µmol) of [11C]CUNBID-101 was injected as an i.v. bolus and emission data were collected for 120 min in 3D mode. We evaluated several different models (one- and two-tissue compartment (1TC, 2TC) iterative and non-iterative kinetic models, likelihood estimation in graphical analysis (LEGA; a bias-free alternative to the graphical method), and basis pursuit), using binding potential (BP = Bmax/KD). To assess the performance of each model, we compared Results: using five different metrics (percent difference and within subject mean sum of squares (WSMSS) for reproducibility, variance across subjects, identifiability based on bootstrap resampling of residuals for each method, and time stability analysis to determine minimal required scanning time). Models were also evaluated in scans performed after injections of the 5-HT1A antagonist WAY100635 (0.5 mg/kg, i.v.) and the 5-HT1A agonist 8-OH-DPAT (2 mg/kg, i.v.).
Results: and Conclusions: All metabolites are more polar than [11C]CUNBID-101 and no significant change in metabolites was observed in the blocking studies. The free fraction is 59 ± 3%. 100 minutes of scanning time is adequate and for ROI-level analysis, LEGA gives best Results:. Median test-retest percent difference for BP is 9.07% ± 4.33% across all regions; WSMSS = 2.22; variance = 4.42; bootstrap identifiability = 0.59. Pre-administration of WAY100635 and 8-OH-DPAT resulted in 87% ± 5% and 76% ± 4% average reductions in BP values respectively across all ROIs. Based on a measurable free fraction, high affinity, selectivity, blood brain permeability and favorable plasma and brain kinetics, [11C]CUNBID-101 is an excellent candidate for imaging high affinity 5-HT1A receptors.
KINETIC MODELING OF THE NET RADIOLIGAND [C-11]MRB IN HUMANS: A TEST-RETEST STUDY
Jean-Dominique Gallezot1, B. Planeta-Wilson1, G.J. Wang2, R.E. Carson1, Y.S. Ding1,2
1PET Center, Department of Diagnostic Radiology, Yale University, New Haven, CT, 2Medical Department, Brookhaven National Laboratory, Upton, NY, USA
Purpose: The ability to quantify in vivo the density of nor-epinephrine transporters (NET) has been hampered by the lack of suitable radioligands. [C-11]MRB is a new radioligand with suitable properties in monkeys [1],[2]. The aim of this study was to evaluate [C-11]MRB in humans, and to investigate possible quantification strategies.
Methods: Five volunteers were included in a test-retest protocol. Each subject was injected twice, 150 minutes apart, followed by a 90-min dynamic acquisition on the HR+ scanner and the measurement of the metabolite-corrected arterial input function. Time-activity curves (TACs) were computed for the occipital, frontal, parietal, and temporal cortices, caudate, putamen, cerebellum and thalamus. TACs were analyzed using the Logan graphical analysis with input function (GA) [3] or with reference region (GAr) [4], compartmental modeling with one (1C) or two (2C) tissue compartments [5], and the SRTM [6] and SRTM2 [7] Methods:.
The main outcome measures were the distribution volume (DV) or the relative DV (DVR). Reproducibility of the measures was estimated by computing the mean (mTR) and standard deviation (σTR) of the relative difference of the DV(R) estimates between test and retest scans.
Results: Fit quality was good to excellent with the 2C model and both SRTM Methods:, but was bad to acceptable with the 1C model, depending more on the individual subject than on the ROI. GA plots could be considered linear from 30 min postinjection on.
For Methods: using the arterial input function, reproducible DV values were obtained with the 1C model (5%<σTR<10%) and the GA (3%<σTR<19%), but not with the 2C model (σTR>75%). DV values estimated with the 1C model were lower (−13±5%) than the GA values, due to the lack of fit observed with that model. GA DV values ranged from 4.2±0.2 (caudate) to 7.3±1.9 (thalamus).
The occipital cortex and/or the putamen may be used as a reference region in monkeys [2]. In this study the occipital and putamen DV values were the same (5.0±1.1 and 5.0±0.8, respectively), and the occipital was chosen as the reference region. Reproducible DVR values were obtained with GAr (1%<σTR<7%) and SRTM2 (2%<σTR<8%) Methods:, but not with the SRTM method (σTR=100% in thalamus). 1C, GA, GAr and SRTM2 Methods: provided very similar DVR values (mean relative difference < 1%). GAr DVR values ranged from 0.80±0.12 (caudate) to 1.44±0.24 (thalamus).
There was no apparent difference between test and retest scans in any regions (mTR<σTR for all analyses).
Discussion: For regional analysis, GA and GAr seem to be the most readily applicable Methods: for the analysis of regional TACs with [C-11]MRB. Further investigation of the SRTM2 method for voxel-based analyses is ongoing. Although the underlying hypotheses of this method are not met, this method provides accurate DVR values, and is likely to be less sensitive to noise than GA.
INCREASES IN STRIATAL FDOPA INFLUX CONSTANTS IN UHRP PATIENTS ARE NOT DUE TO CHANGES IN PLASMA OMFD CONCENTRATIONS
Marie Claude Asselin1, Oliver D. Howes2, Safiye Osman4, Kawai Yau4, Paul R. Stokes3, Paul M. Grasby3
1Wolfson Molecular Imaging Centre, The University of Manchester, Manchester, 2Institute of Psychiatry, London, UK, 3MRC Clinical Sciences Centre, Hammersmith Hospital, 4Hammersmith Imanet, Hammersmith Hospital, London, UK
Background: Quantification of brain PET studies with 6-[F-18]fluoro-L-DOPA(FDOPA) is complicated by its peripheral conversion by catechol-O-methyltransferase(COMT) into the metabolite 6-[F-18]fluoro-3-O-methyl-L-DOPA(OMFD) which is also transported across the blood-brain barrier. Given that genetic variation in the COMT gene may be associated with altered prefrontal cortex function and that the COMT genotypes differ in enzymatic activity[1], the present work aimed to investigate whether changes in peripheral metabolism of FDOPA could account for increases in striatal uptake in patients at ultra high risk of psychosis (UHRP) [2], who have a 20–40% chance of developing a psychotic illness in the next year[3].
Methods: Four healthy subjects (21-31years) and nine patients (20-35years) meeting UHRP criteria[3] underwent a 90-min PET scan following the injection of 150MBq of FDOPA. All subjects were pre-treated with 400mg of entacapone (peripheral COMT inhibitor) and 150mg of carbidopa (amino acid decarboxylase inhibitor) 60min prior to radiotracer injection. Discrete venous blood samples were taken for genotyping (pre-injection) and measurement of radiolabelled metabolites (at 10, 30, 60, 90min) using HPLC. The time course of the OMFD fraction in plasma was fitted to a monoexponential increase to a constant. A Patlak plot was constructed for the total striatal uptake using the cerebellar uptake as the input function and the slope (the influx constant Kic) was estimated from a linear fit of the transformed data corresponding to the time interval 20-90min.
Results: The OMFD fraction in plasma at 60min was significantly higher in the UHRP patient group than in the pilot control group (0.31±0.11 vs 0.16±0.07, p<0.01). Individual measurements and fits are plotted in Fig.1 alongside the mean of those previously measured without(x, solid line) or with(+, solid line) 400mg of entacapone in six Parkinson disease patients[4] who have been found not to differ from controls[5]. Kic values significantly decreased with increasing OMFD fraction (Fig.2, R2=0.27, p<0.04) at a rate (~10% decrease in Kic when OMFD fraction doubled) which is in accordance to prediction from simulated data[6].
Conclusion: Entacapone appears to be less effective in UHRP patients than in controls at reducing peripheral FDOPA metabolism by COMT. Higher plasma OMFD concentrations would not explain increases in Kic but instead underestimate the magnitude of real increases in striatal FDOPA uptake in URHP patients. Confirmation of these preliminary findings awaits expansion of the control group and determination of the COMT genotypes.
PHARMACOKINETICS OF [11C]VERAPAMIL DURING CYCLOSPORIN A INDUCED P-GLYCOPROTEIN INHIBITION
Stina Syvanen1,2, Andrew Hooker2, Obaidur Rahman1, Helena Wilking1, Gunnar Blomquist3, Bengt Langstrom1,4, Mats Bergstrom5, Margareta Hammarlund-Udenaes2
1Uppsala Imanet, GE Healthcare, 2Department of Pharmaceutical Biosciences, 3Department of Oncology, Radiology and Clinical Immunology, 4Department of Biochemistry and Organic Chemistry, Uppsala University, Uppsala, Sweden, 5Department of Clinical Imaging, Novartis Pharma, Basel, Switzerland
Background and aims: [11C]verapamil and the P-glycoprotein (P-gp) inhibitor cyclosporin A (CsA) have been studied as a model system for drug interaction with P-gp (1)–(3). The aim of the present study was to investigate with [11C]verapamil the degree of P-gp inhibition, at a constant concentration of CsA. A second aim was to develop a pharmacokinetic model describing the brain and plasma pharmacokinetics of [11C]verapamil at different levels of P-gp inhibition.
Methods: [11C]verapamil was administered to anesthetized rats as an i.v. bolus dose followed by graded infusions via a computer controlled infusion pump to obtain a steady-state concentration of [11C]verapamil in brain. The inhibitor drug, CsA, was administered as a bolus followed by a constant infusion, 20 min after the start of the [11C]verapamil infusions. The four CsA dosing schemes used were: 7.5 mg/kg + 2.5mg/kg/h, 15 mg/kg + 5mg/kg/h, 22.5 mg/kg + 7.5mg/kg/h and 45 mg/kg + 15mg/kg/h. The [11C]verapamil infusions were stopped at 70 min. The brain uptake of [11C]verapamil was investigated dynamically for 2 hours in a sequence of PET scans in parallel to measurement of [11C]verapamil and CsA concentrations in plasma. Analysis of the metabolites of [11C]verapamil in plasma and brain was performed in a second group of rats to ensure that the measured radioactivity originated from intact [11C]verapamil and not from metabolites carrying the [11C]carbon label. Modeling of the data was performed with the NONMEM software.
Results: The CsA induced inhibition of P-gp was observed as a rapid increase in brain concentrations of [11C]verapamil. Cmax of [11C]verapamil in brain was obtained about 20 minutes after start of CsA administration and was 2, 5, 7 and 9 times higher than the baseline steady state for the four doses of CsA respectively. The CsA levels were also constant from this time point until the end of the investigation. Administration of 45 mg/kg +15 mg/kg/h CsA resulted in a two-fold increase in metabolite concentration in the brain compared to when no CsA was administered. The increase of metabolites in brain could fully be explained by a similar twofold increase in plasma metabolite concentration. The plasma concentrations of [11C]verapamil and CsA were best described with two-compartment models including proportional errors. The [11C]verapamil uptake in brain could be described with a two compartment model with CsA modulating the transport of [11C]verapamil via an indirect effect model.
Conclusions: [11C]verapamil is a good substrate for P-gp interaction studies in the brain with PET as the concentrations in brain and plasma are rapidly equilibrating. This enables fast detection of a changed P-gp function. Another advantage is that the brain-to-plasma ratio of [11C]metabolites is unaffected by the P-gp inhibition. A full pharmacokinetic model describing the plasma kinetics of CsA and [11C]verapamil, the brain kinetics of [11C]verapamil and the effect of CsA was also developed.
EVALUATION OF [C-11]R116301 AS A PET TRACER OF THE NK1 RECEPTOR: A TEST-RETEST STUDY IN HUMAN SUBJECTS
Saskia Wolfensberger1,2, Kaoru Maruyama1, Bart van Berckel1, Mark Lubberink1, Anu Airaksinen1, Ronald Boellaard1, William Carey3, Wieb Reddingius3, Dick Veltman1,2, Albert Windhorst1, Josee Leysen1,3, Adriaan Lammertsma1
1Department of Nuclear Medicine and PET Research, 2Department of Psychiatry, VU University Medical Centre, Amsterdam, The Netherlands, 3Johnson & Johnson Pharmaceutical Research and Development, Beerse, Belgium
Introduction: R116301 is an orally active, potent and selective non-peptide NK1 receptor antagonist. In a previous study [1], size and presence of the specific signal of [C-11]R116301 was demonstrated using a blocking study. Based on the striatum to cerebellum ratio, a specific signal of around 20–50% was found. To assess [C-11]R116301 further as an NK1 receptor ligand, the purpose of the present study was to assess test-retest variability of [C-11]R116301 binding.
Methods: Studies were performed in 8 normal controls. Each study consisted of two [C-11]R116301 scans, 5 hours apart. Individual scan sessions consisted of a 2D transmission scan and a 90 minutes dynamic 3D emission scan following intravenous administration of ~390 MBq [C-11]R116301 [2]. In addition, continuous on-line and discrete manual arterial blood sampling was performed to derive a metabolite corrected arterial plasma input function. A region of interest comprising whole striatum (the structure with the highest density of NK1 receptors) was defined on an individual MRI scan and projected onto both co-registered PET scans. Cerebellum was used as reference tissue. Striatum to cerebellum ratios (60-90 minutes post injection) were used as outcome parameter. In addition, striatum BP was obtained using Receptor Parametric Mapping (RPM), the basis function implementation of the simplified reference tissue model [3]. Data could not be analysed with arterial input compartment models due to severe stickiness of the tracer.
Results: Equilibrium was reached relatively early after injection, and striatum to cerebellum ratios were almost identical for the intervals 20-90 and 60-90 minutes. Test-retest Results: of striatum to cerebellum ratios (Table 1) were very tight (range 0.97–1..06), showing an average difference of 3% between scans. However, this ratio contains both specific and non-specific components. For the specific component (i.e. by subtracting 1 from the ratios), the average difference between the two scans was 10% (excluding subject 3, where no specific signal was observed). This was similar to the 9% average difference in BP between the two scans as measured with RPM (Table 1).
Conclusion: Test-retest variability of striatum to cerebellum ratios was excellent (3%). Despite the relatively high level of non-specific binding, test-retest variability of specific binding (BP) remained acceptable (10%). The large variation in specific signal between subjects needs to be addressed in future studies.
ADVANCES IN THE QUANTIFICATION OF [11C]RACLOPRIDE DYNAMIC PET WITH AMPHETAMINE CHALLENGES
Yun Zhou1, Michael R. Weed2, Ming-Kai Chen3, Arman Rahmim1, Weiguo Ye1, James R. Brasic1, Mohab Alexander1, Andrew H. Crabb1, Jennifer L. McGlothan3, Farah Ali1, Tomas R. Guilarte3, Dean F. Wong1
1Department of Radiology, 2Department of Psychiatry and Behavioral Sciences, 3Department of Environmental Health Sciences, Johns Hopkins University, Baltimore, MD, USA
The objective of the study is to improve the quantification of [11C]raclopride dynamic PET with amphetamine challenges.
Methods: [11C]raclopride was performed on two groups: (1) 19 Rhesus monkeys (3-5 kg) scanned on a HRRT scanner; (2) 26 cynomologous monkeys (5-9 kg) and 6 Papio anubis baboons (17-27 kg) scanned on a GE Advance scanner. [11C]raclopride (18-25 mCi) at high specific activity was administrated by bolus plus continuous infusion at Kbol = 75 min. Ninety-min dynamic PET scanning was started immediately after tracer injection. Forty min post tracer injection, amphetamine (2 mg/kg) was injected intravenously over 2 min. The dynamic images of 30 frames (4×0.25, 4×0.5, 3×1, 2×2, 5×4, 12×5) were reconstructed in each study. A parameter (dt) that represents a latent period in the amphetamine-induced displacement of tracer with an ESRTM (R1, k2, BP0, BP1) (Zhou et al., Neuroimage 2006, 33(2):550-63) called ESRTMdt (R1, k2, BP0, BP1, dt) was proposed for modeling tracer kinetics in the baseline phase [0 T0+dt] and the displacement phase [T0+dt T] as below: where T0 = 40 min, T = 90, and the CT(t) and CREF(t) represent the tracer concentrations at time t for target and reference tissues, respectively. R1 is the target to reference tissue ratio of transport rate constant, k2 is the efflux rate constant of target tissue, BP0 and BP1 are the tracer binding potentials in target tissue in baseline and displacement phases, respectively. A Marquardt algorithm was used to estimate parameters by fitting the ESRTMdt to the measured striatum time activity curve. For comparison, the ESRTM with given dt in [0 30] was also applied to the same data set. The amphetamine-induced percent decrease in BP (BP%) (= 100(BP0-BP1)/BP0) were calculated.
Results: Based on the Akaike information criterion, the ESRTMdt provided the best model fitting as compared to the ESRTM. The estimates of (R1, k2, BP0, BP% and dt) (mean ± SD) from ESRTMdt fitting were (0.86 ± 0.08, 0.13 ± 0.02, 2.85 ± 0.30, 28 ± 9 and 12.43 ± 3.73) for cynomologous monkeys, (0.96 ± 0.09, 0.18 ± 0.02, 4.88 ± 0.66, 28 ± 8, and 14.15 ± 6.17) for Rhesus monkeys, and (0.91 ± 0.04, 0.18 ± 0.02, 2.97 ± 0.32, 46 ± 5, and 6.66 ± 1.69) for baboons. The estimates of BP% from ESRTM fitting decreased monotonically in the given dt.
Conclusions: The ESRTMdt provided better model fitting significantly as compared to the ESRTM with given dt. The estimates of BP% from ESRTM fitting were sensitive to the pre-assumed latent period (dt). The estimated latent period values from the study will be useful for the optimization of experimental design in PET study with amphetamine challenge. Grant support: AA12839, DA00412, NS38927, MH075378, ES07062, ES019075, and Michael J. Fox Foundation for Parkinson's Research.
IN VIVO IMAGING OF AMYLOID PLAQUES IN A TRANSGENIC MOUSE MODEL OF ALZHEIMER'S DISEASE
Andre Manook1, Gjermund Henriksen1, Stefan Platzer2, Frauke Neff3, Marc Huisman1, Behrooz Yousefi1, Markus Settles4, Markus Schwaiger1, Hans Juergen Wester1, Alexander Drzezga1
1Department of Nuclear Medicine, 2Department of Neurology, 3Department of Neuropathology, 4Department of Radiology, Technical University of Munich, Munich, Germany
Background/Aim: Formation of amyloid s-plaques (As) in the brain is one of the earliest and most relevant pathophysiological processes in the development of Alzheimer's disease (AD). Modern treatment strategies are directed towards reduction of cerebral As-load. Thus, non invasive imaging of As may assist the evaluation of treatment effects. Different PET and SPECT tracers for imaging of As-load in humans are currently being evaluated in clinical trials with promising Results:. However, only a marginal retention of the lead As-radiotracers has been reported in transgenic rodent models of AD so far, despite a high expression of As detectable by means of in vitro and ex vivo binding assays. This limitation hampers preclinical As-tracer development and evaluation. The aim of the current study was to establish a feasible protocol for the in vivo As-imaging in transgenic mice, using state of the art small animal imaging technology.
Methods: 11C-PIB-PET using a small-animal PET scanner (Siemens-CTI Focus 120) with 3D acquisition in list-mode (0-60 min) and cranial MRI (Philips Achieva 1.5T with 23mm microscopy coil) were performed in 10 APP/PS1 transgenic mice (age: 9 months, n=5; 23 months, n=5) and age-matched controls. MRI and PET images were manually overlaid for anatomical correlation and definition of regions of interest (ROIs). Count rates per voxel of 11C-PIB-uptake were summed up (30-35min) and ROI-ratios were calculated. The As-free cerebellum was used as a reference region. All animals also received 3H-PIB to compare the in-vivo data with ex-vivo digital autoradiography. In separate studies, the cranial and regional brain uptake of 11C-PIB was determined by means of ex-vivo biodistribution studies.
Results: Ratios between As-rich cortical regions and the cerebellum revealed a significant difference (p< 0.001) between 23-months Alzheimer models and controls (mean ratio cortex/cerebellum 1.85 (Standard deviation (SD)= 0.03) in AD and 0.91 (SD = 0.11) in controls) and also between 9-months Alzheimer models and controls (mean ratio cortex/cerebellum 1.49 (SD = 0.20) in AD and 0.91 (SD = 0.06) in controls). Cortex/cerebellar ratios revealed higher values in all transgenic animals compared to healthy controls. The ex-vivo validation confirmed the in-vivo PET ratios and proved that the overlay of PET and CNS-MRI enabled accurate ROIdefinition and reliable determination of specific binding.
Conclusions: For the first time it has been demonstrated that by means of 11C-PIB small animal PET and a technologically advanced methodology, the amyloid beta-plaques in a transgenic rodent model can be visualized and quantified non-invasively with satisfactory Results:. The established protocol may represent a valuable screening method for As-tracer evaluation and development, and provide an important step towards translational imaging in AD-research.
USE OF IMAGE DERIVED INPUT FUNCTIONS FOR [11C]PIB STUDIES: ASSESSMENT OF ACCURACY AND TEST-RETEST VARIABILITY
Jurgen E.M. Mourik1, Nelleke Tolboom2, Bart N.M. van Berckel1, Mark Lubberink1, Philip Scheltens2, Adriaan A. Lammertsma1, Ronald Boellaard1
1Department of Nuclear Medicine and PET Research, 2Department of Neurology, VU University Medical Centre, Amsterdam, The Netherlands
Introduction: Derivation of input functions directly from dynamic PET images obviates the need for arterial sampling and will enhance clinical applicability of quantitative brain studies. The aim of the present study was to assess accuracy and test-retest variability of [11C]PIB studies when using image derived input functions (IDIF) obtained using reconstruction-based partial volume correction (PVC) [1].
Methods: PET and arterial blood data from five repeat dynamic [11C]PIB scans, acquired using an ECAT EXACT HR+ scanner and an on-line blood sampler, were used in the present study. Test and retest scans were performed on the same day. Scans were reconstructed using standard (no PVC) ordered subset expectation maximization (OSEM, 2 iteration (i), 16 subsets (s)) and a PVC-OSEM reconstruction algorithm, which corrects for the spatial resolution of the scanner. PVC-OSEM scans were reconstructed using 4i, 16s and a Gaussian recovery kernel of 5.5 mm FWHM.
For the region below the base of the skull, regions of interest (ROIs) were drawn semi-automatically over the four hottest pixels of the carotid arteries using a pseudo blood volume image, generated by summation of the early time frames (<60s). These ROIs were projected onto all dynamic frames, thereby generating corresponding IDIFs. Both IDIFs with and without calibration to manual samples taken during the scans were used for further analysis.
Parametric images of volume of distribution (Vd), based on Logan analysis, were generated using both the on-line blood sampler input function (BSIF) and the various IDIFs. For each subject, 15 tissue ROIs were drawn and projected on all Vd images. The validity of each IDIF was assessed by correlating IDIF-based Vd values with the corresponding BSIF-based values. In addition, test-retest variability was investigated.
Results: Image quality of BSIF- and IDIF-based Logan Vd images was similar. A summary of the quantitative comparison of IDIF- and BSIF-based Vd values is given in figure 1 and 2. Calibration of the IDIFs did not result in better correlation coefficients, but significantly reduced differences in Vd. For PVC-OSEM, calibration resulted in a slope of 1.00 (figure 1). No significant differences in test-retest variability were found between BSIF- and IDIF-based Vd values.
Discussion and Conclusions
For [C11]PIB studies, an image derived input function is an accurate alternative to arterial blood sampling, provided that scans are reconstructed using a reconstruction-based partial volume correction method. Manual samples, however, are still needed for calibrating the IDIF, and for determining plasma/blood ratio and metabolites. These samples can probably taken from venous blood.
EFFECTS OF HYPERPHENYLALANINEMIA ON PET MEASUREMENT OF REGIONAL RATES OF CEREBRAL PROTEIN SYNTHESIS WITH CORRECTION FOR RECYCLING IN ADULT MONKEYS
Carolyn B. Smith1, Kathleen C. Schmidt1, Jeffrey T. Bacon2, Shrinivas Bishu1, Thomas V. Burlin1, Michael A. Channing2, Tianjian Huang1, Zhong-Hua Liu1, Mei Qin1, Bik-Kee Vuong2, Zengyan Xia1, Peter Herscovitch2
1Laboratory of Cerebral Metabolism, NIMH, 2PET Department, Clinical Center, NIH, Bethesda, MD, USA
Introduction: We examined the sensitivity of the L-[1-11C]leucine PET method (1) to detect changes in the fraction (lambda) of the precursor pool for protein synthesis derived from arterial plasma. Based on studies in a mouse model of phenylketonuria, in which plasma phenylalanine concentration was increased 20-fold and lambda decreased (to ~60% control) (2), we induced hyperphenylalaninemia in monkeys and determined rCPS and lambda.
Methods: Isoflurane-anesthetized monkeys (n=4) were dynamically scanned with the HRRT scanner for 60 min following injection of 4.8–9..6 mCi of L-[1-11C]leucine. Each animal was scanned two-three times under conditions of normal plasma amino acid concentrations and once 60 min following the commencement of an i.v. programmed infusion of 1.5% phenylalanine that maintained elevated plasma phenylalanine concentrations for the duration of the study. ROIs were placed on MR images and transferred to co-registered PET images to construct tissue time activity curves. Kinetic model rate constants were estimated for whole brain and four grey and two white matter regions. Regional rates of cerebral protein synthesis (rCPS) were computed as [(K1k4)/(k2+k3+k4)]x(Cp/lambda) where Cp is the arterial plasma leucine concentration. K1-k4 are rate constants for transport of leucine from plasma to brain, from brain to plasma, for catabolism of leucine, and for incorporation of leucine into protein, respectively. Lambda was calculated as (k2+k3)/(k2+k3+k4).
Results: Mean (±SD) arterial plasma phenylalanine concentration (nmol/ml) increased from 62±11 in controls to 258±48 in phenylalanine-infusion studies. Arterial plasma leucine (nmol/ml) was similar in control and phenylalanine-infusion studies, 189±36 and 158±26, respectively. In the phenylalanine-infused monkeys, values of lambda were significantly decreased by 10–16% in whole brain and six regions; rCPS was unaffected in all regions examined (Table).
Conclusion: Our Results: show that values of lambda estimated with the L-[1-11C]leucine PET method change in response to altered plasma amino acid concentrations. Effects on lambda are consistent with the competitive inhibition of transport of leucine by increased plasma phenylalanine. The effect on lambda shows that competition for the transporter Results: in a reduction in the fraction of leucine in the precursor pool for protein synthesis coming from plasma. Even under these hyperphenylalaninemic conditions rCPS remains normal due to the compensating increased contribution of leucine from protein degradation to the precursor pool.
Supported by the Fragile X Research Foundation; Intramural Research Program, NIMH; CC, NIH
REGIONAL RATES OF CEREBRAL PROTEIN SYNTHESIS IN NORMAL ADULT YOUNG MEN MEASURED WITH L-[1-11C]LEUCINE AND PET: EFFECT OF SCANNING INTERVAL
Kathleen C. Schmidt1, Shrinivas Bishu1, Thomas V. Burlin1, Michael Channing3, Tianjian Huang1, Zhong-Hua Liu1, Mei Qin1, Bik-Kee Vuong3, Zengyan Xia1, Alan Zametkin2, Peter Herscovitch3, Carolyn B. Smith1
1Laboratory of Cerebral Metabolism, NIMH, NIH, Bethesda, 2MAP, NIMH, NIH, Bethesda, 3PET Department, CC, NIH, Bethesda, MD, USA
Objectives: We measured regional rates of cerebral protein synthesis (rCPS) with L-[1-11C]leucine and PET (1) in conscious, young adult men. The method uses kinetic modeling to estimate lambda, the fraction of the precursor pool for protein synthesis derived from arterial plasma, to correct for recycling of tissue amino acids (2). The effects of scanning interval on determination of kinetic model rate constants, lambda, and rCPS were examined.
Methods: Subjects (21-23y) were dynamically scanned with the HRRT following injection of 19-26 mCi L-[1-11C]leucine. ROIs were placed on MR images and transferred to co-registered PET images to construct tissue time activity curves. Rate constants were estimated over 0-60, 0-75, and 0-90 min intervals for whole brain and five ROIs. Lambda was calculated as (k2+k3)/(k2+k3+k4) and rCPS as [(K1k4)/(k2+k3+k4)]x(Cp/lambda), where Cp is the plasma leucine concentration. K1-k4 are rate constants for transport of leucine from plasma to brain, from brain to plasma, for catabolism of leucine, and for incorporation of leucine into protein, respectively.
Results: Estimates of K1 were constant with estimation interval; values were lower in subjects with higher plasma neutral amino acid concentrations, consistent with competetive inhibition at the carrier. Estimates of k2+k3, k4, and rCPS decreased slightly with time in whole brain and gray matter regions, and increased slightly in white matter; the mean change in the estimates was minimal (<3%) between the 0-60 and 0-75 min intervals in whole brain and gray matter, and between the 0-75 and 0-90 min intervals in white matter. Lambda was constant with estimation interval. Whole brain estimates were K1=0.0287±0.0066 ml/g/min, k2+k3=0.0713±0.0184/min, k4=0.0329±0.0013/min, lambda=0.68±0.05, rCPS=1.64±0.15 nmol/g/min (mean±SD, n=6; 0-75 min). Coefficients of variation in lambda were <10% and in rCPS <20% in all five regions. The figure shows rCPS in a coronal section at the level of the thalamus.
Conclusions: Our Results: indicate that reliable low variance estimates of lambda and rCPS in both gray and white matter can be obtained with a 75 min scanning interval.
Supported by the Fragile X Research Foundation; Intramural Research Program, NIMH; CC, NIH
MURINE BRAIN REGIONAL (+)-PHNO BINDING: D3 AND D2 COMPONENTS
Eugenii Rabiner1, Roger Raymond2, Mustansir Diwan2, Patrick McCormick3, Alan Wilson3, Jose Nobrega2
1Clinical Imaging Centre, GlaxoSmithKline, Hammersmith Hospital, London, UK, 2Neuroimaging Research Section, CAMH, Toronto, ON, Canada, 3PET Centre, CAMH, Toronto, ON, Canada
Background: Although [11C]-(+)-PHNO was developed as a D2 agonist PET ligand, evidence has accrued that it may be a D3-preferring ligand [1]. As [11C]-(+)-PHNO PET signal is a mixture of D2 and D3 binding, accurate interpretation of PET data requires knowledge of regional D2 and D3 components of [11C]-(+)-PHNO binding. An examination of regional [3H]-(+)-PHNO binding in mice lacking D2 or D3 receptors (D2KO or D3KO) should provide data on the brain regional components of the [11C]PHNO signal.
Methods: Three groups of 12 mice each - wild type (WT), D2KO and D3KO- were injected i.v. with [3H]-(+)-PHNO. Half of each group received 10mg/kg of SB-277011 i.p., 60 min before the [3H]-(+)-PHNO injection. SB-277011 is a D3 antagonist with 100-fold selectivity over D2 in vitro. Sixty minutes after the [3H]-(+)-PHNO injection the mice were sacrificed, their brains extracted and [3H]-(+)-PHNO binding measured in striatal and extra-striatal regions using quantitative autoradiography.
Results: (1) In D3KO specific (+)-PHNO binding was completely abolished in extra-striatal regions and in the ventral striatum, while in the caudate-putamen it was reduced by 50% compared to WT mice. In D2KO SB-277011 decreased the specific binding of [3H]-(+)-PHNO by 96% compared to WT mice, while in D3KO no change in specific binding was seen, confirming the high selectivity of SB-277011 for D3 over D2 at this dose. (2) In WT mice regional decreases in specific [3H]-(+)-PHNO binding following SB-277011 varied from 49%, in the caudate-putamen to 70% in the ventral striatum, 87% in the ventral pallidum, and 100% in the habenula, midbrain, cerebellum lobes IX & X and the hypothalamus.
Conclusions: Our Results: are consistent with the data of Narendran [1], indicating that in vivo PHNO has preferential affinity for D3 over D2high receptors, (contra Seeman [2]). In extra-striatal regions [11C]-(+)-PHNO binding represents the D3 component only, and can be used to estimate D3 drug occupancy without confounds of the D2 component.
THE SURFACE D2-BINDING PROFILE OF THE ATYPICAL ANTIPSYCHOTIC DRUG AMISULPRIDE
Ningning Guo1, Wen Guo1, Olivier Guillin2, Marc Laruelle3, Jonathan Javitch1, Stephen Rayport1
1Departments of Psychiatry and Pharmacology, Columbia University and Department of Neuroscience, New York State Psychiatric Institute, New York, NY, USA, 2Service Universitaire De Psychiatrie, Centre Hospitalier Du Rouvray and Neuropsychopharmacology of Depression Unit, CNRS FRE 2735 (IFRMP), Institute for Biomedical Research, University of Rouen, Rouen, France, 3GlaxoSmithKline and Department of Neuroscience, Imperial College, London, UK
Background and aims: All antipsychotic drugs used in the treatment of schizophrenia are D2 antagonists, but have additional actions at 5-HT2A and other receptors. In contrast, amisulpride is a selective D2/D3 antagonist with negligible action at other receptors, making it a unique pharmacological tool for examining the actions of antipsychotic drugs. Amisulpride is a close cousin of sulpiride, which binds selectively to membrane surface D2 receptors (D2R), in contrast to the radiotracers raclopride and spiperone, which freely permeate the plasma membrane and therefore bind to both the surface and internalized D2R. Because of its hydrophilicity, amisulpride is more likely to bind to surface receptors selectively. If a substantial portion of radiotracer binding is to internalized D2R but efficacy Results: from blocking surface D2R, this might explain amisulpiride's therapeutic efficacy at lower measured D2R occupancy. Here, we used a model in vitro system to examine the binding profiles of amisulpride to surface vs. internalized D2R.
Methods: T-REx HEK 293 cells were stably transfected with D2short cDNA in the plasmid pCIN4, and stably co-transfected with pcDNA4-GRK2 and pcDNA5-β-arrestin2 to foster robust agonist-induced D2R internalization. Cells were maintained under continuous selection with 5 µg/ml blasticidin. GRK2 and β-arrestin2 expression were induced by tetracycline and D2R internalization by quinpirole. [3H]Raclopride was used to detect both surface and intracellular receptors. Whole cell competition binding was performed under control and D2R-internalized (addition of quinpirole) conditions. Competition assays were performed at 4oC following standard in vitro binding methodology.
Results: Consistent with our previous report, sulpiride bound selectively to surface receptors in a one-site biding curve (Ki=11.3 ± 4.4 nM). After quinpirole treatment, sulpiride showed a two-site binding curve with Ki1= 10.2 ± 6.8 nM, matching the Ki of surface D2R in control cells (where most receptors are on the surface), and Ki2= 166.6 ± 59.3 µM for internalized receptors. Similarly, amisulpride competition binding gave a two-site binding curve after quinpirole, with Ki1= 2.0 ± 0.8 nM, matching control cells (Ki=1.7 ± 0.7 nM), and Ki2= 20.2 ± 12.1 µM for internalized receptors. D2R internalization had no impact on competition binding curves for raclopride, spiperone, IZBM, fallypride or haloperidol, all of which were fit with a one-site binding curve, and showed no change in Ki following quinpirole.
Conclusions: The present Results: suggest that the atypical antipsychotic drug amisulpride binds to surface receptors selectively, whereas the typical antipsychotic drug haloperidol and several other D2 antagonists bind to both the surface and internalized D2R. If after amisulpride enters the brain it does not permeate the neuronal plasmalemma, then its therapeutic efficacy at an apparently lower receptor occupancy might be explained by its effective blockade of functional D2R on the plasma membrane. Further exploration of amisulpride binding will be useful for (1) development of amisulpride-like drugs that are better transported across the blood brain barrier, which would achieve higher brain levels while remaining membrane impermeant at the neuronal plasmalemma, and (2) development of amisulpride-like PET ligands as better reporters of drug occupancy and synaptic dopamine release.
PET AND MICRODIALYSIS MEASUREMENT OF DOPAMINE RELEASE IN STRIATUM AND EXTRASTRIATAL BRAIN REGIONS FOLLOWING AMPHETAMINE
Mark Slifstein1,2, Holly Moore1,2, Lawrence Kegeles1,2, Marc Laruelle1, Anissa Abi-Dargham1,2
1Department of Psychiatry, Columbia University, 2Department of Functional Brain Mapping, New York State Psychiatric Institute, New York, NY, USA
Introduction:. The amphetamine challenge (AMPH) is a frequently used paradigm for estimating dopamine (DA) release non invasively in the brain by measuring changes in the specific binding of radioligands for the D2/D3 receptors. Previous studies have correlated the change in radioligand binding with DA efflux increases following AMPH as measured by microdialysis using radioligands that can only image the striatum such as C-11 raclopride or I-123 IBZM. Fallypride (C-11 or F-18) provides quantifiable images in striatum and extrastrial regions. In this study, we measured the correlation between changes in [11C] fallypride binding and DA efflux increases following AMPH in baboons.
Methods:. Three female baboons (A, B and C) were studied under isoflurane anesthesia. Data were acquired for 2hr on the HR+ following a bolus injection of C-11 fallypride. A baseline scan was acquired, then after 1 additional hr, amphetamine was injected as a bolus in doses of 1 mg/kg (A, B) or 1.5 mg/kg (C). The post-challenge scan was started 1/2 hr after amphetamine. The parameter V3” was estimated under both conditions using SRTM with cerebellum as reference region. The change following amphetamine was quantified as ΔV3? = V3?_post / V3?_ pre ? 1.
Microdialysis was later performed on 2 of the animals (A and C) using the same anesthesia protocol. Probes were placed in striatum, anterior cingulate and thalamus and perfused with aCSF. After probe equilibration, dialysates were collected every 15 min. After 1 hr, animals received the same AMPH dose as during the PET study, and 3 more dialysates were collected. DA concentration was quantified by HPLC-ED. Post-AMPH values were expressed as % increase relative to mean pre-AMPH samples (ΔDA). Average regional ΔDA was correlated with ΔV3” within and across subjects.
Results: Regional ΔV3? was highly correlated with ΔDA, both within and across animals: ΔDA = −46.4* ΔV3? ? 7.7 for animal A (R_squared = 0.97), −12.6*ΔV3? +2.0 for animal C (R_squared = 0.80) and −21.7*ΔV3? −0.4 across all 3 animals (R_squared = 0.91). Mean ΔV3? was −19 +/− 10%, −9 +/− 7% and −3 +/− 17% and mean ΔDA was 406 +/− 72% 95 +/− 76% 73 +/−31% in striatum, thalamus and cingulate.
Conclusions: : The high correlation between ΔV3? and ΔDA supports the validity of PET measurements of DA release following AMPH in extrastriatal regions as well as striatum using fallypride.
COMPARING IV AMPHETAMINE-INDUCED 11C RACLOPRIDE DISPLACEMENT BETWEEN HRRT AND GE ADVANCE PET: IMPLICATIONS FOR HIGH RESOLUTION/SENSITIVITY ON DOPMAINE RELEASE MEASURES
Dean F. Wong1,2, Olivier Rousset1, Arman Rahmim1, Anil Kumar1, Hiroto Kuwabara1, Ayon Nandi1, Yun Zhou1, Betsy McCaul2, Gary Wand2,3
1Department of Radiology, 2Department of Psychiatry, 3Department of Medicine, Johns Hopkins Medical Institutions, Baltimore, MD, USA
Background and aims: Psychostimulant-induced displacement of [11C]Raclopride (RAC), often interpreted as a measure of intrasynaptic dopamine release, is increasingly used as a novel tool for examining brain physiology and pathophysiology. Brain dedicated high-performance PET scanners such as the HRRT (High Resolution Research Tomograph, Siemens CPS/CTI) promise to enhance measurements carried out on non-PET CTs/PET-only like the Siemens HR+ and the GE Advance (GEA). Here we test the hypothesis that the 2?3 fold increase in spatial resolution and sensitivity of the HRRT over the GEA offers an improvement in the quantification of binding potentials (BPs) and dopamine release (DAR) measures following psychostimulant challenge. Such documentation may justify the technical challenges of increased slice number (207 vs 35 for the GEA) and reconstruction time (>15 hours for a 90-min, 30-frame acquisition).
Methods: Healthy volunteers (age range 22 ? 25, 4 males, 2 females) underwent two 90-minute PET scans each with IV RAC on the HRRT: 1) Scan 1 following an IV bolus of saline, 2) Scan 2 (2.25 hrs later) following 0.3mg/kg IV amphetamine 5 minutes before tracer injection. Studies were reconstructed using the highest resolution (span 3), current Siemens/CPS/CTI reconstruction software, and a 32-node IBM cluster. Images were acquired in list mode and reconstructed to 30 frames. Time-activity curves were generated in the following regions of interest: anterior/posterior caudate and putamen, and ventral striatum. BPs were obtained using the modified simplified reference tissue method (Ichise, et al, 2002). DAR was calculated as the percent change of baseline to post-amphetamine BPs.
All measures obtained on the HRRT were compared to 6 matched (for age, sex and race) subjects who completed amphetamine challenges on the GEA.
Results: In each matched pair, we calculated percent increase in BP for HRRT vs. GEA. There was a significant increase in both baseline (range 9–27%, mean 16.0 + 6.7% SD) and post-amphetamine BPs (range 6–35%, mean 18.3 + 10.8% SD) across all regions; however, these BP increases between HRRT and GEA were not significantly different between saline and amphetamine scans. Specifically, BPs for the HRRT were higher than GEA BPs in anterior and posterior putamen and anterior caudate (p<0.05), while the increase was not statistically significant for posterior caudate and ventral striatum (p>0.1). There was also no significant change in DAR between the two scanners (p>0.1).
Conclusion: These preliminary data sets suggest that the HRRT can successfully measure the relatively robust effect of RAC displacement by IV amphetamine challenge. BPs, as expected by partial volume (PV) effects, were significantly higher with HRRT vs. the GEA. Although the contrast is lower in the post-amphetamine scans, the change in BP between the two scanners was not significantly different between the two scans. Future studies will not only expand the sample size but also apply these comparisons where contrast is even greater between baseline and postchallenge studies, i.e high and low specific activity RAC. Also, formal PV corrections are being developed for the HRRT as they have for the GEA to allow direct comparison between the GEA and the HRRT.
EFFECT OF REGIONAL SPECIFICITY ON PET MEASURES OF RACLOPRIDE BINDING IN THE STRIATUM IN PATHOLOGICAL GAMBLERS
Ericka Peterson1, Anders Rodell1, Jakob Linnet2, Doris Doudet2, Albert Gjedde2, Arne Moller2
1Center of Functionally Integrative Neuroscience, University of Aarhus, Aarhus University Hospitals, 2PET Center, Aarhus University Hospital, Aarhus, Denmark
BACKGROUND: Dopamine transmission in the ventral striatum (VST), a structure which includes the nucleus accumbens, ventral caudate, and ventral putamen, plays a critical role in the pathophysiology of psychotic states and in the reinforcing effects of virtually all drugs of abuse. The release of dopamine in the ventral striatum, as measured by PET or dialysis has been associated with reward and expectation mechanisms in human and animals, and may be associated with sensation or novelty seeking personality traits. High sensation seeking (SS) correlates with diverse addictions, including tobacco, alcohol, and gambling, which are all linked to DA-mediated reinforcement/reward systems.
METHODS: Dynamic emission recordings were obtained for 60 minutes after i.v. [11C] raclopride (190-366 MBq) using the ECAT HR tomography for 7 subjects diagnosed as pathological gamblers during baseline (no decision making, no possibility to win money). Spatial normalization to a stereotactic atlas was performed using a combination of linear and nonlinear registrations. Regions were extracted using model-based segmentation. Maps of binding potentials (pB) were calculated by the Lammertsma Simplified Reference Tissue Method with the cerebellum as reference. Mean pB(s) were calculated for left and right putamen, caudate and ventral striatum. Correlation between the baseline pB and the Zuckerman scores were examined for each striatal region.
RESULTS: There was a significant negative correlation (r = 0.82, r2 = 0.67, p = 0.024) between the baseline pB and the Zuckerman scores in the left ventral striatum only.
CONCLUSION: In this pilot study, we found a significant correlation between baseline levels of D2 receptor occupancy and the Zuckerman score, a score characteristic of an individual's willingness to seek certain experiences even when risks are involved. The relationship was, in this small sample, limited to the VST, a region of the striatum known for its involvement in reward and expectation as well as being heavily implicated in the reinforcing effects of addictive drugs. This suggests that further investigations in the role of DA and receptor occupancy need to be considered on a regionally specific basis in pathological gambling.
COMBINING PET AND EQUILIBRIUM DIALYSIS TO ASSESS BLOOD BRAIN BARRIER TRANSPORT
Roger Gunn1,2, Scott Summerfield3, Cristian Salinas1, Kevin Read4, Graham Searle1, Alberto Ruffo4, Christine Parker1, Alexander Stevens3, Thomas Bonasera1, Phillip Jeffrey3, Marc Laruelle1
1GSK Imaging Department, London, 2Department of Engineering Science, University of Oxford, Oxford, UK, 3Gsk, Dmpk, Harlow, UK, 4Gsk, Dmpk, Verona, Italy
Background and aims: Until recently, passive diffusion across the BBB has been a general assumption for transport of PET radioligands into the brain. However, when there is an active transport process, a free concentration gradient is established at the BBB. To take active transport into account, we propose that the non-displaceable volume of distribution (VN) of a radiotracer be decomposed into non-specific binding and BBB transport components.
VN = (f1/f2) (FT/FP),
where f1 and f2 are the free fractions in plasma and the non-displaceable tissue compartment(s), and FT and FP are the free concentrations in tissue and plasma respectively. Hence, the total non-displaceable distribution volume is defined by the product of non-specific binding ratio (f1/f2) and free partition coefficient at equilibrium across the BBB (FT/FP). For tracers that involve passive diffusion FT/FP=1, and VN simplifies to f1/f2, its usual definition. The FT/FP ratio can be experimentally assessed by comparing VN as measured with PET, with the f1/f2 fraction as measured in vitro with equilibrium dialysate technique[1]. Seven compounds, with known p-glycoprotein (PGP) substrate status, were used to validate the proposed method.
Methods: Estimates of VN were obtained from PET studies in Landrace pigs for (n=7) different PET radiotracers. VN values were estimated from cerebellar regions using a two tissue compartment model, either under baseline conditions when this region was know to be a reference region or after blocking. Free fractions in plasma and brain homogenate were determined by means of 96-well equilibrium dialysis for the same seven unlabelled compounds. Test compounds were incubated in plasma or brain (10 ug/mL, 37oC) and dialysed against phosphate buffered saline through a 12-14 kDa cutoff semipermeable membrane over five hours. Aliquots of tissue and dialysate were collected and analysed for test compound using HPLC/MS/MS analysis. The free fractions were determined as the concentration ratio of analyte in dialysate to that in plasma or brain. The PGP status of the seven compounds were either determined from separate experimental data or via in silico modelling techniques.
Results: For compounds predicted not to be PGP substrates (n=4), a good agreement between the PET and equilibrium dialysis data was observed consistent with passive diffusion (FT/FP = 0.91 +- 0.26, not significantly different from 1 (p=0.57)). For compounds predicted to be PGP substrates, FT/FP = 0.43 +- 0.16, significantly lower than 1 (p<0.05), and consistent with active efflux.
Conclusions: This work supports the combined use of PET and equilibrium dialysis to determine the true free concentration of a radiotracer in brain. This has implications for the assessment of BBB transport, estimation of in vivo KDs and interpretation of biodistribution studies. In addition, the ability to estimate f1 and f2 in vitro may provide a useful screening tool for selection amongst novel candidate radioligands.
PET IMAGE DENOISING THROUGH A SYNERGISTIC WAVELET TRANSFORM BASED MULTI-RESOLUTION ANALYSIS OF STRUCTURAL AND FUNCTIONAL DATASETS
Federico Edoardo Turkheimer1,2, Alexander N. Anderson2, Nicolas Boussion3, Nicola Pavese2, David J. Brooks1,2, Dimitris Visvikis3
1Clinical Neuroscience Department, Division of Neuroscience, Imperial College London, London, UK, 2MRC Clinical Sciences Centre, Hammersmith Hospital, London, UK, 3INSERM U650, Laboratoire Du Traitement De L'Information Medicale (LaTIM), CHU Morvan, Brest, France
Introduction: Positron Emission Tomography (PET) allows the imaging of functional properties of the living tissue as opposed to the imaging of anatomy whereas other modalities (CT, MRI) provide structural information at significantly higher resolution and better image quality. Constraints for injected radioactivity, technological limitations of current instrumentation, and inherent spatial uncertainties on the decaying process affect PET images sensitivity. In this work we illustrate how structural information of matched anatomical images can be used in a multiresolution model to enhance the signal-to-noise ratio (SNR) of PET images. The model states a flexible relation between function and structure in the brain and replaces high resolution information of PET images with appropriately scaled MRI or CT local detail. The method can be naturally extended to other functional imaging modalities (SPECT, fMRI).
Methods: The methodology is born out of previous work [1] and is based on the multi-resolution property of the wavelet transform (WT). Firstly, the co-registered structural image (MRI/CT) is down-graded to the resolution of the PET volume through appropriate filtering. Secondly, a redundant version of the WT is applied to both volumes. Thirdly, a linear model is applied to the set of local coefficients of both image volumes and resulting parameters recorded. The overall set of linear coefficients is then modelled as a mixture of multivariate Gaussian distributions and fitted through a k-means algorithm. Finally, the local wavelet coefficients of the PET image are substituted by the corresponding values of the MRI-CT set calibrated according to the segmentation resulting out of the clustering algorithm.
Results: Initial application to real and simulated data-sets combining CT and MRi structural information with PET images shows very effective noise reduction (20–30% STD) while resolution is preserved. As an example, the figure shows the raw PET parametric map (A), the co-registered MRI (B) and the denoised map (C) for a [11-C]-(R)-PK11195 study of neuroinflammation in a Huntington's disease patient. The method retrieves the well known disease pattern of striatal and frontal microglial activation signal. The technique is also robust to small errors in the co-registration parameters.
Conclusions: The synergistic structural/functional wavelet de-noising is an effective methodology to increase sensitivity of PET, it is practical to implement and computationally fast. This application confirms that multi-resolution modelling in wavelet space is a valuable tool for multi-modality integration. The technique is ideally suited to CT-PET and MRI-PET scanner data whereas, in the latter case, dedicated sequences could be devised to boost combination with functional data.
WAVELET DENOISING FOR PARAMETRIC IMAGING OF THE PERIPHERAL BENZODIAZEPINE RECEPTORS WITH F-18-FEDAA1106
Miho Shidahara, Yoko Ikoma, Chie Seki, Yota Fujimura, Hiroshi Ito, Yuichi Kimura, Tetsuya Suhara, Iwao Kanno
Molecular Imaging Center, National Institute of Radiological Sciences, Chiba, Japan
Background and aims: The statistical noise of time activity curve (TAC) at voxel level causes severe bias and poor precision for estimated binding potential, BP(=k3/k4), images using a nonlinear least square fitting (NLS). The purpose of this study is to evaluate noise reduction capability of wavelet denoising for estimated BP images of the peripheral benzodiazepine receptor (PBR). 18F-FEDAA1106 is a radioligand for PBR, and its BP images should be formed using NLS because no reference regions can be assumed due to the physiological aspects of PBR1. We applied wavelet denoising to simulate data and clinical dynamic image of PBR with 18F-FEDAA1106.
Methods: After administration of 18F-FEDAA1106, three dimensional dynamic PET scans were performed by ECAT EXACT HR+ system (CTI-Siemens, Knoxville, USA) having 41 frames. The wavelet processing was applied in a volume fashion with a three-dimensional discrete dual-tree complex wavelet transformation2 at 4 scales with 112 subbands. The advantage of wavelet denoising is to realize spatially adaptive smoothing. In order to eliminate noise component in wavelet coefficients, real and imaginary coefficients for each sub-band were individually thresholded with NormalShrink3. Simulations were conducted to evaluate the performance of the wavelet denoising for parameter estimation. A simulated dynamic data was consisted of 4 parts of 2 gray matters (K1=0.25, k2=0.078, k3=0.043 or 0.0516, k4=0.0086), a white matter (K1=0.15, k2=0.075, k3=0.043, k4=0.01), and CSF to simulate anatomical structure of the brain (Hoffman brain phantom: 12812855 pixels, 222 mm). Each part had the BPs of 5, 6, 4.3, and 0, respectively. Then the phantom was smoothed with Gaussian filter (2.52.5 mm FWHM) and then Gaussian noise was added to mimic exact measurement at the noise level 20%. The BP images derived from wavelet denoising were compared with true BP image using 156 rectangular ROIs (55 pixel). The Wavelet denoising was also applied to clinical data derived from 3 young normal volunteers. Parametric images of BP were formed using voxel-based NLS fitting, and the Results: were compared with an ordinary ROI averaged estimates1.
Results: In the simulation studies, estimated BP by denoised image showed better correlation against the true BP values (Fig-1A), although no correlation was observed in the estimates. In clinical data, wavelet denoising improved image quality of the estimates (Fig-1B). Originally estimated BP image includes bias against ROI averaged estimates (Y=1.12X+0.94, R2=0.55), however, estimated BP by denoised image improved relationship with ROI analysis (Y=0.91X+0.95, R2=0.57).
Conclusions: Wavelet denoising improves bias and variation of pharmacokinetic parameters, especially, BP.
MULTI-RESOLUTION BAYESIAN REGRESSION FOR PARAMETRIC IMAGING OF [18F]FDG BRAIN STUDIES
Rainer Hinz1, Paul Edison2, David J. Brooks2, Federico E. Turkheimer2
1Wolfson Molecular Imaging Centre, The University of Manchester, Manchester, 2Medical Research Council Clinical Sciences Centre, Hammersmith Hospital, London, UK
Background and aims: For the generation of parametric images of cerebral glucose metabolism, we have so far used the value of the impulse response function (IRF) at a late time point calculated with spectral analysis (SA) as an estimate of the combined forward rate constant KI [1]. This method provides unbiased estimates and images of good statistical quality, however, it also requires the acquisition of an arterial plasma input function with good temporal sampling. The Patlak plot [2] allows the generation of parametric maps utilising a series of discrete blood samples. Because the Patlak plot does not use the entire dynamic data set, the statistical quality of the parametric images is poorer. Here, we investigated the properties of the parametric maps generated with the new wavelet-based multi-resolution Bayesian regression (MBR) implementation of the Patlak plot [3].
Methods: After bolus administration of 185 MBq [18F]FDG, dynamic 3D PET data were acquired for 60 min on the ECAT EXACT HR+ scanner. The arterial plasma input function was generated using an online blood detector and 6 discrete arterial samples. SA was used with 100 basis functions logarithmically spaced between 0.0001053 s-1 and 0.1 s-1 to generate parametric maps of the IRF at 45 min. For the Patlak analysis, data of the last 6 frames between 30 and 60 min were used.
Results: The figure shows parametric images of KI from an Alzheimer's disease subject. In comparison with region of interest based analysis, all three Methods: provided KI estimates in close agreement. In comparison with the original Patlak method on the left handside, SA and MBR Patlak provide images of superior quality, including cold spots in sulci. The Bland Altman plot in the right panel is derived from a voxelwise comparison between the parametric maps generated by SA and MBR Patlak. Apart from very small KI values where the nonnegativity constraint of SA applies, there is no systematic difference between the estimates provided by SA and MBR Patlak. This is expressed by the LOESS regression line in red which is for all KI > 0.01 min-1 a straight line at zero difference.
Conclusions: Parametric maps of the combined forward rate constant KI with low noise and good structural delineation can be generated from [18F]FDG brain PET scans with the MBR implementation of the Patlak plot. The Results: are in good agreement with those of SA but unlike SA can be obtained without a continuously acquired input function.
ELIMINATION OF ARTERIAL BLOOD SAMPLING IN LOGAN PLOT USING INTERSECTIONAL SEARCHING ALGORITHM AND CLUSTERING IN [11C]TMSX PET
Mika Naganawa, Yuichi Kimura, Kenji Ishii, Keiichi Oda, Kiichi Ishiwata
Positron Medical Center, Tokyo Metropolitan Institute of Gerontology, Tokyo, Japan
Background and aims: The adenosine A2A receptors, which mediate an opposite reaction to the dopamine D2 receptors, in the human brain are visualized by PET with [11C]TMSX [1]. The aim of this study is to propose a robust method for estimating a cumulative integral of the plasma time-activity curve (intpTAC) that is used for an input function for the Logan plot (LP). Our method is based on the intersectional searching algorithm originally proposed by Wang et al. [2] and clustering with two criteria reflecting kinetics of a tissue time-activity curve (tTAC). A tTAC and its cumulative integral curve span a plane in a feature space and the intpTAC is estimated as an intersection of the planes. In this study, we applied the proposed method to [11C]TMSX PET data to generate total distribution volume (tDV) image and binding potential (BP) image.
Methods: Dynamic PET scans and serial arterial blood sampling were performed on 5 subjects including 4 normal volunteers and 1 patient with Parkinson's disease with [11C]TMSX. The BP images were calculated by the LP using both estimated and measured intpTACs with the centrum semiovale as the reference region [3].
Results: Figure 1 shows a typical example of the estimated pTAC and the metabolite-corrected pTAC. The pTAC was obtained by differentiating the estimated intpTAC. The estimated intpTAC matched well with the measured intpTAC with metabolite correction in all subjects. Figure 2 shows the estimated BP maps with or without arterial blood sampling. The regression analysis in voxel-by-voxel comparison was y = 1.01x-0.00 (r2 = 0.99).
Conclusion: We conclude that the proposed method estimates robustly the intpTAC and enables to create BP parametric images of adenosine A2A receptor without arterial blood sampling. The intersectional searching algorithm was sensitive for noise in tTAC, and the proposed clustering scheme can reduce noise appropriately.
ANATOMICAL BASED RECONSTRUCTION OF DYNAMIC BRAIN PET IMAGES USING N-[11C]-METHYLPIPERIDIN-4-YL-PROPRIONATE ([11C]-PMP) IN ELDERLY NORMALS
Natalie Nelissen1,3, Kristof Baete2, Mathieu Vandenbulcke1, Guy Bormans4, Johan Nuyts2,3, Virginie Kinnard2, Tjibbe de Groot2,4, Koen Van Laere2, Rik Vandenberghe1, Patrick Dupont2,3
1Department of Cognitive Neurology, 2Department of Nuclear Medicine, KU Leuven, Leuven, Belgium, 3Medical Imaging Center, 4Department of Radiopharmacy, KU Leuven, Leuven, Belgium
Background and aims: Recently, we have developed a new PET reconstruction algorithm (AMAP) based on a maximum a-posteriori algorithm and using anatomical a-priori information obtained from MRI (Baete et al. 2004a,b). This algorithm corrects for partial volume effects during the reconstruction. The aim of this study was to study the effect on the modelling parameters when using data reconstructed with this new algorithm.
Methods: In 10 healthy subjects (6 male, mean age: 67 yr), data were acquired dynamically for one hour (4 × 15s, 4 × 60s, 2 × 150s, 10 × 300s) on a HR+ after the injection of 300 MBq [11C]-PMP. Arterial blood samples were drawn to measure the plasma input function and to determine the metabolite fraction. A T1-weighted volumetric MRI was acquired and segmented according to an optimized VBM approach using a template of elderly normals.
Dynamic PET images were reconstructed using AMAP or 3D-FBP (filtered backprojection). Both reconstruction Methods: included corrections for randoms, scatter and attenuation. A correction for small movements during the PET study was applied and the resulting images were then smoothed with a Gaussian kernel of 6 mm FWHM.
An irreversible two-tissue compartment model was applied in each voxel with non-linear fitting of the parameters. A constrained value of K1/k2 was determined for each subject on the basis of a whole brain analysis using an unconstrained model.
K1 and k3 parametric images were calculated using the data reconstructed with either one of the two Methods and the parametric maps were compared using SPM2 after warping to the MNI space and smoothing with a Gaussian kernel of 12 mm.
Results: A high correlation (see table) between the K1 values obtained with the two reconstruction Methods: was found. The correlation for the k3 values was high in cerebellum, striatum and thalamus but was lower in the neocortex. Furthermore, k3 values calculated with the AMAP data were significantly (pFWEcor < 0.05) higher throughout the whole brain compared to the ones determined from the 3D-FBP data. However, no significant difference was found for the K1-maps. Median values (1/min) for K1 and k3 are given in the table.
Conclusion: Kinetic modelling of [11C]-PMP reconstructed using AMAP is feasible and the k3 (but not the K1) values are higher compared to 3D-FBP.
PET IMAGING OF 11C-RACLOPRIDE FOLLOWING UPTAKE IN AWAKE ANIMALS
Stephen L. Dewey1, Wynne K. Schiffer1, Vinal Patel1, Paul Vaska1, Craig Woody2, David Schlyer1, Dianne Lee1, David L. Alexoff1, Joanna Fowler1
1Department of Chemistry, 2Department of Physics, Brookhaven National Laboratory, Upton, NY, USA
Background and aims: The ability to non-invasively image behavior-induced changes in neurotransmitters in awake freely moving animals may provide insight into the neurochemical mechanisms associated with them (i.e., drug self-administration, sensitization, expression and acquisition of cue-induced conditioned place preference (CPP)). In addition, it will provide for direct comparisons between small animal and human PET imaging data. In an initial series of small animal PET studies, we compared 11C-raclopride (11C-rac) binding in anesthetized animals with identical measures obtained in the same awake freely moving animals (adult male Sprague-Dawley rats). Our previous studies demonstrated that 11C-rac binding is sensitive to drug-induced changes in brain dopamine (Dewey, et al., 1992; 1993). The ultimate goal of these studies is to measure behavior-induced changes.
Methods: On day 1, ketamine/xylazine anesthetized animals received 11C-rac injections (iv or ip) followed by 60 mins of dynamic scanning using a microPET R4 tomograph (CTI microsystems). On day 2, these same awake and freely moving animals received 11C-rac (iv or ip) followed by a 15 or 30 min uptake period, respectively. Animals were then anesthetized with ketamine/xylazine and received a static 20 min scan.
Results: Striatum/cerebellum (ST/CB) ratios obtained from ketamine/xylazine anesthetized animals (iv) averaged 2.08 while the ratios obtained in awake freely moving animals (iv) averaged 3.12. ST/CB ratios obtained from ketamine/xylazine anesthetized animals (ip) averaged 2.84 while the ratios obtained in awake freely moving animals (ip) averaged 2.91. The variability associated with ip injections was greater than with iv injections suggesting that iv administration offers a marked advantage over ip. Methamphetamine pretreatment (1.0 mg/kg) reduced the ST/CB ratio in both anesthetized and awake animals by approximately 20%.
Conclusions: These data suggest that measures of 11C-rac uptake obtained in awake freely moving animals (iv) are higher than in ketamine/xylazine anesthetized rats while values obtained in both groups following an ip injection were similar. Ketamine/xylazine anesthesia did not alter the effects of methamphetamine. Studies using this novel approach together with behavioral measures including a CPP strategy in combination with automated motion tracking software (TopScan, Cleversys. Inc), are currently ongoing in animals trained to associate specific environments with methamphetamine, cocaine, or inhalants. Finally, these data support the development and characterization of the first rat conscious animal PET (RatCAP) which will provide imaging data throughout the entire awake period.
USDOE/OBER DE-AC02-98CH10886; NIH DA015041
IN VIVO MEASUREMENT OF BMAX AND KD OF (R)-ROLIPRAM IN RAT
Masahiro Fujita, Masao Imaizumi, Sami Zoghbi, Jinsoo Hong, Victor Pike, Robert Innis
Molecular Imaging Branch, National Institute of Mental Health, Bethesda, MD, USA
Background and aims: Rolipram is a selective inhibitor of phosphodiesterase 4 (PDE4), the enzyme that catabolizes the second messenger cAMP. The enzyme activity of PDE4 is regulated by phosphorylation with greater activity in the phosphorylated than the non-phosphorylated form. Studies using recombinant DNAs showed the phosphorylated form is more sensitive to inhibition by rolipram. Thus, rolipram binding may provide indirect measurement of the activity of PDE4. We measured Bmax and Kd of rolipram in vivo and compared to in vitro Results: because the phosphorylation status is unlikely to be preserved in processing for homogenate binding.
Methods: In vivo Bmax and Kd were measured by estimating specifically bound (B) and free (F) levels of (R)-[C-11]rolipram under transient equilibrium in PET experiments without blocking (n = 5, mass dose: 0.75 ± 0.12 µg/kg), partially (n = 5, 0.015 ? 0.0015 mg/kg) and fully blocking (n = 2, 0.9 ? 1.0 mg/kg) with non-radiolabeled (R)-rolipram. PET images were acquired with the ATLAS, and metabolite-corrected arterial input function was measured in all scans. B under transient equilibrium was estimated by performing two-compartmental fitting by constraining K1/k2 to the value obtained in full blocking scans. F was estimated from the activity in nondisplaceable compartment and plasma free fraction measured in five rats not used in PET imaging. Bmax and Kd were calculated by nonlinear fitting using one- and two-binding site models for the saturation curve created from B and F obtained above in diencephalon, frontal cortex and hippocampus to compare previously reported in vitro Results: (1) obtained from the animals used in the no blocking scans in this study.
Results: Nonlinear fitting provided excellent fitting for the saturation curves with R square > 0.90 for both one and two binding site models (Fig.). F-test did not show a significant difference in goodness-of-fit between these two models not supporting the presence of two binding sites. One-site model gave Bmax of 61, 64, and 64 nM, and Kd of 1.3, 1.2, and 0.96 nM in diencephalon, frontal cortex and hippocampus, respectively. Compared to previous reports (1), (2), in vitro and in vivo measurements showed similar Bmax values. Kd measured in vivo was ~15% of the in vitro values.
Conclusions: (R)-rolipram has similar binding site density under in vivo and in vitro conditions and may have greater affinity in vivo implying greater levels of PDE4 phosphorylation.
IN VIVO QUANTIFICATION OF BENZODIAZEPINE RECEPTOR DENSITY AND AFFINITY IN RAT BRAIN USING SMALL ANIMAL PET WITH A MULTI-INJECTION PROTOCOL
Mrie-Claude Gregoire1, Stefanie Dedeurwaerdere2, Lucy Vivash2, David Binns3, Peter Roselt3, Tien Pham1, Andrew Katsifis1, Rod Hicks3, Terence O'Brien2, Damian Eric Myers2
1Radiopharmaceuticals Research Institute, Australian Nuclear Science and Technology Organisation, Sydney, NSW, 2The Department of Medicine, The University of Melbourne, Royal Melbourne Hospital, Parkville, 3Centre for Molecular Imaging, Peter MacCallum Cancer Centre, Melbourne, VIC, Australia
In order to follow up receptor density and affinity changes occurring in small animal models of disease, in vivo PET imaging is of great potential. Before creating a simplified non-invasive method, it is recommended to fully understand the kinetics of the radioligand-receptor interactions. In this study, we aim at fully characterizing the [18F]FMZ kinetics in healthy rats, thus identifying all the parameters of the compartmental model that describes the interactions between [18F]FMZ and the benzodiazepine (BZ) receptors in the brain. This study also investigates the feasibility of performing a multi-injection protocol in the rat with small animal PET.
Methods
Five rats underwent 60-minutes PET scans after injection of (i) Tracer dose of [18F]FMZ; (ii) Pre-treatment with FMZ (150 ug/kg), then tracer dose [18F]FMZ; (iii) Tracer dose of [18F]FMZ, at T=20min: displacement with FMZ (150 ug/kg); (iv) Three-injection protocol: T=0: 0.6nmol of [18F]FMZ; T=42min: 2.7 nmol of [18F]FMZ; T=70min: 495 nmol of FMZ. Thirty-seven arterial blood samples were collected during protocol (iv), and whole blood concentration was counted. The parent compound concentration in plasma was assessed and an input function could be derived. A non-linear compartmental model describing the behaviour of [18F]FMZ and FMZ in the tissue was used for kinetics simulation.
Results
The two first protocols show that uptake in brain structures is correlated to BZR density values estimated with in vitro binding and non-specific uptake is similar among most of the brain structures. Displacement studies displayed a rapid and major signal decrease in hippocampus and pons (90% and 79%, resp.).
Multi-injection kinetics are shown in figure 1. The arterial input function could be accurately assessed, and the first model parameters were identified. Receptor density Bmax and affinity KdVr were consistent with values presented in a multiple-experiment study carried out with [11C]Flumazepil (Liefaard, 2005); Bmax ranged from 19.47 to 81.5 pmol/ml and KdVr from15.68 to 22.51 nM, in hippocampus and pons, respectively.
This study demonstrates that the identification of the model parameters (esp. density and affinity) in the rat brain with a multi-injection protocol is feasible using small animal PET and [18F]FMZ.
(+/−)FENFLURAMINE AND (+/−)MDMA ARE EQUIPOTENT SEROTONIN RELEASERS IN LIMBIC AND NON-LIMBIC AREAS: AN MICROPET [18F]MPPF STUDY
Pedro Rosa-Neto, Andrew Goertzen, Antonio Aliaga, Jean-Paul Soucy, Gassan Massarweh, Shadreck Mengeza, Younes Lakhrissi, Alan Evans
Montreal Neurological Institute and Hospital, McGill University, Montreal, QC, Canada
Background and aims: Estimates of serotonin release in neocortical and limbic brain regions may provide relevant information regarding mechanisms underlying the psychopharmacology of serotonergic drugs. The subjective effects of (+/−)MDMA, a widely used recreational amphetamine, \are thought to be in part linked to the release of serotonin. (+/−)Fenfluramine is also a potent serotonin releaser that, unlike (+/−)MDMA, induces minor changes in mood and perception. Thus, it is possible that regional differences of serotonin release may explain the differences between the effects of (+/−)fenfluramine and (+/−)MDMA. We tested the hypothesis that the magnitude of serotonin release evoked in the hippocampus by (+/−)MDMA is higher than that due to (+/−)fenfluramine. We used a rodent Positron Emission Tomography scanner and the radioligand [18F]MPPF for comparing the magnitudes of (+/−)fenfluramine and (+/−)MDMA evoked serotonin release in several brain regions. [18F]MPPF is a selective 5-HT1A receptor antagonist sensitive to fluctuations in synaptic serotonin. Methods: Six adult Sprague Dawley rats were submitted to three [18F]MPPF dynamic studies, each one performed 14 days apart. The first [18F]MPPF scan started after 30 minutes of an iv injection of saline, the second scan after 10mg/kg/iv of (+/−)fenfluramine and the third scan after 10mg/kg/iv of (+/−)MDMA. Binding potentials were calculated using the simplified reference tissue method and the reference tissue graphical method for irreversible ligands, with Results: subsequently compared using ANOVA. Results: We found that, compared to the baseline condition, (+/−)fenfluramine and (+/−)MDMA evoked significant and comparable declines of [18F]MPPF binding in all brain regions studied. For both drugs, the magnitude of this effect was higher in the raphe followed by the hippocampus formation (fig1A,B and C), mesial prefrontal (fig1A) region and cortex. Conclusions: These Results: suggest that the differences in behavioral patterns induced by (+/−)MDMA and (+/−)fenfluramine may not be attributed to global or regional differences of serotonin release.
THE EFFECT OF A P-GLYCOPROTEIN INHIBITOR ON RAT BRAIN UPTAKE AND BINDING OF [18F]ALTANSERIN: A MICRO-PET STUDY
Mikael Palner1, Stina Syvanen1,2, Ulrik S. Kristoffersen3,4, Nic Gillings4, Andreas Kjaer3,4, Gitte M. Knudsen1
1Neurobiology Research Unit and Center of Integrated Molecular Brain Imaging, Rigshospitalet, Copenhagen, Denmark, 2Uppsala Imanet and Uppsala University, Uppsala, Sweden, 3Cluster for Molecular Imaging, Panum Institute, Copenhagen University, 4Department of Clinical Physiology, Nuclear Medicine and PET, Rigshospitalet, Copenhagen, Denmark
Flourine-18-labeled Altanserin {3-(2-[4-(4-fluorobenzoyl)piperidin-1-yl]-ethyl) −1,2-dihydro-2-thioxo- quinazolione} is a widely used 5HT2A receptor-selective positron emission tomography (PET) tracer (Ki 0.13 nM) in humans. Ex vivo studies in the rat show, however, that altanserin has a limited brain uptake and the altanserin binding potential (BP2) is highly variable with a low reproducibility. Therefore, we hypothesized that altanserin ? in parallel to what has been reported for the radiotracers [11C]Carazolol, [11C]Verapamil and [18F]MPPF ? is a substrate for P-glycoprotein (P-gp), an efflux transporter in the blood-brain barrier. We investigated rat brain uptake and specific binding of [18F]Altanserin before and after inhibition of P-gp with Cyclosporin A (CsA).
Small animal PET was used to estimate brain percent standard uptake value (%SUV) of radioactivity in control and CsA treated rats. [18F]Altanserin (1-10 MBq) was injected i.v. and arterial blood samples were obtained from a femoral cathether during the scan. To assess the specific binding before and after CsA treatment, a displacement study with i.v.injection of the potent 5HT2(A/C) antagonist Ketanserin was performed. The area under the arterial input curves (AUC) was computed to compare radioligand levels in plasma and blood. Previous studies in rats have shown only minimal brain and plasma metabolism of [18F]Altanserin and for that reason, no metabolite analyses were carried out. Radiotracer in vivo distribution was compared to in vitro autoradiography images, from 20 µm rat brains sections. Specific binding, in terms of BP2 [(ROI-Cerebellum)/Cerebellum], was determined on the basis of the mean %SUV after 60 minutes and the outcome was compared to ex vivo dissection 60 min post injection of [18F]Altanserin.
When normalized with the corresponding AUC's, brain images of CsA treated rats had a 52 % higher total brain uptake of [18F]Altanserin, and a 6.46 fold increase in frontal cortex BP2 compared to rats injected with [18F]Altanserin alone. As demonstrated in the figure, a substantially higher brain uptake of [18F]Altanserin was seen in the CsA treated (B) compared to controls (A: without CsA, C: with CsA and Ketanserin). The distribution of activity in the brain was found to be similar to the distribution found by in vitro 5HT2A receptor autoradiography. CsA treated rats had lower radioactivity in blood and plasma than controls, Ketanserin treated rats had lower radioactivity levels in plasma but not in blood, probably because of the Ketanserin-induced displacement of [18F]Altanserin from platelet 5-HT2 receptors. It is concluded that, CsA pronouncedly affects [18F]Altanserin uptake and binding in the rat brain. By contrast, [18F]Altanserin shows a high uptake and binding in the human brain. When new PET-tracers are evaluated, these species differences should be kept in mind.
DETERMINATION OF FDG TRANSPORT KINETICS AND PARTITION COEFFICIENT BETWEEN PLASMA AND RED BLOOD CELLS IN MOUSE IN VIVO
Sung-Cheng Huang, Xiaoli Zhang, Hsiao-Ming Wu, Koon-Poon Wong, Gregory Ferl, Chin-Lung Yu, Weber Shao, Waldemar Ladno, Heinrich Schelbert
Department of Molecular and Medical Pharmacology, UCLA School of Medicine, Los Angeles, CA, USA
Background and aims: Quantification of tissue metabolic rates of glucose (MRGlc) in mouse using FDG microPET requires the time activity curve of FDG in plasma. However, time activity curves derived from microPET images are those of whole blood, and FDG does not cross murine RBC membrane freely. To convert blood TAC to plasma TAC requires knowledge of the transport rate constants of FDG across the RBC membrane. We have thus performed in vivo studies to determine the transport rates and partition coefficient of FDG between plasma and red blood cells (RBC).
Methods: Five anesthetized mice (~27g, 1-1.5% isoflurane) were studied. For each animal, after the femoral artery was cannulated, FDG was injected through the tail vein (~400 uCi bolus). Serial arterial blood samples (~45 uL each) were taken manually via the femoral catheter at approximately 0.1, 0.4, 0.8, 1.2, 1.6, 2.0, 2.6, 5, 10, 15, 30, 45, 60, 90 min. For each sample, the first ~10 uL was counted directly in a well counter to give the blood FDG activity. The rest (~35uL) was centrifuged to measure the hematocrit (Hct), plasma activity, and plasma glucose level. Activity levels in RBC were then determined, and were fitted by the convolution of the plasma TAC and a multi-exponential function to determine the FDG transport kinetics and partition coefficient between plasma and RBC.
Results: Hematocrit from early blood samples was 0.44±0.02 and decreased approximately linearly versus time, T, post FDG injection at a rate of 0.0009±0.0002/min due to blood loss and catheter flushing (i.e., Hct = 0.44 ? 0.0009 T). Ratio of measured plasma to blood FDG activity, without adjustment for changes in Hct, decreased following the relationship: 1.09+0.39exp(−0.072 T), with T in minutes post bolus injection. Two exponential components (halftimes: 0.27±0.17 and 18.7±5.2 mins) were required to describe the kinetics of FDG from plasma to RBC. The partition coefficient of RBC FDG relative to plasma was 0.61±0.081, and had no apparent correlation with plasma glucose level over the range of 155 to 276 mg/dl. Using the estimated average transport rate constants, the ratio of plasma to whole blood activity decreased over time with a time constant of 0.058±0.011/min instead of 0.072±0.009/min when no corrections were made for the method-related decrease in Hct.
Conclusions: We demonstrated that determination of the transport kinetics of FDG from plasma to RBC in vivo in mouse is feasible from serial blood sampling. Although the mechanism accounting for the two components in the kinetics of FDG from plasma to RBC still needs to be determined by further experiments, the estimated values of the transport rate constants are useful for converting whole blood TAC to plasma TAC in mouse FDG microPET studies.
SIMULTANEOUS FUNCTIONAL AND MORPHOLOGICAL BRAIN IMAGING WITH AN INTEGRATED MR/PET SCANNER
Wolf-Dieter Heiss1, Heinz-Peter Schlemmer2, Matthias Schmandt3, Bernd J. Pichler2, Ziad Burbar3, Christian Michel3, Ralf Ladebeck3, Claude Nahmias4, David W. Townsend4, Claus D. Claussen2
1Max Planck Institute for Neurological Research, Cologne, Germany, 2Department of Radiology, University of Tuebingen, Tuebingen, Germany, 3Siemens Medical Solutions, Erlangen, Germany, 4Department of Cancer Imaging and Tracer Development, University of Tennessee Medical Center, Knoxville, TN, USA
Objective: Coregistration of morphologic and functional imaging modalities has provided new insights into physiology and pathology of the brain, but is limited by artefacts due to movement and repositioning, different image properties, different distortion, inaccuracies caused by different reconstruction algorithms and by volume effects. Additionally, it does not permit to record different steps of dependent processes and cannot visualize different parameters of function simultaneously in various brain structures. These limitations can only be overcome by hybrid systems integrating various modalities. The combined PET/CT scanner was a successful step to integrated imaging; with this equipment sequential morphological and functional studies can be obtained in one examination, but simultaneous investigations of various morphological and functional parameters cannot be achieved. Additionally, CT of the brain lacks of soft tissue contrast and is often not sufficient to reveal relevant pathology. Comprehensive assessment of morphology, function and molecular mechanisms might be achieved by a newly developed integrated MR/PET system.
Equipment: PET detectors based on lutetium-oxyorthosilicate (LSO) scintillation crystals and avalanche photodiodes (APDs) are insensitive to magnetic fields and invisible for the MR system. 32 cassettes of 5 LSO-APD blocks, each containing 12 × 12 crystals (2.5 × 2.5 × 20 mm3) and 3 × 3 APD arrays form a ring with an inner diameter of 35.5 cm and an axial field of view of 19 cm. This ring is mounted into a standard clinical MR scanner (Trio, Siemens). A standard bird cage transmit / receive headcoil was mounted on the scanner bed and reproducibly phased inside the PET scanner.
Results: Simultaneous MR/PET data acquisition of the Hoffmann brain phantom revealed excellent image quality without any significant distortions or artefacts. Neither the performance of the PET (spatial resolution ~2.5 mm) nor of the MRI was degraded by combining the two systems or by simultaneous data acquisition. In a preliminary study of 3 patients (2 oncology cases without brain metastases, one with mild cognitive impairment), FDG uptake in the brain was recorded simultaneously with various MR pulse sequences, including time-of-flight angiography, diffusion weighted imaging and proton spectroscopy, in which all the relevant peaks could be clearly resolved. The accurate fusion of PET- and the various MR-images permitted to match exactly FDG uptake to small cortical, subcortical and brainstem regions. Whereas the two oncological patients did not show pathological changes in the brain, in the case with mild cognitive impairment multiple white matter lesions were detected which could be matched to areas with slightly decreased FDG uptake and increased ratios of choline to creatine.
Conclusion: This first application of an integrated system demonstrates the feasibility of simultaneous PET and MRI studies in the human brain. The technology combining different imaging modalities might become a powerful tool not only for basic neuroscience but also for human brain research and for clinical neurology. Thus, MR/PET will open-up new possibilities for complementary molecular and functional investigations.
ADEQUACY OF DUAL ADMINISTRATION OF 15O2 AND H215O FOR RAPID AND ACCURATE ASSESSMENT OF CBF AND CMRO2
Iida Hidehiro, Kudomi Nobuyuki, Hayashi Takuya, Inomata Toru, Miyake Yoshinori, Ohta Youichirou, Teramoto Noboru, Koshino Kazuhiro, Piao Nishuku
Department of Investigative Radiology, National Cardiovascular Center, Suita, Osaka, Japan
Introduction: 15O-oxygen (15O2) has a unique feature that the recirculating H215O, a metabolic product of 15O2, has significant influence to PET images. Ability of absolute quantitation of CMRO2 and OEF is largely attributed to the fact that kinetic behavior of recirculating H215O and 15O2 are accurately expressed by a mathematical model. Original technique to determine CMRO2 and OEF by Mintun [1] was based on the 3 step method, in which CBF and CBV were separately determined. Simultaneous determination of CBF, CBV and CMRO2 from a single dynamic scan for 15O2 (1 step method [2]) was attractive, but considered to enhance statistical noise. A novel technique was developed by us, in which CBF, CMRO2 and CBV were determined from a single dynamic scan in conjunction with sequential administration of 15O2 and H215O [3]. The statistical properties has however yet to be evaluated. This study was intended to evaluate the accuracy and statistical properties of this technique. Adequacy and significance of sequential administration of 15O2 and H215O was tested by a Monte Carlo simulation.
Methods: The arterial input function for H215O and 15O2 was obtained for the 3 step, 1 step, and dual administration protocols. Regional tissue time-activity curves (TTAC) were generated, with added realistic Poisson noise, for 3 different physiological conditions, namely normal; CBF=0.5mL/min/g, OEF=0.4, ischemia; CBF=0.25mL/min/g, OEF=0.8, and hyperemia; CBF=0.8mL/min/g, OEF=0.25. OEF, CBF, CMRO2 and the arterial blood volume (Va) were estimated according to each technique COVs as well as systematic errors of estimated parameters were evaluated.
Results: and Discussion: The 1 step approach yielded CBF and CMRO2 that had statistical noise significantly enhanced with systematic bias. The dual input resulted in values which had same magnitude of statistical uncertainty, without systematic bias as compared with the 3 step. These Results: are consistent with our findings that (a) image quality and absolute values by the dual administration were comparable to those by the 3 step in clinical studies, and (b) quantitative OEF obtained by the dual tracer agreed well with those by the arterial-sinus difference of oxygen contents in rhesus monkeys. This technique appeared to be rapid and accurate enough to be applicable to clinical studies in patients with acute stroke.
REQUIREMENTS FOR SUCCESS OF RADIOLABELED COMPOUNDS FOR EMISSION TOMOGRAPHY: COMPARISON OF SEVEN LIGANDS FOR IMAGING THE SEROTONIN REUPTAKE SITE
Betina Elfving1,2, Jacob Madsen3, Gitte M. Knudsen1
1Neurobiology Research Unit N9201, Rigshospitalet, University Hospital Copenhagen, Copenhagen, Denmark, 2Centre for Basic Psychiatric Research, Aarhus Psychiatric University Hospital, Risskov, Denmark, 3PET and Cyclotron Unit, Rigshospitalet, University Hospital Copenhagen, Copenhagen, Denmark
Over the last 15 years, numerous attempts have been made to develop suitable radiolabeled tracers for PET- or SPECT imaging of the serotonin reuptake site (SERT). Several radioligands are currently in use and yield acceptable quantifiable images of the SERT in the human brain. The aim of this study is to define characteristics of a good SERT radioligand and to investigate species differences. We examined seven different selective serotonin reuptake inhibitors (SSRIs) in an in vitro design. Except for one they had all previously been tested as radiolabeled PET- or SPECT-ligands. The outcome of the ligands as imaging tracers was compared to their SERT density and/or ligand affinity in rat and monkey brain. The SERT affinity (Kd) and density (Bmax) values for [3H]-(+)-McN5652, [3H]MADAM, [11C]DASB, [123I]ADAM, [3H]-(S)-citalopram, [3H]fluoxetine, and [3H]paroxetine were determined in rat and monkey cerebrum and cerebellum (reference region) membranes.
[3H]-(S)-citalopram and [3H]-(+)-McN5652 had statistically significantly lower affinity ([3H]-(S)-citalopram 5.1 nM versus 0.71 nM; [3H]-(+)-McN5652 0.23 nM versus 0.11 nM), whereas [3H]paroxetine (0.052 nM versus 0.077 nM) displayed a statistically significantly higher affinity for SERT in monkey cortex as compared to the rat cerebrum (t-test, pKd values, p<0.05). The affinity of [3H]MADAM (0.11 versus 0.13 nM), [123I]ADAM (0.050 versus 0.071 nM), and [11C]DASB (0.34 versus 0.25 nM) for SERT obtained with rat cerebrum and monkey cortex are similar. In monkey cortex Kd and Bmax could not be determined with [3H]fluoxetine.
Surprisingly, out of the seven SSRIs, [3H]-(S)-citalopram, [3H]MADAM and [11C]DASB all showed significant specific binding to SERT in monkey cerebellum with Bmax cortex:cerebellum ratios being 17, 3 and 4, respectively. These ratios were relatively well predicted from data obtained in rat brain tissue.
In Conclusion:, it can be estimated that imaging of the human 5-HT transporter in a high-density region requires radioligands with Kd values between 0.03 and a maximum of 0.3 nM (at 37°C). The differential specific cerebellar binding raises the question of the suitability of cerebellum as a reference region for non-specific binding. Further, our data suggest that SERT radioligands commonly used in imaging studies bind to various classes of SERT, that differ either structurally (e.g., high- or low-affinity states) or in their subcellular organization (e.g., internalization). Alternatively, the ligands bind to other, not yet identified macromolecules.
GRAPHICAL ANALYSIS OF OCCUPANCY STUDIES
Vincent J. Cunningham1, Roger N. Gunn1, Eugenii Rabiner1, Mark Slifstein2, Marc Laruelle1,2
1GSK Clinical Imaging Centre, Imperial College London, London, UK, 2Columbia University, New York, NY, USA
Background and aims: In terms of the standard PET neuroreceptor model, receptor occupancy (RO) by an exogenous drug is measured by the fractional decrease in VS, the volume of distribution of the specifically bound radioligand. This can be derived from VT, the total volume of distribution, if an estimate of VND (free plus non-specifically bound radioligand) is available. It is frequently the case that an ideal reference region (VS = 0, VT = VND) does not exist for the target. Lassen et al. (1) overcame this problem by assuming VND and RO constant across regions. By comparing regional estimates of VT between a baseline scan in the absence of cold drug and after administration of cold drug, RO can be obtained graphically. This approach has previously been used to estimate VND for an ideal reference region (1),(2) and RO (1),(3). We now present a more general equation for graphical analysis of occupancy studies which does not assume the availability of a baseline scan.
Methods: We show that;
for non-baseline scans where the number denotes the scan and C denotes the corresponding concentration/dose of the cold drug. The derivation assumes that the Hill coefficient is 1 and full occupancy is attainable in the limit. This gives an easily interpretable plot of the LHS of equation 1 against VT1, with RO2 given by the slope, and VND by the × intercept. The special case, C1= 0, in which scan 1 is a true baseline scan, gives:
corresponding to re-arrangement of equations in Lassen et al. (1). Plots of (VT1 ? VT2) against VT1, are again easily interpretable with the slope a direct measure of occupancy.
Results: The figures are from an [11C]-WAY100635 study of 5-HT1a receptor occupancy. Scans were performed at baseline and at cold doses of a blocking agent at 1.5, 10 and 150ug/Kg. Analysis using the baseline scan with equation 2 gave corresponding occupancies of 28.4%, 78.1% and 97.0%. Discarding the baseline scan and analyzing according to equation 1 with the lowest dose as C1 gave occupancies of 81.4% and 96.8% at the higher doses. Estimates of the half saturation dose can then be derived, giving 31.5% occupancy at the lowest dose.
Conclusions: All these latter estimates were thus obtained not only without a reference region but also without baseline data.
MULTI-INPUT SPECTRAL ANALYSIS FOR ASSESSING CEREBRAL UPTAKE OF LABELLED METABOLITES: VALIDATION AND APPLICATION TO [11C]PIB AND [18F]FDDNP STUDIES
Mark Lubberink1, Bart N. M. van Berckel1, Gert Luurtsema1, Kevin Takkenkamp1, Nelleke Tolboom2, Maqsood Yaqub1, Adriaan A. Lammertsma1
1Department of Nuclear Medicine and PET Research, VU University Medical Centre, Amsterdam, The Netherlands, 2Alzheimer Centre, VU University Medical Centre, Amsterdam, The Netherlands
Background and aims: Cerebral uptake of radioactive metabolites poses a major challenge for quantitative tracer kinetic PET studies. Although this uptake can be quantified by PET studies of the labelled metabolites themselves [1],[2], it requires extensive chemistry and additional PET measurements, and is not feasible on a routine basis. Use of spectral analysis [3] has previously been suggested as a tool for assessing uptake of labelled metabolites of (R)-[11C]verapamil [4]. The aim of the present study was to investigate the potential use of multi-input spectral analysis (MISA) for assessment of metabolite uptake using simulations. In addition, the method was applied to clinical [18F]FDDNP and [11C]PIB data.
Methods: MISA was implemented using 30 basis functions (BFs) based on tracer concentration in plasma, 30 BFs based on metabolite concentration in plasma, and one whole blood compartment, with time constants between 1/scan length and 1 min-1.
Sets of 1000 noisy time-activity curves (TACs) were simulated, each set being based on different rate constants and plasma kinetic data with fast ([11C]PIB) or slow ((R)-[11C]verapamil) metabolism, with or without metabolite uptake. Volume of distribution (Vd), model order, and percent area under the curve due to metabolites (%AUCmet) were determined using MISA and compared to true values. TACs of 35 grey matter regions of interest (ROIs) of 10 subjects (5 [18F]FDDNP and 5 [11C]PIB) were analysed using MISA to identify possible metabolite uptake.
Results: Using BFs based on [11C]PIB plasma kinetics, good correlation (r2 0.96) was found between %AUCmet obtained by MISA and true %AUCmet. MISA derived %AUCmet significantly > 0 always corresponded to a true metabolite compartment (figure 1).
A metabolite compartment was consistently identified in clinical [18F]FDDNP time-activity curves, with mean ± SD %AUCmet of 29 ± 10 in 175 ROIs, compared to 22 ± 4 for [11C]PIB. According to the Akaike criterion, MISA consistently provided better fits than SA with only parent BFs in the case of [18F]FDDNP.
Conclusions: Simulations indicate that a potential metabolite contribution can be identified accurately using MISA for tracers with fast metabolism. MISA applied to clinical data revealed likely metabolite uptake for [18F]FDDNP and [11C]PIB. Further validation of the method, using tracers for which metabolite kinetics have been measured such as [18F]altanserin [1], will be subject of future studies.
IMAGING NANOPARTICLES WITH PET
Wynne K. Schiffer1, Joseph Carrion1, Mathew Maye2, Barbara Panessa-Warren3, Colleen Shea1, Oleg Gang2, Stephen L. Dewey1, Joanna Fowler1
1Medical Department, 2Center for Functional Nanomaterials, 3Materials Science Department, Brookhaven National Laboratory, Upton, NY, USA
BACKGROUND AND AIMS: Nanoparticles radiolabeled with PET isotopes provide the unique opportunity to track their in vivo pharmacokinetics, regional distribution and accumulation relative to physico-chemical properties such as core material or size. A general radiolabeling strategy was developed and applied to two classes of nanoparticle at two different sizes ? Quantum Dots (11C-CdSe/ZnS/ZnS) or gold nanoparticles (11C-Au) of 2 or 10 nm core size.
METHODS: Paired PET scans were performed on mice, where the first scan measured 2 nm core 11C-CdSe/ZnS followed by a second scan of 10 nm 11C-CdSe/ZnS. Stability was examined with evaporative loss experiments. After the 90 min scans, organs were harvested for light microscopy (LM), transmission electron microscopy (TEM) and inductively coupled plasma atomic emission spectroscopy (ICP-AES) to independently assess tissue and cellular distribution (LM and TEM) and the concentration of metal in tissues (ICP-AES).
RESULTS: TEM indicated that the dispersity and surface chemistry of the nanoparticles were not altered by the isotope (carbon-11). Evaporative loss experiments provided evidence that the isotope was not metabolized into 11C-CO2 or 11C-MeOH. There was a strong linear correlation (R2=0.82) between carbon-11 measured with PET and cadmium or gold concentrations measured with ICP-AES. A two-compartment kinetic model described the kinetics of nanoparticle dispersion. However at 10 nm cores, neither material penetrated the brain (see Figure). Highest uptake at this size was in the liver and kidneys (Figure [a]). With 2 nm particles (Figure [b]), peak brain uptake occurred at 4% of the injected concentration within the first three min of scanning while the highest uptake was in the lungs, liver and kidney, with time to peak of 3 sec in the lung and 2.0 min in the kidney and liver.
CONCLUSIONS: A general radiolabeling strategy for PET was used to show that the kinetics of nanoparticles can be modeled and predicted based on fundamental physico-chemical characteristics.
CEREBRAL BLOOD FLOW MEASURED BY STABLE XENON ENHANCED COMPUTERIZED TOMOGRAPHY COMPARED TO H215O POSITRON EMISSION TOMOGRAPHY
Edwin Nemoto1, Yonas Howard2, Ronda Pindzola3, Hiroto Kuwabara4, Donald Sashin1
1Department of Radiology, University of Pittsburgh, Presbyterian University Hospital, Pittsburgh, PA, USA, 2Department of Neurosurgery, University of New Mexico, UNM Hospital, Albuquerque, NM, USA, 3Department of Neurology, University of Pittsburgh, Presbyterian University Hospital, Pittsburgh, PA, USA, 4Department of Radiology, Johns Hopkins University, Johns Hopkins University Hospital, Baltimore, MD, USA
Background and aims: Both the vasodilatory effect of stable xenon and the limited blood brain barrier permeability of H215O on cerebral blood flow (CBF) measured by CT and PET, respectively, have been extensively studied with a general consensus that PET CBF is the gold standard for comparison (1).
Methods: We measured CBF by stable xenon-CT (Xe/CT) (28% xenon in 40% oxygen for 4.5 min) and within 24 hrs, by dynamic PET with H215O in 12 symptomatic stroke patients which allowed comparison of the CBF values using both Methods: in the major cerebral vascular territories, white matter, and basal ganglia in 4 axial slices in each patient.
Results: The linear regression relationship between PET CBF and Xe/CT CBF was described by the equation: PET CBF = 0.608 (Xe/CT CBF) + 3.204, R=0.693, P<0.001, N=144) showing that Xe/CT CBF overstated CBF relative to PET CBF by 40%. However, Herscovitch et al (2) reported that paired CBF measurements by PET using H215O and 11C-butanol showed a linear regression relationship described by the equation: Butanol CBF = 1.237 (H215O CBF) −0.830, R = 0.989, N=18, P<0.0001. Thus, H215O CBF understated 11C-butanol CBF by 24%. Correcting for the 24% understatement of PET H215O CBF, PET CBF understates Xe/CT CBF by approximately 27% (Fig.1).
Conclusions: In light of studies showing good correlation between Xe/CT CBF and CBF by the Kety-Schmidt nitrous oxide method (3) and iodoantipyrine (4), it is likely that the difference between PET CBF and Xe/CT CBF is due to a combination of partial volume effects, water permeability surface area product and some flow activation by xenon.
Fig. 1; PET CBF corrected by 24% by comparison with 11C-butanol (2) and Xe/CT CBF measured in twelve symptomatic patients with occlusive vascular disease. Linear regression analysis shows that Xe/Ct CBF overstates PET H215O CBF by 27%.
Supported in part by: AHCPR #RO3 HS09021-01, #K01 HL03851-01, and The Walter L. Copeland Fund of The Pittsburgh Foundation.
ERROR ANALYSIS FOR PET MEASUREMENT OF DOPAMINE D2 RECEPTOR OCCUPANCY BY ANTIPSYCHOTICS
Yoko Ikoma, Hiroshi Ito, Ryosuke Arakawa, Masaki Okumura, Miho Shidahara, Chie Seki, Hidehiko Takahashi, Harumasa Takano, Yuichi Kimura, Tetsuya Suhara, Iwao Kanno
Molecular Imaging Center, National Institute of Radiological Sciences, Chiba, Japan
Background and aims: The clinical effect of antipsychotic drugs has been reported to be associated with a dopamine D2 receptor occupancy. This occupancy with antipsychotic drugs has been measured using PET by quantifying the binding potential (BP) of receptors with [11C]raclopride or [11C]FLB457 before and after antipsychotic administration. In this quantitative analysis, the uncertainty in estimated kinetic parameters depends on the signal-to-noise ratio of a time-activity curve (TAC) in tissue, which may differ according to injection dose, scan protocol, sensitivity of a PET camera and so on. In this study, the reliability of BP estimates in PET measurement of dopamine D2 receptor occupancy was investigated by means of the computer simulation and measured PET data in humans. The influence of scan protocol on the error of estimated BP was also evaluated.
Methods: In the simulation study, 90-min scan TAC of the putamen for [11C]raclopride and the temporal cortex for [11C]FLB457 were created using measured TAC of the cerebellum as an input function and assumed k-values (R1=K1target/K1reference=0.97, k2=0.23, BP=3.2, 0.96, and 0.32 for [11C]raclopride, and R1=0.81, k2=0.05, BP=2.1, 0.63, and 0.21 for [11C]FLB457) with the simplified reference tissue model (SRTM) [1]. The noise was assumed to be Gaussian distribution whose variance was proportional to the true TAC, and added having the same level as observed in volume-of-interest (VOI) based TACs. One thousand noise-added TACs were realized, then kinetic parameters including BP were derived using the SRTM from these TACs, in which length of dynamic PET scan used for the parameter estimation were reduced from 90 to 20 min, and the mean value and SD were calculated. In the human study of [11C]raclopride (injected dose 215±16.6 MBq, specific radioactivity 185±49.6 GBq/µmol) and [11C]FLB457 (injected dose 216±16.2 MBq, specific radioactivity 213±22.6 GBq/µmol) with and without antipsychotics, replicated TACs were generated by resampling the fitting residuals of each frame with bootstrap approach [2] [3], and the reliability of BP estimates was evaluated for various scan length.
Results: In [11C]raclopride simulation study, a 30-min short PET scan gave us unbiased BP estimates as well as 90-min scan, and the deviation was under 10% in case of VOI-based estimation. Conversely, in [11C]FLB457, the mean value and SD of the BP estimates increased remarkably in case of the shorter scan than 60-min. In human study with [11C]raclopride both with and without antipsychotic administration, SD of the caudate and putamen was under 10% when the scan length was longer than 32 min. Meanwhile, in [11C]FLB457 study, the mean value increased and the SD of temporal cortex was over 10% when the scan length was shorter than 60 min.
Conclusions: In a VOI analysis of [11C]raclopride study, the scan length can be shorten to 30 minutes. However, in [11C]FLB457 study, 60-min scan is required for a reliable estimation since its kinetics is slower.
ESTIMATION OF IMAGE DERIVED INPUT FUNCTIONS USING A RECONSTRUCTION BASED PARTIAL VOLUME CORRECTION ALGORITHM: METHODOLOGY AND EVALUATION IN [11C]FLUMAZENIL STUDIES
Jurgen E.M. Mourik1, Mark Lubberink1, Ursula Klumpers2, Emile Comans1, Adriaan A. Lammertsma1, Ronald Boellaard1
1Department of Nuclear Medicine and PET Research, 2Department of Psychiatry, VU University Medical Centre, Amsterdam, The Netherlands
Introduction: The availability of image derived input functions (IDIF) obviates the need for arterial blood sampling and thereby facilitates clinical use of quantitative PET studies. The aim of this study was to develop a method for deriving IDIFs using reconstruction-based partial volume correction (PVC) [1].
Methods: PET and arterial blood data from nine dynamic [11C]flumazenil scans, acquired using an ECAT EXACT HR+ scanner and an on-line blood sampler, were used to develop and evaluate the method. Scans were reconstructed using both standard (no PVC) ordered subset expectation maximization (OSEM, 2 iterations, 16 subsets) and a PVC-OSEM algorithm, which corrects for the spatial resolution of the scanner. Number of iterations and width of PVC kernel were varied.
The following regions of interest (ROIs) Methods: were evaluated for defining cerebral arteries: (a) pixel value threshold, (b) variable number of ‘hottest’ pixels per plane, (c) region growing, (d) cluster analysis, and (e) MR-based ROI. ROIs were defined on a pseudo blood volume image, generated by summation of early frames (<60s). ROIs were copied to all frames and IDIFs were extracted from both OSEM and PVC-OSEM images.
For each IDIF the following parameters were derived: (a) area under the curve (AUC) for peak (1-2 min), (b) AUC for tail (2-60 min), (c) volume of distribution (Vd) obtained from parametric Logan images, and (d) Vd and K1 obtained from parametric basis function method (BFM) images. In each case, Results: were compared with those using on-line measured arterial input functions.
Results: For PVC-OSEM, the optimal trade-off between computational time and signal-to-noise ratio was obtained for 4 iterations and 16 subsets. A 5.5 mm Gaussian resolution kernel gave optimal recovery correction. The best IDIF was obtained using the ‘four hottest pixels per plane’ over the blood pool in the region below the base of the skull. Compared with standard OSEM, use of PVC-OSEM improved mean (SD) AUC from 0.46 (0.06) to 1.15 (0.11) for the peak and from 0.82 (0.06) to 0.94 (0.12) for the tail part of the input function, respectively. Results: of the comparison between OSEM and PVC-OSEM for Vd and K1 are shown in Table 1 and Figure 1.
Discussion and Conclusions: Excellent correlations were obtained between Vd and K1 values based on IDIFs and those based on on-line sampled input functions. Definition of an accurate IDIF may be sensitive to patient movement and future studies need to focus on motion correction Methods:. Nevertheless, this study shows the feasibility of deriving accurate IDIFs from dynamic PET scans using reconstruction-based PVC.
EVALUATION OF PARTIAL VOLUME CORRECTIONS FOR (R)-[C-11]PK11195 STUDIES
Marc Kropholler1, Ronald Boellaard1, Alie Schuitemaker2, Bart van Berckel1, Reina Kloet1, Phillip Scheltens2, Claus Svarer3, Adriaan A. Lammertsma1
1Department of Nuclear Medicine and PET Research, VU University Medical Centre, Amsterdam, The Netherlands, 2Department of Neurology, VU University Medical Centre, Amsterdam, The Netherlands, 3Neurobiology Research Unit, Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark
Introduction: Partial volume effects may affect quantification of position emission tomography (PET) scans. The purpose of the present study was to evaluate several partial volume correction (PVC) Methods: in relation to the analysis of (R)-[C-11]PK11195 studies in healthy subjects and in subjects with mild cognitive impairment (MCI) and Alzheimer's disease (AD).
Methods: Young (n = 12; age 28 ± 9 yr) and elderly (n = 18; 69 ± 7 yr) healthy subjects, together with MCI (n = 9; 73 ± 6 yr) and probable AD (n = 15; 69 ± 7 yr) patients, were included. Subjects underwent a 60 minutes dynamic brain scan following bolus injection of 400 ± 60 MBq of (R)-[C-11]PK11195. A metabolite corrected plasma input function was derived using continuous online and discrete manual arterial blood sampling. Eight PVC Methods: were tested [1], including Methods: by Meltzer [2], M?ller-Gartner [3], Rousset [4] and Alfano [5]. Regions of interest (ROI) were defined on a co-registered MRI scans for the following structures: thalamus and orbital frontal, medial inferior, temporal and parietal cortex. MRI segmentation was used to restrict ROI to grey matter and to apply PVC. Binding potential (BP) and volume of distribution (VD) were determined using a two-tissue reversible plasma input compartment model, both with and without fixing K1/k2 to the value for whole cortex (BPfixed, VDfixed and BPunfixed, VDunfixed). In addition, BP was obtained using the simplified reference tissue model (BPsrtm) with whole cerebellum as reference tissue [6].
Results: and discussion: Different ROI showed similar trends. Therefore, only thalamus Results: are presented. PVC increased VDfixed, VDunfixed, BPfixed and BPsrtm, but not BPunfixed (Fig. 1A). Changes in BP and VD values for young, elderly, MCI and AD subjects were similar, although occasionally a smaller increase in BP or VD was seen in MCI and AD patients. More importantly, PVC changed the K1/k2 ratio (Fig. 1B), resulting in changes in BPsrtm and BPfixed but not BPunfixed values. Therefore changed BPsrtm and BPfixed values were considered unphysiological. In addition, PVC reduced BP and VD precision and did not improve discrimination between subject groups. Simulations confirmed clinical Results:. Therefore, PVC cannot be recommended for (R)-[C-11]PK11195 studies.
A DATA ADAPTIVE CRITERION FOR CHOOSING BETWEEN ROBUST AND LEAST SQUARES FITTING
R. Todd Ogden1,2,3, Chung Chang2, J. John Mann1,3, Ramin V. Parsey1,3
1Department of Neuroscience, New York State Psychiatric Institute, New York, NY, USA, 2Department of Biostatistics, 3Department of Psychiatry, Columbia University, New York, NY, USA
Background and aims: In neuroreceptor studies with PET, it is common to apply weighted least squares (WLS) methodology for fitting models to data and thus estimating kinetic parameters. This approach is known to be optimal when the errors are normally distributed, but parameter estimation can be heavily influenced by large errors or outliers in the data. In such a case, we have shown that a robust fitting method such as minimizing the sum of weighted absolute residuals, corresponding to estimation the conditional median rather than the conditional mean, provides more stable estimates of kinetic parameters for a variety of kinetic models and their alternatives (such as graphical Methods:). The strategy of applying robust Methods: to all studies can offer substantial improvements for those in which influential observations are present but estimation is less efficient than WLS Methods: for studies in which errors are approximately Gaussian.
Methods: We propose a data adaptive criterion based on measures of influence adapted for nonlinear regression in order to choose between WLS and robust fitting. This is applied to three common models used in PET neuroreceptor studies.
Results: A simulation study (see Figure) was conducted in which the three Methods: (WLS estimation, robust estimation, and estimation from “better? method as determined by the proposed criterion) were compared as the percentage of outliers increases from 0 to 10%. For each method and percentage of outliers, the median absolute deviation (MAD) of estimated VT values was computed for 1000 datasets. The simulation Results: show that, as expected, the performance of the WLS method (and, to a lesser extent, that of the robust method) worsens as the percentage of outliers increases. The data adaptive method performs best of the three for all but the smallest percentages of outliers. This methodology is applied to real PET data and improvements are shown.
Conclusions: Routine application of the proposed criterion to select between robust and WLS fitting Methods: can offer substantial improvements in estimation of kinetic parameters.
STUDIES ON THE MOLECULAR BASIS OF NON_SPECIFIC BINDING
Sarra Sebai1, Gemma Shearman1, Lula Rosso1, Xavier Mulet1, Magdalena Baciu1, Oscar Ces1, James Clarke1, Vincent J. Cunningham2, Roger Gunn2, Rob Law1, Ian Gould1, Christine Parker2, Christophe Plisson2, Richard Templer1, Antony Gee1,2
1Department of Chemistry, 2GSK Clinical Imaging Centre, Imperial College London, Hammersmith Hospital, London, UK
Non-specific binding is a poorly understood phenomenon despite it being a major reason for the failure of potential radioligands for in vivo imaging. To address this issue we have studied the phenomenon of non-specific binding at a molecular level using a combination of experimental and in silico Methods:
The molecule-membrane interactions of artificially constructed membrane preparations with a small library of compounds (some being successful PET radioligands) were studied using small-angle X-ray scattering, solid-state NMR, HPLC and fluorescence microscopy.
Quantum mechanical Methods: were used to estimate accurately the strength of the electronic interaction between individual drug molecules and a single phospholipid molecule commonly present in brain membranes.
The findings of the above investigations were correlated with the absolute in vivo non-specific binding distribution volumes, physicochemical and pharmacological parameters for the above compounds obtained from literature reports or in house data.
Most of the molecules tested resided in the polar-apolar interface of the artificially constructed membrane system. Surprisingly, many of these compounds caused hydrolysis of the fatty acid chains of the phospholipids in an autocatalytic process to produce free fatty acids and mono chained phospholipids. Further investigation showed that the rate of membrane hydrolysis might be correlated with the degree of non-specific binding measured for these compounds in vivo. The in silico modelling of small molecule-membrane interaction energies are also correlated with the in vivo data.
Put together these data suggest the possible existence of a novel membrane hydrolysis phenomenon, which we speculate could be a fundamental mechanism for molecular transport across membranes and also the degree of non-specific binding of radiolabelled molecules.
A FEASIBILITY STUDY ON QUANTITATIVE [15O]H2O AND [18F]FDG BRAIN PET IN A NEW LARGE-ANIMAL STROKE MODEL IN SHEEP
Henryk Barthel1, Johannes Boltze2, Udo Grossmann3, Andreas Schildan1, Annette Foerschler4, Uwe Gille5, Marianne Patt1, Anita Seese1, Frank Emmrich2, Osama Sabri1
1Department of Nuclear Medicine, University of Leipzig, 2Fraunhofer Institute for Cell Therapy and Immunology Leipzig, 3Translation Centre for Regenerative Medicine, University of Leipzig, 4Department of Neuroradiology, University of Leipzig, Leipzig, Germany, 5Vita34 PLC, Leipzig, Germany
In stroke, stem cell (SC) treatment is currently in the stage of translation from preclinical to clinical testing as a new therapy option. It is estimated to have the potential to overcome the therapeutic nihilism which exists in stroke apart from the 3-hours time window in which thrombolysis is effective. We have developed a new large-animal stroke model with the aim to apply this model for evaluation of the effect of autologous SCs cells on stroke outcome. This present study was initiated to test whether it is feasible to quantify sheep brain perfusion and glucose consumption (as a surrogate parameter of brain viability) after stroke by employing a clinical PET scanner.
In adult male Merino sheep, stroke was induced by complete left middle cerebral artery occlusion (MCAO) using high-frequency bipolar forceps. 24h after MCAO, 4.0mio kg-1 autologous mononuclear bone-marrow cells (containing 1.5–2..0% stromal and hematopoietic SCs) were given i.v. Brain PET was carried out immediately before SC application, 2wks and 6wks after MCAO. For that purpose, the sheep were anesthetized, i.v. injected with ~1000 and ~370 MBq [15O]H2O and [18F]FDG, respectively, and dynamically scanned on a clinical high-resolution ECAT EXACT HR+ scanner. The PET imaging was paralleled by neurological investigations using an established clinical score as well as by MRI (1.5T Gyroscan Integra; T1, T2, T2*, diffusion-weighted sequences, MRA). The PET image data were co-registered with the individual MRI data employing the MultiModality software. The animals were sacrificed 7wks after MCAO and brain histology was obtained.
So far, 13 PET investigations have been carried out in the animals. Radiotracer input function which is required for PET data modeling was no obtainable from arterial blood samples. This was due to the follow-up design of this study and the complex anatomy of the sheep vasculature. As an alternative it was verified that it is possible to derive the input function internally from the maxillary artery as visualized in the dynamic PET images. In the sheep, complete MCAO resulted in reproducible neurological dysfunction (absent/delayed startle reflexes, ataxia, circling, torticollis). We also constantly observed a pathologic signal in the diffusion-weighted MRI sequences in the MCAO-related brain areas 24h after MCAO (volume 15±4ml). This was surrounded by a territory of decreased [15O]H2O uptake but preserved [18F]FDG uptake (35ml). In the territory of the initial diffusion disturbance, the [15O]H2O and [18F]FDG uptake was reduced by 69±7% and 46±9%, respectively (as compared to the contralateral brain hemisphere; p<0.05).
From these Results: it is concluded that it is feasible to quantify brain perfusion and glucose consumption in sheep by means of PET. Combined PET and MRI image data processing enables for discrimination between infarction core and surrounding ischemic penumbra, i.e. potentially salvageable brain tissue. To our knowledge, this is the first report on brain PET imaging in a sheep stroke model. It will be possible in the next step by applying the PET protocol developed in this feasibility study to answer the question whether autologous SCs have a positive effect on brain integrity after stroke.
ERROR PROPAGATION OF INPUT FUCTION TO ESTIMATION OF ANTIHISTAMINE DRUG RECEPTOR OCCUPANCY USING C-11 DOXEPIN PET
Kazuhiko Adachi1, Michio Senda2, Jun Nakakuma1, Keiichi Matsumoto2, Keiji Shimizu2, Hideyuki Tominaga2, Nobuo Kubo3
1Department of Mechanical Engineering, Kobe University, Kobe, Hyogo, Japan, 2Division of Molecular Imaging, Institute of Biomedical Research and Innovation, Kobe, Hyogo, Japan, 3Department of Otolaryngology, Kansai Medical University, Moriguchi, Osaka, Japan
Background and aims: PET imaging with C-11 doxepin, a histamine H1 receptor ligand, has been used to evaluate sedative side effect of antihistamine drugs on human subjects by measuring the receptor occupancy (RO) that reflects the adverse effect. The RO is obtained from the binding potential (BP) of the C-11 doxepin following the administration of the drug and that in the baseline. BP is usually derived from the measured distribution volume (DV) in the cerebral cortex and in the cerebellum. Although accurate estimation of the input function is crucial to estimation of DV and BP, sequential arterial blood sampling with metabolite analysis is not practical, and standard input function with one-point venous blood sampling is often adopted. The purpose of the study is to evaluate the effect of the error in the input function on the estimation of the DV, BP, and RO.
Methods: The subjects consisted of 17 young male subjects (age 20-24) who underwent C-11 doxepin PET study in the baseline and under anti-histamine medication. The input function was obtained from the standard input function calibrated with three-point venous blood sampling for each subject sampled at 5, 10 and 15 min after the intravenous injection of 769+/−41MBq of C-11 doxepin (22+/−4MBq/nmol). The input function was corrected for the labeled metabolites using population data. The time course of the brain tissue radioactivity for each subject was measured with PET for 90 minutes starting the injection. One-tissue compartment model was used to estimate the model parameters: K1, the rate of tracer penetration into the brain; k2, the rate of tracer back diffusion to plasma, to compute DV as K1/k2. DV was also obtained by Logan plot. Using the cerebellum as the reference tissue, the H1 receptor BP in the frontal cortex was obtained. As a simulation, error was introduced to the input function by +/−10 percent variation in gain or +/−10 sec shift in time and the effect on DV, BP and RO was evaluated.
Results: The error analysis indicated that +/−10 percent variations in gain for input function led to +/−10 percent errors in K1, which directly led to about +/−10 percent errors in DV. However, these errors were canceled out for BP and RO. The time shift of +/−10 sec in input function led to +/−3 percent error in K1 and +/−5 percent error in k2, resulting in +/−4 percent error in DV. The errors in DV obtained by Logan plot were approximately 90 percent smaller than that for DV computed from K1 and k2, and the propagation to BP and RO was minimal.
Conclusions: The Results: indicate that, although accurate estimation of the input function is necessary, Logan plot is more robust and reliable than non-linear least squares estimation of the rate constants to obtain DV. The simulation also suggests that the error tends to cancel out for BP and RO and that simplified method of acquiring the input function may provide RO with sufficient precision to evaluate the sedative effect of anti-histamines.
PATTERNS OF BRAIN ACTIVATION DURING CUE-ELICITED CRAVING IN RATS
Wynne K. Schiffer, Joseph Carrion, David L. Alexoff, David Frumberg, Stephen L. Dewey
Medical Department, Brookhaven National Laboratory, Upton, NY, USA
BACKGROUND AND AIMS: Human PET studies have identified a metabolic signature of drug craving in dependent subjects presented with drug-associated cues. We tested the hypothesis that cue-elicited craving for methamphetamine in rats, when modeled with the conditioned place preference (CPP) paradigm, Results: in a similar pattern of brain activation. We developed a behavioral neuroimaging strategy and tested both Forced Exposure (FE; during FDG uptake, animals are placed in a distinct environment where they received methamphetamine, METH) and Free Choice exposure (FC; animals are given a choice of chambers during FDG uptake).
METHODS: PET FDG scanning in male Sprague Dawley rodents was performed during the CPP pre-test. Animals were treated with METH (1.0 mg/kg) or saline in separate chambers for 10 days. On the day of the CPP test, animals received only FDG prior to being placed in the METH-paired chamber (FE) or in the neutral chamber (FC). The extent of locomotor activity in the drug-paired chamber was recorded on videotape and rated with a computerized automated scoring system (TopScan, CleverSync). Animals were anesthetized after 45 min of uptake and scanned for 10 min. FDG data were normalized to stereotaxic space and proportionally scaled to the global mean. Voxel-wise analysis was implemented in SPM2 and verified by Region of Interest (ROI) regression analysis performed using PMOD. The main factor of interest was locomotor activity in the drug paired chamber (distance, in mm) and where possible, preference (time in drug-paired minus time in saline-paired chamber).
RESULTS: Compared to the baseline scan, both conditions produced significant bilateral activations of the motor cortex, temporal cortex, cerebellum and thalamus. In the FC protocol, there were significant bilateral increases in FDG uptake in the medial forebrain bundle and striatum that were not evident in the FE protocol. However, FE significantly activated the amygdala and sensory cortex where FC did not. Patterns of activation detected by SPM were substantiated by ROI analysis, where brain activity from the FC striatal and orbitofrontal ROIs (including the medial forebrain bundle) was positively correlated with conditioning score (R2 = 0.57, p < 0.001).
CONCLUSIONS: These studies support the translational value of small animal PET using this behavioral neuroimaging strategy coupled with voxel-based image analysis (see Figure). Our data provide evidence that the FC approach (Figure [a]) most closely mimics the brain activation documented in human subjects (Figure [b], from Volkow et al., 2003), where common areas of activation were the thalamus, frontal cortex and cerebellum. Our studies support an experimental protocol where FDG is injected away from the tomograph and with minimal impact on behavior. Reproducible data can be obtained following a 10 min scan, increasing throughput and maximizing tomograph availability.
THE INHIBITORY EFFECT OF PK11195 ON GLUCOSE METABOLISM IN QUIN-INJECTED RAT STRIATUM
Misato Amitani1, Akihiro Ohashi1, Antony Gee2, Osamu Inoue1
1Course of Allied Health Sciences, Graduate School of Medicine, Osaka University, Suita, Osaka, Japan, 2GlaxoSmithkline Clinical Research Unit, ACCI, Addenbrookes Hospital, Cambridge, UK
[Background and aims:]In central nervus system, peripheral benzdiazepine receptors (PBRs) are predominantly expressed in glial cells. At the cellular level, localization of PBRs is mainly in the outer mitochondrial membrane. The inhibitory effect of PK11195, a selective PBR ligand, on neuronal damage induced by microinjection of an excitatory amino acid (quinolic acid; QUIN), was recently reported. The mechanism for neuroprotection of PK11195 has been thought to diminish activation of microglia. It has been well known that glucose metabolism was appreciably enhanced by infusion of excitatory amino acid at an early phase after the infusion. In this study, we examined whether PK11195 also inhibited the enhancement of glucose metabolism following excitatory amino acid using QUIN-injected rats.[Methods:]QUIN (60 nmol/microL), PK11195 (10 nmol/microL) and a combination of QUIN and PK11195 were infused into the right striatum of rats for 5 minutes, respectively. After 2 hours, the rats were intravenously injected with [14C]deoxy-D-glucose (DG; 185 kBq/rat), and decapitated 1 minute or 45 minutes after injection of the tracer. The brain sections were prepared and exposed to imaging plates for 1–2 weeks, and autoradiograms were quantitated. Using the next sections, nissl staining were perfomed, and the number of neuronal cells and glial cells were counted. In order to estimate microglial activation at this time point, in vitro binding of [3H]PK11195 was examined.[Results:]The intrastriatal injection of QUIN alone caused a significant enhancement of [14C]DG uptake (45 minutes) in the infused side of striatum (about 160% vs control side), however, the initial uptake of [14C]DG was not increased. The co-injection of PK11195 with QUIN completely blocked the enhancement of [14C]DG uptake induced by QUIN. No significant alterations of [14C]DG uptake were observed in PK11195-infused striatum. By nissl staining, it was found that the number of neuronal cells in striatum was decreased (about 30%) at 2 hours after QUIN infusion; however, the treatment of PK11195 was effective in decreasing neuronal damage. By in vitro autoradiography, there were no significant changes of [3H]PK11195 binding in QUIN-infused rat brain.[Discussion]As initial uptake (1 minute) of [14C]DG in QUIN-injected striatum was not increased, the enhancement of glucose metabolism induced by QUIN seems to be due to the increase of phosphorylation rate by hexokinase (HK) rather than transport process. PK11195 completely suppressed the increase of glucose metabolism in QUIN-infused striatum. The voltage-dependent anion channels (VDACs), a protein of PBR complex, are associated with mitochondrial membrane bound HK; therefore, it seems more likely that PK11195 regulates HK activity through VDACs. The neuroprotective action of PK11195 has been thought to be due to inhibit microglial activation, however, no activation of microglia was observed at 2 hours after QUIN-infusion by [3H]PK11195 binding. This study revealed that PK11195 also showed the depressant effect on changes in energy metabolism in intact brain at an early stage of neuronal damage. The alterations in energy metabolism may be occurred in glial cells, since PBRs are localized in glial cells.
FDG PET MEASUREMENT OF LONG-TERM POTENTIATION IN RAT BRAIN
Trine Hjornevik1, Hong Qu1, Johannes Gjerstad3, Line Jacobsen3, Bent Schoultz2, Jan Gunnar Bjaalie1, Gjermund Henriksen2, Frode Willoch1,4
1Centre for Molecular Biology and Neuroscience, Department of Anatomy, University of Oslo, Oslo, Norway, 2Department of Chemistry, University of Oslo, Oslo, Norway, 3National Institute of Occupational Health, Oslo, Norway, 4Department of Radiology, Aker University Hospital, Oslo, Norway
Background: Neuronal events leading to development of long-term potentiation (LTP) in the nociceptive pathways may be a cellular mechanism underlying central hyperalgesia. Here, we examine induction of LTP in nociceptive dorsal horn neurons (single cell activity) and changes in brain activity (FDG µ-PET).
Methods and Results: Female Sprague?Dawley rats weighing 250?350 g, adapted for minimum 10 days to environment and given food and water ad librum, were for the following experiments anesthetized with isoflurane gas. The left sciatic nerve was given a high-frequency conditioning stimulation (HFS) to induce LTP (5 trains/1 s, 100 Hz, 1 ms pulses, 10 s intervals).
1) Extracellular single cell activity was recorded from neurons at depths of 80?500 µm from the surface of the spinal cord (n=6). A clear LTP of the nociceptive transmission following sciatic nerve HFS was observed in single C-fibre neurons in laminae I?IV of the dorsal horn. The field potential response was 150% of baseline immediately after HFS and increasing to 200% 180 min after stimulation (p<0.001).
2) FDG PET studies were performed in parallel. A normal blood glucose level was confirmed in rats prior to FDG injection. A HFS group (n=10) and a sham group (6) underwent at consecutive days, dynamic scan (Focus 120, Siemens) over 60 min with measured attenuation correction prior to FDG injections. All rats underwent the first day a rest study without any intervention. At the second day, FDG was injected 2.5 hours after the HFS or sham operation. Data were MAP reconstructed and summed between 30-60 minutes. Activity from lacrimal glands was removed before further processing. Statistical comparisons between rest and stimulation condition were performed with SPM5 (http://www.fil.ion.ucl.ac.uk/spm/) after anatomic standardization to a MRI template fitted to the Paxinos and Watson atlas (5th edition, 2006). Level of significance was set at p<0.05. In the HFS group there was a significant relative increased FDG uptake in hypothalamus, ipsilateral amygdala and contralateral amygdala with adjacent insular cortical areas. The sham group demonstrated a slight increase in hypothalamus only. Significant relative decreases were in the HFS group seen in the contralateral and ipsilateral sensorimotor cortices (smc) and in the upper brain stem including the periaqueductal grey. The sham group demonstrated an equivalent pattern, but with a smaller extent of brain stem changes.
Conclusion: The data show that HFS applied to the sciatic nerve lead to the induction of LTP in the dorsal horn C-fibres and remains for 3 hours. At supraspinal level the HFS led to an increased activity in the amygdala bilaterally and contralateral insular area not seen in the sham control group. The HFS group exhibited also a more emphasized activation in hypothalamus and extended deactivation in the upper brain stem, all areas important in nociception. The study provides an in vivo model for studying LTP at supraspinal level and may be important in the study of hyperalgesia and antinociceptive treatment.
UPREGULATION OF PBR BINDING SITES IN BRAIN C6 GLIOMA MODEL: A [11C]PK11195 MICROPET STUDY
Antonio Aliaga, Pedro Rosa-Neto, Shadreck Mzengeza, Barry Beddell
Montreal Neurological Institute and Hospital, McGill University, Montreal, QC, Canada
Background and aims
PK11195 and other peripheral benzodiazepine receptors (PBR) ligands have been shown to induce apoptosis in glioma cells, and, as such, are currently being explored as novel tumour-targeted therapeutic agents. Thus, the characterization of PBR binding in animal models of brain tumor is required in order to verify whether of these drugs reach their pharmacological target.
Methods
C6 glioma spheroids were implanted into the right cerebral hemispheres of 6 male Sprague-Dawley rats (326±78g). Animals underwent PET studies at 14-18 days post-implantation using racemic [11C]PK11195 (266.± 13.5uCi, SA~0.8 1.2Ci/uMol). All scans were performed with rats under anesthesia with isofluorane 2% delivered by a nose cone. The tail vein was catheterized for the injection of pharmaceuticals. Images were obtained during 60 min using a CTI Concorde R4 microPET scanner using list mode file, and subsequently reconstructed using filter-back projection algorithm. Four animals had a two [11C]PK11195 scans the fist one performed in baseline condition and the second scan 20 min after 3mg/kg of (+)PK11195. An additional [18F]FDG (172±42uCi) study was performed 100 minutes after the completion of the second [11C]PK11195 scan. Additional two animals had a [18F]FDG after [11C]PK11195 scan without receiving any cold PK PK11195. Time-activity curves were obtained in both tumor, cortex contralateral to the tumor and cerebellum. [11C]PK11195 binding potential (B.P.) was determined using the simplified reference tissue method [2]. Static [18F]FDG standard uptake value (SUV) were obtained using the same volumes of interest.
Results
The mean [11C]PK11195 baseline binding potential for the brain tumor was pB=1.38±1.4 while in the contralateral cortex the binding potential was pB=0.16±0.04 (p<0.01; t=5.99;df=2). The mean SUV of the tumor was 2.01±0.27 while in the contralateral cortex 1.28±0.04 (p=0.05;t=2.43;df=3). Saturation binding experiments showed a occupancy of nearly 100% following 3mg/kg of (+) PK11195 (p<0.01; t=6.9;df=2). There was a positive correlation between [18F]FDG SUV and [11C]PK11195 binding (r2=0.68, r=0.82, P=00.2).
Conclusions
These experiments suggest that the C6 model overexpress saturable binding sites for [11C]PK11195 compared to the normal cortex. The increase of [11C]PK11195 can be associated with an overexpression of PBR receptors by the tumor cells as well as microglia and infiltrating macrophages. Although this data suggest a positive correlation between tumor cell proliferation, inflammatory infiltrates and metabolism, the mechanism underlying this finding has to be further explored. [11C]PK11195 and microPET provide means to evaluate glioma viability and the pre-clinical evaluation of new antiglioma agents targeting PBR.
SILMULTANEOUS FDG MEASUREMENT BY BETA-MICROPROBE AND ?PET: A COMPARATIVE FOLLOW-UP IN THE RAT BRAIN
Martine Guillermier1, James James Godart4, Marc Dhenain1, Laurent Besret3, Thierry Delzescaux1, Albertine Dubois2, Sebastien Jan2, Frederic Pain4, Regine Trebossen2, Anne-Sophie Herard1
1MIRCen, CEA, Fontenay-Aux-Roses, France, 2Shfj, Cea, Orsay, France, 3Sanofi-Aventis, Vitry-Sur-Seine, France, 4Imnc, Cnrs, Orsay, France
Purpose: µPET and beta-microprobe are complementary Methods: to assess brain metabolism. µPET provides a high spatial resolution (1.4 mm) and information concerning the whole brain while beta-microprobe has a high temporal resolution (1 measurement per second) but is more invasive and only provides information from a restricted region surrounding the probe. We compared the local cerebral metabolic rate of glucose (CMRGlu) measured, in vivo, in the rat striatum by µPET and beta-microprobe approaches. beta-microprobe data were compared to µPET images acquired simultaneously in the same animal. In the implanted striatum, we compared CMRGlu recorded with both Methods:. Comparison of µPET signal from the two striata also allowed evaluating the effect of beta-microprobe implantation on the cerebral metabolism. In addition, the animal was longitudinally followed up during 42 days to determine whether: i) the unilateral chronical beta-microprobe implantation had any influence on CMRGlu ii) chronic implantation modified the beta-microprobe in vivo sensitivity. Method: betamicroprobe was chronically implanted in the left striatum of an adult male Sprague-Dawley rat. Subsequent 3D MR images were recorded at 3T (Inversion recovery fast spin echo protocols (TR/TE/wTE/TI: 2000/10.1/79.4/200 ms, Rare Factor=16, isotropic resolution 300µm). The animal was also chronically implanted with both arterial and venous catheters. At days 14, 28, and 42 after probe implantation, beta-microprobe recording and PET dynamic images (FOCUS 220, Siemens Medical Solutions) were simultaneously performed after 2-[18F]fluoro-2-deoxy-D-glucose (FDG) intravenous injection (0,7 mCi/100 g). An arterial input function was measured for each experiment. The animal was anesthetized using isoflurane (2%). Twenty one dynamic PET frames were acquired over 60 minutes following FDG injection. PET images were reconstructed using FORE and 2D OSEM algorithm (4 iterations and 16 subsets) and registered with MRI using rigid spatial transformation and mutual information as optimization criterion. The beta-microprobe measurements were initially taken every second. The CMRGlu in the ipsilateral and the contralateral striatum were calculated using the Patlak simplified method between 5 to 45 minutes on ROI's derived from MRI data. Results:/Conclusion: Whatever the acquisition mode, the evaluated CMRGlu showed variations between the different exams (Fig. 1). This could be explained by modifications of animal physiology at the time of each examination (weight, anesthesia, blood pO2, pCO2, and pH). Despite this variation, our preliminary data suggest that at each time point the CMRGlu obtained using the beta-microprobe was similar to that measured on the PET images. In addition, comparison between ipsilateral and contralateral CMRGlu at each time point showed that beta-microprobe implantation did not change glucose metabolism in the striatum, and that beta-microprobe sensitivity did not vary, up to 42 days post-implantation.
EFFECT OF TREATMENT ON THE DOPAMINE TRANSPORTER IN A RAT MODEL OF PD; A PET STUDY
Vesna Sossi1, Katherine Dinelle1, Siobhan McCormick3, Rick Kornelsen3, Geoffrey Topping1, Thomas Ruth4, Doris Doudet2
1Department of Physics and Astronomy, UBC, Vancouver, BC, Canada, 2Department of Medicine, UBC, Vancouver, BC, Canada, 3Parkinson's Research Centre, Vancouver, BC, Canada, 4TRIUMF, Vancouver, BC, Canada
Background and aims:. There is speculation about the optimum treatment strategy for Parkinson's disease (PD). A recent study involving DAT imaging and clinical outcomes (1) compared the efficacy of dopamine agonist therapy (DA) to that of levodopa therapy. While imaging showed a relative preservation of DAT in patients on DA, interpreted as slowing of disease progression, patients treated with levodopa showed less motor impairment. There is thus an apparent discrepancy between the clinical assessment of disease progression and that measured by preservation of DAT binding. The question was raised of a possible interaction between treatment itself and DAT levels. In this study we are addressing this question using a unilateral 6-hydroxydopamine (6-OHDA) rat model of PD and a microPET Focus 120 scanner.
Methods:. Six unilaterally lesioned rats underwent a baseline 11C-(+) dihydrotrabenazine (DTBZ- a VMAT2 marker) and 11C-methylphenidate (MP- a DAT marker) scan. Three rats were randomly assigned to levodopa treatment (10 mg/kg; mixed with benserazide) once daily for four weeks, and three rats were given 1mg/kg/day of pramipexole once daily for the same duration. After 4 weeks of treatment the rats are being rescanned with the same imaging protocols ? at present only two out of the six rats (one from each treatment group) underwent the second MP scan. Lesion severity was estimated as LS = (1- BP[lesioned side]/BP[healthy side])*100. Relative changes in LS estimated before and after treatment were assumed to provide an assessment of treatment effects on DAT levels.
Results:. Preliminary Results: from the baseline condition scans show a very good correlation between LS evaluated with DTBZ and LS evaluated with MP: LSMP = 0.79 LSDTBZ + 0.15, r2 = 0.95. The more severe LSMP obtained for milder lesions appears consistent with human studies showing a DAT downregulation in mild disease (barring anesthesia confounding effects ? to be investigated). In the second MP scan for the rat that underwent levodopa treatment the LS estimate changed from 51% to 66%, while for the rat that underwent pramipexole treatment the LS estimate changed from 52% to 39% which is consistent with the hypothesis of treatment influencing DAT levels.
Conclusion:. These are only preliminary data, since up to now we only have Results: for two rats. If the findings are confirmed at the completion of the study, we will show a direct correlation between treatment and DAT regulation and will thus provide some explanation for the apparent contradiction observed in the human studies.
IMAGING VESICULAR ACETYLCHOLINE TRANSPORTER IN RODENTS USING [18F]FLUOROETHOXY-BENZOVESAMICOL AND MICROPET
Pedro Rosa-Neto1, Antonio Alliaga1, Shadreck Mzengeza1, Gassan Massarweh1, Elevine Landry2, Marc-Andre Bedard2, Jean-Paul Soucy1
1Montreal Neurological Institute and Hospital, McGill University, Montreal, QC, Canada, 2Universite du Quebec a Montreal, Montreal, QC, Canada
Introduction
Acetylcholine neurotransmission has an important role in several cognitive processes and seems to be altered in neurodegenerative diseases such as Alzheimer's disease. [18F]fluoroethoxy benzovesamicol [18F]FEOBV is a high affinity positron emitting ligand for the vesicular acetylcholine transporter. Vesicular acetylcholine transporter imaging should provide information on the presynaptic component of the cholinergic system.
Methods
Three male rats (350±10g) were anesthetized with isofluorane 2% provided by a nosecone and placed in the scanner (CTI Concorde R4 microPET). Physiological parameters were monitored during the whole PET session. After a brief transmission scan with a [57Co] point source, animals received an i.v. dose of 11 MBq of [18F]FEOBV with high specific activity. Emission scans were obtained for 60 min. in list mode. Images were subsequently reconstructed using a filtered-back projection algorithm and coregistered to a rat anatomical template. Time-activity curves were obtained at different levels and binding potentials (pB) were determined using the reference tissue graphical method.
Results
[18F]FEOBV behaves as a irreversible ligand in brain areas with high concentration of vesicular acetylcholine transporter. There was a fast washout of radioactivity from the cerebellum. Figure 1 shows averaged binding potential maps of [18F]FEOBV in the rat brain as observed in these animals. The highest binding potentials were detected in the caudate (Fig1C), amygdala/hippocampus (Fig1B) and basal forebrain (Fig 1A). There was high uptake of radioactivity by the harderian glands (Fig 1C).
Conclusions
[18F]FEOBV and MicroPET are capable of imaging the vesicular acetylcholine transporter in the rat brain.
VOXEL-BASED STATISTICAL ANALYSIS OF INTERHEMISPHERIC CMRGLU DIFFERENCES DURING A UNILATERAL VISUAL STIMULATION IN RATS
Albertine Dubois1, Anne-Sophie Herard2, Laurent Besret2, Guillaume Flandin3, Gilles Bonvento2, Thierry Delzescaux1
1Service Hospitalier Frederic Joliot, CEA-UIIBP, Orsay, France, 2Service Hospitalier Frederic Joliot, CEA CNRS URA2210, Orsay, France, 3Service Hospitalier Frederic Joliot, CEA-UNAF, Orsay, France
Purposes: Besides the newly developed PET scanners dedicated to in vivo functional study, autoradiography remains the reference technique widely used to assess cerebral glucose metabolism (CMRGlu) in rodents. Autoradiographs are conventionally analyzed by a region-of-interest (ROI) analysis, which is not suited for a whole-brain analysis. Statistical Parametric Mapping (SPM) is a voxel-based statistical method used to determine differences in brain activity during functional neuroimaging experiments (PET, fMRI). Here, we describe the feasibility of adapting the SPM method to an autoradiographic data set mapping CMRGlu in rats during unilateral visual stimulation.
Methods: After injection of [14C]-2-deoxyglucose, adult rats (n=5) were visually stimulated during 45 minutes with a moving checkerboard with left eye stimulated and right eye unstimulated. Coronal brain sections (about 150 sections, 20µm-thick) encompassing the visual system were cut with a cryostat and exposed to generate autoradiographs. The data acquisition and the 3D reconstruction of the functional volume were performed using the methodology we previously developed (1). To allow intergroup comparisons of activated vs. non-activated hemispheres, the mid-sagittal planes of each 3D reconstructed functional volume were computed and the left hemispheres were flipped over around this plane. The rat hemibrain template and the spatial normalization of the 10 hemibrains were performed as previously described (2). Interhemipheric differences in CMRGlu were tested by comparing the activated hemibrain group with the symmetrized non-activated hemibrain group. The procedure of SPM analysis of covariance (ANCOVA) was used to specifically consider differences linked to unilateral visual stimulation (3). Results: obtained with SPM were compared to those we previously reported (1). We chose a significance threshold P<0.01 (uncorrected for multiple comparisons) for individual voxels within cluster size of 4000 contiguous voxels (~1 mm3).
Results: SPM detected several areas of significant increase in CMRGlu in the activated hemibrains, namely the areas 17 and 18a of the visual cortex (VC; +10% and +14%, respectively), the superior colliculus (SC; +20%) and the lateral geniculate nucleus (LGN; +15%) (Figure). These areas displayed a similar location and spatial extent to those determined using a ROI approach (1).
Conclusion: SPM ANCOVA analysis allowed the detection of interhemispheric differences in CMRGlu that were consistent with physiology, literature and our previous study. The Results: obtained for the VC, which is difficult to accurately delineate manually, underscores one of SPM advantages over the ROI method. SPM allows the detection of activations that would be difficult to predict, and hence may be missed with the ROI approach. This study demonstrated that SPM is a promising and generic approach for analyzing autoradiographic data sets.
RELATIONSHIP BETWEEN SEROTONIN TRANSPORTER AND SEROTNIN 2A RECEPTOR BINDING IN HEALTHY HUMAN SUBJECTS: A PET STUDY
David Erritzoe1, Vibe Frokjaer1, Haroon Arfan1, Claus Svarer1, Olaf Paulson1,2, Gitte Moos Knudsen1
1Neurobiology Research Unit and Center for Integrated Molecular Brain Imaging, Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmark, 2Danish Magnetic Resonance Center, Hvidovre University Hospital, Hvidovre, Denmark
Background and aims: Serotonergic neurotransmission is involved in the regulation of physiological functions such as mood, sleep, memory and appetite, and serotonergic dysfunction is implicated in the patophysiology of a variety of neuropsychiatric disorders. We have previously shown that serotonin and serotonin transporter (SERT) positive fibers colocalized almost entirely (Nielsen et al, 2006 (1)). In addition, endogenous serotonin levels are probably to a large extent determined by the SERT density and the postsynaptic serotonin 2A (5-HT2A) receptor has been shown to regulated by 5-HT2A agonism, including serotonin. The aim of this study was in the same subjects to compare SERT and 5-HT2A receptor levels in order to establish if these two markers of the serotonergic system covariate.
Methods: Seventeen healthy human subjects, 9 males, (mean age 33.1±4.6 years) were included. The SERT binding was imaged with [11C]DASB PET based on 90 minutes dynamic acquisition starting immediately after bolus injection of 5.4±1.6 Mbq [11C]DASB/kg body weight. 5-HT2A receptor binding was assessed using a bolus plus constant (Vbol=1.75) infusion set-up with 40 minutes scan time starting 2 hrs post injection of 3.7 MBq [18F]altanserin/kg body weight. Subjects also underwent a 3T MR-scan.
For both tracers, PET images were co-registered to the corresponding MR images, and volumes of interest (VOI) were applied automatically onto the individual MRIs and PET images. Cerebellum was used as reference region. On the basis of a larger sample consisting of 104 healthy subjects' [18F]altanserin-scans, the regional variability of the binding was assessed with principal component analysis (PCA). A global binding measure was defined from high binding VOIs for both tracers: A neocortical (volume weighted average of 11 cortical VOIs) BP1 was generated for 5-HT2A and a subcortical (volume weigthed average of thalamus, caudate nucleus and putamen) BP2 for SERT. For SERT, BP2 was calculated using MRTM2 with k2′ fixed generated from striatal data using MRTM. For 5-HT2A receptor binding, BP1 (defined as the difference in count rate between VOI and cerebellum divided by count rate in plasma) was used as the outcome measure. In order to assess the relationship between the two parameters, a linear regression analysis was performed using neocortical 5-HT2A binding as the dependent variable, subcortical SERT binding as independent variable and age as a covariate.
Results: The PCA showed that in all the neocortical VOI's the primary PCA-component explains almost all of the in between subject variance in 5-HT2A. This justifies the use of a single parameter for outcome of the 5-HT2A measure. No significant correlation was seen between subcortical SERT and cortical 5-HT2A binding data.
Conclusion: No linear relationship was detected between subcortical SERT and cortical 5-HT2A binding. Further subjects will be included and non-linear relationships investigated.
A PRELIMINARY CHARACTERISATION OF A DOPAMINE D1 PARTIAL AGONIST RADIOLIGAND FOR PET
Sjoerd Finnema1, Benny Bang-Andersen2, Lars Farde1, Morten Jorgensen2, Balazs Gulyas1, Christian Foged3, Hakan Wikstrom4, Christer Halldin1
1Karolinska Institutet, Stockholm, Sweden, 2H. Lundbeck A/S, Valby, Denmark, 3Novo Nordisk A/S, Bagsvaerd, Denmark, 4University of Groningen, Groningen, The Netherlands
Background and aims: So far the dopamine D1 receptor has been studied with PET using antagonist radioligands such as [11C]SCH23390 and [11C]NNC 112. It has been found that these antagonist radioligands are insensitive to amphetamine induced elevated dopamine levels [1], which is in contrast to D2 receptor antagonist radioligands. Recently developed dopamine D2 receptor agonist radioligands, such as [11C]MNPA have shown an increased sensitivity for alterations in dopamine levels compared to D2 antagonist radioligands [2] and we therefore aim to develop a D1 agonist radioligand.
Methods: (S)-[11C]N-methyl-NNC 01-0259 was obtained by N-methylation of (S)-NNC 01-0259 with [11C]methyl iodide, as reported before [3]. Altogether, five PET measurements were performed in two female cynomolgus monkeys. Initially a baseline and an amphetamine challenge study (1.0 mg/kg) were performed to determine the susceptibility of the radioligand binding to elevated dopamine levels. In addition, pretreatment studies with the antagonist MDL 100,907 (0.1 and 1.0 mg/kg) were conducted to study binding to the 5-HT2A receptor. Binding Potential (BP) values were calculated for striatum and neocortex using MRTM2 with the cerebellum as a reference region.
Results: The average i.v. injected amount of (S)-[11C]N-methyl-NNC 01-0259 was 53.8 ± 2.3 MBq and the specific radioactivity was >5000 Ci/mmol. Pretreatment with amphetamine altered the binding potential (MRTM2) in striatum from 1.53 to 1.47 (−4%) and in the neocortex from 0.43 to 0.41 (−5%). Pretreatment with the 5-HT2A antagonist MDL 100,907 resulted in a reduction in striatum and neocortex BP of −2% and −19% (0.1 mg/kg) and −11% and −31% (1.0 mg/kg) respectively.
Conclusion: The small effect of amphetamine on (S)-[11C]N-methyl-NNC 01-0259 binding indicate low sensitivity to endogenous dopamine levels. This observation is in line with earlier reports using a D1 antagonist radioligand [1]. MDL100,907 pretreatment studies indicate that radioligand binding in the neocortex partly corresponds to binding to the 5-HT2A receptor in an extent comparable to the antagonist radioligands [11C]SCH23390 and [11C]NNC 112 [4].
(+)- AND (−)-[18F]NORCHLORO-FLUORO-HOMOEPIBATIDINE - NEW SUITABLE RADIOTRACERS FOR IMAGING NICOTINIC ACETYLCHOLINE RECEPTORS
Peter Brust1, Winnie Deuther-Conrad1, Georg Becker2, Marianne Patt2, Andreas Schildan2, Kai Kendziorra2, Osama Sabri2, Joerg Steinbach1, Joerg Thomas Patt1
1Institute of Interdisciplinary Isotope Research, Leipzig, Germany, 2Department of Nuclear Medicine, University of Leipzig, Leipzig, Germany
Background and aims: Nicotinic acetylcholine receptors (nAChR) are particularly important during age-related cell degeneration. Our recent studies demonstrated the suitability of the nAChR ligand 2-[18F]F-A-85380 for neuroimaging of patients with various types of brain dementias including Alzheimer's disease. However, the slow binding kinetics of this radiotracer limits its suitability for clinical PET studies. Therefore, we have recently developed [18F]norchloro-fluoro-homoepibatidine ([18F]NCFHEB) for nAChR imaging. Taken into consideration the stereoselectivity of this compound, two new radiotracers were derived for this purpose. In this study we compare the brain kinetics of these new radiotracers with 2-[18F]F-A-85380.
Methods: 24 piglets of mixed German domestic breed were anesthetized with 0.25% isoflurane in 65% nitrous oxide and 35% oxygen. 50-150 MBq of (+)/(−)-[18F]NCFHEB (groups 1 and 2) or 2-[18F]F-A-85380 (group 3) in 10 ml saline were intravenously infused within 2 min. Three animals of each group received an additional i.v. injection (1 mg/kg) of the nAChR agonist A-81418 at 10 min prior to radiotracer application followed by a continuous infusion (2 mg/kg/7h) of the same compound. One animal of group 1 was pre-treated with (+)-NCFHEB (0.7 mg/kg i.v.). Regions of interest were drawn on summed PET images. The region with lowest uptake (olfactory bulb) was selected as reference region for the calculation of specific radiotracer binding. During the whole time course of the PET investigations about 50-60 plasma samples were obtained. Selected samples were analyzed with HPLC to determine the amount of radiolabeled metabolites.
Results: Summed PET images obtained during equilibrium of specific radiotracer binding show that both enantiomers of [18F]NCFHEB exhibit a distribution pattern similar to 2-[18F]F-A-85380 with the highest binding in thalamus, moderate binding in striatum, and lowest binding in cerebellum and olfactory bulb. The brain uptake of (+)- and (−)-[18F]NCFHEB is three- and twofold higher than that of 2-[18F]F-A-85380, respectively. All three radiotracers display different binding kinetics in regions with high, moderate, or low specific binding. Generally, the equilibrium of specific binding of (−)-[18F]NCFHEB is reached much earlier than that of (+)-[18F]NCFHEB or 2-[18F]F-A-85380. The peripheral metabolism of (+)-[18F]NCFHEB proceeds somewhat slower than that of the other two radiotracers. Complete inhibition of specific radiotracer binding was observed after pre-treatment with (+)-[18F]NCFHEB. Continuous application of A-81418 by injection and infusion resulted in partial inhibition of the specific binding of all three radiotracers. Interestingly, the peripheral metabolism of 2-[18F]F-A-85380 was considerably increased after application of A-81418 with implications for the plasma input function. This phenomenon was negligible for the other two radiotracers.
Conclusions: We have found that both enantiomers of [18F]NCFHEB are suitable radiotracers for nAChR imaging. Certain features are better than those of the clinically used 2-[F-18]F-A-85380. While
(−)-[18F]NCFHEB might be suitable in regions with high nAChR densities, (+)-[18F-]NCFHEB may offer advantages for regions with lower nAChR expression.
GLOBAL-TWO-STAGE APPROACH FOR : [11C]DASB PARAMETRIC IMAGING: EVALUATION ON SIMULATED DATA
Giampaolo Tomasi, Alessandra Bertoldo, Valentina Bovo, Claudio Cobelli
Department of Information Engineering, University of Padova, Padova, Italy
INTRODUCTION: Parametric imaging often suffers for the low signal to noise ratio present in PET images. To overcome this problem, a method originally developed in pharmacokinetic/pharmacodynamic literature, the Global Two Stage, GTS, could be used. Here GTS is assessed on voxel-based [11C]DASB simulations.
MATERIAL AND METHODS: A 1-tissue compartment model was employed to describe [11C]DASB kinetics.100 repetitions of three planes of an healthy subject image were generated as described in [1]; the distribution volume DV (=K1/k2) was used for the evaluation of performances through Bias and Rmse defined as 100*(1/p_true)*sqrt [sum((p_i-p_true)^2) / N] (N=100). Generalized Weighted Linear Least Squares, GWLLS, was first employed [1]. GTS [2] was then applied to the parameter estimates and variances provided by GWLLS. GTS is an iterative algorithm which computes at each iteration the mean parameter vector p_i, averaging the parameter vectors of the pixels of a population, and its covariance B_i. The pixel population was defined by segmenting each slice into 5 sub-regions (populations) according to the value of DV prior to the application of conventional GTS. The parameter vector of pixel j at iteration i is then computed as inv[inv(A)+inv(B_i)]*[inv(A)*p_0+inv(B_i)*p_i] where p_0 and A are the initial parameter and covariance estimate for pixel j, which are kept fixed throughout the algorithm. The pixel's TAC, therefore, is never employed by GTS itself making the algorithm fast even if the original model is non-linear. A convergence criterion is used to stop the iterations.
RESULTS: We tested GTS applied to the GWLLS parametric images of K1 and k2 in order to generate GTS DV maps from K1_GTS/k2_GST. GTS was also applied directly to DV GWLLS images. The best Results: in terms of DV accuracy were obtained when DV GWLLS maps were used. Figure 1 displays true, GWLLS and GTS maps for DV (upper panel) and Rmse (lower panel) of slice#32. Rmse decreases from 32% to 26%, from 37% to 31% and from 64% to 54% for slice #32, #36 and #47 respectively with a negligible increase of computational time. Similar reductions were also observed for Bias.
CONCLUSION: This study shows that GTS is a fast and accurate method to generate substantially unbiased parametric images of [11C]DASB DV starting from GWLLS estimates. GTS is potentially usable with other PET tracers and models, provided that initial estimates can be obtained by conventional weighted least squares. Further studies with real [11C]DASB data and additional simulations with other PET tracers are needed to better characterize the potentiality of GTS in PET parametric imaging.
BRAIN AND WHOLE-BODY IMAGING OF PERIPHERAL BENZODIAZEPINE RECEPTORS IN NONHUMAN PRIMATES USING [11C]PBR28
Masao Imaizumi, Emmanuelle Briard, Sami Zoghbi, Jinsoo Hong, John Musachio, Robert Gladding, Victor Pike, Robert Innis, Masahiro Fujita
Molecular Imaging Branch, National Institute of Mental Health, National Institutes of Health, Bethesda, MD, USA
Background: Peripheral benzodiazepine receptors (PBRs) are upregulated on activated microglia and are thereby biomarkers of neuroinflammation. We developed a PET ligand with an aryloxyanilide structure, N-acetyl-N-(2-[11C]methoxybenzyl)-2-phenoxy-5-pyridinamine ([11C]PBR28) to image PBRs. Quantification of brain PBRs may be affected by the binding of the PET ligand to the receptors in peripheral organs and its influence to the arterial input function. The objectives of the current study were to evaluate whole body biodistribution, kinetics of [11C]PBR28 and the influence of the peripheral binding to the arterial input function in rhesus monkey.
Methods: Brain (Baseline: N=6, Blocking: N=1) and whole-body PET imaging (Baseline: N=3, Blocking: N=1) of [11C]PBR28 were performed with the measurement of metabolite-corrected arterial input function in all brain and two whole body scans including the two blocking experiments. Regional distribution volumes in brain were calculated with one- and two-tissue compartment models.
Results: The saturating doses of PBR ligands caused a marked increase of parent radiotracer in plasma (~400% increase) and brain activity (~200%) at 2 min by displacing radioligand from PBRs in peripheral organs. By dividing area under the curve of organs by that of total plasma activity, blockage was 91% and 85% in lungs and kidneys, respectively. [11C]PBR28 showed high brain uptake of 300% standardized uptake value in baseline scans (Fig.). The images showed no receptor free region that could be used for reference tissue analysis; thus, quantitation of receptor density required the measurement of parent radiotracer in plasma. [11C]PBR28 binding in brain was fit equally well with metabolite-corrected arterial input function using one- and two-tissue compartment models, suggesting that the nonspecific uptake rapidly equilibrated and was a small percentage of total tissue radioactivity (Fig.). Using the compartmental Methods:, the nondisplaceable uptake was only ~10% of total distribution volume (Table), and the distribution volume of [11C]PBR28 was stably determined with 100 min of scanning.
Conclusions: Arterial input function of [11C]PBR28 was affected by its binding to PBRs in peripheral organs. [11C]PBR28 has high levels of specific binding, which should provide sensitive measurement of changes in PBRs.
QUANTIFICATION OF PERIPHERAL BENZODIAZEPINE RECEPTORS USING A NEW ARYLOXYANILIDE BASED LIGAND [11C]PBR28 IN HUMAN
Masao Imaizumi1, Emmanuelle Briard1, Sami Zoghbi1, Jinsoo Hong1, John Musachio1, Janet Sangare1, Fumihiko Yasuno1, Tetsuya Suhara2, Victor Pike1, Robert Innis1, Masahiro Fujita1
1Molecular Imaging Branch, National Institute of Mental Health, National Institutes of Health, Bethesda, MD, USA, 2Molecular Imaging Center, National Institute of Radiological Sciences, Chiba, Japan
Background and aims: N-acetyl-N-(2-[11C]methoxybenzyl)-2-phenoxy-5-pyridinamine ([11C]PBR28) is a potential PET ligand with highly selective and specific binding to peripheral benzodiazepine receptors (PBRs). It has been reported that the density of PBR in the brain is increased in inflammatory, neurodegenerative and psychiatric disorders. Therefore, a reliable tracer method for evaluating PBRs is useful for clinical research. Our aim of this study was to evaluate the kinetics of [11C]PBR28 in human.
Methods: We performed brain PET imaging of [11C]PBR28 for nine healthy volunteers, measured the concentration of the parent radiotracer and radiometabolites in serial arterial plasma samples, and calculated regional distribution volume (V) with standard compartmental modeling. To determine necessary length to measure V accurately, kinetic analysis was performed by analyzing short portions of the acquired data ranging from 120 to 40 min.
Results: All subjects showed similar blood data. Activity of [11C]PBR28 showed fast decline, decreasing to 50% of the peak in 2.3±0.4 min and to 0.3±0.1% of the peak at 2 h. Seven subjects showed a brain activity of ~200% SUV at 6.0±2.5 min and gradual washout, decreasing to 50% of the peak in 81±21 min and 41±7% of the peak at 2 h. The levels of brain activity were expected from our monkey data considering the lower receptor density in humans than in monkeys. The kinetics was well described by two-tissue compartment model with good fitting and identifiability of V. V was 3.7±0.4 mL/cm3 in several brain regions (Table). The analysis with varying data length showed that the percentage differences of V compared to the 120 min values were less than ±10% after 80 min, indicating that 80 min is minimum required scan length to accurately measure V. Two subjects showed markedly different brain data peaking earlier at 1.7±0.0 min and rapid washout within 7 min after the peak, only the one-tissue compartmental fitting converged and gave lower distribution volumes of 0.7±0.3 mL/cm3.
Conclusions: [11C]PBR28 would be a promising for use in human. Phase I study in healthy subjects with whole-body imaging is currently on-going.
QUANTIFICATION OF 11C-SB207145-PET FOR 5-HT4 RECEPTORS IN THE HUMAN BRAIN: PRELIMINARY RESULTS
Lisbeth Marner1, Nic Gillings2, Roger N. Gunn3, Robert Comley3, William Baare4, Steen G. Hasselbalch1, Gitte Moos Knudsen1
1Neurobiological Research Unit, The Neurocenter, Rigshospitalet, Copenhagen, Denmark, 2PET and Cyclotron Unit, Deparment of Clinical Physiology and Nuclear Medicine, Rigshospitalet, Copenhagen, Denmark, 3GlaxoSmithKline, UK, 4Danish Research Center for Magnetic Resonance, Hvidovre Hospital, Copenhagen, Denmark
5-HT4 receptors are of pharmacological interest in Alzheimer's Disease because of their involvement in learning and memory. The purpose of the study is to evaluate a novel 5-HT4-receptor tracer, [11C]-SB207145, for positron emission tomography (PET). Three healthy subjects (2 females and 1 male) have been included so far in the study. Magnetic resonance imaging (MRI) was performed as well as a 2-hour PET scan after a bolus injection of [11C]-SB207145. Arterial blood samples were drawn regularly during the scan for measurements of radioactivity. A minimum of 5 samples was further processed by high-pressure liquid chromatography to measure the fraction of plasma activity representing unmetabolized parent compound. From these measures a metabolite corrected arterial input function to the brain was established. Volumes of interest (VOI's) were delineated automatically on coregistered MRI images, and partial volume corrected mean VOI values were extracted from the different time points in the PET images to establish the tissue activity curves of each VOI. Three analytical Methods: were used to establish the binding parameters using the arterial input function and the tissue activity curves. The binding potential (BP2) from the one-tissue compartment model (1-TC), two-tissue compartment model (2-TC) and simplified reference tissue model (SRTM) were compared using both the cerebellum and the occipital lobe as reference regions. The occipital lobe was chosen as a blocking study from pigs suggests a potential difference in non-specific binding in the cerebellum and the neocortex1.
In all low binding regions (cortical regions and the hippocampus) the 2-TC model was necessary to describe the tissue activity data. In the high binding regions (the basal ganglia) the 2-TC model only showed a marginally better description of the data compared to the 1-TC model. SRTM showed a 30% underestimation of the BP2 compared to the 2-TC model, which is most likely due to the slow tissue kinetics of the tracer. The underestimation is consistent with earlier findings in pigs1. When using the occipital lobe as a reference region only the striatum and the hippocampus could be investigated. This was due to the fact that the values of neocortical VOI's and the occipital lobe were similar. The striatal and hippocampal BP2 values were lower when the occipital lobe was used as a reference region compared to the cerebellum. The underestimation when using SRTM, however, was still present. Further studies are ongoing to determine the appropriate kinetic modeling approach.
RADIATION ABSORBED DOSE TO THE BASAL GANGLIA FROM PET RADIOPHARMACEUTICALS USED TO STUDY DOPAMINERGIC FUNCTION
William Robeson2, Vijay Dhawan1, Ma Yilong1, Claude Margouleff1, David Bjelke2, David Eidelberg1
1Neuroscience Department, The Feinstein Institute for Medical Research, North Shore University Hospital, New York, NY, USA, 2PET Scanner, The Feinstein Institute for Medical Research, North Shore University Hospital, New York, NY, USA
Objectives: At NSUH, we have employed three agents to study dopaminergic function. 18FDOPA and 18FPCIT dosimetry has demonstrated the urinary bladder to be the critical organ. 11C-Raclopride dosimetry has shown the small intestine (SI) as the critical organ. As these tracers accumulate in the basal ganglia (BG) with high affinity and long residence times, radiation dose to the BG may become significant. We have performed dynamic PET measurements to determine if in fact the BG although not a whole organ receives the highest doses.
Methods: Dynamic PET studies of the brain were performed in seven adult subjects (age range 28-74 years). Three subjects were administered FPCIT, two had FDOPA and two had Raclopride. Regions of interest were drawn over left and right BG structures. Resultant time/activity curves were generated and used to determine residence times for dosimetry calculations. S-factors were computed using the MIRDOSE3 nodule model for a 13.5 gram mass BG.
Results: For FPCIT, BG dose ranged from 0.03 to 0.06 mGy/MBq (Bladder: 0.06 mGy/MBq). For FDOPA, BG dose ranged from 0.009 to 0.014 mGy/MBq (Bladder: 0.16 mGy/MBq). For Raclopride, BG dose equaled 0.0035 mGy/MBq (SI: 0.026 mGy/MBq). For all subjects, BG doses are lower than the published critical organ doses except in a young subject, 28 years old, for FPCIT.
Conclusions: The basal ganglia do not appear to be dose limiting in functional studies using dopamine receptor ligands and PET. However, for young normal volunteer subjects studied with F-18 labeled dopaminergic tracers with high uptake, basal ganglia may exceed bladder dose and become the critical structure.
TOWARDS AN MRI BASED ATTENUATION CORRECTION FOR BRAIN MR-PET
Elena Rota Kops, Hans Herzog
Institute of Medicine, Forschungszentrum Juelich, Juelich, Germany
Attenuation correction in PET is usually based on measured transmission data. Future combined MRI-PET scanners will probably not allow to measure the tissue attenuation directly. Therefore, the anatomical T1-weighted images provided by the MRI may be used. In this work we report a procedure of MRI-based attenuation (MBA) correction in comparison with the original PET-based attenuation (PBA) correction. Preliminary studies [1] with the MBA correction showed that bone tissue tends to be underestimated. Thus, corresponding CT images should help as a reference in the further development of the MBA approach because of their much better delineation of bone.
A complete data set of PET, MRI, and CT measured in the same patient was used. The PET emission data, acquired with a Siemens HR+ scanner after injection of FDG and preceded by a transmission scan of 10 min, underwent PBA and were reconstructed with filtered back-projection. These PBA corrected images served as golden standard. High-resolution T1-weighted (MP-RAGE) MRI data were acquired with a Siemens 1.5-Tesla Sonata Vision scanner. The CT data were acquired with a Siemens Somatom Emotion 6. The MBA correction was derived from the MRI data as follows: After co-registration to the PET images, the MR images were segmented with the software package BrainSuite2 into (1) brain tissue and (2) other (soft) tissues, while the MPItool and the co-registered CT images were used to delineate (3) bone, (4) sinus and nasal areas, and (5) mastoid cells. Knowing the elemental compositions and specific densities of the different absorbing tissues, the attenuation coefficients (AC) valid for photons of 511 keV resulted in 0.099 cm-1 for brain tissue, 0.095 cm-1 for soft tissue, and 0.146 cm-1 for bone. Based on the detailed Hounsfield values of the CT image we could assign different AC numbers to the sinus and nasal area (0.01 cm-1) and to the mastoid cells (0.054 cm-1). Two different combinations of the segmented areas were chosen and attenuation maps were created: M1 contained all five segmented areas with the above ACs, M2 set area 4 equal to area 5 (AC=0.054 cm-1), corresponding to the best attenuation map in [1]. The MRI-based attenuation maps were forward projected yielding the attenuation correction factors (ACF) in sinogram-like files which could directly be applied for correction of the emission data. To compare the PBA and MBA corrections, regions of interest were defined at several brain areas of the PBA-PET image and copied to the MBA-PET images.
The MBA-PET images yielded similar, but slightly lower values of activity concentration for all ROIs compared to PBA-PET images. The mean relative error for M1 was about −5+/−5%, for M2 was −4+/−5%. This is an improvement compared to our preliminary Results: [1] showing an error of −10+/−3%.
The additional information on bone and cavities (areas 4 and 5) derived from CT helped to improve the MBA approach. Future work aims to obtain such information from appropriate MR images as well.
QUANTIFICATION OF CBF AND OXYGEN METABOLISM WITH 3-DIMENSIONAL PET AND OXYGEN-15: VALIDATION BY COMPARISON TO 2-DIMENSIONAL PET
Masanobu Ibaraki1, Shuichi Miura1, Shigeki Sugawara1, Eku Shimosegawa2, Tetsuro Mizuta3, Akihiro Ishikawa3, Masaharu Amano3
1Department of Radiology and Nuclear Medicine, Akita Research Institute of Brain and Blood Vessels, Akita, Japan, 2Department of Nuclear Medicine and Tracer Kinetics, Osaka University Graduate School of Medicine, Osaka, Japan, 3Medical System Division, Shimadzu Corporation, Kyoto, Japan
Introduction: Quantitative PET measurements with oxygen-15 (15O) provide absolute CBF, CBV, CMRO2, and OEF, which are used for assessment of brain pathophysiology. Absolute quantification relies on physically accurate PET measurement that is thus far achieved by 2-dimensional PET (2D-PET), which is regarded as the gold standard for measurement of CBF and oxygen metabolism. Three-dimensional PET (3D-PET), which has the great advantage of efficient detection, can reduce radiation dose; however, some difficulties arise, e.g. an increase in scatter coincidence. The present study investigated whether 3D-PET with 15O is as reliable as 2D-PET.
Methods: 2D-PET and 3D-PET measurements were obtained on the same day in eight healthy men (21 to 24 years of age). 2D-PET was performed with a Headtome-V scanner (Shimadzu Corp., Japan), which consists of BGO detectors and provides 47 sections with a 150-mm axial FOV. For 3D-PET, an Eminence-G SOPHIA scanner (PET/CT; Shimadzu Corp.) dedicated to 3D acquisition mode was used. It consists of GSO detectors and provides 59 sections with a 156-mm axial FOV. Both PET scanners were calibrated by using a cylindrical phantom. Detector normalization, attenuation correction, scatter correction (for 3D-PET only), and image reconstruction were applied similarly to the phantom and the subjects. A hybrid scatter correction method was used in 3D-PET study. This method is based on acquisition in dual energy window combined with a convolution-subtraction method in upper energy widow. Each PET study included three sequential PET scans: 4-min PET scanning that started 3min after continuous inhalation of [15O]-CO gas, 3-min PET scanning that started after slow bolus inhalation of [15O]-O2 gas, and 3-min PET scanning that started after slow bolus infusion of [15O]-H2O. The inhaled (or injected) dose for 3D-PET study was about one-fourth that for 2D-PET study. Arterial input functions obtained with a beta detector system were used to calculate CBF, CBV, CMRO2, and OEF maps.
Results: and Discussion: In 2D-PET study, average CBF, CBV, CMRO2, and OEF for neocortical regions (frontal, temporal, parietal, and occipital lobes) were 54+−13 (mL/100 mL/min), 3.7+−0.5 (mL/100 mL), 3.7+−0.6 (mL/100 mL/min), and 0.37+−0.06, respectively, in agreement with previous PET studies. In 3D-PET study, scatter correction strongly affected the Results:. Without scatter correction, average values were 42+−6 (mL/100 mL/min), 4.7+−0.6 (mL/100 mL), 3.1+−0.4 (mL/100 mL/min), and 0.39+−0.06, respectively. With the exception of OEF, values differed between 2D-PET and 3D-PET. However, average values scatter-corrected by the hybrid method were comparable to those of 2D-PET: 55+−11 (mL/100 mL/min), 3.7+−0.5 (mL/100 mL), 3.9+−0.8 (mL/100 mL/min), and 0.37+−0.06, respectively. We confirmed that, in addition to the absolute values, scatter-corrected 3D-PET yields image contrast and image quality similar to that of 2D-PET. Our Results: showed that scatter correction is crucial in quantitative 3D-PET study and that appropriate scatter correction makes it possible to perform 15O PET study with a lower radiation dose and with the same degree of accuracy as 2D-PET.
APPLICATION OF HARDWARE-BASED MULTIMODAL REGISTRATION SYSTEM TO FUSION OF PET AND MRI IMAGES
Kazuhiro Koshino1, Hiroshi Watabe1, Akihide Yamamoto1, Yasuyuki Ose1, Masaaki Hikake1, Noboru Teramoto1, Hiroshi Sato2, Hidehiro Iida1
1Department of Investigative Radiology, National Cardiovascular Center Research Institute, Suita, Osaka, Japan, 2Laboratory for Diagnostic Solution, Advanced Medical Engineering Center, National Cardiovascular Center Research Institute, Suita, Osaka, Japan
Background and aims: Fusion of PET and MRI images provided complementary information, in addition to physiological information from PET and anatomical information from MRI. Several software-based registration techniques have been proposed. Those techniques tended to depend on 1) distributions of tracers in PET images, in which some distributions did not correspond with anatomical information in MRI image, 2) statistical noises and geometric distortions in images, 3) spatial resolution of images. We have developed a hardware-based registration system, which was independent of modalities, kinds of tracers, qualities and spatial resolutions of images. We applied the system to registration between PET and MRI images of a rat.
Methods: To measure positions of a subject in PET or MRI scanners, we used an optical motion tracking system, POLARIS (Northern Digital Inc., Canada). Reference targets (RTs) were attached on gantries of both scanners. Positions and orientations of the subject in field of views of both scanners were given as relative positions from corresponding RTs. Spatial relation between coordinates of the scanner and the RT could be obtained by position calibration. A transmission scan on a rat was performed with a PET scanner (microPET focus120, Siemens, USA) and a T1-weighted scan on the rat was done using an MRI scanner (Signa Horizon LX 3T, GE, USA). The rat was anaesthetized and was fixed with a fixture for avoiding head movement if recovered from anesthesia. Another target for monitoring positions of the rat was attached on the fixture. We assumed the rigid body model between positions of a head of the rat and the target on the fixture. A transmission image was transformed into the MRI scanner coordinate using position calibration data and measured positions of the rat in both scanners.
Results: A fused axial image of the PET and MRI images was shown in a figure. Registration errors for translations were 1.6, 1.8 and 6.2mm for horizontal and vertical directions in axial plane and transaxial direction, respectively. No remarkable error for rotational angles was observed.
Conclusions: Our registration system aligned PET and MRI images except for error in transaxial direction. It was shown that our system would be applicable to multimodal registration of small animals.
IMPACT OF KINETIC MODEL DESIGN ON THE ANALYSIS OF DYNAMIC [15O]O2 STUDIES
Ronald Boellaard1, Jochem P. Bremmer2, Bart N.M. van Berckel1, Abraham Rijbroek3, Karin Klijn2, Jaap Kappelle2, Adriaan A. Lammertsma1
1Department of Nuclear Medicine and PET Research, VU University Medical Center, Amsterdam, The Netherlands, 2Department of Neurology, Rudolf Magnus Institute of Neuroscience, University Medical Center Utrecht, Utrecht, The Netherlands, 3Department of Surgery, VU University Medical Center, Amsterdam, The Netherlands
Introduction: Various kinetic models have been proposed for calculating oxygen extraction fraction (OEF) and consumption (CMRO2) assuming that, in tissue, oxygen is instantaneously metabolised into water [1],[2] or allowing for different washout rates for oxygen and water [3],[4]. The purpose of this study was to evaluate performance of different kinetic models for analysing dynamic [15O]O2 studies.
Methods: Eleven studies consisting of a 3 min [15O]CO scan and 10 min dynamic [15O]H2O and [15O]O2 (bolus inhalation) scans were performed. Scans were acquired in 3D mode using an HR+ PET scanner. On-line continuous arterial blood sampling was performed to derive an input function. Data were analysed using models specified in Table 1. Type I models used the assumption of instantaneous metabolism of oxygen [1],[2]. In its simplest form CBV was fixed to the value obtained from the CO scan, and CBF and Vd to those from the H2O scan. Variations of this model used CBV, CBF and/or Vd as fit parameters. Model II consisted of two parallel single tissue compartments, allowing for different washout rates for oxygen and water [3],[4]. Model III contained a two-tissue compartment model for oxygen and a single tissue compartment model for water. Model performance was evaluated by goodness of fit using weighted residual squared errors (WRSE) and Akaike criterion (AIC), and by regression analysis of OEF and CMRO2.
Results: and discussion: According to both WRSE and AIC, models II and III provided ‘better’ fits than type I models. For type I models use of Vd as fit parameter provided better fits than reusing it from the [15O]H2O scan, consistent with preference for models II and III that use separate washout rates for oxygen. Use of Vb as fit parameter provided more accurate fits than use of CBV obtained from [15O]CO scans. Regression analysis showed a fair to good correlation of OEF (coefficient = 0.67 to 0.73) and CMRO2 (coefficient=0.77 to 0.80) for most models compared with model 1A, i.e. the model that has been used routinely in the past. Models allowing for different oxygen washout rates, however, provided on average 5 to 20% lower OEF and CMRO2.
Conclusions: Models II and III, which are not based on the assumption of instantaneous metabolism of oxygen into water in tissue, provide more accurate fits of dynamic [15O]O2 studies than conventional (type I) models, which is consistent with experimental findings of Seki et al. [3]. OEF and CMRO2 values may, on average, differ up to 20%, depending on the actual model being used.
EVALUATION OF COMPARTMENT MODELS AND SEMI-QUANTITATIVE MEASURES FOR ANALYSING [18F]FDDNP STUDIES
Maqsood Yaqub1, Ronald Boellaard1, Bart N.M. Van Berckel1, Nelleke Tolboom1,2, Anke A. Dijkstra1, Albert D. Windhorst1, Gert Luurtsema1, Philip Scheltens2, Adriaan A. Lammertsma1
1Department of Nuclear Medicine and PET Research, VU University Medical Centre, Amsterdam, The Netherlands, 2Neurology and Alzheimer Centre, VU University Medical Centre, Amsterdam, The Netherlands
Introduction: [18F]FDDNP is a PET ligand that has been introduced for imaging neurofibrillary tangles and beta-amyloid fibrils in the brain. Increased cerebral uptake of [18F]FDDNP in patients with Alzheimer's disease (AD) has been described[1]. The purpose of the present study was to evaluate quantification of [18F]FDDNP binding using standard compartment models[2]. In addition, potential semi-quantitative measures for routine clinical use were investigated.
Methods: The following kinetic and semi-quantitative Methods: were assessed: plasma input single tissue (1T-2k), two tissue irreversible (2T-3k) and two tissue reversible (2T-4k) compartment models[2], simplified (SRTM) and full (FRTM) reference tissue models[3], standard uptake values (SUV) and SUV ratios (SUVr). Both simulations and clinical studies were used to assess the effects of binding potential (BP=0.1–0..4), flow and fractional blood volume (Vb=0.025–0..075) on precision and accuracy of estimated parameters. Various [18F]FDDNP time activity curves (TAC) were simulated at a 15% noise level using either reference tissue (RI) or plasma input (PI). In addition, [18F]FDDNP studies were performed in 4 control and 3 AD subjects. Methods: were compared using several frontal cortex TACs (typical AD regions). Grey matter cerebellum was used as reference tissue.
Results: Simulations showed best BP accuracies for SRTM (<5% bias), with lower bias for higher BP. For the 2T-4k model, bias reduced (to <29%) with decreasing flow (R1=0.6–0..9), but was still best for SRTM (2.5% bias) and FRTM (5% bias). Effects of Vb on BP accuracy were small for 2T-4k (<4%), SRTM (<10%) and FRTM (<10%), although occasionally SRTM and FRTM showed outliers.
According to the Akaike information criterion (AIC)[4], the 2T-4k model (72%) was preferred over 1T-2k (0%) and 2T-3k (28%) models in healthy subjects. This preference was even 100% in AD subjects. SRTM was preferred (72%) over FRTM in both control and AD subjects. Good correlations were found between BP-SRTM and BP-FRTM (R2=0.92) or SUVr over 60-90 min (R2=0.82). In this limited series of subjects, none of the Methods: provided a significant contrast between AD and healthy subjects, although 2T-4k BP showed a trend (p<0.1) for decreased binding in AD and 2T-3k a trend (p<0.1) for increased Ki in AD. The 2T-4k BP decrease might be explained by an observed increase of k4, possibly caused by metabolites entering the brain. SRTM did not show a significant difference as well. Simulations, however, indicated that reference tissue models are sensitive for changes in k4 between target and reference region.
Conclusions: AIC indicates a preference for 2T-4k for plasma input and SRTM for reference tissue models. This quantification study will be extended to assess the impact of metabolites entering the brain (as potential cause of change in k4) on pharmacokinetic analysis.
SERIAL CHANGES IN HEMODYNAMICS IN ACUTE ISCHEMIC STROKE: CLINICAL APPLICATION OF RAPID DUAL AUTORADIOGRAPHIC METHOD PET
Tomoyuki Ohara1, Chiaki Yokota1, Takuya Hayashi2, Masaki Naganuma1, Junji Kasuya1, Toshiyuki Uehara1, Hiroshi Moriwaki1, Kazunori Toyoda1, Hiroshi Watabe2, Hidehiro Iida2, Kazuo Minematsu1
1Cerebrovascular Division, Department of Medicine, National Cardiovascular Center, Suita, Osaka, Japan, 2Department of Investigative Radiology, Advanced Medical and Engineering Center, National Cardiovascular Center, Research Institute, Suita, Osaka, Japan
Introduction
Identification of ischemic penumbra in acute ischemic stroke is essential to acute therapeutic interventions. Positron emission tomography (PET) is the only technique that provides accurate, quantitative measurement of cerebral blood flow and metabolism in human. However, PET with the conventional 3-step technique requires relatively long time for the examination. Since October 2006, we have been ready to perform PET with a faster technique (a rapid dual autoradiographic method PET, rapid DARG-PET) 1) in patients with acute ischemic stroke within 24 hours after the onset. Total examination time is approximately 25 minutes in the rapid DARG-PET. We examined serial changes in a patient with severe carotid artery stenosis by use of the rapid DARG-PET.
Case Presentation
A 60-year-old man was admitted for mild dysarthria and right hemiparesis. Diffusion-weighted magnetic resonance imaging (DWI) on admission showed high signals in the territory of the left middle cerebral artery (MCA). A rapid DARG-PET, performed 12 hours after the onset, showed slight reduction in cerebral metabolic rate of oxygen (CMRO2) and cerebral blood flow (CBF) in the area corresponding to the high signal lesion on DWI. It disclosed an increase in cerebral blood volume (CBV) without an increase in oxygen extraction fraction (OEF) in the region surrounding the DWI-high signal area. Angiogram on the second hospital-day revealed pseudo occlusion of the left internal carotid artery (ICA) with a main collateral flow through the anterior communicating artery from the right ICA to the left MCA. The second rapid DARG-PET examination performed two weeks later, revealed normalization of increased CBV. It was hypothesized that, in an acute stroke, blood flow supply associated with the collateral circulation is dynamically changing in the area of ischemia, and thus quantitative assessment of ischemic status is important for the patient care.
Conclusions
The rapid DARG-PET is able to identify transient changes in cerebral blood flow and oxidative metabolism in patients with acute ischemic stroke. This technique will be useful for further investigation of pathophysiology and hemodynamic status of stroke in man.
A PARTIAL VOLUME CORRECTION-BASED APPROACH FOR DETECTION OF RECEPTOR/TRANSPORTER DISTRIBUTION IN THE HUMAN BRAIN: INITIAL APPLICATION TO [11C]RACLOPRIDE PET SCANS
Hiroto Kuwabara, Anil Kumar, Olivier Rousset, Weiguo Ye, James Brasic, Mohab Alexander, Dean F. Wong
Department of Radiology, Johns Hopkins University, Baltimore, MD, USA
Background and aims: Voxel-wise identification of receptor/transporter distributions in the brain with PET may be confounded for regions of relatively low densities or when only suboptimal ligands are available for the site. We tested whether a new partial volume correction (PVC)-based approach identifies regions of relatively low densities of dopamine D2/D3 receptors with [11C]raclopride, a relatively low affinity ligand, compared to published Results: using higher affinity D2/D3 radioligands or autoradiography.
Methods: Twenty-one normal subjects (age: 31 ± 8) underwent one 90-min dynamic [11C]raclopride scan. Binding potential (BP) maps were constructed by the bolus-plus-infusion transformation (BPIT; Kuwabara et al., 2002). For PVC, tissue segments were identified on MRI, transferred to PET space, and smoothed by PET camera specific Gaussian kernels to reproduce the smearing effect. The resulting geometric transformation matrix (M, Rousset et al., 1998), is an n by m matrix where n and m are numbers of voxels to perform PVC and tissue segments, respectively. PV-corrected BP values of tissue segments (T, an m by 1 matrix) were obtained by left matrix division (M\T) using Gaussian elimination. The PVC-predicted BP map was constructed by inserting the M*T product to the PVC voxels and −1 to the remaining voxels. Manually defined VOIs for putamen, caudate nucleus, and cerebellum, and automatically defined (using SPM2) white matter, gray matter, and cerebrospinal space segments (m=8) were used for the initial iteration. PVC-predicted BP maps represent weighted mean BP values of the segments at the voxel level assuming a uniform BP value for each segment. Comparison of observed BP maps to PVC-predicted BP maps, using SPM2, allows exploration of regions of significant deviation from the homogeneity assumption. New segments that correspond to regions of significant deviation were included for the second iteration as described below.
Results: The first iteration of SPM analysis identified clusters in posterior putamen (peak t=33.95 at [-28,-10,6]), ventral striatum (t=-11.64 at [20,8,0]), thalamus (t=25.23 at [4,-14,8]), globus pallidus externa (t=19.66 at [20,-8,8]), internal capsule between putamen and caudate nucleus (t=13.94 at [-22,14,14]), and rectus gyrus (t=13.32 at [20,22,-12]) bilaterally at a p<0.05 level, corrected for multiple comparisons and setting the minimal cluster volume to 1 ml. Inclusion of above clusters as separate tissue segments in a second iteration of PVC (m=20) revealed no clusters in SPM analysis at the same criteria. The identified regions generally agreed with those reported using high affinity ligand [18F]fallypride (e.g., Mukherjee et al., 2002) except for amygdala. The globus pallidus externa and internal capsule clusters, which were not reported with [18F]fallypride, are supported by in vivo autoradiography (e.g., Tupla et al., 2003).
Conclusions: The proposed approach successfully identified extra striatal region with moderate dopamine D2/D3 receptor density with [11C]raclopride PET that agreed with published Results: using higher affinity ligands and in vivo materials. Although the approach appears promising, further validation will be required to apply this method to receptor and transporter studies with diffuse distributions, as well as to identify their changes in clinical populations.
Grant support: AA12839, MH078175, and K24 DA00412 (DFW)
TRACER KINETIC DATA ANALYSIS BY EXPECTATION-MAXIMIZATION OPTIMIZATION OF THE TISSUE IMPULSE RESPONSE FUNCTION: THEORY AND APPLICATIONS
James E. Holden1, John L. Graner1, Vesna Sossi2
1Department of Medical Physics, University of Wisconsin-Madison, Madison, WI, USA, 2Department of Physics, University of British Columbia, Vancouver, BC, Canada
We have created a method for the analysis of tracer kinetic data that depends on a single assumption, that the uptake of tracer from arterial plasma into tissue can be represented as a linear system. The response of a tissue to the delivery of tracer with arterial concentration time course Cp(t) thus is expressed as the convolution of that input time course with the impulse response function I(t) that completely characterizes the response of the tissue under investigation to that tracer. In our method the I(t) is estimated using expectation-maximization (E-M), with the optimization condition that convolution of the estimated I(t) with the measured Cp(t) yield the measured time course in the target tissue, and with I(t) constrained to be a monotonically decreasing function of time. Although this is not the maximum likelihood solution, the method yields the same suppression of noise as seen in E-M image reconstruction. The integral over all time of I(t) is the distribution volume (DV) of that tracer in that tissue. This leads to the impulse-response (I-R) Logan plot, shown in the equation below. The integral under the input function apparently missing from the first term on the right side is in fact present, given that the integral under an impulse is equal to one for all times greater than zero. When I(t) is reduced to a single decaying exponential function at later times post-injection, the second term on the right becomes constant in time, thus allowing the estimation of DV by graphical Methods:.
If the plasma time course is replaced by the time course in a reference tissue, the E-M optimization yields the reference-input I-R function. The integral over all time is then the distribution volume ratio (DVR), which can thus be estimated from the reference-input I-R Logan plot.
If I(t) becomes constant in time before the end of the study, the value of that constant provides an estimate of the rate constant Ki for the irreversible trapping of tracer from plasma.
Implementation of both the impulse-response and conventional Logan Methods: on the same data using a sliding window for the linear regression fitting range showed that the I-R method reaches a final and more stable value earlier in the study than does the conventional method for several reversible tracers.
Finally, the I-R functions provide powerful insight into the approach to steady state of the various components of tracer in tissue with each other and with tracer in plasma. For 11C-labeled methylphenidate, the integral under the reference-input I-R function (DVR(t)) often attains a constant value by 30 minutes post-injection, whereas the integral under the plasma-input I-R function (DV(t)) is still increasing at the end of a 60-minute study.
PERFORMANCE EVALUATION OF VARIOUS PARAMETRIC METHODS: FOR [11C]PIB
Maqsood Yaqub1, BNM Van Berckel1, Nelleke Tolboom1,2, Anke A. Dijkstra1, Albert D. Windhorst1, Gert Luurtsema1, Philip Scheltens2, Adriaan A. Lammertsma1, Roland Boellaard1
1Department of Nuclear Medicine and PET Research, 2Neurology and Alzheimer Centre, VU University Medical Centre, Amsterdam, The Netherlands
Introduction: [11C]PIB is a ligand for imaging amyloid fibrils in brain. Price et al.[1] and Lopresti et al.[2] have evaluated Methods: for quantification of [11C]PIB studies, including performance of various parametric Methods:. The aim of the present study was to further investigate performance of parametric Methods: in which the reference tissue efflux rate (k2′) is fixed to improve quantification. Performance of parametric Methods: was evaluated using simulations and clinical data.
Methods: The following parametric Methods: were studied: RPM1[3], Reference Logan[4], MRTMo and MRTM[5]. In addition, RPM2[6], MRTM2[5] and two newly developed multi-linear Methods: (MRTM3 and MRTM4), with fixed k2′, were included. Both simulations and clinical data were used to determine effects of flow (R1=0.6–0..9), fractional blood volume (Vb = 0.025–0..075) and binding potential (BP=0.2–3.) on precision and accuracy of parametric BP and DVR. For simulations various [11C]PIB time activity curves (TACs) were generated at a 15% noise level using both reference tissue input (RI) and plasma input (PI). In addition, test-retest studies (two scans with arterial blood sampling on one day) were performed in 6 control and 5 AD subjects. TACs were derived from 8 regions of interest in the frontal cortex. Grey matter cerebellum was used as reference tissue. BP obtained with SRTM and DVR obtained with two-tissue reversible plasma-input model (2T-4k DVR) were used as reference.
Results: RI simulations indicated that differences in bias between the various Methods: reduced with increasing BP (>0.2). Overall, lowest BP biases were observed for MRTM2 (<4.3±26%) and RPM2 (<7.4±24%), i.e. when k2′ was fixed. Simulations at different flow settings (R1=0.6–0..9) indicated lowest BP biases for MRTM2 (<3±12%), RPM1 (<1±23%) and RPM2 (<4±11%). PI simulations indicated a similar performance compared with RI simulations. Increasing Vb increases bias for all Methods: For clinical data all Methods: showed good correlation with SRTM (R2=0.89–95.). The correlation coefficient (R2) for RPM1 was 0.95 and it was 0.92 for MRTM, MRTM2 and MRTM3. Both simulations and clinical data indicated a larger bias (about −15%) for reference Logan, MRTMo, MRTM3, MRTM4 than for MRTM, RPM1, RPM2 and MRTM2 (bias <5%) compared with both BP-SRTM and/or 2T-4k DVR. The difference in average BP between AD and control subjects was (P<0.001), however, slightly smaller for RPM1, MRTM2 and RPM2 than for the other Methods:. Finally, best DVR test-retest variability was found for MRTM3 (2.7%), but was slightly worse for other Methods: (max 4.1%).
Conclusions: MRTM, RPM1, RPM2 and MRTM2 were the most accurate parametric Methods: for [11C]PIB studies. Fixing k2′ had a small beneficial effect on the performance of these Methods:. Visual inspection of parametric images also revealed a small improvement of image quality using k2′ fixed, consistent with simulations. RPM2 is the recommended method for [11C]PIB.
IMPROVED GLLS METHOD FOR PARAMETER ESTIMATION WITH A PRIOR DISTRIBUTION VOLUME AND A NEW GRAPHICAL PLOT
Lingfeng Wen1,2, Stefan Eberl1,2, Dagan Feng1,3
1School of Information Technologies, University of Sydney, Sydney, NSW, Australia, 2Department of PET and Nuclear Medicine, Royal Prince Alfred Hospital, Sydney, NSW, Australia, 3Department of Electronic and Information Engineering, Hong Kong Polytechnic University, Kowloon, Hong Kong-S.A.R., China
Background and aims
Parametric images, derived from voxel-by-voxel parameter estimates, can demonstrate physiological function without the prior knowledge required to define regions of interest (ROIs). The generalized linear least square (GLLS) method has been proposed to address the biased parameter estimates of the linear least square method while retaining its computational efficiency required for parametric image generation. GLLS has been successfully applied to ROI-derived tissue time activity curves (TTAC) in brain, myocardial and liver PET studies. However, in parametric image generation, the high level of random noise in voxel-wise TTACs potentially leads to unsuccessful fits with GLLS. In this work, an improved GLLS approach is proposed to enhance estimation reliability by incorporating a new graphical plot and a prior distribution volume (Vd) estimate.
Methods
The improved GLLS consists of four steps: 1) estimating K1 and k2 by GLLS for 3-compartment and 4-parameter model; 2) obtaining prior Vd as detailed below; 3) deriving k3 from the slope of a new graphical plot; 4) estimation of k4 by k4 = K1*k3/(Vd*k2-K1). Two approaches were used to derive prior Vd: 1) Logan plot (Logan-Vd-aided GLLS) and 2) GLLS fitting for a 2-compartment model (2comp-Vd-aided GLLS). Curve fitting was performed using the original and proposed GLLS Methods: for 1000 curves with FDG kinetics at ten levels of Gaussian distributed noise. Estimates with negative constants, indicating unsuccessful fits, were set to zero. The percentage bias and coefficient of variation (CV) for Ki (K1*k3/(k2+k3)) were derived by comparing the mean and standard deviation with the reference values for each noise level.
Results
As shown in the figures of percentage bias and CV versus noise level, all three studied Methods: suffered from increasing bias and poorer reliability with increasing noise level. The Logan-Vd-aided GLLS achieved less bias than the original GLLS with reduced number of unsuccessful fits. Relatively lower Vd derived from two-compartment model, gave rise to overestimation of Ki. The two proposed improved GLLS Methods: demonstrated lower CVs than the original GLLS.
Conclusion
The proposed improved GLLS Methods: clearly demonstrated improvements in reliability and bias over the original GLLS, opening the way to successfully using GLLS for parametric image generation. Further investigations are warranted such as regressive formation of Vd from GLLS fitting and evaluation with clinical data.
QUANTIFICATION OF NEURORECEPTORS WITH IRREVERSIBLE BINDING RADIOLIGANDS AND MULTIPLE LINEAR ANALYSIS
Su Jin Kim1, Jae Sung Lee1, Yu Kyeong Kim1, James Frost2, Gary Wand3, Mary E. McCul3, Dong Soo Lee1
1Department of Nuclear Medicine, Seoul National University College of Medicine, Seoul, South Korea, 2Department of Diagnostic Radiology, Yale University School of Medicine, New Haven, CT, USA, 3Department of Psychiatry and Medicine, Johns Hopkins University School of Medicine, Baltimore, MD, USA
Objectives: A novel method for the quantification of net accumulation (Ki) of radioligands in an irreversible compartment (MLAIR: Multiple Linear Analysis for Irreversible Radiotracer) was developed and compared to the conventional Patlak graphical analysis (PGA).
Methods: Multiple linear regression analysis is performed on the linearized differential equation of irreversible 2-tissue compartment model in which the Ki is a macro parameter of the least squares solution. Computer simulations were performed to assess the statistical reliability of the parameter estimation using this method. Dynamic PET data acquired with several irreversible binding tracers (i.e. N1′-([11C]methyl)naltrindole for δ-opioid receptor density and 18F for bone metabolism) were analyzed and parametric images of Ki were compared to demonstrate the superior properties of this method over the PGA.
Results: In computer simulation, the MLAIR showed less variability and errors in Ki estimation for various levels of noise and specific binding compared to the PGA. For real PET data, MLAIR estimates showed good correlation with PGA and NLS (r > 0.9). The quality of parametric images was also considerably improved using MLAIR method in comparison with the PGA.
Conclusion: This new method will be useful in the quantitative analysis of PET and SPECT obtained with irreversible binding radiotracers.
Figure 1. Ki parametric images generated using PGA (A) and MLAIR (B)
COMPARISON OF DIRECT AND INDIRECT PARAMETRIC ESTIMATION METHODS
Charalampos Tsoumpas1,2, Federico Turkheimer2,3, Kris Thielemans1,2
1Hammersmith Imanet Limited, Part of GE Healthcare, 2Clinical Sciences Division, Faculty of Medicine, 3Neuroscience Division, Faculty of Medicine, Imperial College London, London, UK
Introduction
We evaluate a methodology to directly reconstruct parametric images in 3D PET. The reconstruction algorithm models the physics, the statistics and the kinetics of the measured counts. The whole implementation is built using the STIR library.
Methodology
Both real and simulated data for FDG brain studies will be presented. The measurements were performed with an ECAT HR+ in frame histogram mode. During the human subject study the radioactivity in blood was monitored. Both the radioactivity in the blood and the scanner counts were cross-calibrated such that they are in the same radioactivity units.
i. Indirect parametric image estimation
The data was reconstructed using filtered back projection (FBP) and ordered subsets expectation maximization (OSEM). The model parameters were estimated using the method proposed by Patlak et al. (i.e. linear regression) which relates the plasma counts with the last six frames obtained after their reconstruction. The slope and the intercept of the regression represent the “FDG Uptake” and the “Free FDG Fraction”, respectively.
ii. Direct reconstruction
Instead of applying the reconstruction and then the modelling, this algorithm has a similar concept to the PIRA algorithm to the PIRA algorithm of Matthews J et al (PMB 1997). It fits the parametric image directly to the projection data, thus we called it POSEM.
Results
Sample parametric images from real data are shown in the following figures. The noise reduction with POSEM is transparent, especially for the “Free FDG Fraction”. Further investigation on its convergence and statistical properties (e.g. bias, variance, covariance etc.) will be presented for real and simulated studies.
DISCREPANCY BETWEEN IN VIVO AND IN VITRO IOMAZENIL BINDINGS IN A RAT MODEL OF FOCAL CEREBRAL ISCHEMIA
Yoshihiro Tohyama1,2,3, Kazuhiho Sako4, Toshihide Sugimura5, Joji Nakagawara2, Yuji Kuge3, Nagara Tamaki3
1Kotoni Royal Hospital, Sapporo, Hokkaido, Japan, 2Department of Neurosurgery, Nakamura Memorial Hospital, Sapporo, Hokkaido, Japan, 3Department of Nuclear Medicine, Hokkaido University School of Medicine, Sapporo, Hokkaido, Japan, 4Department of Neurosurgery, Nayoro City Hospital, Nayoro, Hokkaido, Japan, 5Department of Neurosurgery, Abashiri Neurosurgical Hospital, Abashiri, Hokkaido, Japan
[Background] The binding sites of benzodiazepine (BZ) are rich distributed widely throughout the central nervous system on a variety of cell types associated with GABA(A) receptors. The measurement of BZ bindings in the ischemic brain could possibly be a good indicator for the viability after ischemic insult. We evaluated the relationship between iomazenil (IMZ), BZ receptor antagonist, binding and cerebral blood flow (CBF) in focal ischemic rat brain. [Method] Male Wistar rats with 3, 24 or 72 hours after occlusion of middle cerebral artery (MCA) and ipsilateral common carotid artery (CCA) were used. (1) In vivo binding: Three hours after the injection of 125I-IMZ, 123I-N-isopropyl-p-iodoamphetamine (IMP) was administered for CBF. (2) In vitro binding: CBF was measured using 14C-iodoantipyrine (IAP) autoradiographic method. The frozen sections, adjacent to those for local CBF, were incubated in the buffer solution with 125I-IMZ for in vitro IMZ binding. [Results:] In vivo study: The decreasing of IMZ binding was observed in the severe ischemic cortex. In vitro study: The binding of IMZ increased in the moderate and severe ischemic area in the rat brains with 3, 24 and even 72 hours ischemia. The Results: obtained by in vivo and in vitro were contradictory. [Conclusion:] IMZ in vitro binding increases in rat cortex even with irreversible histological ischemic changing by MCA and CCA occlusions. IMZ binding dose not accurately reflect the viability of neuron in the early stage of cerebral ischemia due to the elevated binding of IMZ by ischemic insult. It is hard to be detected this increasing of IMZ binding in the measurement of distribution of IMZ in vivo. In vivo IMZ binding in ischemic brain should be analyzed in consideration of ligands delivery and the matter of equilibrium.
CONSTRUCTION OF A HUMAN MR-PET SCANNER AT 9.4T CAPABLE OF SIMULTANEOUS DATA ACQUISITION
N. Jon Shah
Institute of Medicine, Research Centre Juelich, Juelich, Germany
A stand-alone 9.4T MRI machine has a significant number of advantages for neuroscientific research and the MR developments that underpin the methodology. These include the following: higher signal-to-noise ratios; higher spectral dispersion ? leading to better resolved spectra; a stronger BOLD effect which can be utilised in fMRI; and a platform for future technological developments.
Notwithstanding its ubiquitous success and because of its non-invasive character, MRI is not as sensitive as PET. Two of the major advantages of PET are its sensitivity and its ability to deliver metabolic information.
A combination of the two methodologies offers a number of unique possibilities, not only for technological developments, but also for the application of these developments. Scientific and technological novelty is contained in the following: development of new Methods: for MR-PET at 9.4T; molecular imaging employing simultaneous MRI and PET; imaging of novel, dualpurpose contrast agents; development of PET detectors for use in magnetic fields; and the development and testing of components for ultra high-field MRI.
The scientific rationale, the planning, the required infrastructure, and potential scientific projects for such an instrument will be presented and discussed. Results: from a combined 3T MR-PET scanner, from which many components for the 9.4T MR-PET scanner will be taken, will be presented as a basis for discussion.
A SIMPLIFIED PROCEDURE FOR MEASURING THE REGIONAL CEREBRAL METABOLIC RATE OF GLUCOSE IN MICE USING FDG-PET IMAGING
Hsiao-Ming Wu, Hong-Dun Lin
Department of Molecular and Medical Pharmacology, David Geffen School of Medicine, University of California, Los Angeles, CA, USA
Background and aims: To date, autoradiography remains the major tool for measuring the regional cerebral metabolic rate of glucose (rCMRglc) in mice. It has the advantage of better image resolution but the measurement was limited to end-point. Small animal FDG-PET imaging can be a better tool in providing rCMRglc in vivo if the traditional quantification method is simplified. In this study, we generated a mouse brain FDG template with region-of-interests (ROIs) for different brain regions. We simplified the measurement of rCMRglc by generating the parametric images of mouse brain and normalizing the images to the template. We compared the values of rCMRglc obtained from this simplified method with those obtained from the traditional manually-drawing-ROI method. Methods: Three C57/BL6 mice (21-25g) were studied by PET. Anesthetized with 1.5% isoflurane, the mouse was scanned for 30 minutes. 14 blood samples (0.2 µ? each) were taken from the femoral artery by a microfluidic blood sampling device. The FDG in plasma were estimated from the blood samples by using an equation reported previously (Wu et al., 2005). The parametric images of the glucose metabolic rate (µmol/min/100g) were generated by applying the Patlak analysis to the 3-22 minutes PET data. The FDG template of the mouse brain was generated by using the head images of the three mice based on the multiresolution 3D optical flow estimation method (OFEM). Six ROIs (2 frontal, 2 parietal and 2 occipital cortex ROIs) were determined on the template. Two sets of rCMRglc values, 18 values for each set, were calculated from the 3 mice images. One set used the normalized parametric images and the ROIs on the template; the other set used the 6 corresponding ROIs that were manually-drawn on the individual non-normalized parametric images. Results: Three sets of parametric images were successfully normalized to the FDG brain template. The brain template has lower noise and better resolution than the individual mouse image. There is no significant (P > 0.5 by using paired student's t test; Pearson correlation=0.88) difference between the two sets of rCMRglc values (18.8 ± 3.7 vs. 19.3 ± 6.2 µmol/min/100g, respectively). The rCMRglc values were comparable to those reported in the literature. Conclusions: We simplify the quantitative analysis of mouse brain FDG-PET imaging by generating the template of brain structures. The rCMRglc can be calculated reliably from FDG-PET images without drawing ROIs and performing model fitting. More studies are underway to improve the statistics in brain template so that more ROIs can be determined. The work is supported by UC bio05-10510 and DOE DE-FC03-02ER63420 grants.
BRAIN AND WHOLE BODY IMAGING OF PHOSPHODIESTERASE 4 USING (R)-[C-11]ROLIPRAM IN HUMAN AND RHESUS MONKEY
Masahiro Fujita, Yong Ryu, Sami Zoghbi, Jinsoo Hong, Robert Gladding, Victor Pike, Robert Innis
Molecular Imaging Branch, National Institute of Mental Health, Bethesda, MD, USA
Introduction: Rolipram binds to phosphodiesterase 4 (PDE4), the enzyme that catabolizes the second messenger cAMP. cAMP cascade is a major signal transduction pathway in neurons and inflammatory cells, and inhibitors of PDE4 are expected to have therapeutic effects in mood disorders and inflammatory diseases. (R)-[C-11]rolipram has been successfully used to image brain PDE4 in rat (1), non-human primate (3), and human (2). The purposes of this study were to quantify (R)-[C-11]rolipram binding in rhesus monkey and human brain, examine the test retest reproducibility of the measurement, and to study whole body distribution of radioactivity and radiation-absorbed doses.
Methods: Following (R)-[C-11]rolipram scans were performed. Monkey brain: n = 2, human brain test retest: n = 9, monkey whole body baseline: n = 3, monkey whole body with co-injection of non-radiolabeled (R)-rolipram (1 mg/kg): n = 1. HRRT and GE Advance cameras were used for 2 h brain and whole body scans, respectively. The injection activity was ~185, ~370, and ~740 MBq in monkey, human brain, and human whole body scans, respectively. Arterial blood samples were obtained in all monkey and human brain scans to measure metabolite-corrected arterial input function. Plasma free fraction (f1) was also measured in human brain scans. Distribution volume (V) in brain was calculated by one- (1C) and two-tissue compartment (2C) models. Radiation-absorbed doses were estimated using the MIRD scheme.
Results: As expected from postmortem studies, both monkey and human brain scans showed widespread distribution of (R)-[C-11]rolipram binding. Peak brain activity was 150–300% standard uptake value with higher levels in monkeys than in humans. Fraction of (R)-[C-11]rolipram in arterial plasma decreased to 50% in 10 and 75 min in monkeys and humans, respectively. In both of these species, 2C showed significantly better goodness-of-fit than 1C. 2C well identified V with ~5% COV and showed fairly uniform distribution in cortical and subcortical areas with average of 4.9 and 0.66 mL/cc in monkeys and humans, respectively. In humans, average V/f1 was 10 mL/cc. Test retest reproducibility of V/f1 was ~15% in eight subjects while one subject who reported job-related stress at the time of one scan showed big variability of ~65%. In whole body scans, brain, heart contents, lungs, liver, kidneys, and bladder were visible. The monkey blocking scan did not show apparent specific binding in peripheral organs. In human whole body scans, the organ with the highest dose was urinary bladder wall (17 µGy/MBq) and the effective dose was 5.5 µSv/MBq.
Conclusion: V in human brain scans was consistent with the Results: of a previous study (2) and smaller than in rats (1) as expected from known binding site density. V/f1 showed modest reproducibility in most subjects. However, there may be confounding factors such as stress. Radiation-absorbed doses were low enough for human use. (R)-[C-11]rolipram is a promising ligand to image PDE4 in human brain.
FLUORINE-18 LABELED SA4503 AS A SELECTIVE SIGMA1 RECEPTOR LIGAND FOR PET
Kazunori Kawamura1, Kiichi Ishiwata2, Hideo Tsukada3, Kazuhiro Shiba4, Chieko Tsuji5,
Norihiro Harada3, Yuichi Kimura2
1Center for Integrated Human Brain Science, Brain Research Institute, University of Niigata, Niigata, Japan, 2Positron Medical Center, Tokyo Metropolitan Institute of Gerontology, Tokyo, Japan, 3Center for Central Research Laboratory, Hamamatsu Photonics K.K., Hamamatsu, Japan, 4Advanced Science Research Center, Kanazawa University, Kanazawa, Japan, 5NARD Institute, Ltd., Hyogo, Japan
Introduction: Sigma receptors are implicated in neurological, psychiatric and movement disorders. Thus, several sigma receptor ligands have been synthesized and evaluated for PET. One of these, SA4503 has been labeled with carbon-11, and was applied to mapping sigma1 receptors in the brain of humans by PET [ref. 1, 2, 3]. Clinically a 90-minutes scan was adapted for [C-11]SA4503-PET due to the half-lives of carbon-11 labeled tracer (half-lives, 20 minutes), however, a longer scan may be more preferable for PET measurement. Therefore, the analogs of SA4503 labeled with fluorine-18 (half-lives, 110 minutes) may be more preferable for PET, if they have similar pharmacokinetic properties as [C-11]SA4503. In this study, we evaluated a novel [F-18]fluoromethyl analog of SA4503 ([F-18]FM-SA4503) as a selective sigma1 receptor ligand for PET. Methods: The affinities of FM-SA4503 for sigma receptors were investigated by membrane binding assay. Biodistribution of [F-18]FM-SA4503 was investigated in mice by tissue dissection. Sigma1 receptor-specific uptake of [F-18]FM-SA4503 in the mouse brain was evaluated in blocking studies with cold FM-SA4503 and haloperidol (2000 nmol/kg co-injection). A conscious male rhesus monkey underwent 180-minutes PET scan with [F-18]FM-SA4503 using a model SHR-7700. After a week, haloperidol-pretreatment measurement with [F-18]FM-SA4503 was performed in the same monkey over a period of 180 minutes. Haloperidol (1 mg/kg) was injected into the monkey through a venous catheter 30 minutes before tracer injection. The displacement measurement was also performed in the other monkey over a period of 180 minutes. Results: FM-SA4503 was a selective affinity for sigma1 receptor (Ki for sigma1 receptor, 6.4 nM; Ki for sigma2 receptor, 250 nM). In mice, the uptake of [F-18]FM-SA4503 in the brain was gradually increased for 30 minutes after injection, and then decreased. In the blocking study, the uptake the brain was significantly decreased by co-injection of haloperidol (32 % of control) and of FM-SA4503 (52 % of control). In the PET study of the monkey brain, high uptake was found in the cerebral cortex, thalamus, and striatum. The radioactivity level of [F-18]FM-SA4503 in the brain regions gradually increased over a period of 120 minutes after injection, followed by a stable plateau phase until 180 minutes after injection. In the pretreatment with haloperidol measurement of the monkey brain, the radioactivity level was 16% of the baseline at 180 minutes after injection, suggesting the high receptor-specific binding. Conclusion: [F-18]FM-SA4503 showed specific binding to sigma1 receptors of the brain in the mice and monkeys. Therefore, [F-18]FM-SA4503 has potential for mapping sigma1 receptors in the brain.
IN-VIVO IMAGING OF MICROGLIAL ACTIVATION USING A PERIPHERAL BENZODIAZEPINE RECEPTOR LIGAND, C-11-CB148 AND ANIMAL-PET FOLLIWING ETHANOL INJURY IN RAT STRIATUM
Fumitaka Ito1, Gen Kudo1, Hiroshi Toyama1, Kentaro Hatano2, Hiromi Suzuki3, Masanori Ichise4, Hiroshi Yamaguchi2, Katsuhiko Sekimata2, Takashi Kato2, Kazuhiro Katada1, Mariana Serra5, Giuseppe Trapani5, Makoto Sawada3, Kengo Ito2
1Department of Radiology, Fujita Health University, Toyoake, Japan, 2Brain Science and Molecular Imaging, National Institute for Longevity Sciences, Obu, Japan, 3Department of Brain Life Science, Research Institute of Environmental Medicine, Nagoya University, Nagoya, Japan, 4Department of Radiology, Columbia University, New York, NY, USA, 5Bari University, Bari, Italy
Background and aims: Microglia plays an active part in brain injury and degeneration. The transition of microglia from the normal resting state to the activated state is associated with an increased expression of peripheral benzodiazepine receptors (PBRs). Increased PBRs may be used as a marker for detecting activated microglia in vivo. PET imaging of microglial activation using [C-11]-PK11195(PK), a PBRs ligand, has been investigated in several neurological conditions. Based on our basic mice experiments, a new PBR ligand, [C-11]-CB148(CB), has higher affinity than does PK. We investigated if CB PET can show activated microglia in a rat brain injury model.
Methods: On day 1, under pentobarbital anesthesia, 8 µl ethanol was injected into the rat right striatum through Hamilton syringe using a steotaxic operative procedure. On day 3, MRI scans were performed to evaluate brain morphologically. On day 4, dynamic PET scans were performed for 64 min with an animal PET scanner (SHR-2000 animal PET scanner, Hamamatsu Photonics) under chloral hydrate anesthesia after a bolus injection of 39-64 MBq of CB through tail vein. Nine ethanol-injured rats and 7 non-injured control rats were evaluated. To avoid arterial sampling, we evaluated the PBRs binding with a method to estimate normalized PBRs distribution volume (V*) with ‘reference’ tissue containing PBRs by Ichise (Neroimage2004;22 supple2:T149-50). V* was defined as the distribution volume (V) normalized by the ‘reference’ region tracer delivery (K1′). V*= V/ K1′= R1/ k2 and R1= K1/ K1′ were estimated voxel-wise with two tissue parameter multilinear reference tissue model by Ichise(JCBFM2003;23:1096-112). On the coronal PET image, regions of interest (ROI) were placed on bilateral striatum (ST) guided by a rat stereotaxic atlas and Harderian glands ('reference’ region). The animals were euthanized after the PET scanning and immunohistochemical staining was performed for confirmation of the presence of activated microglia. Densities of activated microglia in each rat striatum were measured.
Results: The PBRs V* values (min) in injured right ST were significantly higher by 12% than in non-injured left ST (14.82±3.63vs.12.67±6.24, p<0.05). Similarly, the PBR V* right/left striatum ratios were significantly higher by 11% than in control rats (1.1±1.3 vs.1.0±0.1, p<0.05). On immunohistochemical staining, IB4-lection positive cells showing activated microglia were shown around the ethanol injected tract in the right striatum but not in the non-injured left striatum. V* values (min) correlated highly with densities of activated microglia (y=0.0019x ? 0.263, R2=0.79, P=0.0002).
Conclusion: These Results: suggest that CB PET imaging and V* analysis would be a useful tool to evaluate microglial activation(brain inflammation) in rat brain.
THE 5-HT4 RECEPTOR IN THE GOTTINGEN MINIPIG: IN VIVO AND IN VITRO RECEPTOR BINDING
Birgitte R. Kornum1, Nanna M. Lind2, Nic Gillings3, Flemming Andersen3, Gitte M. Knudsen1
1Neurobiology Research Unit and Center for Integrated Molecular Brain Imaging, Copenhagen University Hospital Rigshospitalet, 2Department of Experimental Medicine, Panum Institute, University of Copenhagen, 3PET and Cyclotron Unit, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark
Background: The 5-hydroxytryptamine 4 (5-HT4) receptor is a G-protein-coupled receptor positively linked to adenylate cyclase activity. The receptor is found in the brain of several species, including rat, mouse, pig, monkey and human. The highest 5-HT4 receptor densities in the brain are found in the limbic system including hippocampus and striatum, and several recent studies have suggested involvement of the 5-HT4 receptor in cognitive and emotional processes. The use of the pig in neuroscience and in particular for PET studies is increasing. The pig brain is the size of a macaque brain and it is also a gyrencephalic. Further, animal facilities fulfilling all the pig's needs are more easily established and cheaper to run than primate facilities.
Aim: Here we report the distribution of the 5-HT4 receptor in the Gottingen minipig brain assessed by PET scanning as well as homogenate binding studies where Bmax and Kd were determined in 5 regions.
Methods: Eight Gottingen minipigs, 13-14 months, were PET-scanned (GE Discovery LS scanner) with a dynamic protocol including arterial blood sampling after i.v. bolus-injection of the tracer 11C-SB207145. The simplified reference tissue model with cerebellum representing non-specific binding was used to quantitate PET data. Immediately after sacrification, the brains were taken out, quickly dissected into several regions, and frozen on dry ice. Receptor Bmax and Kd values were obtained for 3H-SB207145 in tissue homogenates from: striatum, hippocampus, frontal cortex, remaining cortex and cerebellum.
Results: We find that the distribution of the 5-HT4 receptor in the pig brain is similar to the distribution in the human brain, with the highest binding in striatum (Fig.1). We detected no specific binding of 3H-SB207145 in cerebellar homogenate. The Kd of 3H-SB207145 was 0.39±0.02 nM in striatum. There was no significant difference between Kd across regions. Accordingly, for Bmax calculation Kd was fixed across regions and pigs to a value of 0.39. We find a good correlation between receptor density and PET-determined receptor binding (Fig. 2).
Fig.1: 11C-SB207145 PET image of a pig brain co-registered to a magnetic resonance picture.
Fig.2: Comparison of BP2 calculated from 11C-SB207145 PET and Bmax in tissue homogenates (3H-SB207145). Means ± s.d.
IMAGING SEROTONIN TRANSPORTER WITH 4-[18F]-ADAM IN HUMAN BRAINS
C-Y Shiue1, C-J Peng1, K-H Ma2, H-S Wang1, S-Y Huang3, R-S Liu4, W-S Huang1
1Department of Nuclear Medicine, Tri-Service General Hospital, 2Department of Anatomy, National Defense Medical Center, 3Department of Psychiatry, Tri-Service General Hospital, 4Department of Nuclear Medicine, National Yang-Ming University, Taipei, Taiwan-R.O.C.
Background: Abnormalities in serotonin transporter (SERT) has been implicated in several neuropsychiatric disorders and is the target for antidepressants. For the last decade, [11C]-(+) McN 5652 has been the most promising PET agent for studying SERT in humans (Szabo et al., 1995). However, this agent has high nonspecific binding and has only moderate signal contrast in human PET studies. Additionally, its pharmacokinetics is not optimal because of its slow brain uptake and the short half-life of 11C. We recently reported that N,N-dimethyl-2-(2-amino-4-[18F]fluorophenylthio)benzylamine (4-[18F]-ADAM) may be a potent SERT imaging agent (Shiue et al., 1003). We report here the preliminary Results: of the imaging of SERT with 4-[18F]-ADAM in human brains.
Methods: 4-[18F]-ADAM was synthesized as reported previously (Shiue et al., 2003). The PET scanner used for this study was ECAT EXACT HR+ (Siemens Medical Solutions USA, Inc). The radioligand was injected i.v. as a bolus to the subject and dynamic scan for 90 min. Binding parameters were determined with Logan plots using cerebellum as the reference region. Results: PET studies in humans showed that 4-[18F]-ADAM passed the blood-brain-barrier and peaked (0.006%/ml) at 1 hr post-injection. It had high uptake in raphe nucleus (RN), thalamus (Th) and striatum (Str), moderate uptake in frontal cortex (FC) and occipital cortex (OC), and low uptake in cerebellum (CB). The distribution volume ratio (DVR) for RN, Th, Str, FC and OC was 2.89±0.55, 2.15±0.50, 1.99±0.35, 1.39±0.15 and 1.36±0.20, respectively (n = 3). Conclusions: 4-[18F]-ADAM is a potent SERT imaging agent for human studies.
EFFECT OF REPETITIVE TRANSCRANIAL MAGNETIC STIMULATION ON A1 ADENOSINE RECEPTORS: A [18F]CPFPX STUDY
David Elmenhorst, Oliver H. Winz, Roland Sparing, Luciano Minuzzi, Andreas Bauer
Institute of Medicine, Research Centre Juelich, Juelich, Germany
Adenosine A1 receptors (A1AR) are widely distributed on cortical and subcortical neurons and modulate neurotransmission. The extracellular concentration of adenosine is amongst others coupled to cellular activity or metabolism. During in vivo animal experiments electrical stimulation of brain tissue increased adenosine levels accessed with microdialysis up to 5 or 10 fold. PET studies in humans showed that repetitive transcranial magnetic stimulation (rTMS) of cortical areas induced changes in blood flow, glucose metabolism or dopamine release at the site of stimulation and/or in distant brain areas (e.g. the basalganglia).
This study examines the impact of excitatory (high frequency) rTMS on the binding of the A1 selective radioligand [18F]CPFPX to A1ARs with PET in humans. So far 5 healthy male subjects received during a 140 min bolus plus constant infusion experiment three blocks of rTMS during the steady state phase of ligand delivery (50 to 140 min) inside the scanner.
Each rTMS block (20 trains of 1 sec stimulation at 10 Hz, interval between trains 12 sec, 90 % of motor threshold) lasted 4 min. There was a resting period of 10 min following each block. The PET photo multipliers were shielded against the magnetic field of the TMS coil by a grounded metal tube. Positioning of the coil over the left medial dorsolateral prefrontal cortex according to probabilistic coordinates superimposed on individual normalized MRI was done by frameless neuronavigation. The actual position of the coil and the stimulated area was confirmed ex post on transmission scans.
Plasma radioactivity concentration and metabolite analysis were performed in arterialised venous blood samples. Parametric total distribution volume (DVt) images were generated as tissue to plasma ratio and used for statistical analysis using SPM2. The 4 min blocks of stimulation were compared to 4 min-frames before and after stimulation by collapsing the parts of the three strains into one bin using a voxelwise paired t-test.
The main finding was a reduced [18F]CPFPX binding during stimulation compared to baseline in the left dorsal caudate nucleus. Based on this SPM Results: a region of interest analysis showed a relative DVt decrease of 24%. No changes were detected in the putamen, nucleus accumbens or the right side of the caudate. This decrease persisted to a lower degree over the 4 min after stimulation. Cortical binding at the site of stimulation was not affected.
However, it should be noted that the examination of a different stimulation site as control condition is ongoing. Until then, these preliminary Results: have to be discussed with care.
SYNTHESIS, RADIOSYNTHESIS, AND BIOLOGICAL EVALUATION OF 18F-LABELLED PYRAZOLOPYRIMIDINES FOR IMAGING A1-GABAA RECEPTORS
Winnie Deuther-Conrad1, Alexander Hoepping2, Matthias Scheunemann1, Steffen Fischer1, Achim Hiller1, Florian Wegner3, Jorg Steinbach1, Peter Brust1
1Institute of Interdisciplinary Isotope Research, Leipzig, 2ABX Advanced Biochemical Compounds, Radeberg, 3Department of Neurology, University of Leipzig, Leipzig, Germany
Introduction: Gamma-aminobutyric acid (GABA) is the major inhibitory neurotransmitter in the vertebrate central nervous system. GABA induces its hyperpolarizing response on neurons by binding to ionotropic pre-, post- and extrasynaptic receptors. The GABAA receptor is a pentameric ligand gated ion channel, that is a major neuropharmacological target in the therapy of diseases such as depression, epilepsy, schizophrenia, and sleep disorders. More than half of the central GABAA receptors consist of a1, b2, and g subunits, and the a1-containing GABAA receptor mediates the sedative, amnestic, and anticonvulsant action of benzodiazepines. Novel pyrazolopyrimidines, supposedly a1-selective GABAA receptor agonists, were developed as sedative-hypnotics for the treatment of insomnia. To study the involvement of a1-containing GABAA receptors in the function of the normal and diseased brain, a new radioligand based on subtype-specific pyrazolopyrimidines was developed. We describe here the synthesis and evaluation of a high affinity 18F-labelled pyrazolopyrimidine GABAA radiotracer and its reference compound in vitro and in vivo in mice.
Methods: 2-[18F]Fluoro-N-methyl-N-{3-[3-(thiophene-2-carbonyl)-pyrazolo[1,5-a]pyrimidin-7-yl]-phenyl}-acetamide (1) was synthesized by heating the corresponding Br-precursor in anhydrous [18F]fluoride/18-Crown-6/acetonitrile for 20 min. Purification was accomplished by HPLC, and solutions of [18F]1 were formulated in sterile saline. Binding affinities of 1 were determined by competitive homogenate assays vs. [3H]flunitrazepam on GABAA receptors isolated from adult rat cerebellum. To compare the distribution of neuronal binding sites of [18F]1 and [3H]flunitrazepam, autoradiographic examinations on sagittal rat brain slices were performed in vitro. Organ distribution and metabolism studies of [18F]1 (~300 kBq, 200 µl saline, i.v.) were carried out in awake female NMRI mice. An ex vivo investigation of brain GABAA receptor occupancy after peripheral administration of [18F]1 was performed by autoradiography.
Results: [18F]1 was synthesized in good yield (35–46% EOB), high specific activity (? 250 GBq/µmol EOS), and radiochemical purity (>99%). In adult rat cerebellar membranes, 1 showed nanomolar affinity for GABAA receptors labelled by [3H]flunitrazepam. The distribution of [18F]1 binding in rat brain slices in vitro was comparable with [3H]flunitrazepam, and high densities of [18F]1 binding sites were observed in the olfactory bulb, cerebral cortex, and cerebellum. Organ distribution studies in NMRI mice indicated a peak in the whole brain uptake of 2.72% ID/g at 15 min post injection. Autoradiographic ex vivo examination of brain distribution revealed no specific localisation of [18F]1 at 15 and 60 min post injection. Metabolite analysis at 5 and 30 min post injection indicated a high accumulation of [18F]1 metabolites in the brain.
Conclusion: Despite promising characteristics in vitro, [18F]1 is not suitable for imaging a1-containing GABAA receptors because of its strong metabolism combined with the accumulation of 18F-labelled metabolites in the brain. The in vivo data strongly indicate the necessity to develop compounds with a much higher metabolic stability. Investigations on structurally modified 18F-labelled pyrazolopyrimidines are in progress.
PET IMAGING OF NEUROKININ-1 (NK1) RECEPTORS WITH [18F]SPA-RQ IN HUMAN SUBJECTS: ASSESSMENT OF REFERENCE TISSUE MODELS AND THEIR TEST-RETEST REPRODUCIBILITY
Fumihiko Yasuno1, Sandra Sanabria2, Donald Burns2, Richard Hargreaves2, Subroto Ghose3, Masanori Ichise1, Federick Chin1, Cheryl Morse1, Victor Pike1, Robert Innis1
1Molecular Imaging Branch, National Institute of Mental Health, Bethesda, MD, 2Merck Research Laboratories, West Point, PA, 3Department of Psychiatry, University of Texas Southwestern Medical Center, Dallas, TX, USA
Background: [18F]SPA-RQ labels the substance P-preferring (NK1) receptor in human brain. A prior study showed that [18F]SPA-RQ brain uptake can be quantified with a reference tissue method and thereby avoid invasive blood sampling. The purposes of this study were to compare three different reference tissue Methods and to assess test-retest reproducibility.
Methods: Eight healthy subjects underwent two [18F]SPA-RQ scans. We calculated binding potential (BP), which is proportional to receptor density, from both regional volume of interest and voxel-wise data. We compared three reference tissue Methods: simplified reference tissue model (SRTM), multilinear reference tissue model (MRTM) and its two-parameter version (MRTM2) (Table 1).
Results: The three Methods: generated equivalent values of regional BP, but MRTM2 was the most resistant to noise. Temporally stable values of BP were obtained with 240 min of imaging data. MRTM2 had excellent test-retest reproducibility, with high reliability (intraclass correlation >0.9) and low variability (<10%). In addition to regional volume of interest analysis, we also created parametric images of BP, variability, and reliability based on voxel-wise time-activity data. The reproducibility of parametric BP was also good, with variability <20% and reliability >0.7 in gray matter regions (Figure 1).
Conclusions: A two-parameter reference tissue method (MRTM2) provided reproducible and reliable measurements of [18F]SPA-RQ brain uptake using 240-min of both regional and voxel-wise data.
DOPAMINE OCCUPANCY AND DOPAMINE TRANSPORT INHIBITION: LESSONS FROM A DYNAMIC MODEL
Sune Jespersen, Albert Gjedde
Center of Functionally Integrative Neuroscience, Aarhus University, Aarhus, Denmark
Introduction: We combined the known elements of dopaminergic neurotransmission in the striatum of the brain in order to reveal the role of individual relaxation constants in the regulation of dopaminergic neurotransmission. The model consists of the tyrosine compartment, fed from the bloodstream by the blood brain barrier facilitated diffusion transporter, and the dopa compartment, fed by tyrosine hydroxylase and decaying to the bloodstream, to 3-O-methyldopa, and to the main conduit in the form of dopa decarboxylation to dopamine. The intracellular dopamine compartment is also fed by the dopamine transporters linking the extracellular and intracellular dopamine compartments. Intracellular dopamine is trapped in vesicles and released upon activation to the intrasynaptic cleft, where dopamine diffuses to the extrasynaptic space. From here it returns to intracellular space by facilitated diffusion. Another fraction of intracellular dopamine is converted to DOPAC, which in turn is converted to HVA. We wish to predict the response to specific disturbances in dopaminergic neurotransmission on the basis of an appropriate compartment representation of diffusion dynamics. Here we examine the validity of describing the extrasynaptic space as a number of hidden, virtual compartments by comparing to the exact solution of a model system for two types of physiological perturbations.
Model: We model the extrasynaptic space (eDA) as a narrow slab of tissue attached to the intrasynaptic space (iDA) at one end, and kept at the free intracellular dopamine concentration at the other. The action of the dopamine transporter surrounding the extrasynaptic space is described by concentration dependent loss of extrasynaptic dopamine characterised by a rate constant k given by the PS/V product. Assuming then that the dopamine concentrations in the intrasynaptic and intracellular compartments are kept constant during the time of observation, the time dependent diffusion equation can be solved exactly. We focus on two physiological perturbations starting from the above mentioned steady-state: 1) The intrasynaptic dopamine concentration is doubled, and 2) the relaxation constant for transport to intracellular dopamine is reduced to half of its normal value. The response of the extrasynaptic dopamine to these two perturbations is compared to that of an approximate compartment model with hidden compartments (1,10,100) based on a discretization of the diffusion equation.
Discussion: The compartmental model consisting of 10 compartments provides a good approximation at most times, corresponding to the exact solution to within less than 12%. Using 100 compartments, the agreement is even better, and the relative deviation is less than about 5%. On the basis of these Results:, it is possible to incorporate the present representation of extra synaptic space in the 11 compartments of the model of dopaminergic neurotransmission in the striatum, and obtain realistic estimates of the responses in dopamine content to various physiological perturbations.
TRACER KINETIC MODELLING OF BETA-AMYLOID LOAD IN TRANSGENIC MICE USING [C-11]6-OH-BTA-1 MICROPET
Isabelle Stangier1, Andre Manook1, Gjermund Henriksen1, Marc Huisman1, Alexander Drzezga1, Nassir Navab2, Sibylle Ziegler1
1Department of Nuclear Medicine, 2Computer Aided Medical Procedures and Augmented Reality, Technical University Munich, Munich, Germany
Background and aims: The purpose of this study was to investigate the applicability of compartmental and graphical Methods: on a transgenic mouse model of Alzheimer's disease using [C-11]6-OH-BTA-1 (PIB) micropet. Methods: The animal population consisted of 4 APP/PS1 mouse models aged 23 months and 4 control mice (genotype: C57BL/6J). Dynamic imaging of all animals was performed on a Siemens Focus 120 micropet scanner with a total scan duration of 60 min after injection of 14.3±5.9 MBq of PIB. The images were reconstructed using filtered backprojection with corrections for radioactive decay and dead time. Volumes of interest (VOI) were drawn on whole brain, frontal cortex and cerebellum. An image-based metabolite corrected input function was derived from VOI definition in the animals' left ventricle. The tracer kinetic data was analysed using a 1 and 2 tissue compartmental (1T, 2T) model and Logan graphical analysis (LGA). Outcome measures were distribution volumes (DV) which were normalised to the cerebellum as reference region to obtain distribution volume ratios (DVR). Results: Both, in transgenic and control mice it was possible to obtain robust estimates of DVR values using the 1T model and LGA, but not for the 2T model. For transgenic mice the DVR values corresponded well with distribution pattern of amyloid load in the brain of the APP/PS1 mouse model as determined by immunostaining (DVR(1T): whole brain: 1.09±0.08, frontal cortex: 1.11±0.06; DVR(LGA): whole brain: 1.33±0.05, frontal cortex: 1.50±0.04). For the control mice the DVR values did not show considerable amyloid loads above the cerebellar level (DVR(1T): whole brain: 1.02±0.04, frontal cortex: 0.95±0.05; DVR(LGA): whole brain: 0.98±0.04, frontal cortex: 0.97±0.05). Significant differences in DVR values of amyloid load between transgenic and control mice were found in whole brain and frontal cortex (1T: whole brain: p=0.1, frontal cortex: p<0.007; LGA: whole brain: p<0.00003, frontal cortex p<0.000003). Conclusions: This study shows that tracer kinetic Methods: like the 1T model and LGA have the potential to quantify beta-amyloid load in a transgenic mouse model of Alzheimer's disease.
TOWARDS IMPROVED TRACERS FOR IN VIVO IMAGING OF AMYLOID-BETA PLAQUE
Behrooz. H. Yousefi, Alexander Drzezga, Hans-Juergen Wester, Gjermund Henriksen
Department of Nuclear Medicine, Klinikum Rechts Der Isar, Technical University of Munchen, Munchen, Germany
Background/Aim
Amyloid-plaque formation is one of the earliest and most relevant physiological processes in the development of Alzheimer's disease (AD). In order to be useful for disease staging and therapy monitoring, a quantification of the amyloid plaque load is required. Despite the promising initial Results: from clinical trials, the regional brain uptake is not correlating completely to the established regional pattern of plaque load as measured by post-mortem quantification. We present a report on our efforts to develop tracers with enhanced selectivity for AB40 and AB42, respectively, and possessing improved in vivo properties.
Methods
Apart from altered substitution pattern on the aromatic rings of BTAs and IMPYs the extent and nature of N-anilino alkylation was varied. In addition, structures have been selected allowing for C-11 and radiohalogenation. Affinity was determined using competition binding assays (Ki) on synthetic AB40 and AB42 amyloid fibrils. Together with structural overlays and modeling calculations, the affinity and selectivity data were used for the identification and assessment of relevant substructures and their influence on AB-binding.
Results
For IMPYs, the binding experiments to fibrils of synthetic AB42- and AB40-peptides showed up to 92 % inhibition of binding of the 3H-labeled reference at the 100 nM level to AB42 and up to 83 % inhibition to AB40 resulting in AB42-to-AB40-selectivity ratios of up to 41. The lipophilicity (logPoctanol/PBS) of the compounds was in the range of 0.5 to 3.5. From our library of BTAs, 6 were selected and showed Ki-values in the range of 1.3–12..4 on AB40, with selectivity ratios of 2-10 over AB42. The uptake of BTAs in the brain of mice (wild type) at 2 min was in the range of 6.1 to 19.3% ID/g (PIB: 10.3). The 2 min-to-30 min uptake ratios were in the range of 1.9–9..8 (PIB: 10.7). By overlays and trend analyses, four distinct substructures with a differentially, pronounced impact on the binding were identified (Fig.1). These substructures apparently independently determine binding and thus allow subtle adjustment of molecular properties. In addition, as recently demonstrated for a series of BTA-analogues, this model enables tailoring of lipophilicity, brain uptake, plasma half-life and in vivo stability.
Conclusions
Twenty-five novel IMPYs and 15 novel BTAs substituted with radiohalogens or C-11 in different positions have been synthesized and evaluated as AB-ligands. Changes of the substitution pattern can be used to improve the selectivity for the two relevant AB-proteins. Work is in progress to further explore the identified key molecular descriptors for further optimization of the combination of radiolabel and the identified lead structures. Verification of the in vitro data is currently being performed in a double transgenic mice model ex vivo and in vivo and compared with aged-matched controls.
CONSTRUCTION OF BINDING POTENTIAL MAPS OF PITTSBURGH COMPOUND B BY BOLUS-PLUS-INFUSION TRANSFORMATION (BPIT): COMPARISON TO POPULAR REFERENCE TISSUE APPROACHES
Hiroto Kuwabara1, Anil Kumar1, Michael Rafii3, Daniel Holt1, Robert Dannals1, Susan M. Resnick2, Dean F. Wong1
1Department of Radiology, 2National Institute on Aging, Baltimore, MD, USA, 3Department of Neurology, Johns Hopkins University, Baltimore, MD, USA
Background and aims: Pittsburgh Compound B ([11C]PIB), an amyloid imaging agent, is being investigated in dementia and aging research. Construction of accurate binding potential (BP) maps is needed as voxel-wise statistical approaches are increasingly employed for detecting regional increases in amyloid deposition. We tested an approach to transfer bolus-only dynamic PET experiments to bolus-plus-infusion experiments in theory such that BP was estimated as voxel-to-reference region ratio minus 1 in a hypothetical steady-state (BPIT).
Methods: Thirty-four Baltimore Longitudinal Study of Aging participants (age: 79.9 ± 7.6) underwent one 90-min dynamic PET scan following bolus injection of PIB (15mCi). Participants had normal cognition (n=28) or mild cognitive impairment (informant-based clinical dementia rating of 0.5). Volumes of interest (VOIs) were defined manually on a spoiled-gradient MRI for putamen, caudate nucleus, thalamus, and cerebellum. A standard VOI template, including 28 VOIs for cortical and subcortical structures, was adjusted to each individual's MRI using SPM2 spatial normalization. VOIs were transferred to PET space according to MRI-PET coregistration parameters from SPM2. BP maps of PIB were generated by BPIT, reference tissue graphical analysis (RTGA; Logan et al., 1996), and multi-linear reference tissue method with 2 parameters (MRTM2; Ichise et al., 2002) using the cerebellum as the reference region. Regional BP values were obtained by applying the VOIs on BP maps (Map analysis). Separately, regional time-activity curves (TACs) were obtained by applying the VOIs on dynamic PET frames to estimate regional BP values for each of the three approaches (TAC analysis).
Results: Regional BP values of the three Methods: agreed for the TAC analysis (BPIT (=y) vs. RTGA (=x): y = 0.991?x + 0.00288, r2=0.992; BPIT vs. MRTM2: y = 0.995?x + 0.000147 r2=0.988). In the comparison of map (=y) vs. TAC (=x) analyses, BPIT showed no deviations of regional BP values (y = 1?x + 0.0000, r2=1.000) while RTGA showed bias and greater variability (y = 0.83?x + 0.0629, r2=0.928). MRTM2 displayed minimal bias and variability in this comparison (y = 0.989?x + 0.00353, r2=0.993). Computational times for constructing BP maps were 19.2 ± 1.4 seconds for BPIT, 26.5 ± 2.0 seconds for RTGA, and 586.3 ± 15.5 seconds for MRTM2 using a 3.4 MHz Pentium 4 PC. No outliers were observed (voxel values 10 times greater than mean of 6 surrounding voxels) for BPIT and RTGA but 34.5 ± 1.9 % of voxels (range: 30.6 ? 39.6%) were outliers with MRTM2 although the majority was outside the brain.
Conclusions: This study validated the BPIT approach for obtaining BP of PIB scans against widely used RTGA and MRTM2 approaches on the basis of the TAC analysis. BPIT suffered no bias and variability in the map vs. TAC comparisons, which was a substantial improvement over RTGA. BPIT and RTGA displayed no outliers but MRTM2 suffered noticeable numbers of outliers. BPIT substantially shortened image construction times. Thus, BPIT appears advantageous over RTGA and MRTM2 in constructing BP maps for PIB scans in aging and dementia research.
Support: Intramural Research Program, National Institute on Aging, NIH; N01-AG-3-2124; K24-DA00412(DFW)
REDUCED STANDARD DEVIATION OF BINDING POTENTIAL VALUES WITH AUTOMATIC SEGMENTATION OF STRIATUM IN PET - IMPLICATIONS TO STATISTICAL POWER
Esa Wallius1, Jussi Tohka1, Jussi Hirvonen2, Jarmo Hietala3, Ulla Ruotsalainen1
1Institute of Signal Processing, TUT, Tampere, Finland, 2Turku PET Centre, 3Department of Psychiatry, UTU, Turku, Finland
Background and aims:. Volume of interest (VOI) delineation is an essential prerequisite for the data analysis of PET images, which serves e.g. drug development. The VOIs are traditionally extracted manually, using an MRI-aided technique. The manual VOI extraction is time-consuming and produces intra- and inter-observer variability to the Results:. Automatic segmentation of VOIs (the caudate and putamen) could remedy these problems. We investigated the standard deviation (SD) of regional binding potential (BP) and normalized absolute difference (NAD) values in a test-retest setting with both manual and automatic Methods:.
Methods:. The test-retest dataset consisted of eight healthy male volunteers, who underwent two 11C-raclopride PET scans during the same day. Regional BP values, NAD reproducibility measures (|BPscan1-BPscan2|/BPscan1) and their SDs were computed using three automatic and two manual VOI extraction Methods:. The VOIs considered were the left and right caudate and putamen. The differences in BP and NAD SDs among the examined brain regions were statistically tested with paired t-tests. The automatic extraction of the striatum was done with deformable model based method “DSM-OS” (dual surface minimization ? outer surface) [1]. The Methods: “pairwise DSM-OS” and “DSM-OS coreg” are modifications of the “DSM-OS” method. In these Methods: the segmentation of the first study is utilized for the segmentation of the second study. The manual VOI extractions were performed by an expert. The “manual orig” method was based on co-registered MR image segmentation. The “manual coreg” method relies on both MR-to-PET and PET-to-PET co-registration in this study setting.
Results:. The table shows that the average SDs were always smaller with the automatic Methods: than with the manual Methods:. Statistically significant differences in NAD SD were found between DSM-OS coreg and manual orig (p=0.0175, 95 % confidence interval [0.48,3.18]) and DSM-OS coreg and manual coreg (p=0.0453, [0.03,2.06]). With BP SD significant differences were seen between DSM-OS and manual orig (p=0.0064,[0.03,0.11]) and DSM-OS coreg and manual orig (p=0.0271,[0.01,0.11]).
Conclusions:. Reduced standard deviation with automatic Methods: implies better statistical power (sensitivity) of the analysis when e.g. the effects of a drug or the differences between groups are examined. The automatic Methods: may be especially useful in large multi-center clinical trials, where it is necessary to reduce any kind of confounding variation in order to estimate the true within- or between-subjects variation.
A SIMULATION OF A NEW IMAGE RECONSTRUCTION METHOD FOR 3-D PET LIST MODE DATA
Shoji Kawatsu1,2, Noboru Ushiroya3
1Department of Radiology, Kyoritu General Hospital, Nagoya, 2Department of Brain Science and Moleculat Imaging, National Center for Geriatrics and Gerontology, Obu, Aichi, 3Department of General Education, Wakayama National College of Technology, Japan
Background and aims
Traditionally, three-dimensional (3-D) positron emission tomography (PET) image reconstruction is performed using the filtered back projection (FBP) method or the iterative reconstruction Methods:. The FBP method is the digital implementation of the Radon transform, while the iterative reconstruction Methods: includes algebraic reconstruction techniques and statistically based iterative Methods:, such as Shepp and Vardi's maximum likelihood algorithm. These Methods: do not directly make use of a geometrical property of the line of response (LOR), namely that two LORs, which originate from two radioactive isotopes located within a sufficiently small voxel, are separated by a distance of a few millimeters. We formerly introduced a new 3-D PET image reconstruction method that directly applies this geometric property of LORs, named the pair of near-missing lines of response (PNMLOR) method. The aim of this study is to varidate the effectiveness of this method by performing a simulation using GATE software.
Methods
A simulation was performed using the PNMLOR image reconstruction method. Data was created by GATE. In the simulation, a CdTe detector was taken as the gamma-ray detector. Each CdTe detector element was assumed to have a cross-section of 2 mm × 2 mm and a depth of 5 mm. The detector array was arranged as a cylinder having a diameter of 18 cm and a length of 20 cm. The CdTe detector elements were located all around the lateral face of the cylinder. List mode data acquisition was assumed and the coincidence-time was set to be 10 ns by the software. The duration of data acquisition was 20 seconds. The cubic voxel size was set 1 mm3. Two LORs separated by a distance of 3 mm were assumed to make a pair of near missing LORs. The sum of the number of PNMLORs, whose midpoint of the shortest line segment connecting these two LORs were within 1.5 mm from the center point of the voxel, was assumed to be proportional to the activity in the voxel.
Result
Positron sources were point sources of activities of 1×10-2, 2×10-2,…., 9×10-2MBq and were positioned in a line x=y=0 and z=10,20,…,90mm. where origin of the coordinate was the center of the cylinder and the Z-axis was the longitudial axis of the cylinder. FWHM was 2.5mm.
Conclusion
We formerly proposed, PNMLOR, a new image reconstruction method for 3-D PET. The proposed method makes direct use of an elementary property of LORs which originate from radioactive isotopes located in close proximity to each other. The proposed method differs from the FBP method and the iterative reconstruction method. A simulation using the proposed method demonstrated good resolution and linearity for data acquired over a short time period.
3-D DIGITAL RODENT BRAIN ATLAS
Trine Hjornevik1, Christian Pettersen1, Dmitri Darin1, Trygve B. Leergaard1, Jan Gunnar Bjaalie1, Frode Willoch1,2
1Centre for Molecular Biology and Neuroscience (CMBN), Department of Anatomy, Institute of Basic Medical Sciences, University of Oslo, Oslo, Norway, 2Department of Radiology, Aker University Hospital, Oslo, Norway
Background
Conventional atlases contain 2-D diagrams of sections cut perpendicular to standard axes through the brain. We present a method for 3-D reconstruction of series of such atlas diagrams, and a tool for visualization and analyses of the reconstructed structures.
Methods and Results
Atlas diagrams (here taken rat atlas from Paxinos & Watson, 5th edition,2005) are digitalized and organized in separate layers in one file. Each layer / diagram is aligned on basis of the given coordinate system. The aligned atlas images are organized as a stack of images (imported into Amira 4.0), using distance from Bregma to define through-plane position. Anatomical structures are segmented and respective segments are assigned in colors, and reconstructed in 3-D. Following surface rendering triangulation, individual structures are exported as VRML (Virtual Reality Modeling Language) files. For visualization we use our tools (m3d) in combination with a database system, The Rodent Brain Navigator, developed in our laboratory (Bjaalie et al., 2006).
Successfully, data from different sources imported were imported in Navigator. The different images sources were brought into the same sterotactic atlas ranging from histological sections to tomographic volumes such as PET, CT and MRI. In addition, to giving stereotactic coordinates, the atlas represents the different anatomic volumes in the brain and are used to identify correct brain regions on the imported image data (figure) and can be used for further analyses. Reslicing allows the user to generate customized atlas diagrams, at any chosen angle. Volumes and surface areas can be extracted from the reconstruction.
Figure. For illustration purposes we show a) MRI, b) segmented, color-coded atlas, and c) superimposition of MRI and atlas.
Conclusion
We present a method for producing a 3-D digital atlas from conventional atlas diagrams. The 3-D reconstruction maintains the stereotaxic standard space of the atlases. The reconstruction facilitates understanding of size, shape and spatial relationships, helping us to database and analyse rat PET images with data from different image sources.
DEVELOPMENT OF OXYGEN-15-LABELED-GAS SYNTHESIZER FOR MEASUREMENTS OF QUANTITATIVE CMRO2 AND CBF BY RAPID PROTOCOL USING PET
Toru Inomata1, Yoshinori Miyake2, Yoichiro Ota1, Tomoyuki Asano3, Kazuhiro Miyamoto3, Hidehiro Iida1
1Department of Investigative Radiology, National Cardiovascular Center, Suita, Osaka, Japan, 2Department of Radiology and Nuclear Medicine, National Cardiovascular Center, Suita, Osaka, Japan, 3Stella Chemifa Corporation, Izumiootsu, Osaka, Japan
Oxygen-15 has been used to quantify the myocardial blood flow (MBF), regional cerebral blood flow (CBF) and cerebral metabolism rate of oxygen (CMRO2) in positron emission tomography (PET). By taking advantage of short half-life of two minutes of oxygen-15, multiple measurements dosing two or more oxygen-15 labeled medicines such as [15O]CO and [15O]CO2 at interval of ten minutes have been performed especially for diagnosis of cerebral infraction. Also, the quantitative determinations of the CBF and CMRO2 were confirmed by this new method using animals.
However, the preparation time of the gases in the PET treatment takes more than one hour to raise the temperature of charcoal in an oven by the conventional method. Therefore, the rapid supply of the gases has been a serious demand, especially for the diagnosis of acute stroke.
Recently, a new method for synthesis of [15O]CO and [15O]CO2 at room temperature have been developed in our group. Then, the rapid supply of the gases can be achievable. In this new method, the synthesizer can be assembled in smaller dimensions than those of conventional one.
We have been developing the Oxygen-15-labeled-gas synthesizer for quantitative measurement of CMRO2 and CBF based on the conventional synthesis method. The figure shows the arrangement of the piping, valves and so on of the synthesizer. The synthesized [15O]CO2, for example, can be rapidly changed from [15O]O2 to the rapid protocol even at the location where the synthesizer unit would not be located close to the PET scanner.
Further modifications can be available to adapt the system to rapid synthesis method which does not require heating equipments in the synthesizer unit.
ARTIFACTS IN A STATISTICAL IMAGE ANALYSIS CAUSED BY THE MISMATCH IN THE SPATIAL RESOLUSION
Takashi Kato1, Kengo Ito1, Yutaka Arahata2, Yukihiko Washimi2, Katsunari Iwai2, Takako Yamada2, Yuji Abe2, Akinori Nakamura1, Kentaro Hatano1, Atsuko Nagano-Saito1, Shoji Kawatsu1, Takashi Nihashi3
1Department of Brain Science and Molecular Imaging, 2Department of Neurology, 3Department of Radiology, National Center for Geriatrics and Gerontology, Obu, Aichi, Japan
Background and aims: The purpose of this study was to investigate artifacts in a voxel-by-voxel statistical result owing to mismatched image smoothness. It is important to understand such errors, for example, in sharing a database which is made with a different imaging apparatus or different image processing conditions. Smoothness (spatial resolution) is one of the characteristics of an image.
Subjects and Methods: The subjects were 38 healthy normal volunteers (23 males, 15 females; mean age 59+/−10 years old) were recruited. Brain PET was performed in 2D acquisition mode with an ECAT HR scanner (Siemens/CTI) after 40-60 min after the injection of 370 MBq [F-18] FDG (fluorodeoxyglucose). Images were reconstructed with filtered back projection into 128×128 matrices and voxel size of 1.8 × 1.8 × 3.125 mm. Spatial resolution of original reconstructed PET images was approximately 6 mm in FWHM. Attenuation correction was applied with using transmission scan data obtained prior to FDG. injection. The PET images were spatially normalized and smoothed with Gaussian kernel filter 4, 8, 12mm in FWHM. Two sample t-test was performed to compare the data sets of the different smoothness. The image smoothing and statistical analyses were done using SPM2. Original PET images were also anatomically standardized with iNeurostat (MediPhysics, Japan) to be three dimensional stereotactic surface projection (3D-SSP) (Minoshima S, Washington University) after they were smoothed with Gaussian kernel filter 4, 8, 12mm in FWHM. Voxel values were normalized with that of reference areas, global mean, thalamus, cerebellum, and pons. 3D-SSP Z score mappings were made to compare smoothed data sets.
Results: SPM analyses showed the statistical difference caused by smoothing appeared in intermediate areas between gray matter (higher FDG uptake value) and white matter (lower FDG uptake value) or between gray matter and cerebral spinal fluid (lower FDG uptake value). In 3D-SSP analyses, smoothing made the Z value (increase compared to smaller smoothing) larger in areas of the primary moto-sensory area, around Sylvian fissures, medial temporal, cerebellar hemispheres, and around cerebral ventricles. Smoothing also made the Z value (decrease compared to smaller smoothing) larger in areas of the parietotemporal association cortices, frontal association cortices, posterior cingulate, striatum, and thalamus. The distributions of decrease Z were overlapped with the area where decreased glucose metabolism is seen in Alzheimer's disease (AD), although the Z value itself was not so big when the difference in the spatial resolution was smaller.
Conclusions: Voxel-based statistical Results: are affected by mismatched spatial resolution in compared data sets of images. It might cause artifacts which are similar to AD-likely change. The spatial resolution must be matched in statistical image analyses.
6-(4-[11C]METHOXY-PHENYL)-3,4-DIHYDRO-2H-[1,4]DIAZEPINO-[6,7,1-HI]INDOL-1-ONE, A POTENTIAL PET TRACER FOR POLY(ADP-RIBOSE) POLYMERASE
Miyake Yoshinori1, Ohta Youichirou2, Watabe Hiroshi2, Teramoto Noboru2, Kurokawa Maki2, Ishida Yoshio3, Morishita Hideki1, Iida Hidehiro2
1Department of Pharmacology, 2Department of Radiology and Nuclear Medicine, 3Department of Investigative Radiology, National Cardiovascular Center, Suita, Osaka, Japan
OBJECTIVES: To evaluate whether it is possible to detect excessive activation of Poly ADP-ribose Polymerase (PARP) with positron emission tomography (PET) in cerebral tissues, we synthetized 6-(4-[11C]Methoxy-phenyl)-3,4-dihydro-2H-[1,4]diazepino-[6,7,1-hi]indol-1-one ([11C]MDDI), a potent PARP inhibitor, and evaluated the pharmacokinetics of [11C]MDDI in the brain of normal rats.
METHODS: Radiolabeling of [11C]MDDI was prepared by O-[11C]methylation of 6-(4- hydroxy-phenyl)-3,4-dihydro-2H-[1,4]diazepino-[6,7,1-hi]indol-1-one, obtained by demethylation of MDDI with [11C]methyl triflate. [11C]MDDI (72 MBq) was administrated into the left femoral vein of rat. Dynamic scanning with micro-PET was started at the time of [11C]MDDI injection. Cerebral blood flow of the rat was also measured.
RESULTS: Total synthesis time from EOB was 35 minutes. The radiochemical yield (EOB) based on [11C]carbon dioxide was 38.3±1.7% (n=2). The final product had a specific activity of 76 GBq/µmol at EOS. The radiochemical purity of [11C]MDDI was over 92%. Although the uptake of [11C]MDDI in the brain of normal rats was low, accumulation and washout of [11C]MDDI in the brain was visible using Micro PET scanner (Siemens, Chigago, USA) (see Figure 1). In addition, [11C]MDDI gave a distribution image which differed from that of cerebral blood flow obtained with intravenous [15O]water-PET.
DISCUSSION: Poly(ADP-ribose) polymerase (PARP) is a nuclear enzyme which is activated by DNA strand breaks and participates in DNA repair. Excessive PARS activation can deplete tissue stores of NAD+ and the depletion of NAD+, an important co-enzyme in energy metabolism, leads to cell death due to depletion of ATP. PARS inhibitors dramatically protect tissues from ischemic damage in focal cerebral ischemia and myocardial infarction. The foregoing Results: suggested that [11C]MDDI, a positron-labeled PARP inhibitor, may be useful for imaging ischemic injury regions in tissues with positron emission tomography (PET).
CONCLUSION: The measurement of [11C]MDDI uptake with PET may be useful for detecting PARP activity in ischemic injury region of brain.
Figure 1. Typical image of [11C]MDDI accumulation, and its time-activity curve in normal brain of a rat.
FLUORINE-18 LABELED DIHYDROTETRABENAZINE (DTBZ) DERIVATIVES FOR PET IMAGING OF VMAT2 IN THE BRAIN
Michael Kilbourn1, Brian Hockley1, Lihsueh Lee1, Catherine Hou2, Rajesh Goswami2, Datta Ponde2, Mei-Ping Kung2, Hank Kung2
1Department of Radiology, University of Michigan, Ann Arbor, MI, 2Department of Radiology, University of Pennsylvania, Philadelphia, PA, USA
Introduction:. PET imaging of the vesicular monoamine transporter 2 site using [11C]dihydrotetrabenazine ([11C]DTBZ) has proven useful in a number of neurologic and psychiatric diseases (1). Availability of a fluorine-18 labeled radioligand would allow more widespread application of this imaging method.
Methods:. (+)-[11C]Dihydrotetrabenazine was synthesized by established Methods:. The fluorine-18 substituted derivatives (9-(±)fluoroethyl, 9-(±)fluoropropyl, and 9-(+)fluoropropyl) were synthesized via appropriate mesylate precursors. Mouse brain distribution studies were done using tail vein injection and sacrifice at designated times afterward. Rat brain equilibrium distributions were done using a bolus+infusion approach (2). Rhesus PET imaging was done using a Concorde Microsystems P4 scanner (isoflurane anesthesia). Distribution volume ratios (DVR) were calculated from Logan plot analyses of tissue time-radioactivity curves, with cerebellum as reference region (3).
Results:. Studies in mouse brain (4) had shown that 9-fluoropropylDTBZ gave better striatum/cerebellum ratios than the fluoroethyl derivative, and was evident in the rat infusion studies (striatal DVR: fluoropropyl, 4.5 +/− 0.3; fluoroethyl 2.9 +/− 0.07). Studies were then focused on evaluation of (+)-FP-DTBZ, the high affinity stereoisomer (Ki values: (+)-isomer 0.10 +/− 0.01 nM, (−)-isomer >3000 nM).
Comparison of the striatal DVR values obtained from PET imaging of [11C]DTBZ (DVR = 4.5), (±)-[18F]FE-DTBZ (DVR = 2.5), (±)-[18F]FP-DTBZ (DVR = 5.1) and (+)-[18F]FP-DTBZ (DVR = 5.6) in monkey brain showed superiority of the (+)-isomer of the fluoropropyl derivative. Reversibility of binding of (+)-[18F]FP-DTBZ was demonstrated by administration of 2 mg/kg unlabeled tetrabenazine at 40 minutes, which produced a rapid clearance of radiotracer from the striatum (Fig. 2).
Conclusions:. (+)-[18F]FluoropropylDTBZ is an excellent radioligand for in vivo PET imaging of the VMAT2 site in the mammalian brain.
Acknowledgements. This work was supported by NIH grants NS-015655 (M.R. K.) and EB-002171 (H.F. K.).
RADIOSYNTHESIS OF [F18]FEOV AND IN VIVO PET IMAGING OF ACETYLCHOLINE VESICULAR TRANSPORTER IN THE RAT
Shadreck Mzengeza1, Gassan Massarweh1, Pedro Rosa1, Jean-Paul Soucy1, Marc-Andre Bedard2
1McConnell Brain Imaging Centre, MNI, McGill University, 2Department of Psychology, Universite Du Quebec A Montreal, Montreal, QC, Canada
We have recently achieved the radiosynthesis of [18F]fluoroethoxy-benzovesamicol ([F18]FEOV), basing our approach on that of Mulholland (Synapse 1998 (30) 263-274). However, the preparation of the precursor was a major obstacle, and in particular the resolution of the diastereoisomers of the Mosher camphor ester was challenging and of very low yield. The precursor was then labelled with [F18] using the GE Tracerlab module with good yield and high specific activity. We administered 12 MBq of that product to a Sprague-Dawley rat via tail vein injection, and imaged the animal with an R5 MicroPET. We observed high binding of [18F]FEOBV in areas of the brain with known high levels of cholinergic innervation, the most active areas being the basal forebrain, medial prefrontal cortex and the striatum, with much lower uptake in cerebellum and occipital cortex.
IMAGE QUALITY IMPROVEMENT OF BRAIN PET IMAGES BY A NEW RECONSTRUCTION ALGORITHM BASED ON MODIFIED ORDERED SUBSETS AND METZ FILTER
Jun-Cheng Lin1, Kuan-Hao Su1, Ren-Shyan Liu1,3, Jyh-Cheng Chen1
1Dept. of Biomedical Imaging & Radiological Sciences, National Yang-Ming University, Taipei, Taiwan, 3National PET/Cyclotron Center, Department of Nuclear Medicine, Taipei Veterans General Hospital, Taipei, Taiwan
Background and aims: PET can provide in vivo, quantitative and functional information for diagnosis. Iterative algorithms such as maximum likelihood expectation maximization (MLEM) algorithm are rapidly becoming the standard for image reconstruction in emission computed tomography. However, MLEM algorithm has very slow convergence speed because a large number of iterations may be required. The objective of our study was to develop a fast and better reconstruction algorithm.
Methods: First, we simulated rat brain microPET images. Second, we created a two-dimension digital noisy Shepp-Logan phantom of 256 × 256 matrixes, and we generated a lesion in the central part of the phantom. In part one, we developed a new algorithm called modified ordered subset expectation maximization (MOSEM) algorithm. In the algorithm, the number of projections in each subset among different iterations is variable to achieve fast convergence. We designed four kinds of MOSEM: Scheme A, Scheme B, Scheme C and Scheme D. Scheme A:8->8->8…; Scheme B: 4->8->16->32->96 …; Scheme C: 4->8->8->16->16->16->32…; Scheme D : 4->4->8->8->16… The first number in each scheme is the number of projections in each subset in first iteration, and so on. We note that Scheme A is equivalent to the regular OSEM. In part two, we further implemented and evaluated an advanced version of the MOSEM for microPET image reconstruction, which is called inter-update Metz filtered MOSEM (IMF-MOSEM). The IMF-MOSEM incorporates filtering action (The parameter is FWHM=1.8 mm and Metz power =3) into the image updating process to improve the quality of the reconstruction. In part one, we evaluated the best Scheme for MOSEM with rat brain microPET images. In part two, we used the IMF-MOSEM to achieve better image quality and evaluated contrast recovery (CR) and coefficient of variation (CV) value calculated with Shepp-Logan phantom.
Results: In part one, the convergence rate was evaluated with root mean square error (RMSE). We used four kinds of MOSEMs to reconstruct rat brain microPET images. The image would be accepted when the RMSE is small than the threshold. We measured the reconstruction time (second) for Scheme A (35.16), Scheme B (86.52), Scheme C (37.27) and Scheme D (20.51), respectively. Therefore, Scheme A and Scheme D were chosen for part two evaluation. In part two, image qualities were evaluated with CR and CV. The performance of IMF-MOSEMs were compared with ordinary OSEM, post-filtered OSEM, IMF-MOSEM (Scheme A) and IMF-MOSEM (Scheme D). The CRs (percentage) of ordinary, post-filtered OSEM, Scheme A and Scheme D were 130.1%, 139.6%, 148.1% and 152.2%, respectively. The CVs (percentage) of ordinary, post-filtered OSEM, Scheme A and Scheme D were 22.4%, 22.8%, 44.9% and 36.9%, respectively.
Conclusions: The quality of reconstructed images depends on the number of projections in a subset and it is better not to use less than four projections per subset. Furthermore, our Results: demonstrate that an appropriate choice of Metz filter parameters can improve the contrast-noise balance of certain regions of interest. Consequently, the reconstruction performed by IMF-MOSEM is fast, and has better image quantity in our simulated rat brain microPET image.