1. PVEOut project, EU IBB, 2004 Cortical GABA-A CBZR and rCBF in Alzheimer`s disease: SPET studies with Partial Volume Effect Correction Mario Quarantelli Biostructure and Bioimaging Institute – CNR Naples - Italy HBM2004 - PVEOut Satellite Meeting Budapest, 12 June 2004

2. PVEOut project, EU IBB, 2004 CBF The progression from Mild Cognitive Impairment (MCI) to Alzheimer’s Disease (AD) is characterized by the increase in severity and extension of cerebral hypometabolism and regional cerebral blood flow (rCBF) [Nestor 2003], already detectable in posterior cingulate, hippocampus and temporal neocortex in MCI patients. Concurrently, gray matter (GM) loss, essentially limited to hippocampal cortex in MCI [Chetelat 2002], progresses into a more widespread cortical loss. Chetelat G, et al. Neuroreport 2002; 13:1939-1943 Nestor PJ, et al. Ann Neurol 2003;54:343–351

3. PVEOut project, EU IBB, 2004 CBF It is not currently known weather this pattern of progression is characterized by a strict coupling of these two phenomena (i.e. GM loss explains the metabolism reduction as detected by FDG-PET and rCBF-SPET studies), or if metabolic impairment exceeds the rate of GM loss. Aim of our study was to compare rCBF decrease independently of GM loss in two groups of MCI and AD patients, using a ROI-based method for partial volume effect (PVE) correction which takes into account both WM and CSF.

4. PVEOut project, EU IBB, 2004 CBF - METHODS 12 MCI subjects (mean age 74.8 yrs, MMSE 27.8 range 24.9-30) 10 AD patients (mean age 79.5 yrs, MMSE 21.2, range 16.9-24) SPECT: –64-slice brain-dedicated SPECT –20-30 min after 740-1110 MBq of 99m Tc-HMPAO –120 steps, 3° angular step, 15 sec per step, 128 x 128 matrix, pixel size 1.673mm, Butterworth filter (cut-off = 0.9 cm -1, order=10), attenuation correction (0.120 cm-1) using Chang’s algorithm [Chang, 1987]. –Final FWHM 10.4mm (6mm axial) MRI: –Magnetization-prepared 3D T1-weighted fast-GrE images, TR/TE/TI 11/2/600ms, 1.5T, voxel size 0.98x0.98x1.2mm Probabilistic MRI segmentation [Ashburner J, 2000] and co- registration to SPECT studies [Friston KJ, 1995]. Ashburner J, Friston KJ, NeuroImage 2000; 11:805-21 Chang, L. IEEE Trans Nucl Sci 1987;25: 638-643 Friston KJ, et al. Hum Brain Map 1995; 2:165–189

5. PVEOut project, EU IBB, 2004 CBF - METHODS VOI including cerebral lobes and hippocampus for each side, and a single region for cerebellum and posterior cingulate, were defined in the MNI space and adapted to each co- registered segmented GM using normalization parameters derived from the SPM99 affine normalization matrix [Berkouk 2003]. For each VOI, uncorrected and PVE-corrected mean tracer concentrations were calculated [Rousset OJ, 1998; Quarantelli M, 2003] and normalized by corresponding cerebellum values. Comparison between MCI and AD groups was carried out for each region by Student's T-test. Significance level was set to P<0.05. Berkouk K, et al. 2003 HBM meeting; Abs. 1017 Quarantelli M, et al. J Nucl Med. 2004;45:192-201 Rousset OG, et al. J Nucl Med 1998;39:904-911

6. PVEOut project, EU IBB, 2004 77yo MCI

7. PVEOut project, EU IBB, 2004 77yo AD

8. PVEOut project, EU IBB, 2004 CBF - Methods

9. PVEOut project, EU IBB, 2004 CBL-normalized rCBF (AD-MCI) / MCI Mean percentage reduction in AD compared to MCI patients. Data are reported before (uncorrected) abd after (corrected) correctionfor partial volume effects. Significant differences (P<0.05) are in yellow.

10. PVEOut project, EU IBB, 2004 CBF – CONCLUSIONS Metabolic/CBF changes in AD are paralleled by GM loss [Baron JC, 2001a; Karas GB, 2003]. Accurate voxel-based comparisons of these changes have shown that the atrophy explains the GM hypometabolism with the exception of the posterior cingulate [Baron JC, 2001b], an area known to be affected precociously in AD [Minoshima S, 2000]. Our results integrate these findings, showing in PVE-corrected rCBF- SPET data from AD patients, as compared to MCI, a similar pattern of posterior cingulate involvement independent of GM loss. Reduced metabolism in these regions may be related to remote functional disruption and/or by larger neuronal loss than expected from structural changes. While larger cohorts are needed to confirm these preliminary findings, longitudinal studies are needed to confirm this pattern of progression of the disease. Baron et al, NeuroImage 2001b; 13:S771 Baron JC, et al. Neuroimage 2001a ; 14:298-309 Karas GB, et al. Neuroimage 2003; 18:895-907 Minoshima S, et al. Ann Neurol 1997, 42 :85-94

11. PVEOut project, EU IBB, 2004 Cortical GABA-A CBZR loss measured with 123I-Iomazenil in Alzheimer’s disease: a SPECT study with Partial Volume Effect correction Cortical synaptic/neuronal loss has been reported in neuropathological studies of patients with AD and may contribute to GM density changes described recently in AD using VBM analysis of structural MRI. PET/SPECT markers of GABA-A central benzodiazepine receptors (cBZR) have been proposed to detect in vivo early cortical synaptic/neuronal loss in AD. Only few studies have been reported with discordant results [Meyer M et al, Arch neurol 1995; 52:314 - Soricelli A, et al. Eur J Nucl Med 1996;23:1323]

12. PVEOut project, EU IBB, 2004 GABA-A CBZR Aim of the study was to investigate whether cortical synaptic/neuronal loss may be detected in vivo in AD using SPECT with 123I-Iomazenil and to assess how the measurement of these receptor changes is affected by partial volume effect (PVE). For this we used a voxel- based method for PVE-correction which takes into account both WM and CSF. Müller-Gartner HW, et al. J CBF Metab 1992;12:571-583 Quarantelli M, et al. J Nucl Med. 2004;45:192-201

13. PVEOut project, EU IBB, 2004 GABA-A CBZR

14. PVEOut project, EU IBB, 2004 GABA-A CBZR 5 AD patients (NINDS-ADRDA criteria; mean age 70 ± 11 yrs, MMSE: 19±4) 8 patients with MCI (74±6.5 yrs, MMSE: 27±2) 3 healthy volunteers (60±5 yrs, MMSE: 30) 20 minute SPECT acquisition, 180 min after injection –brain-dedicated camera (Ceraspect; 64 transaxial slices; voxel size 1.67 3 mm –Attenuation correction (Chang’s algorithm, attenuation factor 0.120 cm-1) These 'late images' were assumed to reflect essentially receptor binding. MRI: 1.5T, CSE sequences providing 32 T1w (600/15) and PD/T2w (2400/15- 90) 4mm-thick axial images; Multiparametric segmentation GM, WM and CSF maps were co-registered to SPECT studies. VOI including frontal, parietal, temporal, and occipital lobes, hippocampus and posterior cingulate cortex for each side, and a single region for cerebellum, was defined in the MNI space and adapted to each co-registered segmented GM using the SPM99 affine normalization matrix. Uncorrected and PVE-corrected mean tracer concentrations were normalized by corresponding cerebellum values. Two tailed Student's T-test on CBL-normalized values. Significance level was set to P<0.05.

15. PVEOut project, EU IBB, 2004 CBZR preliminary results Before PVE-correction AD vs NV reduction bilaterally in all cortical regions except the occipital and right frontal cortices and left hippocampus MCI vs NV reduction in the left posterior cingulate and a trend in the right posterior cingulate. AD vs MCI reduction of 123I-Iomazenil bilaterally in the parietal and temporal cortices and posterior cingulate and in the right hippocampus.

16. PVEOut project, EU IBB, 2004 CBZR preliminary results After PVE-correction AD vs NV reduction in the posterior cingulate bilaterally and in the left hippocampus MCI vs NV No significant differences AD vs MCI reduction in the posterior cingulate bilaterally

17. PVEOut project, EU IBB, 2004 * ## * # ** * # # # # # * ## * # ** * # # # # # * ## * # ** * # # # # # * ## * # ** * # # # # # 123-Iomazenil before PVE correction in AD, MCI and Controls (C). Significant reduction in AD vs Controls= * p<0.05: ** p<0.01 Significant reduction in AD vs MCI s= # p<0.05; ## p<0.01.

18. PVEOut project, EU IBB, 2004 123-Iomazenil after PVE correction in AD, MCI and Controls Significant reduction in AD vs Controls= * p<0.05: ** p<0.01 Significant reduction in AD vs MCI = # p<0.05; ## p<0.01

19. PVEOut project, EU IBB, 2004 These preliminary results provide two interesting observations: 1) In AD patients the GABA-A-cBZD receptors are significantly reduced as compared to Controls and MCI in associative cortical areas including the hippocampus. 2) This reduction survives the PVE- correction in some cortical areas particularly vulnerable in AD (posterior cingulate cortex and, to a lesser extent, the hippocampus). Overall these findings suggest that GABA-A-cBZR alterations in AD parallel but also exceed structural changes as measured by MRI segmentation. Further data are required to confirm these findings in larger groups of subjects CBZR - Conclusions