Document Sample
ScahillRI2001Novelmet Powered By Docstoc
					Neurobiology of Aging 22 (2001) 333–344

Neuroimaging for detection, diagnosis, and monitoring of Alzheimer’s disease and other dementias: proceedings of the Alzheimer’s Imaging Consortium
Edited by M. Weiner, C. DeCarli, M. deLeon, N. Fox, C. Jack, P. Scheltens, G. Small, Z. Katchaturian, R. Peterson, and L. Thal. The Institute for Research on Aging, Director Dr. Howard Fillit, provided a grant to support this meeting. Administrative support was provided by Karen Blackwell and Daphne Dunn. Their support is very gratefully acknowledged. decline in gray matter, white matter, and total brain volume, as well as increases in CSF volume were demonstrated in MRI scans obtained at intervals of 2 and 4 years. Adalsteinsson et al performed magnetic resonance spectroscopic imaging (MRSI) studies in a group of 12 AD patients and 14 controls at two points in time separated by one year. Using anatomic MR images to factor out partial volume averaging effects in each MRSI voxel, the authors computed the rate of change of N-acetyl-aspartate (NAA) in cortical gray matter. A significantly greater rate of NAA loss was identified in AD patients vs controls. In summary, the four presentations in this session demonstrated that rates of longitudinal change in a variety of imaging measures (functional, regional anatomic, global anatomic, and biochemical) can be made in vivo with a reasonable degree of precision. Each of these imaging approaches shows the potential to serve as a surrogate measure for therapeutic efficacy in drug trials.



Weiner M. The Alzheimer’s Imaging Consortium (AIC) held a one day symposium on July 8 2000, just prior to the onset of the World Alzheimer’s Congress. The AIC is an informal organization, open to anyone who wishes to participate. Our goal is to promote and facilitate use of imaging techniques for detection, diagnosis, investigator, and monitoring or dementia. This publication includes all of the abstracts submitted for this meeting, as well as brief summaries of each session, provided by the session chairs. Taken together, there was a general consensus that imaging, especially nuclear and MR techniques, is occupying an increasingly important role in assessment of dementia. With growing interest in pharmaceutical trials of drugs aimed to slow the progression of Alzheimer’s, imaging techniques appear to have practical use for assisting with diagnosis, staging, and monitoring treatment effects.



Reiman EM, Caselli RJ, Chen K, Alexander GE, Bandy D, Frost J, Gonzalez-Lima F. Dept Psychiatry, Univ Arizona, Depts Psychology and Mathematics, Arizona State Univ, Dept Psychology and Inst Neurosci, Univ Texas at Austin, Dept Neurology, Mayo Clinic Scottsdale, PET Center, Harrington Arthritis Research Center, Good Samaritan Regional Medical Center, and the Arizona Alzheimer’s Disease Research Center, Phoenix, AZ, USA Using fluorodeoxyglucose (FDG) positron emission tomography, we previously found that late middle-aged, cognitively normal carriers of the apolipoprotein E ε4 allele, a common Alzheimer’s susceptibility gene, have abnormally low brain activity in the same brain regions as patients with Alzheimer’s dementia, the largest of which is in posterior cingulate cortex. We now find that the ε4 heterozygotes have significant two-year declines in regional brain activity and that these reductions are significantly greater than those in the ε4 non-carriers. Our studies suggest that candidate prevention therapies could be tested after only one or two years of treatment using small samples of cognitively normal, late-middle-aged ε4 heterozygotes. Using FDG autoradiography, we find that aged PDAPP transgenic mice have preferentially and progressively reduced activity in the posterior cingulate cortex and relative sparing in visual, sensorimotor, cerebellar, and white matter regions. The reduction in posterior cingulate activity could be used to track disease progression, help clarify disease mechanisms, and screen candidate treatments in these and other suitable laboratory animals. Functional brain imaging techniques could be used to track the progression of Alzheimer’s disease in the absence of symptoms, help screen candidate treatments in transgenic mice, and test candidate prevention therapies without having to study thousands of individuals or wait many years to determine whether or when they develop symptoms.

Session I 2

Fox N, Jack C. Eric Reiman described results from on going studies of fluorodeoxyglucose (FDG) PET in AD. This group has previously shown that late middle-aged cognitively normal individuals who are APOE 4 homozygotes display a pattern of decreased glucose uptake in the brain that is similar to AD patients, and significantly different from age-matched individuals who do not carry an 4 allele. In the current presentation Reiman et al report on the results of serial imaging studies with FDG PET in 4 heterozygotes. They found that 4 heterozygotes demonstrate declines in glucose uptake in various regions of the brain which are significantly greater than that found in 4 non-carriers. Reiman et al also described the results of FDG autoradiography studies in PDAPP transgenetic mice. The autoradiographic studies demonstrate preferentially reduced glucose uptake in the posterior cingulate cortex (similar to this group’s findings in human APOE 4 homozygotes) with relative sparing of primary sensory motor areas. Scahill et al compared the rates of hippocampal atrophy derived from serial MRI scans in 15 AD patients and 15 controls using three methods: 1) traditional manual outlining; 2) pasting the baseline region derived from manual segmentation onto the registered repeat scan, followed by manual editing of the repeat scan; 3) fluid registration, (nonlinear 3D matching which propagates the baseline region onto the repeat scan). The authors found that methods two and three produced calculated rates of volume loss that were more or less equivalent to that derived from the traditional manual editing approach. In addition, however, methods two and three demonstrated superior reproducibility. Resnick et al described a method for measuring the change in global brain and CSF volume over time from serial MRI studies. In addition, selective rates of atrophy for both the gray matter and the white compartments were computed. Significant rates of



Scahill RI, Crum WR, Fox NC. Dementia Research Group, Department of Clinical Neurology, Institute of Neurology, 8-11 Queen Square, London WC1N 3BG, United Kingdom This study compares two novel methods for hippocampal segmentation with conventional manual segmentation. A group of 15 Alzheimer’s disease (AD) patients and 15 age- and sex-matched controls underwent two volumetric T1-weighted MR scans. The right hippocampus was segmented on the baseline scan using intensity thresholding and manual tracing. The scans were randomised and the operator was blind to the individual’s details. Three methods were applied to outline the right hippocampus on the repeat scans: 1. manual segmentation;

0197-4580/01/$ – see front matter © 2001 Elsevier Science Inc. All rights reserved.

2. pasting the baseline region onto the registered repeat scan, followed by manual editing; 3. fluid registration—voxel-level non-linear three-dimensional matching which propagates the baseline region onto the repeat scan. Both novel methods produced volume differences that were equivalent to or better than the reproducibility errors associated with manual segmentation. The coefficient of variation of manual segmentation was 0.026, with a mean absolute volume difference of 94 ( 103) mm3. The mean hippocampal volumes were 2290 ( 589) mm3 in the AD group and 2753 ( 304) mm3 in the controls. The mean absolute difference between manual segmentation and the alternative methods was 76 ( 60) mm3 for pasting and 99 ( 67) mm3 for fluid registration, less than 4% of hippocampal volume. Manual segmentation is used widely in studies on hippocampal atrophy, but is time-consuming and subject to observer error. The novel methods described produce volumes which are consistent with manual segmentation, yet greatly reduce operator time. Both methods have greater reproducibility than manual segmentation, and fluid registration has the additional benefit of being fully automated and unbiased, thereby reducing subjectivity.

Session II
neuronal integrity with AD progression and possibly of improvement in neuronal function with treatment.

Session II 7

deLeon M, Small G. Accuracy of a clinical diagnosis of dementia can vary from as low as 55 percent to as high as 90 percent, depending on the experience of the diagnosticians and the population being studied [1–3]. A critically important application of modern brain imaging techniques is in adding value to clinical diagnosis and differential diagnosis of dementia. Each imaging technique provides a signal of brain structure or function or both; the challenge is to produce a signal with biological meaning and limited noise. This group of presentations offers a glimpse at several technologies and how each investigative team is addressing the methodological challenges to diagnosis. Magnetic resonance spectroscopy has generated considerable enthusiasm in recent years, with its advantage of providing quantitative regional measures of biochemical and physiological processes. Schuff and associates used proton MR spectroscopic imaging (1H MRSI) to show that 1H MRSI detects regional patterns of N-acetyl aspartate (NAA) loss in Alzheimer’s disease, subcortical vascular dementia, and mild cognitive impairment consistent with known neuropathology. Functional MRI provides measures of signal intensity that are associated with relative cerebral blood flow during memory or other cognitive tasks. In their functional MRI studies, Sperling and associates suggest that activation patterns may distinguish normal aging from early Alzheimer’s disease, but they acknowledge the methodological challenges of using this application in patients with dementia, including atrophy correction, motion artifact, and monitoring task performance. A large number of studies focus on cortical brain volume as a diagnostic aid, and the study of Clark and co-workers combines such anatomical measures with cerebrospinal fluid measures of tau and -amyloid. Their findings suggest a relationship between lower cortical brain volume and higher levels of CSF tau. The work of Mega and colleagues also uses structural MRI data to construct an average Alzheimer’s disease atlas from 36 patients incorporating 60 manually defined subvolumes of interest. Their probabilistic atlas can be used for both structural and functional imaging analysis. Rombouts and co-workers report on their structural measures of the whole brain and note unexpected areas of gray matter volume loss, including symmetric atrophy in the insula and the caudate nucleus. The challenge for investigators and clinicians is to determine the utility of a particular technique or combination of techniques for different diagnostic questions. The emphasis, to date, has been on early differential diagnosis of dementia so that effective symptomatic treatments can be applied. As the field continues to emerge, applications in preclinical stages of age-related memory loss, as well as more advanced disease stages when behavioral disturbances are more common, will be additional areas wherein these techniques may assist in differential diagnostic and treatment decisions. REFERENCES [1] Mayeux R, Saunders AM, Shea S, et al. Utility of the apolipoprotein E genotype in the diagnosis of Alzheimer’s disease. N Eng J Med 1998;338:506 –11. [2] Ryan DH. Misdiagnosis in dementia: comparisons of diagnostic error rate and range of hospital investigation according to medical specialty. Int J Geriatr Psychiatry 1994;9:141–7. [3] Small GW, Rabins PV, Barry PP, Buckholtz NS, DeKosky ST, Ferris SH, Finkel SI, Gwyther LP, Khachaturian ZS, Lebowitz BD, McRae TD, Morris JC, Oakley F, Schneider LS, Streim JE, Sunderland T, Teri LA, Tune LE. Diagnosis and treatment of Alzheimer disease and related disorders: consensus statement of the American Association for Geriatric Psychiatry, the Alzheimer’s Association, and the American Geriatrics Society. JAMA 1997;278:1363–71.



Resnick Sma, Davatzikos Cb, Kraut Ma2, Zonderman Aba. aNational Institute On Aging, Bethesda, MD, USA bJohns Hopkins University, Baltimore, MD, USA Recent stereological investigations have indicated that there is little neuron loss and only modest decreases in neocortical volume with aging. These findings contrast with numerous reports from cross-sectional neuroimaging studies, which reveal greater atrophy in older compared with younger individuals. As both the post-mortem and imaging studies have been limited by their cross-sectional nature, we are performing a longitudinal brain imaging study in older participants in the Baltimore Longitudinal Study of Aging (BLSA). High resolution magnetic resonance imaging (MRI) was performed annually for 94 nondemented individuals (52 men, 42 women; mean age 70.3 yrs.), who completed 5 annual assessments. Volumetric MRI scans obtained at Years 1, 3 and 5 provided measures of total brain, gray, and white matter volumes, and ventricular cerebrospinal fluid (V-CSF). Analysis of 2-year and 4-year change revealed significant longitudinal decreases in both gray and white matter volumes, as well as increases in V-CSF. The mean annual rates of change for the 94 individuals were (in cm3): total brain (gray white) 5.49, gray 2.44, white 3.05, and V-CSF 1.39. Rate of change for the V-CSF, but not gray and white matter volumes, was greater in older individuals. These findings demonstrate substantial longitudinal decreases in brain volume in older adults over intervals as short as 2 and 4 years and provide a background against which pathological brain changes can be identified.



Adalsteinsson Ea, Sullivan EVb, Kleinhans Nb, Spielman DMa, Pfefferbaum Ac. Departments of aRadiology, Stanford University School of Medicine, Stanford, CA, USA, bPsychiatry and Behavioral Sciences, Stanford University School of Medicine, Stanford, CA, USA, cNeuropsychiatry Program, SRI International, Menlo Park, CA, USA. Alzheimer’s disease (AD) patients (7 men, 5 women, 74 7 years; NINCDSADRDA criteria) and healthy elderly volunteers (7 men, 7 women, 73 4 years) underwent MRSI (TE/TR 144/2000) and MRI (FSE, TE1/TE2/TR 20/80/3000) scanning and neuropsychological testing, at baseline and one year later. Separate N-acetyl-aspartate (NAA) measures were estimated for the gray and white matter brain compartments with the help of segmented FSE images. Repeated measures analysis of variance (ANOVA) revealed a significant group effect (p .002) and interaction (p .025), where the gray matter NAA in AD was lower than that of the controls at both times and declined significantly over time in AD but not in controls. This group difference in gray matter NAA concentration was also reflected in a comparison of percent change/year (mean SE: AD 12.36 4.29; control .94 2.18), which accounted for individual differences in interscan interval. ANOVA effects were not significant for white matter NAA. In contrast to NAA, structural gray matter volumes showed only nonsignificant declines (p .06, one-tailed) in the AD relative to the control group, although the AD group had significantly smaller gray matter volumes than the control group at both scan times (p .01 for each time). This longitudinal study provides evidence that decline in NAA gray matter concentration is sensitive to neuronal and functional deterioration in mild to moderate AD over a one-year interval. Thus, NAA gray matter concentration may be a suitable metric for the detection and monitoring of deterioration of



Schuff N, Amend A, Du AT, Jagust W, Chui H, Yaffe K, Weiner MW. VA Medical Center and University of California, San Francisco, CA, USA Objectives: 1) To compare the concentration of N-acetyl aspartate ([NAA], a marker of neuronal integrity) in the cortex and hippocampus of patients with Alzheimer’s disease (AD), subcortical vascular dementia (SIVD), and mild cognitive impairments (MCI) with normal aging; 2) To demonstrate that [NAA] loss measured with proton magnetic resonance spectroscopic imaging (1H MRSI) together with volume loss measured with MRI improves classification of AD and SIVD.

Session II

23–33), ten healthy elderly subjects (ES, age 67– 88), and seven patients (AD, age 71– 89) meeting NINDS-ADRDA criteria for probable AD with mild dementia severity (MMSE 20 –25) were studied. Whole brain EPI-BOLD fMRI was performed with a 3T GE Scanner, while subjects were shown photographs of unfamiliar faces paired with common first names in blocks of novel and repeated face-name pairs. Encoding of novel face-name associations produced robust paradigm-linked activation (p0.00001) in the fusiform gyri, hippocampal formation (HF), superior parietal and dorso-lateral prefrontal cortices. YS and OS showed similar activation in the fusiform gyri and HF, but differed in the pattern of activation in superior parietal and prefrontal regions. AD patients showed significant activation in fusiform regions, but did not show paradigm-linked activation in the HF. Our laboratory has also obtained very similar results with a different group of healthy older subjects and AD patients using a novel scene encoding paradigm. Our preliminary findings suggest that fMRI activation patterns may distinguish normal aging from early AD, however, there are multiple challenges still facing the clinical application of fMRI in the AD population, including motion artifact, atrophy correction, and monitoring task performance.

[NAA] Frontal Cortex Left Right Parietal Cortex Left Right Hippocampus Left Right





0.80 0.82 0.72 0.79 1.21 1.20

0.14 0.16 0.22 ( 17)† 0.17 ( 10)* 0.40 0.40 ( 15)*

0.69 0.73 0.57 0.67 1.23 1.45

0.11 ( 18)† 0.21 0.13 ( 34)† 0.25 ( 24)* 0.39 0.57

0.79 0.80 0.76 0.77 1.28 1.22

0.15 0.15 0.17 ( 13)† 0.16 ( 12)† 0.39 0.25 ( 13)†

0.84 0.85 0.87 0.88 1.28 1.42

0.16 0.14 0.17 0.20 0.40 0.49

* p .05; † p .001; In parentheses are percentage differences to HC; [NAA] in arbitrary units (notice, higher hippocampal [NAA] compared with cortical [NAA] is primarily an artifact of different experimental parameters).

Methods: 1H MRSI were performed on 41 AD, 11 SIVD, 35 MCI, and 52 cognitive normals (CN) of comparable age in addition to volumetric MRI. Areas under the curve of the receiver operator characteristic (AUC-ROC) were evaluated to quantify the classification power of MRSI and MRI. Results: The table lists [NAA] by region and group. AD when compared with CN had [NAA] losses in the left and right parietal cortex and right HP, but normal [NAA] in the other regions. SIVD when compared with CN had [NAA] losses in the left frontal cortex and bilaterally in the parietal cortex. In contrast to AD, however, SIVD showed no [NAA] losses in HP. MCI when compared with CN had [NAA] losses in the parietal cortex and right HP, similar to AD. Finally, AUC-ROC values of an AD or SIVD classification using global atrophy or HP volume measurements from MRI increased by up to 20% when cortical [NAA] was included in the analysis. However, HP [NAA] did not substantially improve classification. Conclusion: 1H MRSI detects regional patterns of [NAA] loss in AD, SIVD, and MCI that are consistent with the known distributions of neuropathology. Furthermore, [NAA] losses provide complementary information to MRI measured volume losses, which may aid the early diagnosis of dementia.



Clark CMa,b,f, McGowan Jh, Moonis Gh, Lee Vd,i, Sturm Mj, Ewbank Dc, Karlawish JHTa,b,e,g, Trojanowski JQa,b,d,i, DeCarli Cj. aAlzheimer’s Disease Center, bInstitute On Aging, cPopulation Center, dCenter For Neurodegenerative Disease Research, eCenter For Bioethics, and the Departments of fNeurology, gMedicine, hRadiology and iPathology, University of Pennsylvania and the jDepartment of Neurology, University of Kansas Cortical brain volume and cerebrospinal fluid (CSF) measurements of tau and -amyloid have been proposed as pathologic markers of Alzheimer’s disease (AD). The association between these anatomic and biochemical measures of neurodegeneration has yet to be investigated. To do this, we evaluated the correlation between these candidate biomarkers using data from a series of 13 patients with a clinical diagnosis of Alzheimer’s Disease (average age at symptom onset 71, average duration of symptoms at the time of brain volume measurement 4 years). In all patients the CSF tau levels were elevated (mean value 808 pg/ml, range 296 –2119), the average percent -amyloid 1– 42 was reduced (mean value 5.4%, range 1.3%– 8.5%) and brain volume (expressed as percent of cranial volume) was reduced (mean value 71.2%, range 66.6%–75.7%). Increasing levels of CSF tau were associated with decreasing cortical brain volume (rs 0.69, p .001). There was no association between cortical brain volume and the CSF percent -amyloid 1– 42 (rs 0.38, p 0.35). These findings suggest a logical relationship between decreasing cortical brain volume and increasing CSF tau levels. This adds to the evidence that these measures are biomarkers that mark a common pathological endpoint in AD neurodegeneration.



Sperling Ra,b, Bates Jb, Cocchiarella Aa, Rentz Da, Schacter Db, Rosen Bb, Albert Mb. Department of Neurology, Brigham and Women’s Hospital, Boston, MA, USA, b NMR Center, Massachusetts General Hospital, Boston, MA, USA Functional MRI is a relative newcomer to the growing field of neuroimaging markers for Alzheimer’s disease (AD). FMRI may be particularly useful in examining the neuroanatomic substrate subserving complex memory processes, and alterations in the patterns of functional activation during memory tasks that occur in normal aging and in early AD. Eight healthy young subjects (YS, age




Session III


Session III 13

Mega MSa,b, Thompson PMa, Dinov IDa, Toga AWb, Cummings JLb. aLaboratory of Neuro Imaging and bAlzheimer’s Disease Research Center Department of Neurology, UCLA School of Medicine, Los Angeles, CA, USA Objective: We develop, evaluate and validate a probabilistic atlas for the elderly and demented brain. Background: Developments [1] in brain mapping instigated construction of probabilistic sub-volume population atlases (SVPA). Disease specific SVPA’s overcome the structural mismatch confounding most multisubject imaging studies in aging and dementia. After construction of a continuum mechanical average atlas [2] manually outlined regions on individual brains, registered to that atlas, produce stochastic probabilistic sub-volumes encoding anatomic, registration, and functional variability. Methods: SPGR MRIs of 36 Alzheimer’s disease (AD) subjects were used to construct an average AD atlas incorporating 60 manually defined subvolumes of interest (SVIs). Two probabilistic representations were constructed—a linear-SVPA based on 12-parameter registration [3] and a nonlinear-SVPA based on 6th-order warping [3]—reflecting the chance any given subject’s SVI occurs at each Atlas voxel location in the elderly and demented populations. Validity of linear and nonlinear SVPAs for automated GM, WM, and CSF tissue counts was evaluated using native SPGR data. Application of the AD SVPA with a statistical mapping technique [4] designed for use in probabilistic atlases is demonstrated for SPECT and PET functional analysis. Results: Correlation of manual native space counts to automated Atlas counts was excellent for the linear 12-parameter probability gradients (r2 0.983). Statistical mapping of functional differences between AD treatment groups and within MCI subgroups is provided. Conclusion: A deformable probabilistic atlas, appropriate for structural and functional imaging analysis of the elderly and demented populations, is now available. REFERENCES [1] Mazziotta JC, Toga AW, Evans AC, Fox P, Lancaster J. A probabilistic atlas of the human brain: theory and rationale for its development. Neuroimage 1995;2: 89 –101. [2] Thompson PM, Woods RP, Mega MS, Toga AW. Mathematical and computational challenges in creating deformable and probabilistic atlases of the human brain. Human Brain Mapping 2000;9:81–92. [3] Woods RP, Grafton ST, Watson JDG, Sicotte NL, Mazziotta JC. Automated image registration: II. Intersubject validation of linear and nonlinear models. J Comput Assist Tomogr 1998;22:153– 65. [4] Dinov ID, Mega MS, Thompson PM, et al. Analyzing functional brain images in a probabilistic atlas: a validation of Sub-volume thresholding. J Comput Assist Tomogr 2000;24:128 –38.

Scheltens P, Soinenen H. Many transgenic (Tg) mice lines, e.g. overexpressing b*-amyloid precursor protein (APP), have been developed as models for Alzheimer’s disease (AD). Currently, treatment strategies for AD strongly focus on therapies designed to slow down the progression of the disease and reduce b*-amyloid accumulation and/or promote clearance of senile plaques. The evaluation of efficacy of these therapies has been limited by lack of methods to detect amyloid load in the brains of living AD patients. The need for in vivo monitoring of the progression of the pathological changes has prompted search for new tools and techniques applicable in animals and/or humans, such as ligands that bind b*-amyloid and can be detected using MR or PET. Gary Small reported about a novel approach to in vivo imaging of senile plaques (SPs) and neurofibrillary tangles (NFTs), the hallmark lesions of AD. In a PET study, a hydrophobic radiofluorinated derivative of 1,1-dicyano-2-[6(dimethylamino)naphtalen-2-yl]propene (FDDNP) was used to detect SPs and NFTs in 7 AD patients and 3 controls. Greater accumulation and slower clearance was observed in SP and NFT rich areas of the brain and these findings correlated with performance in memory tests. In a greater sample of 10 AD patients and 7 controls a complete separation between groups was reached. The ligand was also used with fluorerence microscopy of AD brain specimens. These in vitro studies showed clear visualization of SPs ans NFTs compared to standard techniques (immunostaining for b*-amyloid and tau as well as Thioflavin S staining). The result also correlated with autoradiography data obtained in vitro in the same specimens. Dr Kung MP and coworkers have developed a probe, [(trans,trans)-1-bromo-2, 5-bis-(3-hydroxycarbonyl-4-hydroxy)styrylbenzene (BSB)] to be used for detection of SPs in vivo. It was reported to have excellent fluorescent properties, high affinity for fibrillar b*-amyloid aggregates in vitro and inhibit further b*-amyloid fibrillization. BSB labeled SPs in AD brain sections comparable to thioflavin S. BSB crosses the blood brain barrier and following intravenous injection into living tg mice (Tg2576), BSB labeled AD-like SPs throughout the brain of these tg mice. To quantitate plaque labeling with BSB, the radioiodinated derivative was prepared. In autogradiography this derivative clearly labeled the SPs in postmortem brain sections from AD patients. The potential of BSB as marker of b*-amyloid for in vivo studies in AD are under investigation. Dr Anguiano described single photon computed tomographic imaging strategy to detect b*-amyloid aggregates. The amyloid-binding dyes Congo Red and Chrysamine G were modified to allow their conjugation to the monoamine-monoamide bis(thiol) ligand which complexes technetium(V) in its neutral oxo form. Rhenium oxo complexes, which are non-radioactive, were also synthesized. These complexes bound to b*-amyloid fibrils produced in vitro and stained amyloid plaques and vascular amyloid in AD brain sections. Dr. Alice M. Wyrwicz and coworkers examined the brains of V717F Tg mice with MR imaging and parallel behavioral measurements at 11–14 months of age. T2-weighted spin echo images were acquired using 9.4T imager and 500 ? thick slices to detect alterations in tg mouse brain that might result from AD-like neuropathological changes. Eyeblink conditioning was used to measure learning in these animals. Interestingly, hippocampal volumes were significantly smaller in V717F APP tg mice by 27% when compared to wild-type mice. The finding is in line with reduction in hippocampal volume in Alzheimer patients. Instead in T2-weighted spin-echo MR images, relaxation times were similar for all brain regions studied in Tg and wild-type mice. Fewer transgenic mice exhibited learning relative to controls in eyeblink conditioning experiments: nearly 75% of the wild-type mice were learners by day 6 and 90% by day 12, only 20% of the Tg mice were learners by day 6 and 30% by day 12. These results showed similarities in MR volumetry and learning in V717F b-APP transgenic mice and Alzheimer’s disease in humans and encourage further studies using MR volumetry in the evaluation of the pathology and possible effects of potential drug treatments. These approaches clearly show that AD-related pathological changes can be visualized in vivo with ligands such as FDDP, BSB, Congo Red and Chrysamine G. Further efforts to optimize the ligands and imaging techniques and to evaluate their sensitivity to detect changes in vivo animal models and humans are needed. These tools will be extremely valuable, e.g. in development of new treatments and evaluation of their effect on disease process. REFERENCES [1] Skovronsky DM, Zhang B, Kung M-P, Kung HF, Trojanowski JQ, Lee VMY. In vivo detection of amyloid plaques in a mouse model of Alzheimer’s disease. Proc Natl Acad Sci USA 2000;97:7609 –14. [2] Zhen W, Han H, Anguiano M, Lemere CA, Cho CG, Lansbury PT. Synthesis and amyloid binding properties of rhenium complexes: preliminary progress toward a reagent for SPECT imaging of Alzheimer’s disease brain. J Med Chem 1999;42: 2805–15.



Rombouts SARB, Barkhof F, Witter MP, Scheltens P. Graduate School for Neurosciences Amsterdam, Research Institute Neurosciences, Vrije Universiteit, Amsterdam, The Netherlands Medial temporal lobe (MTL) atrophy as seen in Alzheimer’s disease (AD) can be assessed with volumetric measurements on MRI scans. Typically, such volumetric measurements are based on volume-of-interest comparisons, and thus limited to certain regions determined a priori. Here we apply an unbiased whole-brain analysis method to assess gray matter reductions in AD. Whole brain MRI scans of 7 mild to moderate AD patients (4 men, 3 women, mean age 65 years, range 53–75, mean Mini Mental State Examination (MMSE) score 21.6 (range 16 –28)) and 7 elderly healthy subjects (4 men, 3 women, mean age 57 years, range 51– 68, all MMSE score of 30) were analyzed with SPM99 (freely available at Images were reoriented into standard anatomic space and then segmented to compare gray matter volume in the whole brain between controls and patients. Symmetrical reductions of gray matter volume were detected in AD patients in the bilateral hippocampus (z 4.35), caudate nucleus (z 4.12) and insula (z 3.73). Asymmetric increase in volume was seen in the right superior temporal gyrus (z 5.57), right precentral gyrus (z 3.97), left fusiform gyrus (z 3.63), right inferior parietal lobe (z 3.62), left postcentral gyrus (z 3.54), and the right medial temporal gyrus (z 3.46). Symmetrical gray matter reductions in the hippocampus have been reported in both the pathology and imaging literature in AD. Our whole-brain analysis revealed unexpected areas of gray matter volume loss, including symmetric atrophy in the insula and the caudate nucleus. Further research may be directed towards these areas to study their exact role in the symptomatology and early diagnosis of AD.

Session IV

non-radioactive rhenium analogues have been tested for affinity to A (1– 40) fibrils produced in vitro, and they have affinities comparable to that of Congo red.



Small GW, Huang S-C, Cole G, Satyamurthy N, Agdeppa ED, Kiziloglu Z, Kepe V, Petric A, Vinters HV, Phelps ME, Barrio JR. Departments of Psychiatry and Biobehavioral Sciences, Molecular and Medical Pharmacology, and Pathology, UCLA School of Medicine, Los Angeles, CA, 90095 USA Our group recently discovered a novel approach to in vivo imaging of amyloid senile plaques (SPs) and neurofibrillary tangles (NFTs), the pathognomonic lesions of Alzheimer’s disease (AD). A hydrophobic radiofluorinated derivative of 1,1-dicyano2-[6-(dimethylamino)naphthalen-2-yl]propene (FDDNP) was used in conjunction with positron emission tomography (PET) to visualize (NFTs) and b-amyloid rich SPs in the living brain of AD patients (N 7) and controls (N 3). Greater accumulation and slower clearance was observed in SP- and NFT-rich brain areas and correlated with lower memory performance scores. Since DDNP and its derivatives are fluorescent and have sensitivity for hydrophobic environments, these compounds were also used with fluorescence microscopy to elucidate their binding in AD brain specimens. In vitro results compared with those obtained with standard labeling techniques (i.e., immunostaining with b-amyloid- and tau-antibodies or Thioflavine S) and showed a clear visualization of tangles and b-amyloid plaques in all brain areas involved in disease progression (e.g., entorhinal cortex, hippocampus, and temporal cortex). These results also correlated with digital autoradiography data obtained in vitro with [F-18]FDDNP in the same specimens. Our initial findings support the potential utility of FDDNP-PET imaging as a non-invasive approach to longitudinal evaluation of SP and NFT deposition, permitting more accurate diagnosis and evaluation of therapies. We postulate that the multimeric array of protofilaments organized as tubular fibrils in Ab (SP) and tau (NFT) aggregation lead to the specificity of the highly hydrophobic FDDNP for SPs and NFTs.




Venkatasubramanian PNa, Weiss Cb, Li La, Disterhoft Jb, Chen KSc, Wyrwicz AMa. Center for MR Research, ENH Research Institute, Evanston, IL, USA; bDepartment of Cell and Molecular Biology, Northwestern University Medical School, Chicago, IL, USA, cElan Pharmaceuticals, S. San Francisco, CA, USA Many transgenic (Tg) mice overexpressing -amyloid precursor protein have been generated and proposed as models for Alzheimer’s disease (AD). However, the diagnostic potential of Tg mice has not been realized because characterization of the development of AD-like pathology in these mice has been dominated by invasive techniques. We have examined non-invasively the brains of V717F Tg mice with MR imaging and parallel behavioral measurements at 11–14 months of age. T2-weighted spin echo images were acquired to detect alterations in transgenic mouse brain that might result from AD-like neuropathological changes. Eyeblink conditioning was used to measure learning in the animals. All imaging experiments were done on a 9.4T vertical bore imager. Images of brain slices 500 thick with an in-plane resolution 62.5 125 were acquired from anesthetized mice. Images were analyzed by measuring the number of pixels that comprise hippocampus. Further, contrast to noise ratio (CNR) was measured in cortex, hippocampus and subcortical brain. Ratios of CNR between cortex and subcortical brain as well as between hippocampus and subcortical brain were compared. Hippocampal volumes were significantly smaller in V717F APP transgenic mice by 27% when compared to wild-type control mice. Reduction in hippocampal volume has been linked with Alzheimer’s disease pathology in humans. None of the brain regions in the two groups of mice differed significantly in contrast-to-noise ratio seen in T2-weighted spin-echo MR images, signifying similar relaxation times for the brain in Tg and wild-type mice. Fewer transgenic mice exhibited learning relative to controls in eyeblink conditioning experiments. While nearly 75% of the wild-type mice were learners by day 6 and 90% by day 12, only 20% of the Tg mice were learners by day 6 and 30% by day 12. Our results from MR imaging and eyeblink conditioning experiments indicated that the V717F -APP transgenic mice may have pathological and behavioral similarities to Alzheimer’s disease in humans.



Kung M-Pa, Zhuang Z-Pa, Skovronsky DMb, Hou Ca, Zhang Bb, Lee VM-Yb, Trojanowski JQb, Kung HFa,c. Departments of aRadiology, bPathology and Laboratory Medicine and cPharmacology University of Pennsylvania, PA, USA Alzheimer’s disease (AD) is a neurodegenerative disorder characterized by accumulations of neurofibrillary tangles and senile plaques (SPs). Since SPs are extracellular deposits of amyloid fibrils formed by amyloid- peptides (A ), they can serve as suitable targets for early detection and monitoring of disease progression. Development of specific ligands for detecting A fibrils would provide important tools for novel in vivo and in vitro studies of A amyloidogenesis. For these reasons, we synthesized a novel ligand, (E,E)-1-bromo-2,5-bis-(3-hydroxycarbonyl-4-hydroxy) styryl-benzene (BSB). This highly conjugated molecule has excellent fluorescent properties ( ab 340 nm; em 520 nm), displays high affinity for fibrillar A aggregates in vitro (Ki 0.4 M) and inhibits further A fibrillization. Moreover, BSB fluorescently labeled SPs in AD brain sections with sensitivity comparable to that of thioflavin S. Following intravenous injection into plaque-bearing APP transgenic mice (Tg2576), BSB labeled AD-like amyloid plaques in living mice. To quantitate plaque labeling with BSB, the radioiodinated derivative, [125I]ISB, was prepared. [125I]ISB displayed good affinity to aggregated A (1– 40) (Kd 0.2 M). Amyloid plaques in postmortem brain sections from AD patients were clearly visualized with [125I]ISB using quantitative autoradiography. The potential of [123I]ISB as a marker for in vivo studies of A amyloidogenesis in AD is currently under investigation.

Session IV 18

DeCarli C. Rapid growth of the elderly population amongst industrialized countries has brought new attention to the consequences of various dementing illnesses as increasing age is the single greatest risk factor for dementia. Multiple epidemiological studies also show that Alzheimer’s disease (AD) is the leading cause of dementia. Current therapies for AD are estimated to prolong independence between 6 months to 1 year and are in current use throughout the world. Given the likelihood that new medications will be developed and use of currently available medications will grow, accurate diagnosis of AD is becoming increasingly necessary. Clinical research utilizing structural and functional imaging has identified the earliest brain changes associated with AD. As described below, these techniques may assist clinicians with the diagnosis of AD and focus treatment on those individuals most likely to receive benefit. Two studies from the research group headed by Dr. Marilyn Albert reported results of structural and functional brain imaging from a large group of elderly individuals followed longitudinally for three years. Repeated evaluation revealed four subgroups of individuals, those who maintained normal cognition, those who developed mild cognitive impairment (MCI) and those who developed dementia. Magnetic resonance imaging (MRI) and single positron emission computed tomography (SPECT) were obtained at initial evaluation and analyzed to identify significant predictors of progression to MCI or dementia. Multiple MRI regions including the entorhinal cortex, the banks of the superior temporal sulcus and the anterior cingulate were most useful in predicting subject follow-up status. Individuals with mild memory impairment who progressed to AD within the follow-up period were identified with 93% accuracy using these measures. Conversely, these regions also identified those individuals who did not progress with 85% accuracy. Similar results were found when relative perfusion coefficients were obtained from the same anatomical areas. These results suggest that specific neocortical regions as well as the hippocampi are affected early in the course of AD. Not only does identification of these regions help with early detection of AD, but identification of these regions may also indicate the possibility of unique brain memory systems susceptible to the disease.



Anguiano Ma,b, Zhen Wc, Lansbury PTb. aDepartment of Chemistry, Massachusetts Institute of Technology, Cambridge MA 02139, USA, bCND, Brigham and Women’s Hospital and Harvard Medical School, HIM, 77 Ave Louis Pasteur, Boston, MA 02115, USA, cPerkin-Elmer, Foster City, CA 94404, USA Alzheimer’s Disease (AD) is currently diagnosed based on the clinical observation of cognitive decline. The confirmation of the clinical diagnosis can only be made by the observation of amyloid deposits in the postmortem brain. A method for the non-invasive in vivo quantification of A amyloid deposits would be useful to early diagnose AD, and to monitor the effectiveness of therapies targeted at preventing amyloid accumulation. Here, we describe the initial stages of a strategy to measure brain amyloid, in vivo and non-invasively, by single photon emission computerized tomography (SPECT). The A amyloid binding dyes Congo Red and Chrysamine G have been modified to allow their conjugation to the monoamine-monoamide bis(thiol) ligand, which complexes technetium (V) in its neutral oxo form. The

Further examination of anatomical brain changes associated with the very early AD was reported by Csernansky et al. In this study, the investigators examined differences in hippocampal size and shape between a small group of early AD patients and age matched controls. Control MRIs were used to create a normal brain atlas. Fitting each AD brain to the brain atlas allowed to investigators to calculate deformation patterns as well as quantify size. Hippocampal size of AD patients was highly significantly different from healthy controls and identified the presence of AD with moderate accuracy. Changes in hippocampal shape, however, was also significantly different from healthy controls and added additional accuracy in the prediction of AD based solely on these hippocampal measures. The authors conclude that both size and shape changes of the hippocampus result from the AD process and can be used to detect the presence of AD very early in the course of the illness. Finally, Doraiswamy et al. reported results of magnetic resonance hydrogen spectroscopy (1H-MRS) on a large group of healthy elderly. Earlier research by this group show that n-acetyl-aspartate (NAA) levels decline in untreated AD patients over time. In addition, the ratio of myo-inositol (MI) to NAA is the most sensitive 1H-MRS measure to determine the presence of AD. These investigators presented preliminary cross-sectional analyses of data from the healthy elderly cohort. Agerelated differences in MI/NAA ratios were compared between apolipoprotein epsilon 4 (ApoE4) carriers and non-carriers. The slope of MI/NAA differences with age was significant greater for the ApoE4 carriers as compared to the non-carriers. Moreover, performance on delayed free and cued word recall was significantly correlated with MI/NAA ratios. From this data, the investigators suggest that frontal lobe dysfunction may occur before the onset of symptomatic AD. These studies show significant functional and structural brain changes do occur very early in the course of AD, possibly before the onset of clinical signs and symptoms. If these results are confirmed, these measures may serve to help the clinician to better identify individuals in the earliest phase of the disease or to detect those at highest risk, targeting those individuals most likely to benefit from treatment. REFERENCES [1] Johnson KA, Jones KJ, Becker JA, Satlin A, Holman BL, Albert MS. Preclinical prediction of Alzheimer’s disease using SPECT. Neurology 1998;50:1563–71. [2] Killiany R, Gomez-Isla T, Moss M, Kikinis R, Sandor T, Jolesz F, Tanzi R, Jones K, Hyman BT, Albert MS. The use of structural MRI to predict who will get Alzheimer’s disease. Ann Neurol 2000;47:430 –9. [3] Haller JW, Banerjee A, Christensen GE, Gado M, Joshi S, Miller MI, Sheline Y, Vannier MW, Csernansky JG. Three-dimensional hippocampal MR morphometry with high-dimensional transformation of a neuroanatomic atlas. Radiology 1997;202:504 –10. [4] Lazeyras F, Charles HC, Tupler LA, Erickson R, Boyko OB, Krishnan KR. Metabolic brain mapping in Alzheimer’s disease using proton magnetic resonance spectroscopy. Psychiatry Res 1998;82:95–106.

Session IV



Csernansky JG, Wang L, Joshi S, Miller JP, Gado M, Kido D, McKeel D, Morris JC, Miller MI. aDepartments of Neurology, Psychiatry, Radiology, Pathology, and Anatomy and Neurobiology, Washington University School of Medicine, St. Louis, MO, USA, Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD, USA Substantial volume losses in medial temporal lobe brain structures, such as the hippocampus, have been reported using high resolution magnetic resonance (MR) scans and manual techniques for brain structure outlining in patients with mild-tomoderate dementia of the Alzheimer type (DAT). The substantial losses of hippocampal volume in subjects with even mild symptoms of DAT suggest that even earlier stages of DAT might be detectable if more sensitive methods for brain structure analysis were available. In the present study, we sought to test this hypothesis in elder subjects who had the very mild symptoms of DAT. The structure of the hippocampus was evaluated in these subjects and in healthy elder controls using computerized algorithms for brain structure analysis that provide a precise and quantitative understanding of neuroanatomical volumes and shapes, despite the variability inherent to normal neuroanatomies. These algorithms employ newly developed tools of computational anatomy to represent the typical structures of the brain via the construction of templates, and their variability by probabilistic transformations applied to the templates. Local application of the transformations throughout the brain (i.e. high dimensionality) makes these methods uniquely valuable for defining small deformities in brain structures, such as the hippocampus. Distinct patterns of hippocampal shape change and volume loss discriminated subjects with very mild symptoms of DAT from healthy elder controls. The pattern of hippocampal deformity observed was asymmetric and implicated the CA1 hippocampal subfield as being affected early in the course of DAT. An analysis of both local deformities as provided by high dimensional brain mapping, as well as an analysis of volume differences, was needed to optimally discriminate subjects with the earliest detectable symptoms of DAT from other subject groups. Healthy elder controls were also discriminated from younger controls by hippocampal shape changes that did not resemble the pattern characteristic of DAT and were not associated with volume losses. These results suggest that structural deformities of the hippocampus are detectable with high dimensional brain mapping even at the threshold of clinical diagnosis of DAT. Moreover, early DAT and healthy aging were associated with different patterns of hippocampal deformity, suggesting that these two phenomena are biologically distinct.

21 19


Johnson KA, Killiany R, Jones KJ, Becker JA, Hilson J, Sandor T, KikinisR, Jolesz F, Albert MS. Brigham And Women’s Hospital, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA Identifying individuals in the prodromal phase of AD is difficult but increasingly important as disease modifying treatments become available. SPECT and volumetric MR have previously been used separately to seek evidence of brain abnormality in prodromal AD. We registered SPECT and MRI volumetric data in order to combine the strengths of functional and structural imaging methods. A total of 119 subjects were imaged with both SPECT and MR at baseline and then followed clinically for a minimum of 3 years. Thirty-two subjects were normal at both baseline and follow-up, 61 were classified as questionable AD (CDR 0.5) at baseline and remained questionable at follow-up, and 26 subjects with questionable AD at baseline converted to fulfill NIH criteria for AD. MRI volumes for 4 brain regions (anterior and posterior cingulate, thalamus, and hippocampus) were determined by investigators blinded to clinical status and SPECT image findings. SPECT perfusion images were then sampled using the MRI-derived volumes from registered SPECT/MRI datasets to calculate volumetric perfusion. Subjects who subsequently converted to meet clinical criteria for AD could be distinguished from those who remained either questionable or normal at follow-up on the basis of their volumetric perfusion data (p 0.0001; hit rate 88%). Regional decreases in perfusion were greatest in cingulate regions. SPECT data analyzed with the aid of MRI volumetry was successful in preclinical prediction of AD. The combination of structural and functional imaging methods may be particularly useful in identifying individuals in the earliest pathophysiologic stages of AD. REFERENCES [1] Johnson KA, Jones KJ, Becker JA, Satlin A, Holman BL, Albert MS. Preclinical prediction of Alzheimer’s disease using SPECT. Neurology 1998;50:1563–71. [2] Killiany R, Gomez-Isla T, Moss M, Kikinis R, Sandor T, Tanzi R, Jones K, Hyman BT, Albert M. The Use of Structural MRI to Predict Who Will Get Alzheimer’s Disease. Ann Neurol (in press).

Killiany R, Gomez-Isla T, Moss M, Kikinis R, Sandor T, Jolesz F, Tanzi R, Jones K, Hyman B, Albert M. Boston University, Massachusetts General Hospital, Brigham & Women’s Hospital, Boston, MA, USA Magnetic resonance imaging (MRI) was used to determine whether persons in the prodromal phase of AD could be accurately identified before they developed clinically diagnosed dementia. Normal subjects (n 24) and those with mild memory difficulty (n 79) received an MRI scan at baseline and were then followed annually for 3 years to determine which individuals subsequently met clinical criteria for AD. Patients with mild AD at baseline were also evaluated (n 16). Nineteen of the 79 subjects with mild memory difficulty ‘converted’ to a diagnosis of probable AD after 3 years of follow-up. Baseline MRI measures of the entorhinal cortex, the banks of the superior temporal sulcus and the anterior cingulate, were most useful in discriminating the status of the subjects on follow-up. One hundred percent (100%) of normals and patients with mild AD could be discriminated from one another based on these MRI measures. When the normals were compared to the individuals with memory impairments who ultimately developed AD (the ‘converters’), the accuracy of discrimination was 93%, based on the MRI measures at baseline. The discrimination of the normals and the individuals with mild memory problems who did not progress to the point where they met clinical criteria for probable AD over the 3 years of follow-up (the ‘questionables’) was 85% and the discrimination of the questionables and converters was 75%.



Doraiswamy PM, Barboriak D, Krishnan KRR, Chen G, Smith J, Charles HC. Duke University Medical Center, Durham, NC 27710, USA Proton magnetic resonance spectroscopy (1H-MRS) is an application of magnetic resonance (MR) that allows for the noninvasive assessment of regional brain

Session V
concentrations of N-acetyl aspartate (NAA), glutamine (Glu), GABA, myo-inositol (Ino) compounds, glycine, mobile choline moieties (Cho), and creatine/phosphocreatine (Cr). Although the function of NAA is unclear, it is generally accepted as a neuronal marker. Several studies have shown a reduction in NAA levels in Alzheimer’s disease (AD), although the magnitude of reduction has varied by study design and brain region assessed. MRS changes have been correlated in two studies with in vitro AD neuropathology and in one post-mortem study with the ApoE genotype. In a preliminary study, we recently reported that baseline NAA measures may predict cognitive decline at one year follow up in patients with early stage AD. In addition, we have also found that NAA levels decline in many untreated AD subjects over time. We recently completed a 1H-MRS PROBE study (with a voxel in the left frontal lobe) in 152-elderly nondemented volunteers stratified by apolipoprotein E genotype and family history. The ApoE groups did not differ in age, gender, and cognitive performance. Reference normalized NAA levels and NAA/MI ratios in the left frontal lobe declined significantly with age. Normalized levels of Cr as well as the ratio of NAA/Cr did not decline with age suggesting that MI/NAA is a better marker for age-related neuronal changes than NAA/Cr. The slopes of decline in NAA with age and in the ratio of NAA/MI tended to be greater among ApoE4 carriers than noncarriers with the greatest decline seen among subjects with both the E4 allele and a positive family history. Poor performance on delayed free and cued recall (a potential marker for prodromal Alzheimer’s disease) was also statistically significantly correlated with lower left frontal lobe NAA levels. Our findings suggest that the age-related decline in frontal lobe neuronal functioning may be more severe in Apoe4 carriers and in subjects with mild memory impairment before the onset of symptomatic Alzheimer’s disease. Supported by AFAR.

from a spoiled gradient echo pulse sequence (slice thickness 1.5 mm) and ANALYZE software on a UNIX-based Sun SPARCstation 2. Hippocampal and amygdala volumes were normalized for intersubject variation in head size by dividing structure volume by the total intracranial volume of that particular subject. The ε4 AD patient group was younger and had an earlier age of onset than the ε4 AD patient group. Analysis of covariance revealed that the normalized volumes of amygdala were significantly smaller in ApoE ε4 than ε4 AD patients (left: p .004, right: p .004, MMSE covariate). When ApoE ε4 “dosage” was considered, this effect appeared to accrue from a difference between the 0ε4 and each of the other two AD groups, with no significant difference between the 1ε4 and 2ε4 AD groups. Hippocampal volumes did not differ between the two AD genotypic groups (left: p .276, right: p .458, age covariate). These results are interesting in light of the importance of the amygdala for memory.



Rombouts SARB, Orij PCM, Scheltens P. Graduate School for Neurosciences Amsterdam, Research Institute Neurosciences, Vrije Universiteit, Amsterdam, The Netherlands In functional brain imaging, brain function is typically compared between different populations (patients vs. controls). With functional magnetic resonance imaging (fMRI), it is also possible to determine brain activation in single patients. This allows a comparison of brain function in a single patient to a population of controls or patients. This study concerns a comparison of brain activation during a memory encoding task in a demented patient to an elderly control group and a group with mild to moderate Alzheimer’s disease (AD). Brain activation was measured with fMRI scanning during encoding of color pictures, a task often shown to activate the posterior part of the medial temporal lobe. The task was applied in 11 mild to moderate Alzheimer patients, 10 healthy elderly control subjects and one patient, who was suspected to suffer from fronto-temporal dementia. We tested for differences in regional brain activation during encoding between the patient and the AD population and between the patient and the controls. The control group did not show a significant increase in encoding activation in any brain area compared to the patient. All activated brain regions, including parahippocampal gyrus, were significantly more active in the patient than the AD group (p 0.0005, uncorrected). These results show in this particular patient no functional evidence for AD, but a more normal functional activation pattern, in agreement with the clinical findings. This study warrants further exploration of this method in the diagnosis of dementia in individual patients using a set of functional MR scans of various demented patients for comparison.

Session V 23

Weiner M. Invited Speakers: Ranjit Mani, M.D., Food and Drug Administration; Paul Leber, Neuro-Pharm Group, LLC; Ravi Anand, MD, Novartis Pharmaceuticals Corporation This session was devoted to presentations concerning industry and regulatory agency perspectives, regarding the role of imaging in pharmaceutical trials for Alzheimer’s. Ranjit Mani from the Food and Drug Administration provided an overview and historical review of the use of neuroimaging as a surrogate marker for pharmaceutical trials. He explained the definition of a surrogate marker, as used by FDA. He reviewed the use of imaging for assessment of drugs for treatment of multiple sclerosis. Paul Leber MD, currently Director Neuro-Pharm Group, LLC, worked for FDA for many years. He provided an iconclastic perspective concerning the approach used by FDA to assess surrogate markers. Following Drs Mani and Leber, there was by a lively discussion concerning whether MRI, MRS, or PET could be used to provide surrogate marker information for Alzheimer’s treatment trials. Ravi Anand, representing Novartis provided a detailed view of the advantages problems of conducting multisite MRI/MRS studies for dementia. After the presentations, there was general discussion concerning the future of the AIC. There seemed to be general agreement that this symposium successfully achieved its goal of providing a forum for information exchange and discussion concerning imaging of dementia. Attendees also agreed that a future meeting would be useful, perhaps in 2 years at the next World Alzheimer’s Congress in Stockholm, Sweden.



Rapoport SIa, Chang MCa, Connolly Ka, Bokde Aa, Carson REb, Eckelman WCb. Section Brain Physiology and Metabolism, National Institute on Aging, NIH, Bethesda, MD 20892, USA, bPET Department, NIH, Bethesda, MD 20892, USA Evidence exists of abnormal brain phospholipid metabolism in Alzheimer disease (AD), but no in vivo imaging method exists to examine this defect. The defect may underlie abnormal cholinergic signaling via phospholipase A2 (PLA2). PLA2 activation via muscarinic M1 and M3 receptors releases the important second messenger, arachidonic acid (AA), from the stereospecific numbered 2 position of brain phospholipids. In rats and monkeys, we are able to localize this release in brain, at rest and in response to cholinergic or functional activation, by injecting labeled AA intravenously and measuring regional brain radioactivity by autoradiography or PET, respectively. Tracer entering brain is 90% incorporated into phospholipids within 1 min, in proportion to unlabeled AA released by PLA2. Incorporation is blood flow-independent. Using the positron-emitting isotope [1-11C]AA with PET in 4 human volunteers with eyes open, the tracer incorporation coefficient k* (brain radioactivity/integrated plasma radioactivity) averaged 1.0 –1.4 10 4 ml 1 sec 1g 1 in gray matter and half that in white matter. k* in thalamus and primary striate cortex was 30% higher than in other gray matter regions, suggesting that visual activation increased PLA2 signaling in these regions. Thus, in vivo PET imaging of brain [1-11C]AA might be used to identify defective brain signal transduction in AD. REFERENCE [1] Rapoport et al. Neurochem Res 1999;24:1403–15.

Poster Presentations 24

van Dyck CH, Basso M, Yang J, Half LM, MacAvoy MG, Varma P, Gelernter J. Yale University School of Medicine, New Haven, CT 06520, USA Previous investigators have reported atrophy of the hippocampus and amygdala in Alzheimer’s disease (AD) as well as greater atrophy of these structures in AD patients who carry the apolipoprotein E ε4 (ApoE ε4) allele. We performed MRI-based volumetric measurements of the hippocampus and amygdala in 55 patients with probable AD (26 M, 29 F; 71.3 8.5 years) and 42 cognitively normal elderly control subjects (22 M, 20 F; 73.2 6.7 years). The diagnosis of AD was made according to NINDS/ADRDA criteria. Subjects were scanned with the 1.5 T GE Signa system. Volumetric MRI measurements of the hippocampus and amygdala were made


Poster Presentations
DIFFERENTIAL PATTERNS OF HIPPOCAMPAL AND TEMPORAL LOBE ATROPHY IN ALZHEIMER’S DISEASE AND FRONTOTEMPORAL DEMENTIA Eight healthy volunteers of age range of 59 to 77 years were studied along with 4 MCI patients of comparable age. Subjects underwent fMRI using a delayed response working memory task with an S1–S2 interval of 18-sec. The task consisted of a visual stimulus (cue) of one or three faces, followed by a delay, followed by a test face (probe), which matched or did not match a face in the original stimulus. Epochs from 12 seconds prior, to 33 seconds after stimulus were excised from the time series. The epochs over 18 trials were signal averaged separately for the one and three face conditions. Activated voxels in the middle frontal gyri were interrogated and the hemodynamic response compared. All subjects completed the fMRI experiment satisfactorily. Differences in activation magnitude between the one and three face conditions were found during the encoding and retrieval phases for the volunteers (p 0.05) and MCI groups. MMI patients showed a trend (p 0.11) toward decreased activation during encoding and retrieval, compared to elderly controls. There were no significant differences in task performance between groups.


Galton CJa, Patterson Kb, Hodges JRa,b. aUniversity Department of Neurology, Cambridge, UK, bMRC Cognition and Brain Sciences Unit, Cambridge, UK Introduction: Hippocampal atrophy is well described in Alzheimer’s disease (AD) using many methods of measurement, however the specificity of this finding is debated. Frontotemporal dementia (FTD) can present with a temporal syndrome (SD) and the neuroimaging findings are not well established. Methods: The 3d MRI scans of 27 early stage AD, 18 SD patients and 21 controls were assessed using a manual tracing method with ANALYZE software. This volumetric method measured the following regions bilaterally; the temporal pole, amygdala, hippocampus (including subiculum) parahippocampal gyrus (encompassing the entorhinal cortex), fusiform gyrus, inferior and middle temporal gyri and superior temporal gyrus. Neuropyschological measures of episodic and semantic memory were assessed. Results: Neuropsychological profiles of the patient groups confirmed the expected cognitive profiles: episodic memory was impaired in the AD group, and relatively preserved in the SD group. Semantic memory, was impaired in the SD group but not in the AD group compared with the controls. The volumetric results confirmed the expected bilateral hippocampal atrophy in AD relative to controls with involvement of the amygdalae and right parahippocampal gyrus. SD patients had asymmetric hippocampal atrophy more extensive than AD on the left. Distinguishing the SD group was more severe involvement of the temporal pole bilaterally and the left amygdala, parahippocampal, fusiform and inferior and middle temporal gyri. The relationship of these findings to the cognitive measures are discussed. Conclusions: Contrary to our expectations we found significant hippocampal atrophy in the SD patients. These results have implications for the diagnosis and understanding the cognitive deficits in AD and SD. The relationship between performance on the episodic memory tests and medial temporal atrophy across groups does suggest a linear relationship between these measures and size of the hippocampus.



Rombouts SARB, Barkhof F, Veltman DJ, Machielsen WCM, Witter MP, Bierlaagh MA, Lazeron RHC, Valk J, Scheltens P. Graduate School for Neurosciences Amsterdam, Research Institute Neurosciences, Vrije Universiteit, Amsterdam, The Netherlands We applied functional magnetic resonance imaging (fMRI) with a learning task in healthy elderly subjects and patients with Alzheimer’s disease (AD) to study brain activation during memory performance. The purpose was 1) to determine the feasibility of fMRI during a learning task in healthy elderly subjects and patients with AD and 2) to test our hypothesis that brain activation is decreased in the medial temporal lobe (MTL) memory system in AD patients compared to controls. In 12 mild to moderate AD patients and 10 elderly control subjects, activation of the MTL memory system was studied. We used two learning tasks that required encoding of new information into memory and applied a random effects data analysis. After the fMRI experiment, subjects were tested for recognition of the encoded objects. In elderly control subjects, activation during memory encoding was observed in medial and lateral temporal lobe structures (fusiform, parietal and occipital parts and the hippocampal formation), and frontal cortex, as reported previously in young controls. Focusing on the MTL, we found that activation was significantly decreased in AD patients compared to controls in the left parahippocampal gyrus and hippocampus bilaterally during the first encoding task, but not during the second (p 0.05, uncorrected). fMRI with a learning task appeared feasible in elderly subjects and AD patients. The measured functional signal decrease in MTL areas warrant further exploration of the (early) diagnostic utility of fMRI in AD and other dementias.



Small SA, Wu EX, Nava AS, Perera GM, DeLaPaz R, Stem Y. Departments of Neurology and Radiology, Columbia University College of Physicians and Surgeons, New York, NY, USA Background: A major goal for the early detection of Alzheimer’s disease is being able to evaluate the functional integrity of individual hippocampal subregions. Despite limited success, most fMRI protocols have difficulty evaluating all hippocampal subregions because of inadequate spatial resolution. Objective: We have recently developed an MR protocol designed to generate T2* maps of the hippocampal formation with sub-millimeter resolution. Here, we test this protocol with transgenically modified mice, and in humans with Alzheimer’s disease. Methods/Results: Mice study: Transgenically modified mice, expressing a mutant protein causing hippocampal deficits, were scanned in a 4 tesla high-field magnet adapted for rodent imaging. Using the high-resolution protocol, significantly different signal in the hippocampal formation of mutant mice was detected compared to wild-type littermates. Human study: 5 patients with probable/possible AD and 5 healthy elderly participated in the study. Scanning was performed on a 1.5 tesla magnet. Using the high-resolution protocol, subregional analysis detected diminished hippocampal signal intensity in AD patients compared to controls, and this difference was most prominent in the entorhinal cortex. Conclusion: The high-resolution protocol presented here can detect neuronal dysfunction at the level of the hippocampal subregion. We are currently evaluating whether this protocol can detect early AD, prior to the onset of dementia.



Budinger TF. Lawrence Berkeley National Laboratory, Berkeley CA, USA; Univ of California, Berkeley and San Francisco, CA, USA Recent advances in imaging technologies have given us the potential to study metabolism, neuroreceptor concentrations and gene expression in animal models of Alzheimer’s disease. Over the past 10 years, transgenic manipulations of mice have created phenotypes with the potential of linking specific genes to molecular, cellular and organ functions. Application of the transgenic or gene “knock-out” mouse mutant technology to studies of the biochemistry and physiology of the brain, heart, musculosketetal and metabolic systems has historically been through necropsy studies, which require the sacrificing of many animals and the performance of tedious assays. Contemporary noninvasive imaging methods of radionuclide emission tomography [i.e., positron emission tomography (PET); single photon emission tomography (SPECT)]; magnetic resonance imaging (MRI); magnetic resonance spectroscopy (MRS); and optical imaging (i.e., near-infrared imaging and tomography, fluorescence imaging, optical tomography and bioluminescence imaging)) have great potential for the study of transgenic animals as small as the 30-g mouse [1] and specifically in such animal preparations as recently reported for a model of AD [2]. Three modes of noninvasive imaging of small animals relevant to Alzheimer’s disease are radiolabeled reporter gene imaging by PET or SPECT, MRI and MRS of enzyme kinetics or contrast agent accumulation, and detection of fluorescence and bioluminescence reporter genes (luciferase and green fluorescence protein). This review first presents a technological description of the characteristics of the image data that can be acquired from current or feasible instruments and gives examples of how noninvasive imaging methods have been applied to studying physiology and gene expression in small animals, particularly transgenic animals.



Petrella JR, Bucher L, Jha A, Doraiswamy PM, McCarthy G. Duke University Health System, Durham, NC, USA We performed a pilot study in a small group of elderly volunteers and patients with mild cognitive impairment (MCI) to established the feasibility of using event-related fMRI during a delayed response memory task in these populations. Such a design was used so that activation during a memory cue (S1), sustained over a delay period, and evoked by a memory probe (S2) could be more easily dissociated.

Poster Presentations
REFERENCES [1] Budinger TF, Benaron DA, Koretsky AP. Imaging transgenic animals. Annual Review of Biomedical Engineering 1999;1:611– 48. [2] Guenette SY, Tanzi RE. Progress toward valid transgenic mouse models for Alzheimer’s disease. Neurobiology of Aging 1999;20(2):201–11. IN VITRO NMR SPECTROSCOPY IN AN ANIMAL MODEL OF DOWN SYNDROME AND ALZHEIMER’S DISEASE

ability to quantitate metabolites’ levels from the entire brain. This would potentially represent the most direct measure of neuronal loss available. We studied (thus far) 4 aged normal volunteers and 4 patients with Alzheimer’s Disease (AD) with a non-localized spectroscopic acquisition technique employing a novel phase cycling scheme that provides both water and fat suppression along with removal of T1 and T2 weighting. Results were normalized by brain volume measurements based upon MRI, allowing the calculation of absolute NAA levels, dubbed whole-brain NAA (WBNAA). We compared these findings with results in younger normals and patients with multiple sclerosis (MS), where the WBNAA was found to be both age-invariant and highly sensitive to neuronal loss in MS. Aged (65–78 yo) normal volunteers exhibited WBNAA values that were not different from younger (17–52) normals. In 3 of 4 AD patients, the WBNAA values were also normal. This is consistent with the hypothesis that brain loss proceeds in parallel with neuronal loss, a reasonable though untested idea. Only in the oldest AD patient did WBNAA appear to drop off. This suggests the intriguing idea that in end-stage or later-stage AD neuronal loss begins to outstrip parenchymal loss, representing perhaps a more severe stage of the disease. REFERENCE [1] Gonen et al, MRM 1998;40:684 –9. LACK OF PET ACTIVATION OF POSTERIOR LEFT TEMPORAL LOBE CORRELATES WITH ANOMIA IN ALZHEIMER DISEASE


Caserta MTa,b, Yao Fb, Wyrwicz Ac,d. aDept. of Psychiatry, University of Chicago, Chicago, IL 60637, USA, bDept. of Psychiatry, Northwestern University Medical School, Chicago, IL, USA, cDept. Anesthesiology, Northwestern University Medical School, Chicago, IL, USA, dThe Center for MR Research, Evanston Northwestern Healthcare Research Institute, Evanston, IL 60208, USA NMR Spectroscopy was used to evaluate cytosolic compounds and membrane phospholipids simultaneously in murine trisomy 16 (ts 16), a model of Down Syndrome (DS) and Alzheimer’s Disease (AD). Fetal trisomy 16 brains and their euploid littermate controls were used to examine the cellular biochemistry that underlies the neurodevelopmental consequences of chromosome triplication in a model of DS. Proton NMR spectroscopic analysis of brain tissue extracts demonstrated decreased levels of choline and increased levels of myo-inositol in ts 16 brains compared to controls. These data are consistent with the cholinergic deficits and elevated myo-inositol levels previously described in ts 16 by other methods. Ts 16 brains also demonstrated higher levels of creatine, adenosine and tyrosine than their euploid littermates. NAA, a neuronal marker, is not measurable at this stage. Increased levels of myo-inositol and creatine, compounds localized predominantly in glia, implicate abnormalities in the glial or trophic environment of ts 16 brain. Phosphorus NMR spectroscopic analysis of these extracts revealed some surprising results. Anionic phospholipid membrane components, such as phosphotidylinositol and phosphotidylethanolamine, were elevated in ts 16 brains, while sphingomyelin and phosphotidylcholine were comparable to controls. Since these compounds are confined to the inner leaflet of the cellular membrane, the results suggest that membrane composition is altered specifically in the cytosolic bilayer on ts 16 brain at this stage. Together, the proton and phosphorus spectroscopic results indicate multiple biochemical pathways are affected in ts 16 brains. Utilizing NMR spectroscopy’s comprehensive nature, in vitro NMR spectroscopic identification of affected compounds in ts 16 brain extracts provides evidence for involvement of multiple biochemical [pathways that would affect complex neuronal and glial functions. Understanding the long-term effects of the multiple cellular and membrane anomalies found in ts 16 may help to elucidate the processes that lead to neuronal dysfunction and AD neuropathology in DS individuals. In vivo proton and phosphorus NMR spectroscopic analysis of very mildly cognitively impaired individuals with probable AD and healthy, cognitively intact controls are now in progress to test the applicability of these findings in affected individuals. INVESTIGATION OF GLOBAL ABSOLUTE N-ACETYL ASPARTATE LEVELS IN ALZHEIMER’S DISEASE


Chertkow H, Whatmough C, Murtha S, Evans A. Bloomfield Centre for Studies on Aging, Lady Davis Institute, Jewish General Hospital, and McGill University Dept. of Neurology and Neurosurgery, Montreal, Quebec, Canada Background: AD subjects are often anomic, due to deterioration of semantic memory. The brain correlate of this anomia remains obscure, as does the neural substrate of semantic memory. Method: We carried out oxygen 15 PET imaging in 15 normal elderly and 15 mild AD individuals during picture naming. Subjects named blocks of line drawings of animals normed for high accuracy (easy) and lower accuracy (hard). Subtraction images and Regression maps (assessing regression of cerebral blood flow onto each subjects naming accuracy) were analyzed in comparison to a baseline set of conditions. Results: Looking at subtractions of easy from hard blocks of animals (or looking at negative regression results) the normal subjects demonstrated increased cerebral blood flow in left frontal and posterior left temporal regions during naming of harder blocks of items. The AD subjects showed frontal activation, but no posterior left temporal activation. This was most striking for the most anomic subjects. Conclusion: PET can establish a brain-behaviour correlation in anomia which explains the naming failure in AD, and supports a neural localization for semantic memory in the left posterior temporal area. LEUKOARAIOSIS IS SIGNIFICANTLY CORRELATED WITH REGIONAL CORTICAL ATROPHY IN ALZHEIMER’S DISEASE



McGowan JCa, Clark CCa, GE Ya, Udupa J, Grossman RIa, Gonen Ob. aDepartment of Radiology, University of Pennsylvania, Philadelphia, PA, USA, bFox Chase Cancer Center, Philadelphia, PA, USA Topic area: Monitoring Disease Progression 1 H Magnetic Resonance Spectroscopy (MRS) offers the potential to identify chemical compounds in CNS tissue, including N-acetyl aspartate (NAA), which is considered to be a marker of neuronal integrity. Recent advances in 3D spectroscopic imaging provide metabolite information from a large portion of the brain, but lack the

Furman M, Bekinschtein T, Acion L, Gomila H, Capizzano AA, Kremer J, Starkstein ´ SE. Raul Carrea Institute of Neurological Research (FLENI), Buenos Aires, Argentina ´ Purpose: To examine the correlations between tissue volume changes and severity of leukoaraiosis (LA) in Alzheimer’s disease (AD) using volumetric high resolution MRIs. Methods: Eighteen patients with probable AD were scanned on a 1.5T system (Signa, GE). The protocol included 3D SPGR (0.88 1.3 mm in plane resolution, 1.5 mm slice thickness) and 3D dual echo FSE (TR 3,500 ms, TE1 32, TE2 96; 3 mm slice thickness). Image processing was carried out on the BRAINS software (MH-CRC, University of Iowa). Images were normalized to the Talairach atlas and segmented into



gray matter, white matter, and CSF. LA was manually edited by a radiologist (AAC). Tissue volumes normalized to the intracranial volume were calculated for each lobe. Results: There were significant negative correlations between LA and cortical volumes in specific regions: left frontal cortex (r 0.55, p 0.05) and bilateral temporal cortex (r . 0.52, p 0.05). Cortical gray matter and sulcal CSF volumes were also correlated (r 0.73, p 0.001). Conclusions: There was a significant correlation between global LA and cortical atrophy in specific brain regions, suggesting that LA may contribute to the atrophy process in AD.

Poster Presentations



Minoshima S, Cross DJ, Foster NL, Henry TR, Kuhl DE. Departments of Internal Medicine and Neurology, The University of Michigan, Ann Arbor, MI, USA PET imaging with [18F]fluorodeoxyglucose in very early AD patients before the onset of dementia showed an initial metabolic reduction in the posterior cingulate cortex as well as association cortices. Neurofibrillary tangles are known to appear initially in the transentorhinal cortex. These findings indicate clearly discordance in the distribution between metabolic reduction and traditional pathologic changes. Temporal lobectomy (including the entorhinal cortex) in epilepsy patients resulted in a small metabolic reduction in the posterior cingulate cortex and thalamus, but no reduction in the lateral parietal or frontal association cortices as seen in very early AD. Basal forebrain cholinergic neurons are also known to degenerate severely. Image-based multivariate correlational analysis between a cortical cholinesterase activity measured by N-[11C]methyl-4-piperidinyl propionate PET, glucose metabolism, and cortical atrophy measured by MRI in mildly demented AD patients revealed that cholinergic changes seen in the cerebral cortex did not correlate with regional metabolic changes or atrophy. These results suggest that neither loss of entorhinal efferents nor cholinergic deficit explains the metabolic features seen in very early AD. Given recent immunocytochemical evidence of massive glutamatergic synaptic alteration in early AD cortex and insights into neuronal and glial mechanisms of glucose metabolism, very early metabolic changes in AD probably reflect significant impairment of cortico-cortical glutamatergic systems in a preclinical stage of the disease.

the postmortem brain in Alzheimer’s disease relative to age-matched controls showed a decrease in NAA concentration, this presumed marker of neuronal density has received a literature of almost 980 articles on the interpretation of NAA in the brain in AD, stroke, brain injury, aging and neurodegenerative disorders. In almost every article the assumption is that NAA is a marker of neuronal density with a notable exception [1]. We have proposed the concept that NAA is not a neuronal density marker based on our observations that whereas at necropsy the ratio of gray to white matter NAA was 4:1 [2] in vivo proton spectroscopy shows the ration to be 1:1. We recognized that there might be post mortem changes but erroneously assumed these changes would be at the same slow rate in the Alzheimer’s brain as with non-demented age-matched controls and this fact of decline in concentration was not an important consideration. However in this previous work we did not recognize the possibility of a differential metabolic activity between white and gray matter. Over 20 years ago experiments in the developing rodent showed that N acetylasparatic acid serves as a source of acetyl groups for lipogenesis in the developing brain [3]. Other observations from early literature give evidence for a very active metabolic role of NAA (e.g. one day turnover, second most abundant amino acid after glutamate). In view of these observations we performed experiments to demonstrate the dynamic role of NAA in the mammalian brain and the activity of N-acetyl-L-aspartate amidohydrolase in the normal rabbit brain white and gray matter. We present our data on ex vivo dynamics of NAA metabolism from the rabbit gray and white matter measured on tissue extracts at 400 MHz. Data are taken from about 20 minutes post sacrifice to demonstrate the rate of decline of NAA in white vs gray matter. From white and gray matter of the rabbit brain and other evidence in the literature we conclude that NAA is an active metabolite essential for energy during metabolic stress and that the enzyme releasing acetate from NAA, N-acetylaspartominase, is 3 times more active in white matter than in gray matter. These observations should be taken in context of Prof. Pettegrew’s hypotheses regarding membrane metabolism and energy requirements of the Alzheimer’s brain which might be struggling for metabolites in its jeopardized state [4]. REFERENCES [1] Rubin Y, Cecil K, Wehrli S, McIntosh TK, Lenkinski RE, Smith DH. Highresolution 1H NMR spectroscopy following experimental brain trauma. Journal of Neurotrauma 1997;14(7):441–9. [2] Kwo-On-Yuen PF, Newmark RD, Budinger TF, Kaye JA, Ball MJ, Jagust WJ. Brain N-acetyl-L-aspartic acid in Alzheimer’s disease: a proton magnetic resonance spectroscopy study. Brain Research 1994;667:167–74. [3] D’Adamo AF, Smith JC, Woiler C. The occurrence of N-acetylaspartate amidohydrolase (aminoacylase II) in the developing rat. Journal of Neurochemistry 1973;20(4):1275– 8. [4] Pettegrew JW, Klunk WE, Panchalingam K, McClure RJ, Stanley JA. Magnetic resonance spectroscopic changes in Alzheimer’s disease. Annals of the New York Academy of Sciences 1997;826:282–306.



Capizzano AA, Martınez A, Acion L, Furman M, Bekinschtein T, Gomila H, Kremer ´ ´ J, Starkstein SE. Raul Carrea Institute of Neurological Research (FLENI), Buenos ´ Aires, Argentina Purpose: To examine the correlation between a semi-quantitative and a quantitative method of grading leukoaraiosis (LA). Methods: Ten probable Alzheimer’s disease (AD) patients were evaluated with MRI (Signa, GE) using 3D SPGR (0.88 1.3 mm in plane resolution, 1.5 mm slice thickness) and 3D dual echo FSE sequences (TR 3,500 ms, TE1 32, TE2 96; 3 mm slice thickness). Quantitative image analysis was carried out using BRAINS software (MH-CRC, University of Iowa). Images were normalized to the Talairach atlas and segmented into gray matter, white matter, and CSF. LA was manually edited by a radiologist (AAC). Volumes of LA were calculated for each lobe. Semiquantitative scoring of LA was performed by a radiologist (AM) according to a modification of the method of Scheltens et al. Periventricular and deep white matter hyperintensities were scored on axially reformatted PD and T2-W images based on their number and size. Results: There was a significant correlation between quantitative and semi-quantitative estimates of global LA: (r 0.94, p 0.05) as well as in the frontal (r 0.84), parietal (r 0.76), occipital (r 0.76) and left temporal (r 0.83) lobes (p 0.05 for each one). Conclusions: Quantitative and semi-quantitative estimates of LA in AD are strongly correlated allowing comparisons across different studies.





Budinger TFa,b, Shepkin VDc, Brennan KMa. aLawrence Berkeley National Laboratory, Berkeley, CA, USA, bUniv of California, Berkeley and San Francisco, CA, USA, c Univ of Illinois, Urbana, IL, USA The use of proton spectroscopy for neuroscience studies has emphasized measurements of N-acetyl-aspartate (NAA), creatine, choline-related protons and the ratios of these metabolites. The interest in NAA in the brain stems from its uniquely high level of 5– 6 mM in the brain after myelination. This concentration results in a prominent peak in the NMR proton spectrum from in vivo studies. Because detailed studies of

Jenkins R, Fox NC, Rossor MN. Dementia Research Group, Department of Clinical Neurology, Institute of Neurology, 8-11 Queen Square, London, UK Raised global and regional brain atrophy has been shown in previous studies of Alzheimer’s disease. We examined the sensitivity and specificity of rates of global atrophy in a group of unselected AD and controls using registered serial MRI. Seventy-five individuals—34 controls and 41 subjects with clinically diagnosed probable AD (27 sporadic, 14 familial)— had 2 volumetric MRI with a 12 month

Poster Presentations


interval. Each individual’s pair of scans was accurately registered and rates of atrophy were calculated using the Brain Boundary Shift Integral (BBSI). The mean rate of atrophy in the AD group of 2.5( 1.4)% was significantly greater (p 0.001) than the mean rate of 0.3( 0.4)% in the control group. Receiver Operating Characteristic analysis showed sensitivity and specificity of 93% and 94% respectively. One control subject had a rate that was greater than 2 SD away from the control mean. Further investigation revealed that this subject had been administered steroids during the scan interval. Three AD subjects had rates of atrophy within the control range. A review of these cases found that one had recently been re-diagnosed as non-AD, one had been scanned across a major scanner upgrade and one has a diagnosis currently under review. Re-analysing the data after the removal of these outliers showed 100% sensitivity and specificity. Rates of atrophy of less than 1%/year seen in suspected AD need to be examined further whereas atrophy of more than 1%/year may be used as a marker of abnormal atrophy suggestive of AD or other degeneration.



Scheltens P, Boscher L. Graduate School for Neurosciences Amsterdam, Research Institute Neurosciences, Vrije Universiteit, Amsterdam, The Netherlands Medial temporal lobe atrophy (MTA) as seen in Alzheimer’s disease (AD) can be assessed with volumetric and visual measurements on MRI scans. We conducted an evidence-based review of the utility of MRI assessment of MTA to distinguish AD from normal aging (aim 1) en from other types of dementia (aim 2). We conducted a Medline search with the key words: Alzheimer’s disease, MRI, hippocampus, volumetric-, area-, linear measurement and visual assessment. Studies were included if they used an objective clinical diagnostic reference as the gold standard, usually the NINCDS-ADRDA criteria and/or DSM III-R or IV, with or without follow up. We only included studies meeting criteria for Class I or II or “class I or II” evidence and when a likelihood ratio (LR) could be calculated. We found 27 studies meeting our inclusion criteria. The majority of studies involve small and selected groups. The overall LR for aim 1 was approximately 10 and the overall LR for AD vs. non-AD is around 6, based on 9 studies. Studies using visual analysis yield better results than those in which volumetry analysis was used. Of the 10 studies using visual analysis only 8 had class 1 (A or B) evidence. Volumetric studies have an aura of greater reliability and accuracy, but the data in this review do not support this notion. Probably due to the large variety in the techniques used in studying hippocampal and other medial temporal lobe volumes, a strict and direct comparison of studies is problematic. The evidence coming from these studies was of class 1 (A or B) strength in 11 of the 19 studies reviewed. Two studies that directly compared the two methods showed results in favour of visual analysis with class 1 evidence. Of the 9 studies comparing AD patients to patients with other dementias 3 showed no significant difference in MTA. In the other 6 a difference was found but specificity was lower for distinguishing AD from other dementias than from healthy controls. We conclude that data are still limited, especially concerning the value of MTA in differentiating AD from other dementias and mild cognitive impairment, probably the most relevant question for clinical practice. No studies actually used postmortem data as the golden standard and sample sizes were generally small, with a few exceptions. As diagnostic marker to adequately differentiate AD from normal aging, assessment of MTA has a reasonably high LR Given the fact that an MRI study will increasingly being made in clinical practice, adding coronal slices will yield valuable extra information at no extra cost or burden for the patient. In this respect, making one MRI examination during the course of a dementia evaluation can hardly be judged to be optional anymore. Volumetric quantification of MTA is still confined to research settings because of the poor generalisability and agreement between centres.



Frisoni GBa, Resconi Gb, Colosio Lb, Geroldi Ca, DeCarli Cc. aLaboratory of Epidemiology & Neuroimaging, IRCCS San Giovanni di Dio FBF, Brescia, Italy, and b Department of Mathematics and Physics, Universita Cattolica del Sacro Cuore, Brescia, ` Italy, cAlzheimer’s Center, Kansas University Medical Center, Kansas City, KS, USA Background: MR-based volumetric analysis of regional brain volumes is gaining widespread acceptance in clinical and epidemiological settings. Regional analyses are usually accomplished with techniques such as operator guided tracing, thresholding or topologic sampling. These techniques are generally accurate, but time consuming. The availability of automated segmentation algorithms would allow greater utility for MRI quantification. Aim: To develop an algorithm based on artificial intelligence (AI) that automatically separates brain from non-brain tissues as the first step to a completely automated regional segmentation algorithm. Method: The algorithm is divided into 3 parts. Part I uses an AI-based morphogenetic neuron model to generate a mathematical model of the brain image. Part II uses local splines to identify the brain borders and Part III extracts the brain from the skull. The algorithm is completely automatic. Results: An example of the algorithm is shown below. The raw image is in the center, to the left the image after processing with the artificial intelligence algorithm, and to the right is a segmented image based on the threshold method of DeCarli et al., 1992 after manual separation of brain from non-brain tissues. The algorithm works equally well on different slices of the same subject and on homologous slices of different subjects. Further validation of the method is ongoing. Conclusions: The method is promising to automatically segment the brain from non-brain tissue without operator intervention.


Poster Presentations


Wahlund L-Oa, Julin Pa, Johansson S-Ec, Scheltens Pa,b. aDepartment of Clinical Neuroscience, Occupational Therapy and Elderly Care Research, Karolinska Institute, Huddinge University Hospital, Huddinge, Sweden, bDepartment of Neurology, Academisch Ziekenhuis Vrije Universiteit, Amsterdam, The Netherlands, cDepartment of Statistics, Stockholm University, Stockholm, Sweden Objectives: It has been shown that atrophy of medial temporal lobe structures such as hippocampus and entorhinal cortex on MRI may distinguish Alzheimer’s disease patients (AD) form healthy controls. However, the diagnostic value of visual inspection and volumetry of medial temporal lobe atrophy (MTA) on MRI in a clinical setting is insufficiently known. Methods: MTA from 143 subjects was visually rated from hard copies, using a 0 – 4 rating scale and a comparison was made with the volumes (ccm) of the medial temporal lobe as estimated with volumetry, using a stereological method. All subjects were recruited in an unselected way in a clinical setting in the center for memory impairments at the Huddinge University Hospital. AD patients (n 41), patients with other dementias (OD) (Vascular Dementia, Fronto-Temporal Dementia and Unspecified Dementia) (n 36) as well as non-demented subjects (n 66) were included. MTA and volumetry were evaluated as a diagnostic tool by performing logistic regression analysis including age, sex and MMSE score and calculating the sensitivity and specificity and percentage correct classification. Results: Visual and volumetric analysis yielded statistical significant differences between AD patients and non-demented subjects, as well as between the OD and non-demented subjects. Combining MMSE scores and visually rated MTA ratings yielded a sensitivity of 95% for AD, 85% for OD. Non-demented subjects were identified with a specificity of 96%. Volumetry did not have an added value over the MMSE score alone. Conclusions: Visual rating of MTA is a clinically useful method for differentiating AD from controls and is both quicker and more accurate than volumetry.

Shared By: