515. Preserved serotonergic projections but widely reduced brain serotonin 2A receptors in patients and in a mouse model of Alzheimer's disease
L. Marner1,2, A. Ettrup1,2, P.V. Jensen1,2, V.G. Frokjaer1,2, J. Kalbitzer1,2, S. Lehel3, W.F.C. Baaré2,4, S. Aznar1,2, G.M. Knudsen1,2 and S.G. Hasselbalch1,2,5
1Neurobiology Research Unit, University Hospital Rigshospitalet; 2Center for Integrated Molecular Brain Imaging; 3PET and Cyclotron Unit, Copenhagen University Hospital Rigshospitalet; 4Danish Research Center for Magnetic Resonance, University Hospital Hvidovre; 5The Memory Clinic, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark
Background and aims: Post mortem studies suggest serotonergic dysfunction in Alzheimer's disease (AD), and serotonin 2A (5-HT2A) receptors are globally reduced early in the disease.1 The objective of the study was to investigate serotonin transporters (SERT) as a measure of serotonergic projections2 in patients with AD and in a mouse model of amyloid plaque deposition.
Methods: We included 12 patients (mean age 73.7±7.6 years, 8 males) with Alzheimer's disease (average MMSE of 24, range 19 to 26) and 11 healthy age-matched subjects (mean age 72.5±6.8 years, 6 males). Subjects were investigated with a 90 mins dynamic [11C]DASB-PET to measure SERT and a 40 mins steady-state [18F]altanserin-PET to measure 5-HT2A receptors. Partial volume correction was applied to correct for atrophy.
For autoradiography, we included 42 mice, of which 21 were APP/PS1 transgenic mice with increased age-related amyloid load but no pronounced cell death. The mice were 4, 8, or 11 months old, corresponding to pre-, mid- and late-plaque stages. Using [3H]-escitalopram, we quantified levels of SERT in the hippocampus, and in the same mice, 5-HT2A levels were assessed in hippocampus, prefrontal and somatosensory cortices using [3H]MDL100907.
Results: In AD patients, SERT binding was 34% (P = 0.0003) lower in the hippocampus, while cortical regions and midbrain were unaffected (Figure 1). The 5-HT2A receptors were markedly reduced (25 to 66%) in AD patients in most regions (Figure 1). In the 8 and 11 month old transgenic AD mice SERT binding was normal in hippocampus and 5-HT2A binding in somatosensory and prefrontal cortices was decreased in 8 months (P = 0.06) and 11 months (P = 0.01) old mice by 7 to 19%.
Conclusions: We showed a marked decrease in 5-HT2A receptor binding in patients with AD and in the APP/PS1 mice. This suggests that the decreased 5-HT2A binding may be associated with amyloid deposition. The SERT binding was unaffected in most cortical regions and in midbrain, suggesting that the serotonergic projections and the neuron bodies in dorsal nucleus raphe are intact. The SERT reduction in hippocampus in patients could be secondary to the neuron death taking place in hippocampus of AD patients3 and not reported in the transgenic mice.
126. In vivo diagnosis of Alzheimer's disease by means of [18F]BAY 94–9172 and PET
H. Barthel1, J. Luthardt1, M. Patt1, E. Hammerstein2, G. Becker1, K. Hartwig2, A. Schildan1, S. Hesse1, P. Meyer1, J. Reischl3, U. Hegerl2, C. Reininger3, B. Rohde3, H.-J. Gertz2 and O. Sabri1
1Department of Nuclear Medicine; 2Department of Psychiatry, University of Leipzig, Leipzig; 3Bayer-Schering Pharma AG, Berlin, Germany
There is optimism that β-amyloid (Aβ) PET tracers may shift the time-point for accurate diagnosis of Alzheimer's disease (AD) from postmortem to antemortem. BAY 94–9172 is a new Aβ-binding 18F-labeled stilbene derivative currently in clinical development. Within the following the results of a proof of principle study investigating the ability of PET brain scanning with BAY 94–9172 to differentiate subjects with AD from healthy controls (HCs) are presented.
10 subjects with AD (DSM-IV and NINCDS-ADRDA criteria, age = 69±7 yrs, Clinical Dementia Rating (CDR) score = 1 to 2, Mini-Mental State Examination (MMSE) score = 19±7, ApoE4: 6 positive), 10 age-matched healthy controls (HCs, MMSE score ≥28, CDR score = 0, ApoE4: 2 positive), as well as 3 patients with fronto-temporal dementia (FTD, Neary consensus criteria as well as MRI and/or [18F]FDG findings compatible with FTD, age = 59 to 78 yrs, MMSE score = 9 to 14, CDR score = 1 to 2, ApoE4: 1 positive) were included. Dynamic brain PET images (ECAT EXACT HR+, ∼300MBq [18F]BAY 94–9172 i.v., 3D mode, Neuroshield, correction for measured transmission, filtered back-projection) were acquired. The 70 to 90 mins post injection images were analyzed: (1) visually by three blinded experts, (2) by volume of interest (VOI) analysis: VOIs anatomically defined via MRI (1.5 Tesla, 3D T1w volumetric MPRAGE sequence) co-registration, resulting standard uptake value (SUV) ratios using the cerebellar cortex as reference region, (3) via voxelwise, single-case comparisons with a database created with the PET data from the 10 HCs (SPM2, cluster size >30 voxel, P<0.001).
Visual analysis revealed 9 of the 10 AD, but only 1 of the 10 HC brains to be positive for Aβ (P<0.001). In FTD, 2 of the 3 patients were rated positive for Aβ. For all subjects, the frequency of ApoE4-positive genotypes was higher in the Aβ-positive compared with the Aβ-negative PET scans (P = 0.05). When compared to these of the HCs, the SUV ratios in AD subjects were significantly higher in frontal (1.73±0.37 versus 1.35±0.18, P = 0.0098), anterior cingulate (1.72±0.36 versus 1.41±0.15, P = 0.022), posterior cingulate (1.79±0.48 versus 1.36±0.13, P = 0.012) and lateral temporal (1.49±0.30 versus 1.24±0.10, P = 0.020) cortices, as well as in caudate heads (1.49±0.30 versus 1.19±0.13, P = 0.010). Voxelwise analyses were in complete agreement with the results of the visual rating. Here, a newly developed Aβ load index combining information on extent and severity of pathological tracer uptake was negatively correlated to the MMSE score (r = −0.609, P = 0.004).
Typical [18F]BAY 94–9172 PET images in AD and HC.
These results indicate that [18F]BAY 94–9172 PET scanning has the potential to non-invasively provide accurate and quantitative measures of Aβ load during life. As such, this new tracer may facilitate an antemortem diagnosis of AD. Further evaluation of the diagnostic efficacy of [18F]BAY 94–9172 in global, multi-center trials is ongoing.
This trial is sponsored and supported by the Bayer-Schering Pharma AG.
38. Identification of human ‘non-binders’ and measurement of specific binding with [11C](R)-PK 11195: implications for PET imaging of brain inflammation
W.C. Kreisl, Y. Fujimura, N. Kimura, J.S. Hong, C.L. Morse, S.S. Zoghbi, R.L. Gladding, V.W. Pike, M. Fujita and R.B. Innis
Molecular Imaging Branch, National Institute of Mental Health, Bethesda, Maryland, USA
Objective: Ten percent of subjects scanned with [11C]PBR28 show unusual behavior in that the radioligand is unable to bind to the 18 kDa translocator protein (TSPO, formerly known as the peripheral benzodiazepine receptor) in brain and in peripheral organs.1 ‘Non-binders’ have not previously been reported using the older radioligand [11C](R)-PK 11195, possibly because of low specific signal. We measured the displaceable (i.e., specific) binding of [11C](R)-PK 11195 in monkey brain to compare to reported values for [11C]PBR28. We also sought to determine if [11C](R)-PK 11195 could distinguish binders from non-binders in human subjects.
Methods: Monkey brain study: Two rhesus monkeys had dedicated brain PET imaging with [11C](R)-PK 11195, both at baseline and after TSPO-blockade using 5 mg/kg of non-radiolabeled PBR28. Total brain activity was measured by calculating total distribution volume (VT) in each scan using an unconstrained two-tissue compartment model and metabolite-corrected arterial input function. Human whole-body study: Eight healthy human subjects received whole body PET imaging with both [11C](R)-PK 11195 and [11C]PBR28. Four subjects had previously been identified as non-binders with [11C]PBR28. Concentration of radioactivity in organs that express TSPO was measured and corrected for body weight and injected dose. The area under the time-activity curve from 5 to 20 mins was used to measure uptake of radioligand in tissue.
Results: Monkey brain study: VT values for [11C](R)-PK 11195 decreased by 52% after pre-treatment with unlabelled PBR28. Specific binding (VS), calculated as the difference in VT at baseline and blocked conditions, was 57-fold lower than that reported for [11C]PBR28.2 Human whole-body study: In the [11C]PBR28 scans, non-binders had less uptake than binders in brain, heart, lung, kidney, and spleen (P<0.001). In the [11C](R)-PK 11195 scans, non-binders had less uptake (SUV·min) than binders in the heart [42.6±3.0 (SD) versus 52.8±2.8, P<0.03] and lung (16.2±2.5 versus 23.1±2.8, P<0.03). However, [11C](R)-PK 11195 could not distinguish binders and non-binders in the brain, kidney, or spleen. Specific binding of [11C]PBR28, operationally defined as the difference in organ uptake between normal subjects and non-binders, was several-fold higher than that of [11C](R)-PK 11195 in the brain (41-fold), heart (4-fold), lung (4-fold), kidney (15-fold), and spleen (8-fold).
Conclusions: In monkey brain, the specific binding of [11C](R)-PK 11195 was 57-fold lower than that of [11C]PBR28. In humans, [11C]PBR28 clearly distinguishes binders from non-binders in each organ studied. We cannot determine if non-binding is a phenomenon shared by [11C](R)-PK 11195, since this radioligand distinguishes non-binders only in the heart and lung, which may be a result of multiple comparisons or of high amounts of nonspecific binding of this ligand. However, [11C]PBR28 has much greater specific binding in monkey and man than [11C](R)-PK 11195 and [11C]PBR28 is preferred in the 90% of human subjects who display usual receptor kinetics.
355. Preclinical predictors of performance of amyloid-imaging tracers in humans
J.C. Price1, W.E. Klunk2, N.S. Mason1, B. Lopresti1, M.L. Debnath2, M. Berginc1, W. Bi1, S.T. Dekosky3,4 and C.A. Mathis1
1Radiology; 2Psychiatry; 3Neurology, University of Pittsburgh, Pittsburgh, Pennsylvania, USA; 4Neurology, University of Virginia, Charlottesville, Virginia, USA
Objectives: Desirable properties for amyloid-imaging tracers include good specific retention and rapid clearance of nonspecific binding (given sufficient initial brain entry). The difficulties involved in testing tracers in humans necessitate a robust preclinical protocol for predicting success in humans. Here we show the utility of measurements of in vitro binding affinity (1/Kd) and nonspecific clearance in animals for predicting relative performance of two amyloid-imaging tracers: PiB and a new [F-18]-labeled derivative, 2′F-PiB.
Methods: Two controls (CN: 75 yrs, 66 yrs, MMSE = 30) and 1 Alzheimer's disease (AD: 57 yrs, MMSE = 28) subject were studied with PET (HR+ 10 to 15 mCi) over 90 (PIB) or 120 (2′F-PiB) min (no blood sampling). Anesthetized baboons were studied similarly (PiB: n = 8; 2′F-PiB: n = 4). Tissue uptake (SUV) and tissue ratios (SUVR) were computed (40 to 90 mins) for frontal cortex (FRC) and cerebellum (CER: nonspecific uptake reference region). Computer simulations were performed to simulate AD time-activity curves for PiB and 2′F-PiB. Simulations utilized a legacy1 PiB input function for an ‘average’ AD subject and average (n = 9) 2-tissue compartment AD PiB model parameters for FRC (K1 = 0.254 mL cm−3 min−1, k2 = 0.122 min−1, k3 = 0.049 min−1, k4 = 0.017 min−1) and CER (K1 = 0.283 mL cm−3 min−1, k2 = 0.145 min−1, k3 = 0.012 min−1, k4 = 0.016 min−1). Experimental values for in vitro Kd of 1.4 nmol/L (PiB) and 5.5 nmol/L (2′F-PiB) and the human PiB k4 values (above) were used to estimate 2′F-PiB k4 values.
Results: The 2′F-PiB k4 values for AD FRC and CER were estimated to be 0.067 and 0.063 min−1, respectively. The simulations predicted an AD PiB FRC SUV value that was 2.0-fold greater than that predicted for 2′F-PiB. This suggested that the specific retention of 2′F-PiB would be much lower than that of PiB. However, differences in nonspecific clearance could offset this effect when tissue ratios are calculated. This was examined via simulations for CER and in vivo baboon imaging data. The simulations predicted a PiB CER SUV value 1.6-fold higher than that for 2′F-PiB. This was consistent with PET data acquired in amyloid-free baboons for which the experimentally determined PiB CER SUV was 1.6-fold higher than that measured for 2′F-PiB. These results suggest that the PiB FRC SUVR would be ∼1.3-fold greater that of 2′F-PiB (i.e., 2.0/1.6). The findings in humans were very consistent with these predictions. PET studies in the AD patient resulted in a PiB FRC SUV value that was 2.3-fold higher than that measured for 2′F-PiB, while the PiB CER SUV was 1.6-fold higher (and similar across all 3 subjects: 1.7±0.2). PET studies in the AD subject also yielded a PiB FRC SUVR value of 3.1, a value that was 1.4-fold higher than the 2′F-PiB FRC SUVR value of 2.2.
Conclusions: This comparison of experimentally-driven Kd simulations with animal studies of nonspecific clearance was robust relative to human in vivo PET data results and shows the value of experimentally-determined affinity and clearance in predicting in vivo performance of amyloid-imaging tracers in humans.
754. Evaluation of reference models for [11C]ABP688 targeting the metabotropic glutamate receptor 5 in rats—application to an epilepsy model
D. Elmenhorst1,2,3, G. Biagini1,4, L. Minuzzi3, A. Aliaga1, G. Massarweh1, M. Diksic1, M. Avoli1, A. Bauer2 and P. Rosa-Neto3
1Montreal Neurological Institute, McGill University, Montreal, QC, Canada; 2Institute for Neurosciences and Biophysics INB-3, Research Center Juelich, Jülich, Germany; 3Translational Neuroimaging Laboratory, Douglas Hospital, McGill University, Montreal, QC, Canada; 4Department of Biomedical Sciences, University of Modena and Reggio Emilia, Modena, Italy
Objectives: An increase of metabotropic glutamate receptors type 5 (mGluR5) immunoreactivity was found in hippocampal specimens from patients suffering from pharmaco-resistant temporal lobe epilepsy. Further, pharmacologically induced myoclonic seizures could be suppressed by the systemic application of mGluR5 antagonists in mice. Based on this evidence, the primary objective of the present study was to determine the changes in receptor density of mGluR5 possibly occurring in the pilocarpine model of epilepsy in rats. Reference tissue methods for [11C]ABP688 were evaluated by in-vivo blocking experiments in control rats in order to avoid invasive blood collection procedures for quantifying receptor changes and to allow repeated scans in the same animal.
Methods: In order to evaluate if the rat cerebellum is a suitable reference region, 2 Sprague-Dawley (SD) rats were scanned twice in a microPET R4 scanner under isoflurane anesthesia for 60 mins. Following a baseline scan, the selective antagonist M-MPEP (6 mg/kg) was administered before the 2nd scan to block specific binding. Twelve arterial blood samples were drawn per each scan and corrected for metabolites. To compare changes in binding potential (BPnd) between control (n = 9) and chronically epileptic (n = 7) animals, SD rats were scanned for 60 mins without blood sampling. The simplified reference tissue model (SRTM) and the 2 parameter version of the multi-linear reference tissue model (MRTM2) were used in a ROI based and voxel wise manner, respectively. The average time delay between pilocarpine induction of epilepsy and scanning was 88 days.
Results: Total distribution volumes (Vt) of cerebellum changed after M-MPEP blockade on average +1% and −7% using 2 tissue compartment modeling (2TCM) and Logan's non-invasive graphical analysis (NIGA), respectively. Binding in caudate-putamen was reduced by −66% and −67% to a Vt of 3.2 to 4.2 mL/mL (for comparison: cerebellum Vt: 2.9 to 3.4 mL/mL) for 2TCM and NIGA, respectively. BPnd based on 2TCM correlated well with BPnd derived from various reference models (SRTM, MRTM and its 2 parameter versions, NIGA). Time stability evaluation showed that BPnd and Vt determinations were stable after 50 mins of scanning. Chronic epilepsy did not reveal significant changes in the ROI based analysis in the present sample. In the cingulate cortex—prelimbic cortex area, BPnd increased by 11% increase showing a trend towards significance (t-test, P = 0.11). Voxel wise statistical analysis showed significant clusters with decreased BPnd on the posterior ventral part of the hippocampus and with increased BPnd in mesial ventral frontal cortex.
Conclusions: The cerebellum is a suitable reference region for the quantification of mGluR5 availability with [11C]ABP688. Blood sample based quantification of BPnd correlates well with reference region based analyses. Scan duration of 50 mins is required. The observed increase in receptor density in humans (hippocampal specimens) suffering from temporal lobe epilepsy was not observed in the pilocarpine model, but the stastical power has to be increased to confirm these results. The present results in the pilocarpine model contrast with the immunohistochemical studies in humans but are in agreement with receptor binding studies conducted by our group.
