voxel-based morphometry in alzheimer’s disease

12
Author Proof 1 Review www.expert-reviews.com ISSN 1473-7175 © 2008 Expert Reviews Ltd 10.1586/14737175.8.11.xxx Major improvements in neuroimaging tech- niques have been seen during the past decades, including both technological advances for data acquisition and development of new models for image processing and analysis. Such achieve- ments have allowed increasingly more detailed assessments of structural and functional aspects of the human brain in vivo. As a consequence, a great deal of novel information has been gath- ered about the normal development and aging of the human brain. Also, the patterns of brain abnormalities associated with several psychiatric, neurological and neurodegenerative disorders have been progressively uncovered. Among the neurodegenerative disorders, Alzheimer’s disease (AD) is the most frequent cause of dementia in the general population. Its prevalence is estimated to grow substantially in the near future, as the life expectancy is con- tinuously increasing in several countries. This is predicted to cause a major impact in health systems across the world [1–3] . AD is a highly heritable condition [4] . Major efforts have been made to map its biological markers in order to enable the development of secondary preventive treatments for this disorder [5] . Modern MRI techniques are now widely used to investigate regional volume def- icits and other structural brain abnormalities associated with the vulnerability to develop AD and its progression. Morphometric brain abnor- malities, as assessed with MRI, are among the potential biological markers for AD that have attracted the greatest deal of research interest in recent years. Both in morphometric MRI studies and func- tional imaging investigations of AD, regional brain indices are usually obtained in groups of subjects affected by this disorder in compari- son to control groups of healthy individuals, matched for demographic variables. The two approaches most widely employed to perform such quantitative measurements and between- group comparisons are region-of-interest (ROI) and voxel-by-voxel analysis methods. The ROI methods involve the delineation of selected anatomical brain structures, most often manually, in order to obtain quantita- tive volumetric indices for such regions. Thus, ROI-based approaches do not allow the mea- surement of regional brain volumes in a whole brain fashion. In morphometric MRI studies Geraldo F Busatto , Breno S Diniz and Marcus V Zanetti Author for correspondence Centro de Medicina Nuclear, 3º andar, LIM-21,Rua Dr. Ovídio Pires de Campos, s/n 05403– 010, São Paulo, SP, Brazil Tel.: +55 113 069 8193 Fax: +55 113 082 1015 [email protected] Recent morphometric MRI studies have investigated brain volume abnormalities associated with the diagnosis of Alzheimer’s disease (AD) using voxel-based morphometry (VBM). This technique allows the assessment of gray matter volumes in subjects with AD or related conditions compared with healthy controls in an automated fashion, across the whole brain. This article reviews VBM findings related to different AD stages and its prodrome, mild cognitive impairment. These findings include not only gray matter deficits in medial temporal structures as seen in former MRI studies of AD conducted using manual region-of-interest measurements, but also volume changes in several other brain regions not assessed in previous MRI studies. We also discuss potential applications of VBM to improve AD diagnostic accuracy in routine clinical practice. Finally, we highlight future research directions in this field, including: investigations on the relationship between VBM findings of multifocal gray matter deficits and changes in white matter tracts that interconnect such regions; the need for population- based VBM studies using large AD samples; and the potential of studies combining VBM measurements with other potential biological markers (such as brain imaging indices of amyloid- β deposition and cerebrospinal fluid AD markers) to further advance our knowledge about the physiopathology of AD. KEYWORDS: Alzheimer’s disease • MRI • mild cognitive impairment • neurodegenerative disease • neuroimaging • voxel-based morphometry Voxel-based morphometry in Alzheimer’s disease Expert Rev. Neurother. 8(11), xxx–xxx (2008)

Upload: agachamento

Post on 12-Jan-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Author Pro

of

1

Review

www.expert-reviews.com ISSN 1473-7175© 2008 Expert Reviews Ltd10.1586/14737175.8.11.xxx

Major improvements in neuroimaging tech-niques have been seen during the past decades, including both technological advances for data acquisition and development of new models for image processing and ana lysis. Such achieve-ments have allowed increasingly more detailed assessments of structural and functional aspects of the human brain in vivo. As a consequence, a great deal of novel information has been gath-ered about the normal development and aging of the human brain. Also, the patterns of brain abnormalities associated with several psychiatric, neurological and neurodegenerative disorders have been progressively uncovered.

Among the neurodegenerative disorders, Alzheimer’s disease (AD) is the most frequent cause of dementia in the general population. Its prevalence is estimated to grow substantially in the near future, as the life expectancy is con-tinuously increasing in several countries. This is predicted to cause a major impact in health systems across the world [1–3].

AD is a highly heritable condition [4]. Major efforts have been made to map its biological markers in order to enable the development of secondary preventive treatments for this

disorder [5]. Modern MRI techniques are now widely used to investigate regional volume def-icits and other structural brain abnormalities associated with the vulnerability to develop AD and its progression. Morphometric brain abnor-malities, as assessed with MRI, are among the potential biological markers for AD that have attracted the greatest deal of research interest in recent years.

Both in morphometric MRI studies and func-tional imaging investigations of AD, regional brain indices are usually obtained in groups of subjects affected by this disorder in compari-son to control groups of healthy individuals, matched for demographic variables. The two approaches most widely employed to perform such quantitative measurements and between-group comparisons are region-of-interest (ROI) and voxel-by-voxel analysis methods.

The ROI methods involve the delineation of selected anatomical brain structures, most often manually, in order to obtain quantita-tive volumetric indices for such regions. Thus, ROI-based approaches do not allow the mea-surement of regional brain volumes in a whole brain fashion. In morphometric MRI studies

Geraldo F Busatto†, Breno S Diniz and Marcus V Zanetti†Author for correspondenceCentro de Medicina Nuclear, 3º andar, LIM-21,Rua Dr. Ovídio Pires de Campos, s/n 05403–010, São Paulo, SP, Brazil Tel.: +55 113 069 8193 Fax: +55 113 082 1015 [email protected]

Recent morphometric MRI studies have investigated brain volume abnormalities associated with the diagnosis of Alzheimer’s disease (AD) using voxel-based morphometry (VBM). This technique allows the assessment of gray matter volumes in subjects with AD or related conditions compared with healthy controls in an automated fashion, across the whole brain. This article reviews VBM findings related to different AD stages and its prodrome, mild cognitive impairment. These findings include not only gray matter deficits in medial temporal structures as seen in former MRI studies of AD conducted using manual region-of-interest measurements, but also volume changes in several other brain regions not assessed in previous MRI studies. We also discuss potential applications of VBM to improve AD diagnostic accuracy in routine clinical practice. Finally, we highlight future research directions in this field, including: investigations on the relationship between VBM findings of multifocal gray matter deficits and changes in white matter tracts that interconnect such regions; the need for population-based VBM studies using large AD samples; and the potential of studies combining VBM measurements with other potential biological markers (such as brain imaging indices of amyloid-β deposition and cerebrospinal fluid AD markers) to further advance our knowledge about the physiopathology of AD.

Keywords: Alzheimer’s disease • MRI • mild cognitive impairment • neurodegenerative disease • neuroimaging • voxel-based morphometry

Voxel-based morphometry in Alzheimer’s diseaseExpert Rev. Neurother. 8(11), xxx–xxx (2008)

Author Pro

of

Expert Rev. Neurother. 8(11), (2008)2

Review Busatto, Diniz & Zanetti

of AD, the ROI-based approach has been employed to provide volumetric indices most often of the hippocampus and other selected medial temporal lobe portions [6–9]. Manual measure-ments are usually performed in several thin brain slices along the extension of the brain structure of interest, and are therefore labor intensive, time consuming and subject to observer bias. Recently, automated methods have been developed for measur-ing hippocampal volume in a faster way, including the determi-nation of shape models of the hippocampus through parametric deformation algorithms [5,8,10]. However, these automated meth-ods for hippocampal volumetry still need further evaluation. Also, variable anatomical borders have been used across different manual and automated ROI-based MRI studies of AD [6–9], and this may limit data comparisons between separate studies and the conduction of meta-analyses of the results obtained.

As an alternative to ROI-based measurements, many neuroim-aging studies over the past two decades have employed automated methods that allow hypothesis-free voxelwise comparisons of regional brain indices across the entire brain, without requir-ing the a priori selection of anatomical ROI borders and with perfect repeatability. This approach was originally developed for PET studies of brain glucose metabolism and regional cerebral blood flow, based on the use of the Statistical Parametric Mapping (SPM) program [11]. More recently, there has been widespread application of the same concept for morphometric and shape measurements using volumetric MRI sequences, as well as for the analysis of diffusion tensor imaging (DTI) and functional MRI data [12–15].

The application of the voxel-based approach for comparisons of regional gray matter volumes between different subject groups studied with MRI is referred to as voxel-based morphometry (VBM) [16,17]. The VBM technique initially involves spatial transformations of the MRI scans of all subjects included in a study onto a common anatomical space, in order to remove interindividual variations in brain size and shape. In such spatial normalization processes, images are conformed by linear and nonlinear transformations to a standard template (based on databanks of images normalized to a stereotactic atlas), or to a customized template created specifically for the study (obtained from a pool of images from the population under investiga-tion). Subsequently, the images of each subject are automati-cally segmented into gray matter, white matter and cerebral spinal fluid compartments, and smoothed with a Gaussian filter [16,17]. Gray matter segments are then compared statistically between groups on a voxel-by-voxel basis, and statistical maps are produced in standardized brain space showing the location of voxel clusters where significant between-group differences in mean grey matter volumes are present, at a predefined statistical level of inference. The voxel-based approach presents advan-tages in comparison to ROI methods as it is fully automated, rater independent and capable of investigating the presence of AD-related morphometric gray matter abnormalities across the whole brain (rather than solely on selected brain regions) in large samples of AD subjects relative to healthy controls, with high reproducibility.

This article aims to provide a comprehensive review of VBM studies of AD conducted to date. We discuss the current VBM-based knowledge about the patterns of gray matter abnormalities related to AD at its different stages, as well as to mild cognitive impairment (MCI), accepted nowadays as a prodromal condition of AD [18]. We also discuss the potential applications of VBM in the diagnosis of AD, as well as future research directions in this field.

Profile of gray matter abnormalities in ADSeveral VBM studies have assessed differences in regional gray matter volumes in AD samples compared with healthy con-trols. The most consistent finding across such studies is the presence of atrophy in medial temporal lobe structures: hip-pocampus, amygdala, entorhinal cortex and parahippocam-pal gyrus [14,19–41], which are already detectable at early stages of the disease [20,22,27,31,32,38,41]. However, with a lesser degree of consistency, the volumes of other brain regions have also been frequently identified to be reduced in AD, including: the temporal neocortex [14,20,22–26,32–36,40,41], parietal neocortex [20,24,26,28,33,34,36,37,40,41], insula [14,19,22,25,32–35,41], precuneus [20,22–24,26,28,33,41], anterior cingulate cortex [14,22,32,33,41], pos-terior cingulate cortex [20,22–24,33–35,41], frontal cortex [14,20,22,24,28,32,33,35,37,41], thalamus [14,20,24,25,33,35,37] and caudate nucleus [14,19,20,22,25]. Finally, AD-related volumetric reductions of the putamen [20,33], basal forebrain [14,35], cuneus [41], occipital cor-tex [34,41] and hypothalamus [20] have been occasionally reported in VBM studies.

The variability of results in the VBM studies that have inves-tigated brain volumetric abnormalities in association with AD to date may relate to methodological differences. For instance, a number of recent MRI studies of AD have added steps to the standard VBM protocol in order to optimize its performance and minimize image processing artifacts in the stages of spatial normalization and segmentation [17,24,42]. Also, highly variable sta-tistical thresholds have been employed for voxelwise comparisons across different VBM studies. With regard to sample character-istics, several studies have enrolled patients with heterogeneous clinical features and at different stages of AD. Finally, there may be added heterogeneity of findings depending on the criteria used for selecting healthy control subjects, as well as due to difficul-ties in defining what a ‘healthy’ elderly cohort is. Most studies have selected, as healthy control subjects, cognitively intact elderly individuals with no history of neurological or general medical conditions that could grossly affect brain structure. Such healthy individuals may frequently suffer from clinical conditions that could potentially affect brain anatomy and function, such as arterial hypertension, thyroid dysfunction and others, but the exclusion of those subjects would lead to the selection of healthy control groups that are not representative of the general elderly population. Also, healthy control groups selected based on clinical criteria may lead to the inclusion of subjects who carry genetic mutations that are known risk factors for dementia and which may be associated with anatomical variations in the brain, even in the absence of objective cognitive impairment.

Author Pro

of

www.expert-reviews.com 3

ReviewVBM in Alzheimer’s disease

A number of VBM studies have attempted to relate find-ings of gray matter abnormalities in AD patients to specific clinical features and stages of the disease. Such an approach is potentially helpful to delineate more clearly the profiles of brain abnormalities that may be present in association with AD. For instance, cross-sectional studies on samples of very mild AD patients revealed gray matter decreases affecting pre-dominantly medial temporal lobe structures relative to con-trols at such early disease stages [22,27,31,38]. Using a different study design, Matsuda et al. (2002) prospectively assessed 15 mild AD patients and 25 controls, and found that AD-related decreased gray matter volumes were present at the outset in the hippocampus, amygdala, parahippocampal gyrus, superior and inferior temporal gyri, as well as in the posterior cingulate gyrus and precuneus bilaterally [23]; after 1 year, AD-related volumetric changes spread to surrounding areas including the thalamus, anterior cingulate cortex, caudate nucleus and lateral parietal cortex.

Morphometric MRI studies have also investigated the pres-ence of correlations between patterns of gray matter loss and the degree of cognitive impairment in subjects with dementia. VBM studies in AD and MCI patients have shown that volume reduc-tions of the whole brain or in circumscribed brain regions such as the hippocampus, entorhinal cortex, parahippocampal gyrus, middle temporal gyrus, anterior cingulate cortex and posterior parietal cortex, may be associated with worse global cognitive performance, or specific impairments in learning and recent epi-sodic memory, along with visuo-constructional, executive and arithmetic deficits [37,43,44].

Also, a number of VBM studies have investigated differ-ences in the patterns of gray matter atrophy between early- and late-onset AD individuals. Frisoni et al. found that early-onset AD subjects presented atrophy in the hippocampus, tempo-ral-parietal cortex and precuneus, whereas late-onset patients showed predominant involvement of the inferior temporal lobe, as well as a greater degree of hippocampal atrophy [26]. Ishii et al. (2005) reported gray matter reductions in the right para-hippocampal gyrus, inferior frontal gyrus and precuneus, the inferior parietal lobule bilaterally, and the left operculum in association with early-onset AD [28]; differently, late-onset AD subjects displayed atrophic findings circumscribed to the hip-pocampus bilaterally. Consistent with the later findings, Karas et al. (2007) found that a younger age of AD onset was associ-ated with lower gray matter density in the precuneus [45].In addition, Kinkingnéhun et al. (2008) prospectively followed 23 subjects with mild AD and 18 controls for 3 years and, at the end of this period, dichotomized AD subjects in slow and fast decliners based on Mini-Mental State Examination (MMSE) score decrements over time [41]. Compared with slow decliners, fast declining AD patients showed greater tissue loss in the pre-cuneus, lingual gyrus, cuneus and the cortex surrounding the parieto-occipital sulcus bilaterally. Taken together, the evidence from these studies suggest that specific patterns of gray matter abnormalities vary depending on different clinical aspects of AD, such as age of onset and rate of progression over time.

Thus, distinct patterns of regional volume deficits may represent biological markers, for example, endophenotypes, character-izing different forms of AD. Interestingly, Shiino et al. (2006) found four different patterns of gray matter atrophy in a group of 40 patients with mild-to-moderate AD [34]. Patients with brain volume deficits predominantly located in medial temporal structures (amygdala, anterior and posterior hippocampus, and uncus) had a probability of 100% for a disease duration of less than 36 months, whereas those with additional involvement of posterior lateral cortices (in inferior temporal, parietal and occipital regions) and the posterior cingulate gyrus/precuneus had a 90% probability for a disease onset earlier than 65 years of age (early-onset AD). In conclusion, the VBM-based evidence to date, although still limited, suggest that AD patients presenting with greater involvement of posterior associative cortices and precuneus/posterior cingulate gyrus are more likely to have an earlier onset and a faster rate of disease progression.

Finally, studies employing the VBM technique have also assessed the profile of gray matter abnormalities in groups of AD subjects relative to patients suffering from other neurodegenera-tive conditions, such as dementia with Lewy bodies [36,46–48], frontotemporal lobar degeneration [40,49], Parkinson’s disease dementia [47] and posterior cortical atrophy [35]. These studies, although as yet limited in number and with some heterogene-ity in the reported results, suggest that a pattern of gray matter atrophy involving mainly the medial temporal lobe, insula and temporoparietal cortices is specific for AD.

VBM studies assessing gray matter deficits in association with MCI & its conversion to ADThe presence of amnestic MCI implies an increased risk for the development of dementia/AD and may be seen, in some cases, as a prodromal AD stage [18,50]. Given this assumption, patients with amnestic MCI should share some common neuroimaging features with AD patients, although in lesser intensity. Indeed, a number of VBM studies have demonstrated MCI-related regional gray matter abnormalities similar to those seen in AD patients relative to healthy controls. These include: atrophy of hippocampus and amygdala; gray matter reductions in other brain regions (although more circumscribed than those seen in AD), affecting predominantly the frontal and temporoparietal neocortices; and less consistent findings of gray matter deficits in the cingulate gyrus, insula, thalamus and caudate nucleus [29,34,51–55]. When compared directly with AD patients, MCI subjects showed relative volume preservation of the posterior association cortex [51,52], anterior and posterior cingulate cortex, and left amygdala [52].

More recently, the concept of MCI has been broadened to include not only single amnestic deficits, but also single non-am-nestic deficits (single non-amnestic MCI) and multiple domain (amnestic and non-amnestic) MCI [56].It has been hypothesized that these distinct MCI subtypes may have different outcomes, with the amnestic (single or multiple-domain) MCI subtype being associated with a higher risk of progression to AD [57]. Recent VBM studies have investigated whether subclassifications

Author Pro

of

Expert Rev. Neurother. 8(11), (2008)4

Review Busatto, Diniz & Zanetti

of MCI subjects are associated with different neuroanatomical substrates [53]. Compared with multiple-domain MCI patients, subjects with amnestic MCI have been found to display sig-nificant volume loss of the left entorhinal cortex and inferior parietal lobe, while patients with multiple-domain MCI showed significantly reduced volumes of the right inferior frontal gyrus, right middle temporal gyrus and bilateral superior temporal gyri [53]. Whitwell et al. investigated gray matter abnormalities in 145 patients with different MCI subtypes (amnestic, single cognitive domain; amnestic, multiple domain; non-amnestic, single or multiple domain – subclassified in language, atten-tion/executive, and visuospatial subgroups) compared with controls and found different patterns of atrophy across such groups [58]. Subjects in the amnestic MCI groups showed sig-nificant gray matter reductions localized to the medial and inferior temporal lobes, while those in the multiple-domain group also had involvement of the posterior temporoparietal association cortex, and posterior cingulate gyrus; subjects in the non-amnestic single-domain group with language impairment presented with atrophy in the left anterior inferior temporal lobe, whereas those with attention/executive deficits displayed significant gray matter loss in the basal forebrain and hypothala-mus. Taken together, these findings demonstrate that amnestic MCI (including its single- and multiple-domain subtypes) is associated with a prominent reduction of medial temporal lobe structures (mostly the hippocampus and entorhinal cortex). Notwithstanding, amnestic multiple-domain MCI subjects display a more widespread pattern of volume reduction involv-ing neocortical areas, resembling the pattern often seen in AD patients. These findings may reflect the anatomical substrate of a continuum from mild amnestic deficits to more widespread cognitive impairments and, finally, to AD [59,60]. By contrast, patterns of anatomical abnormalities in non-amnestic MCI vary depending on the most prominent cognitive deficits, probably reflecting distinct pathological processes across MCI subtypes, with different outcomes.

Prospective studies using the VBM approach have longitu-dinally assessed MCI patients over time, with an aim to iden-tify gray matter abnormalities potentially associated with the conversion to AD. In this regard, it is interesting to note that Resnick et al., in the only prospective VBM study of normal aging published to date, found significant gray matter loss in cingulate and insular cortex, orbital and inferior prefron-tal cortex, inferior parietal region and, to a lesser extent, the mesial temporal cortex, in 92 non-demented adults (mean age 70.4 ± 7.0 years) over a 4-year period [61]. Chételat et al. found that in comparison to non-converters, MCI subjects who con-verted to AD after 18 months already presented, at baseline, findings of reduced gray matter encompassing the right pos-terior hippocampus, parahippocampal gyrus, and lingual and fusiform gyri [62]. Moreover, the MCI subjects who later con-verted to AD demonstrated, relative to non-converters, greater gray matter loss over the 18-month period in the hippocampus and parahippocampal region, inferior and middle temporal gyri, fusiform gyrus, posterior cingulate gyrus and precuneus

bilaterally. Conversely, Bell-McGinty et al. reported that MCI subjects who converted to AD (over a mean interval of 2 years) revealed greater atrophy in the left entorhinal cortex, bilateral superior temporal gyri, and right inferior frontal gyrus com-pared with non-converters [53]. Bozzali et al., when compar-ing converters and non-converting MCI patients with healthy controls (over a mean follow-up period of 28.7 ± 5.7 months), found that the converters showed more widespread areas of reduced gray matter density than non-converters, with a pattern of abnormalities similar to that seen in patients with AD [33]. Moreover, the converting MCI subjects displayed significant gray matter decreases in the inferior frontal gyrus bilaterally, the left supramarginal gyrus and the right hippocampus relative to non-converting MCI individuals. Whitwell et al. also found a pattern of bilateral gray matter loss affecting the medial and inferior temporal lobe, temporoparietal association cortex and frontal lobe of patients with amnestic MCI who progressed to AD compared with controls [63]. Those same converting MCI subjects presented greater gray matter loss in the medial and inferior temporal lobes, temporoparietal cortex, anterior and posterior cingulate gyrus, precuneus and frontal lobe than amnestic MCI patients who remained stable for 3 years. Another study, conducted by Karas et al., reported that MCI subjects who converted to AD showed significantly greater atro-phy in left parietal regions (encompassing the angular gyrus and inferior parietal lobule) and in the lateral temporal neocor-tex (involving the superior and middle temporal gyri) relative to non-converters [64]. Whitwell et al. assessed longitudinal changes in gray matter volumes in 33 amnestic MCI patients who converted to AD after 3 years and in 33 healthy controls with serial MRI scanning [65]. At the outset, converting MCI subjects displayed localized foci of gray matter atrophy in the medial temporal lobe (involving the amygdala, anterior hip-pocampus and entorhinal cortex) and fusiform gyrus bilaterally relative to control individuals. At 1 year before conversion to AD, such atrophic changes spread to the middle temporal gyrus and the entire extent of the hippocampus. Finally, at the time of clinical diagnosis of AD, the group of patients showed more severe involvement of the medial temporal lobes, temporopari-etal association cortices and frontal lobes compared with con-trols. Finally, a recent longitudinal VBM study demonstrated that MCI subjects who presented a rapid conversion to AD over an 18-month follow-up period exhibited an accelerated atrophy rate in the hippocampus, mainly involving its lateral-superior portion (CA1 subfield), compared with non-converting MCI subjects [66].

Despite the relatively large number of longitudinal VBM inves-tigations published in the past 3 years, it should be noted that the specific brain regions related to the conversion from MCI to AD are not yet completely clear, as there has been some degree of inconsistency in the findings reported across different studies. Moreover, differences in the clinical follow-up length of MCI subjects in each MRI study may be an important source of vari-ability of VBM findings. With longer follow-up periods, there is a greater degree of certainty that non-converting MCI subjects

Author Pro

of

www.expert-reviews.com 5

ReviewVBM in Alzheimer’s disease

will not progress to AD shortly after the end of the study, and this is likely to afford better discrimination between converters and non-converters in regard to regional brain volumes. Finally, the use of variable operational criteria for MCI may also influence the rate of conversion to AD and lead to inconsistent findings across separate imaging studies [67].

In summary, both mild AD and MCI are associated with gray matter reductions affecting mainly medial temporal lobe structures (hippocampus, amygdala, entorhinal cortex and parahippocampal gyrus). Over the disease course, and at the progression from MCI to clinical AD, such atrophic changes may spread to other brain regions, preferentially to the fron-tal and temporoparietal association cortices, cingulate gyrus, insula and precuneus, as well as to the thalamus and basal ganglia. The documentation of volumetric brain changes in several separate brain regions are a specific major contribution of VBM studies to our understanding of the physiopathology of AD, as no previous ROI-based MRI study of this dement-ing disorder had been able to simultaneously assess such a wide range of brain structures.

Diagnostic performance of VBM in ADApart from their contribution to clarify the pathophysiology of AD, morphometric MRI studies of AD have attempted to ascer-tain whether in vivo volume measurements of key brain regions can be applied in clinical practice. ROI-based MRI studies have shown that volumetric measures of medial temporal structures (hippocampus, amygdala, entorhinal, perirhinal and parahip-pocampal cortices) may be useful to aid clinicians in their diag-nostic work up, with an accuracy to differentiate AD patients from age-matched healthy controls of up to 85% (particularly for the parahippocampal gyrus) and greater than 90% if combined measurements of such structures are used [68]. Recently, the diag-nostic accuracy of VBM-based measures of brain atrophy in AD has also been investigated in a number of studies. These studies have assessed the diagnostic performance of VBM employing receiver operating characteristic (ROC) curve ana lysis with indi-vidual Z scores, i.e., the deviation of each individual regional gray matter from the mean of the control group, in selected regions derived from the voxelwise comparison maps between AD patients and controls [28,30,32,39,69]. This allows the plotting of the Z value in a true positive rate versus false positive rate graphic display, aiming to obtain the discrimination accuracy for each brain region, i.e., the value at the point where the sensitivity is the same as the specificity on the ROC curve. Such a VBM-derived approach affords good diagnostic performance in differentiating AD patients from healthy controls, with sensitivity, specificity, accuracy and areas under the ROC curve (Az) values ranging from, respectively, 74 to 85%, 80 to 92%, 83 to 88% and 0.83 to 0.96 [28,30,32,39,69], even at early stages of the disease [28,32,39]. Moreover, Ishii et al. (2005) found similar diagnostic perfor-mances of VBM in both early- and late-onset AD cases [28]; Az values equaled 0.9435 for early-onset AD patients and 0.9018 for late-onset AD patients, with no significant between-group statistical differences.

It is highly relevant to assess the performance of VBM-based morphometric MRI measurements in the diagnosis of AD in direct contrast against other in vivo imaging approaches that have been used in clinical practice. For instance, one study found VBM to have superior diagnostic accuracy (Az = 0.96) than ROI-based hippocampal volumetry (Az = 0.89) [69]. However, the combination of both methods resulted in the best performance (Az = 0.99). Two other studies directly compared the diagnostic performance of VBM with that of 18F-fluoro-deoxyglucose PET (FDG-PET). Kawachi et al. (2006) found that FDG-PET yielded a slightly higher diagnostic accuracy (89%) compared with VBM (83%) in individuals with very mild AD, but the highest accuracy was achieved by the com-bination of both techniques (94%) [32]. Matsunari et al. also reported a better diagnostic performance of FDG-PET (sensitiv-ity, specificity and accuracy of 100, 92 and 95%, respectively) relative to VBM (74, 92 and 85%) in AD, but this superiority was only significant in the early-onset subgroup [39]. Only one study to date assessed the diagnostic performance of VBM in AD compared with a different form of neurodegenerative dis-order [49]. With the use of a discriminant function analysis, this study demonstrated that VBM-based atrophy findings in posterior cingulate, parietal, amygdalar and anterior temporal lobe regions correctly classified 96% of the patients with AD relative to subjects with semantic dementia, a variation of fron-totemporal lobar degeneration.

Importantly, all of the above studies were conducted with samples of modest size. Thus, further studies with larger sam-ples and enrolling patients with different neurodegenerative conditions – a situation closer to the real clinical setting – are needed to better delineate the diagnostic performance of VBM in AD. Moreover, while these results provide strong support to the notion that specific patterns of gray matter changes can be employed to reinforce the diagnosis of AD, it is important to stress that VBM methods were designed to perform mean group comparisons for research purposes, and are not suitable for diag-nosis of individual cases. Novel automated methods are being developed to fulfill the specific aim of improving the diagnostic performance of morphometric MRI images on an individual basis, and this will be discussed in greater detail in subsequent sections of this article.

Relationship between morphological brain changes & other biological markers of ADStudies combining volumetric MRI measurements and biochem-ical analyses may help to integrate different aspects of the brain pathological profile of AD, and generate combined strategies that may yield improved AD diagnostic performance for use in clinical practice. A number of morphometric MRI studies have investigated such issues using ROI-based approaches, and it is expected that VBM studies will follow the same path in the near future.

Amyloid-β and Tau abnormalities are the pathological hall-marks of AD. Amyloid-β

42 protein is a cerebrospinal fluid (CSF)

marker which may reflect the deposition of tau protein into

Author Pro

of

Expert Rev. Neurother. 8(11), (2008)6

Review Busatto, Diniz & Zanetti

amyloid plaques, while total Tau and phosphorylated Tau CSF levels may reflect the degree of neuronal degeneration and tangle pathology in AD [70,71]. In recent ROI-based MRI investiga-tions, higher CSF levels of phosphorylated Tau were negatively correlated with decreased hippocampal volumes in AD patients [72]. Also, increased hippocampal atrophy rates were correlated with increasing levels of Tau protein and decreased levels of amyloid-β

42 in the CSF of AD patients in longitudinal assess-

ments [72]. The CSF profile characteristic of AD pathology, i.e., low amyloid-β

42, high total Tau and phospho-Tau levels [73], has

been found to be associated with significantly decreased hip-pocampal volumes in MCI subjects [74,75], in direct proportion to the degree of memory impairment displayed by these sub-jects [75]. Such findings indicate that the regional brain atrophic changes which are consistently related to AD, mainly in medial temporal lobe structures, are correlated to biochemical indices of pathogenic AD mechanisms even in patients at predementia stages. Also, there are data indicating that the association of the AD-related CSF pathological pattern and MRI-based assessments of medial temporal lobe atrophy may be of higher predictive value than each biomarker alone in the prediction of conversion to AD in MCI subjects on follow-up [76].

One other highly promising approach to investigate patho-logical AD indices in vivo has emerged with the development of selective ligands for amyloid-β imaging in the brain, such as the Pittsburgh Compound-B (PIB) for PET. One recent study combined PIB-PET imaging and VBM in eight AD patients, 17 amnestic MCI subjects and 20 cognitively intact individu-als. In AD patients, there was significantly higher global cortical PIB retention and decreased hippocampal volumes relative to the cognitively normal controls, whereas amnestic MCI individuals displayed an intermediate profile of both PIB retention and hip-pocampal atrophy [77]. Interestingly, there were differences in the topographical distribution of gray matter loss and PIB retention across separate brain regions in AD subjects. Relative to controls, AD patients displayed a pattern of little gray matter loss and high PIB retention in the frontal lobes, high gray matter loss and little PiB retention in the medial temporal lobes, and significant gray matter loss and PiB retention in the temporoparietal association cortices [77]. Consistent with such complementariness between the two methods, diagnostic accuracy using both of them in com-bination was higher compared with either imaging modality in isolation [77].

In view of the development of therapeutic strategies with potential disease-modifying properties [78], there is growing interest in the identification of subjects in the preclinical stages of dementia, or at higher risk of AD development. It would seem plausible, then, that the association of different assess-ment methods would give complimentary information about disease-related processes, thus increasing the accuracy of the early identification of such subjects. As seen above, only a small number of studies have directly addressed these topics to date. Despite their promising results, further investigations are neces-sary to address which are the best complimentary strategies to diagnose incipient AD and to identify those at the higher risk

to develop dementia. The VBM-based MRI approach may be particularly helpful to fulfill this aim, as it allows whole brain correlation of regional gray matter volumes with other potential biological markers of AD.

Indices of brain atrophy & the genetics of ADCurrently, the presence of the APOEε4 alleles is considered the genetic risk factor most consistently associated with late-onset AD [79]. The exact mechanisms whereby it increases the risk for AD is not known, but changes in the metabolism of cholesterol, Tau protein and amyloid-β

42 (and its brain clearance), have all

been hypothesized to be mediated by expression of APOEε4 alleles [80]. ROI-based MRI studies have shown that the presence of APOEε4 alleles is associated with decreased hippocampal and entorhinal volumes in AD patients as compared with non-carriers of this allele [81,82]. Similar findings have been reported in patients with MCI and non-demented elderly subjects carry-ing the APOEε4 allele [83,84]. However, in the only VBM study that assessed the effects of APOEε4 alleles on regional gray matter differences between early-onset AD, late-onset AD and control subjects [26], the inclusion of the APOE polymorphism as a nuisance covariate did not bring any significant changes to the voxelwise statistical maps displaying between-group gray matter differences.

Studies using VBM have also addressed whether the APOE genotype significantly affects regional brain volumes in non-demented subjects. Wishart et al. (2006) demonstrated that in cognitively intact adults (ranging from 18 to 80 years of age), the presence of a single copy of the APOEε4 allele determines volumetric reductions in the prefrontal cortex, right medial tem-poral region and other temporal lobe structures as compared with APOEε3 homozygous subjects [85]. They also found that these effects were more prominent in younger than older sub-jects. Lamaître and colleagues studied a large, population-based cohort of cognitively unimpaired elderly adults (ranging from 63 to 75 years of age) and found significant volume reductions bilat-erally in the medial temporal lobe (including the hippocampus and extending toward the superior temporal gyrus) in APOEε4 homozygous subjects in comparison to APOEε4 heterozygous and APOEε4 non-carrier subjects [86]. Differently, APOE4 heterozy-gous subjects did not display significant differences from APOEε4 non-carrier subjects.

In MCI subjects, the presence of the APOEε4 allele also seems to exert a major influence on regional brain areas vol-umes. Using VBM, Pennanen and colleagues (2006) showed that MCI patients heterozygous for the APOEε4 had signifi-cant volume reductions in the right parahippocampal gyrus and entorhinal cortex compared with MCI subjects not carrying the APOEε4 allele [87]. In the comparison of MCI cases homozygous for the APOEε4 allele versus APOEε4 non-carriers, the former group showed gray matter atrophy most prominently in the right amygdala, followed by the right parahippocampal gyrus, left amygdala and left medial dorsal thalamic nucleus. In a different investigation by the same group, Hämäläinen et al., MCI sub-jects that carried the APOEε4 allele and who later progressed to

Author Pro

of

www.expert-reviews.com 7

ReviewVBM in Alzheimer’s disease

dementia were found to present with significant volume reduc-tions in the left inferior frontal gyrus and the banks of the left intraparietal sulcus at baseline, compared with MCI patients who progressed to dementia and were APOEε4 non-carriers [88]. Therefore, despite minor inconsistencies, the VBM literature to date provides an indication that the presence of the APOEε4 allele has major effects on the volumes of cerebral areas which are known to be affected by the AD degenerative process, cor-roborating the fact that the presence of one or two APOE4 alleles confers an increased risk of AD. The influence of the APOEε4 allele on regional brain volumes is noticeable not only in subjects in the preclinical dementia stages (e.g., MCI subjects), but also in healthy young and older adults.

Genetic mutations in the presenilin 1 and 2 and amyloid pre-cursor protein (APP) genes determine an increased production of amyloid-β

42 peptides and are associated to familial early-onset

AD [89]. In a study investigating whole brain and hippocampal volumes using semi-automated techniques other than VBM, carriers of familial AD-related gene mutations (n = 9; five with APP and four with presenilin 1 mutations) were found to have smaller mean volumes than age-matched controls already in pre-symptomatic stages, and showed an accelerating whole-brain and hippocampal atrophy rate while progressing to MCI and clinical AD [10].

Limitations of the VBM approach Despite the advantages of the VBM methodology in allowing automated, easily applicable volume measurements across the whole brain, the drawbacks of this technique must be acknowl-edged. There may be biases in the segmentation process in brain areas where tissue contrast is poorly defined in MRI scans, and this may be particularly prevalent in the aging brain [90]. Also, there may be systematic registration errors during spatial normal-ization [91], especially in the proximity of brain regions where there may be large differences in size and shape between groups, such as the lateral ventricles [90,92]. Such registration errors may affect the validity of statistical inferences in VBM studies and lead to false-positive results [91]. Moreover, although image smoothing with Gaussian filters is necessary to make data more normally distributed for statistical parametric analyses, the use of large-sized filters may lead to problems in localizing between-group volumetric differences in brain regions of small size [93]. Thus, in contemporary VBM studies of degenerative dementias evaluat-ing small-sized medial temporal structures, it is highly relevant to conduct analyses using Gaussian filters of small size (e.g., 4 mm full-width at half-maximum). The danger of an uncritical interpretation of VBM findings has been made clear in recent large-scale studies of aging differences, which have documented discrepancies in the results obtained with VBM as compared with manual volumetry [90].

In order to address the limitations of the VBM approach, there have been recent improvements in models for spatial normaliza-tion, including mathematical transformations aimed at better preserving image size and shape, even in brain regions showing large differences between groups [94]. Also, the latest version of

the SPM program (SPM5) provides improved routines for image segmentation by unifying and optimizing the processes of tissue classification and registration to templates [95].

Expert commentary & five-year viewAs reviewed previously, recent VBM studies have provided robust evidence of gray matter abnormalities affecting several brain regions in association with the development of AD and its course. Findings of atrophy affecting medial temporal struc-tures, with progression to frontal and temporoparietal associative cortices, cingulate gyrus, insula, precuneus, thalamus and basal ganglia, have been replicated by several VBM studies that assessed gray matter abnormalities in AD. Important findings have also emerged from neuroimaging studies relating indices of brain atro-phy with other potential biological markers of AD and etiological risk factors for this disorder. However, there is still some degree of inconsistency in such studies, and further research is clearly needed in this field over the next few years.

In order to account for the clinical heterogeneity of AD and MCI, it will be mandatory to carry out studies with larger samples than those included in most VBM studies of AD to date. Future large-scale VBM studies of AD could employ epi-demiological designs to allow the recruitment of large samples of incident cases of AD and MCI from defined geographical regions, as performed in recent, population-based studies evalu-ating other brain disorders [96]. Population-based neuroimaging studies may also take account of local socioeconomic and other environmental influences on brain anatomy, by recruiting large groups of healthy neighbors as control subjects. Multicentric MRI studies are an interesting alternative to allow the investiga-tion of large samples of subjects with AD and related disorders [97]. However, there may be variable and complex environmental influences on brain structure across separate countries. Also, inter-scanner reliability may be low for specific brain regions [97–99], and calibration of different scanners may be difficult to perform.

One important topic to be further explored in the VBM literature on AD in the next few years relates to the investiga-tion of white matter abnormalities possibly associated with this disorder. While the vast majority of VBM studies to date were restricted to the investigation of the gray matter compartment, recent studies have provided evidence of white matter volume deficits in AD using the same voxel-based approach [100,101]. VBM studies exploring both gray and white matter compart-ments may help to clarify whether the findings of gray matter deficits in multiple brain areas in AD are related to abnormali-ties in the white matter tracts that interconnect such regions. Accordingly, one recent MRI study found that hippocampal atrophy in AD subjects was significantly related to deficits in the white matter fibers of the cingulum bundle, as well as with hypometabolism (as assessed with FDG-PET) in the posterior cingulate gyrus and several other interconnected brain regions [102]. One approach that is likely to provide most useful infor-mation to clarify hypotheses of white matter disconnection influencing on multifocal patterns of gray matter deficits in AD

Author Pro

of

Expert Rev. Neurother. 8(11), (2008)8

Review Busatto, Diniz & Zanetti

is the combination of VBM indices with white matter integ-rity measures using the novel MRI technique of DTI. Initial DTI studies have already provided evidence of changes in the microstructural integrity of white matter regions that intercon-nect gray matter foci of key relevance to the physiopathology of AD [14,15].

With regard to clinical applications, the VBM studies of AD reviewed herein demonstrated good diagnostic accuracy in dis-criminating patients with AD from healthy controls, even at early stages of the disease. One other key issue is that the dif-ferential diagnosis between specific causes of dementia can be very difficult in the early stages of the disease and may have therapeutic and prognostic implications in clinical practice. Unfortunately, however, data indicative that VBM indices of gray matter atrophy yield satisfactory accuracy in discriminat-ing AD from other causes of dementia are limited [49]. It must be borne in mind that VBM methods were designed to perform mean group comparisons for research purposes, and are not readily applicable for individual diagnostic assessments in rou-tine clinical practice.

In this regard, it is relevant to note that alternative ana lysis methods for structural MRI data have been developed to sup-port the diagnosis of neuropsychiatric conditions, enabling a fully automated categorization of individual structural or functional brain images based on machine-learning techniques, such as support vector machines (SVMs). The image processing routines involved in the SVM approach overlap significantly with those of VBM, including automatic image segmenta-tion and spatial normalization. Conversely, SVM methods apply multivariate, pattern recognition techniques to detect subtle and spatially complex patterns of morphological differ-ences using all the information in a brain scan simultaneously [103,104]. SVM is a supervised classification method, including a ‘training step’ about differences between the groups to be classified (e.g., AD or healthy). Individual scans are then tested against trained datasets, in order to be categorized as member of a particular group (e.g., AD). Recent studies have demon-strated the reliability and validity of SVM-based techniques, as well as its good diagnostic performance in discriminating between MCI and healthy control individuals [105], between AD subjects and healthy controls [106,107], and between AD sub-jects and patients with frontotemporal lobar degeneration [106]. Moreover, Klöppel et al. successfully used this method in large AD and healthy control samples pooled from data acquired at different scanning sites, thus suggesting the usefulness of this approach for use in a multicenter context [106]. Further SVM studies are needed in the next few years in order to further evaluate the diagnostic performance of this technique for the discrimination of AD from other neurodegenerative disorders, such as frontotemporal lobar degeneration and its subtypes (frontotemporal dementia, semantic dementia and progres-sive nonfluent aphasia), dementia with Lewy bodies and pro-gressive supranuclear palsy, among others. Nevertheless, one should bear in mind that obtaining a large, population-specific sample of ‘normal’ brains that could be used as a reference for

the evaluation of individual cases in different medical centers worldwide can be problematic for any method aiming to mea-sure brain morphometric parameters.

In the near future, it is expected that prospective morphomet-ric MRI studies with serial volumetric brain assessments will help to further delineate the patterns of progression of regional brain changes in the transition from normal aging, MCI and dementia. In this regard, it is interesting to highlight recent large-scale international controlled multicenter trials (such as the US, European, Australian and Japanese Alzheimer’s Disease Neuroimaging Initiative and the German Dementia Network) that have been engaged in the Phase III development of core feasible imaging biomarker candidates for AD, along with CSF and plasma biomarkers candidates [108,109,201]. These initiatives should help to further clarify the neurobiological bases of AD and other neurodegenerative disorders, and improve tools for individual diagnosis [110]. Preliminary results demonstrated the methodological adequacy and feasibility of patient enrollment for these studies [111,112], and future results will certainly add new and important information upon the use of neuroimaging tools associated with CSF and plasma biomarkers in the early diagnosis and prediction of AD.

Finally, as currently acknowledged, the accepted diagnostic criteria for MCI yields a heterogeneous population with great variability in the progression rates for AD [67]. This shortcoming may negatively influence the outcome of clinical trials address-ing the disease-modifying properties or the capacity to reduce progression rates from MCI to AD of novel treatment strategies. The addition of biological markers of the disease process may be highly desirable to reduce sample heterogeneity in studies of this kind. Indeed, the recent proposal for revision of the National Institute of Neurological Disorders and Stroke-Alzheimer Disease and Related Disorders (NINCDS-ADRDA) diagnos-tic criteria for AD includes volume loss in the hippocampus, entorhinal cortex or amygdala, as evidenced on MRI, as one of the three possible abnormal biomarkers (along with PET and CSF findings) that must be present (at least one) as a supportive feature of the early AD diagnosis [113]. In addition, there are recent indications of the potential usefulness of structural neu-roimaging biomarkers to track longitudinal changes associated with therapeutic interventions and also as outcome measures in clinical trials on AD samples [114,115]. The use of automated image ana lysis methods for such purposes is likely to add key informa-tion on the disease-modifying properties of novel therapeutic interventions for AD.

Financial & competing interests disclosureThis work was funded by grants from the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP), Brazil (CINAPCE program).

The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this review manuscript.

Author Pro

of

www.expert-reviews.com 9

ReviewVBM in Alzheimer’s disease

Key issues

Voxel-based morphometry (VBM) is a data processing technique for MRI that allows the detection of regional brain volume • abnormalities in groups of Alzheimer’s disease (AD) subjects compared with control groups in an automated fashion, across the entire brain.

VBM studies have consistently detected gray matter deficits in medial temporal structures in AD, extending findings previously obtained • with volumetric MRI measurements using conventional regions-of-interest (ROI) manually placed on selected brain portions.

By assessing the entire brain simultaneously, VBM studies have also allowed the detection of AD-related gray matter deficits in multiple • brain regions not assessed in previous ROI-based MRI studies.

In research studies, there is some VBM variability in the patterns of gray matter deficits seen in brain regions other than the medial • temporal lobe, possibly as a function of different clinical subtypes of MCI, age of onset of AD, rates of disease progression over time, and presence of gene mutations that act as risk factors for AD.

VBM findings of atrophy in posterior associative cortices and the precuneus/posterior cingulate gyrus are associated with an earlier • onset of AD and a faster rate of disease progression.

The VBM approach is sensitive to uncover significant gray matter reductions even in early AD stages or in subjects with minor cognitive • impairment (MCI) before conversion to AD, predominantly located in medial temporal lobe structures.

There are indications that regional brain volume measurements in individual cases obtained using VBM may improve AD diagnostic • accuracy, but it is not possible to clearly envisage the incorporation of this approach for routine use in clinical practice. Novel methods that overlap significantly with VBM, namely support vector machine techniques, are more likely to fulfill this aim, enabling a fully automated categorization of individual structural brain images, through the application of multivariate statistics to detect subtle and spatially complex patterns of morphological differences.

Future VBM studies with larger samples of MCI and AD patients are expected to disentangle the sources of variability of the regional • volume deficits that may emerge in association with such conditions, and thus provide valid biological markers (endophenotypes) characterizing different forms of AD.

Future research studies should also combine VBM measurements with other potential biological markers for AD, such as indices of • regional amyloid-β deposition obtained usingPET, as well as cerebrospinal fluid markers of AD pathology.

Such combined measurements may help to integrate different aspects of the brain pathology of AD, generate combined strategies to • improve AD diagnostic performance for use in clinical practice, and refine inclusion criteria for patient selection in clinical trials aimed at tracking longitudinal brain changes triggered by novel, disease-modifying therapeutic interventions for AD.

ReferencesFerri CP, Prince M, Brayne C 1 et al. Alzheimer’s Disease International. Global prevalence of dementia: a Delphi consensus study. Lancet 366(9503), 2112–2117 (2005).

Kawas CH, Corrada MM. Alzheimer’s and 2

dementia in the oldest-old: a century of challenges. Curr. Alzheimer Res. 3(5), 411–419 (2006).

Jönsson L, Eriksdotter Jönhagen M, 3

Kilander L et al. Determinants of costs of care for patients with Alzheimer‘s disease. Int. J. Geriatr. Psychiatry 21(5), 449–459 (2006).

Gatz M, Reynolds CA, Fratiglioni L 4 et al. Role of genes and environments for explaining Alzheimer disease. Arch. Gen. Psychiatry 63(2), 168–174 (2006).

Teipel SJ, Meindl T, Grinberg L, Heinsen H, 5

Hampel H. Novel MRI techniques in the assessment of dementia. Eur. J. Nucl. Med. Mol. Imaging 35(Suppl. 1), S58–S69 (2008).

Callen DJ, Black SE, Gao F, Caldwell CB, 6

Szalai JP. Beyond the hippocampus: MRI volumetry confirms widespread limbic atrophy in AD. Neurology 57, 1669–1674 (2001).

Jack CR Jr, Shiung MM, Gunter JL 7 et al. Comparison of different MRI brain atrophy rate measures with clinical disease progression in AD. Neurology 62, 591–600 (2004).

Wang L, Miller JP, Gado MH 8 et al. Abnormalities of hippocampal surface structure in very mild dementia of the Alzheimer type. Neuroimage 30, 52–60 (2006).

Apostolova LG, Dutton RA, Dinov ID 9 et al. Conversion of mild cognitive impairment to Alzheimer disease predicted by hippocampal atrophy maps. Arch. Neurol. 63, 693–699 (2006).

Ridha BH, Barnes J, Bartlett JW 10 et al. Tracking atrophy progression in familial Alzheimer’s disease: a serial MRI study. Lancet Neurol. 5, 828–834 (2006).

Friston KJ, Frith CD, Liddle PF, Dolan RJ, 11

Lammertsma AA, Frackowiak RS. The relationship between global and local changes in PET scans. J. Cereb. Blood Flow Metab. 10(4), 458–466 (1990).

Wright IC, McGuire PK, Poline JB 12 et al. A voxel-based method for the statistical analysis of gray and white matter density applied to schizophrenia. Neuroimage 2(4), 244–252 (1995).

Friston KJ, Holmes A, Poline JB, Price CJ, 13

Frith CD. Detecting activations in PET and fMRI: levels of inference and power. Neuroimage 4(3 Pt 1), 223–235 (1996).

Xie S, Xiao JX, Gong GL 14 et al. Voxel-based detection of white matter abnormalities in mild Alzheimer disease. Neurology 66(12), 1845–1849 (2006).

Rose SE, Janke AL, Chalk JB. Gray and 15

white matter changes in Alzheimer’s disease: a diffusion tensor imaging study. J. Magn. Reson. Imaging 27(1), 20–26 (2008).

Ashburner J, Friston KJ. Voxel-based 16

morphometry – the methods. Neuroimage 11(6), 805–821 (2000).

Good CD, Johnsrude IS, Ashburner J, 17

Henson RN, Friston KJ, Frackowiak RS. A voxel-based morphometric study of ageing in 465 normal adult human brains. Neuroimage 14(1), 21–36 (2001).

Plassman BL, Langa KM, Fisher GG 18 et al. Prevalence of cognitive impairment without dementia in the United States. Ann. Intern. Med. 148(6), 427–434 (2008).

Rombouts SA, Barkhof F, Witter MP, 19

Scheltens P. Unbiased whole-brain analysis of gray matter loss in Alzheimer’s disease. Neurosci. Lett. 285(3), 231–233 (2000).

Author Pro

of

Expert Rev. Neurother. 8(11), (2008)10

Review Busatto, Diniz & Zanetti

Baron JC, Chételat G, Desgranges B 20 et al. In vivo mapping of gray matter loss with voxel-based morphometry in mild Alzheimer’s disease. Neuroimage 14(2), 298–309 (2001).

Ohnishi T, Matsuda H, Tabira T, Asada T, 21

Uno M. Changes in brain morphology in Alzheimer disease and normal aging: is Alzheimer disease an exaggerated aging process? AJNR Am. J. Neuroradiol. 22(9), 1680–1685 (2001).

Frisoni GB, Testa C, Zorzan A 22 et al. Detection of grey matter loss in mild Alzheimer’s disease with voxel-based morphometry. J. Neurol. Neurosurg. Psychiatry 73(6), 657–664 (2002).

Matsuda H, Kitayama N, Ohnishi T 23 et al. Longitudinal evaluation of both morphologic and functional changes in the same individuals with Alzheimer’s disease. J. Nucl. Med. 43(3), 304–311 (2002).

Busatto GF, Garrido GE, Almeida OP 24 et al. A voxel-based morphometry study of temporal lobe gray matter reductions in Alzheimer’s disease. Neurobiol. Aging 24(2), 221–231 (2003).

Karas GB, Burton EJ, Rombouts SA 25 et al. A comprehensive study of gray matter loss in patients with Alzheimer’s disease using optimized voxel-based morphometry. Neuroimage 18(4), 895–907 (2003).

Frisoni GB, Testa C, Sabattoli F, 26

Beltramello A, Soininen H, Laakso MP. Structural correlates of early and late onset Alzheimer’s disease: voxel based morphometric study. J. Neurol. Neurosurg. Psychiatry 76(1), 112–114 (2005).

Ishii K, Sasaki H, Kono AK, Miyamoto N, 27

Fukuda T, Mori E. Comparison of gray matter and metabolic reduction in mild Alzheimer’s disease using FDG-PET and voxel-based morphometric MR studies. Eur. J. Nucl. Med. Mol. Imaging 32(8), 959–963 (2005).

Ishii K, Kawachi T, Sasaki H 28 et al. Voxel-based morphometric comparison between early- and late-onset mild Alzheimer’s disease and assessment of diagnostic performance of z score images. AJNR Am. J. Neuroradiol. 26(2), 333–340 (2005).

Pennanen C, Testa C, Laakso MP 29 et al. A voxel based morphometry study on mild cognitive impairment. J. Neurol. Neurosurg. Psychiatry 76(1), 11–14 (2005).

Hirata Y, Matsuda H, Nemoto K 30 et al. Voxel-based morphometry to discriminate early Alzheimer’s disease from controls. Neurosci. Lett. 382(3), 269–274 (2005).

Hirao K, Ohnishi T, Matsuda H 31 et al. Functional interactions between entorhinal cortex and posterior cingulate cortex at the very early stage of Alzheimer’s disease using brain perfusion single-photon emission computed tomography. Nucl. Med. Commun. 27(2), 151–156 (2006).

Kawachi T, Ishii K, Sakamoto S 32 et al. Comparison of the diagnostic performance of FDG-PET and VBM-MRI in very mild Alzheimer’s disease. Eur. J. Nucl. Med. Mol. Imaging 33(7), 801–809 (2006).

Bozzali M, Filippi M, Magnani G 33 et al. The contribution of voxel-based morphometry in staging patients with mild cognitive impairment. Neurology 67(3), 453–460 (2006).

Shiino A, Watanabe T, Maeda K, Kotani 34

E, Akiguchi I, Matsuda M. Four subgroups of Alzheimer’s disease based on patterns of atrophy using VBM and a unique pattern for early onset disease. Neuroimage 33(1), 17–26 (2006).

Whitwell JL, Jack CR Jr, Kantarci K 35 et al. Imaging correlates of posterior cortical atrophy. Neurobiol. Aging 28(7), 1051–1061 (2007).

Whitwell JL, Weigand SD, Shiung MM 36

et al. Focal atrophy in dementia with Lewy bodies on MRI: a distinct pattern from Alzheimer’s disease. Brain 130(Pt 3), 708–719 (2007).

Di Paola M, Macaluso E, Carlesimo GA 37

et al. Episodic memory impairment in patients with Alzheimer’s disease is correlated with entorhinal cortex atrophy. A voxel-based morphometry study. J. Neurol. 254(6), 774–781 (2007).

Samuraki M, Matsunari I, Chen WP 38 et al. Partial volume effect-corrected FDG PET and grey matter volume loss in patients with mild Alzheimer’s disease. Eur. J. Nucl. Med. Mol. Imaging 34(10), 1658–1669 (2007).

Matsunari I, Samuraki M, Chen WP 39 et al. Comparison of 18F-FDG PET and optimized voxel-based morphometry for detection of Alzheimer’s disease: aging effect on diagnostic performance. J. Nucl. Med. 48(12), 1961–1970 (2007).

Josephs KA, Whitwell JL, Duffy JR 40 et al. Progressive aphasia secondary to Alzheimer disease vs FTLD pathology. Neurology 70(1), 25–34 (2008).

Kinkingnéhun S, Sarazin M, Lehéricy S, 41

Guichart-Gomez E, Hergueta T, Dubois B. VBM anticipates the rate of progression of Alzheimer disease: a 3-year longitudinal study. Neurology 70(23), 2201–2211 (2008).

Whitwell JL, Jack CR Jr. Comparisons 42

between Alzheimer disease, frontotemporal lobar degeneration, and normal aging with brain mapping. Top. Magn. Reson. Imaging. 16(6), 409–425 (2005).

Thomann PA, Toro P, Santos V, Essig M, 43

Schröder J. Clock drawing performance and brain morphology in mild cognitive impairment and Alzheimer’s disease. Brain Cogn. 67, 88–93 (2008).

Schott JM, Crutch SJ, Frost C, Warrington 44

EK, Rossor MN, Fox NC. Neuropsychological correlates of whole brain atrophy in Alzheimer’s disease. Neuropsychologia 46, 1732–1737 (2008).

Karas G, Scheltens P, Rombouts S 45 et al. Precuneus atrophy in early-onset Alzheimer’s disease: a morphometric structural MRI study. Neuroradiology 49(12), 967–976 (2007).

Burton EJ, Karas G, Paling SM 46 et al. Patterns of cerebral atrophy in dementia with Lewy bodies using voxel-based morphometry. Neuroimage 17(2), 618–630 (2002).

Burton EJ, McKeith IG, Burn DJ, 47

Williams ED, O’Brien JT. Cerebral atrophy in Parkinson’s disease with and without dementia: a comparison with Alzheimer’s disease, dementia with Lewy bodies and controls. Brain 127(Pt 4), 791–800 (2004).

Brenneis C, Wenning GK, Egger KE 48 et al. Basal forebrain atrophy is a distinctive pattern in dementia with Lewy bodies. Neuroreport 15(11), 1711–1714 (2004).

Boxer AL, Rankin KP, Miller BL 49 et al. Cinguloparietal atrophy distinguishes Alzheimer disease from semantic dementia. Arch. Neurol. 60(7), 949–956 (2003).

Morris J. Mild Cognitive Impairment is 50

early-stage Alzheimer’s disease: time to revise diagnostic criteria. Arch. Neurol. 63, 15–16 (2006).

Chételat G, Desgranges B, De La Sayette V, 51

Viader F, Eustache F, Baron JC. Mapping gray matter loss with voxel-based morphometry in mild cognitive impairment. Neuroreport 13(15), 1939–1943 (2002).

Karas GB, Scheltens P, Rombouts SA 52 et al. Global and local gray matter loss in mild cognitive impairment and Alzheimer’s disease. Neuroimage 23(2), 708–716 (2004).

Bell-McGinty S, Lopez OL, Meltzer CC 53

et al. Differential cortical atrophy in subgroups of mild cognitive impairment. Arch. Neurol. 62(9), 1393–1397 (2005).

Saykin AJ, Wishart HA, Rabin LA 54 et al. Older adults with cognitive complaints show brain atrophy similar to that of amnestic MCI. Neurology 67(5), 834–842 (2006).

Author Pro

of

www.expert-reviews.com 11

ReviewVBM in Alzheimer’s disease

Sorg C, Riedl V, Mühlau M 55 et al. Selective changes of resting-state networks in individuals at risk for Alzheimer’s disease. Proc. Natl Acad. Sci. USA 104(47), 18760–18765 (2007).

Winblad B, Palmer K, Kivipelto M 56 et al. Mild cognitive impairment – beyond controversies, towards a consensus: report of the International Working Group on Mild Cognitive Impairment. J. Int. Med. 256, 240–246 (2004).

Petersen R. Mild cognitive impairment as a 57

diagnostic entity. J. Int. Med. 256, 183–194 (2004).

Whitwell JL, Petersen RC, Negash S 58 et al. Patterns of atrophy differ among specific subtypes of mild cognitive impairment. Arch. Neurol. 64(8), 1130–1138 (2007).

Rozzini L, Chilovi BV, Conti M 59 et al. Conversion of amnestic mild cognitive impairment to dementia of Alzheimer type is independent to memory deterioration. Int. J. Geriatr. Psychiatry 22, 1217–1222 (2007).

Tyas SL, Salazar JC, Snowdon DA 60 et al. Transitions to mild cognitive impairments, dementia, and death: findings from the Nun Study. Am. J. Epidemiol. 165, 1231–1238 (2007).

Resnick SM, Pham DL, Kraut MA, 61

Zonderman AB, Davatzikos C. Longitudinal magnetic resonance imaging studies of older adults: a shrinking brain. J. Neurosci. 23, 3295–3301 (2003).

Chételat G, Landeau B, Eustache F 62 et al. Using voxel-based morphometry to map the structural changes associated with rapid conversion in MCI: a longitudinal MRI study. Neuroimage 27(4), 934–946 (2005).

Whitwell JL, Shiung MM, Przybelski SA 63

et al. MRI patterns of atrophy associated with progression to AD in amnestic mild cognitive impairment. Neurology 70(7), 512–520 (2008).

Karas G, Sluimer J, Goekoop R 64 et al. Amnestic mild cognitive impairment: structural MR imaging findings predictive of conversion to Alzheimer disease. AJNR Am. J. Neuroradiol. 29(5), 944–949 (2008).

Whitwell JL, Przybelski SA, Weigand SD 65

et al. 3D maps from multiple MRI illustrate changing atrophy patterns as subjects progress from mild cognitive impairment to Alzheimer’s disease. Brain 130(Pt 7), 1777–1786 (2007).

Chételat G, Fouquet M, Kalpouzos G 66 et al. Three-dimensional surface mapping of hippocampal atrophy progression from

MCI to AD and over normal aging as assessed using voxel-based morphometry. Neuropsychologia 46(6), 1721–1731 (2008).

Visser PJ, Scheltens P, Verhey FRJ. Do MCI 67

criteria in drug trials accurately identify subjects with predementia Alzheimer’s disease? J. Neurol. Neurosurg. Psychiatry 76, 1348–1354 (2005).

Teipel SJ, Pruessner JC, Faltraco F 68 et al. Comprehensive dissection of the medial temporal lobe in AD: measurement of hippocampus, amygdala, entorhinal, perirhinal and parahippocampal cortices using MRI. J. Neurol. 253(6), 794–800 (2006).

Testa C, Laakso MP, Sabattoli F 69 et al. A comparison between the accuracy of voxel-based morphometry and hippocampal volumetry in Alzheimer’s disease. J. Magn. Reson. Imaging 19(3), 274–282 (2004).

Hesse C, Rosengren L, Vanmechelen E 70

et al. Cerebrospinal fluid markers for Alzheimer’s disease evaluated after acute ischemic stroke. J. Alzheimer Dis. 2, 199–206 (2000).

Blennow K, Zetterberg H, Minthon L 71 et al. Longitudinal stability of CSF biomarkers in Alzheimer’s disease. Neurosci. Lett. 419, 18–22 (2007).

Hampel H, Bürger K, Pruessner JC 72 et al. Correlation of cerebrospinal fluid levels of tau protein phosphorylated at threonine 231 with rates of hippocampal atrophy in Alzheimer disease. Arch. Neurol. 62, 770–773 (2005).

Diniz BS, Pinto JA, Forlenza OV. Do CSF 73

total tau, phosphorylated tau, and β-amyloid 42 help to predict progression of mild cognitive impairment to Alzheimer's disease? A systematic review and meta-analysis of the literature. World J. Biol. Psychiatry (2007) (Epub ahead of print).

de Leon MJ, DeSanti S, Zinkowski R 74 et al. Longitudinal CSF and MRI biomarkers improve the diagnosis of mild cognitive impairment. Neurobiol. Aging 27, 394–401 (2006).

Fjell AM, Walhovd KB, Amlien I 75 et al. Morphometric changes in the episodic memory network and tau pathologic features correlate with memory performance in patients with mild cognitive impairment. AJNR Am. J. Neuroradiol. 29, 1183–1189 (2008).

Bouwman FH, Schoonenboom SN, van der 76

Flier WM et al. CSF biomarkers and medial temporal lobe atrophy predict dementia in mild cognitive impairment. Neurobiol. Aging 28(7), 1070–1074 (2007).

Jack Jr CR, Lowe VJ, Senjem ML 77 et al. 11C PiB and structural MRI provide complementary information in imaging of Alzheimer’s disease and amnestic mild cognitive impairment. Brain 131, 665–680 (2008).

Cummings JL, Doody R, Clark C. 78

Disease-modifying therapies for Alzheimer disease challenges to early intervention. Neurology 69, 1622–1634 (2007).

Mahley RW, Weisgraber KH, Huang Y. 79

Apolipoprotein E4: A causative factor and therapeutic target in neuropathology, including Alzheimer’s disease. Proc. Natl Acad. Sci. USA 103(15), 5644–5651 (2006).

Huang Y. Apolipoprotein E and Alzheimer 80

disease. Neurology 66(Suppl. 1), S79–S85 (2006).

Juottonen K, Lehtovirta M, Helisalmi S, 81

Riekkinen PJ Sr, Soininen H. Major decrease in the volume of the entorhinal cortex in patients with Alzheimer’s disease carrying the apolipoprotein E epsilon4 allele. J. Neurol. Neurosurg. Psychiatry 65, 322–327 (1998).

Hashimoto M, Yasuda M, Tanimukai S 82

et al. Apolipoprotein E epsilon 4 and the pattern of regional brain atrophy in Alzheimer’s disease. Neurology 57, 1461–1466 (2001).

Fleisher A, Grundman M, Jack Jr CR 83 et al. Sex, apolipoprotein E ε4 status, and hippocampal volume in Mild Cognitive Impairment. Arch. Neurol. 62, 953–957 (2005).

den Heijer T, Oudkerk M, Launer LJ, van 84

Duijn CM, Hofman A, Breteler MMB. Hippocampal, amygdalar, and global brain atrophy in different apolipoprotein E genotypes. Neurology 59, 746–748 (2002).

Wishart HA, Saykin AJ, McAllister TW 85

et al. Regional brain atrophy in cognitively intact adults with a single APOE ε4 allele. Neurology 67, 1221–1224 (2006).

Lemaître H, Crivello F, Dufouil C 86 et al. No ε4 gene dose effect on hippocampal atrophy in a large MRI database of healthy elderly subjects. Neuroimage 24, 1205–1213 (2005).

Pennanen C, Testa C, Boccardi M 87 et al. The effect of apolipoprotein polymorphism on brain in mild cognitive impairment: a voxel-based morphometric study. Dement. Geriatr. Cogn. Disord. 22, 60–66 (2006).

Hämäläinen A, Tervo S, Grau-Olivares M 88

et al. Voxel-based morphometry to detect brain atrophy in progressive mild cognitive impairment. Neuroimage 37, 1122–1131 (2007).

Author Pro

of

Expert Rev. Neurother. 8(11), (2008)12

Review Busatto, Diniz & Zanetti

Hardy J. The Alzheimer family of diseases: 89

many etiologies, one pathogenesis? Proc. Natl Acad. Sci. USA 94, 2095–2097 (1997).

Kennedy KM, Erickson KI, Rodrigue KM 90

et al. Age-related differences in regional brain volumes: a comparison of optimized voxel-based morphometry to manual volumetry. Neurobiol. Aging (2008) (Epub ahead of print).

Bookstein FL. “Voxel-based morphometry” 91

should not be used with imperfectly registered images. Neuroimage 14(6), 1454–1462 (2001).

Duran FSL, Valente AA, Miguel EC, 92

Castro CC, Busatto GF. Risk of artifacts due to enlarged ventricles using voxel-based morphometry studies. Presented at: Organization for Human Brain Mapping Conference. Florence, Italy, 11–15 June 2006. Neuroimage (Suppl.), 31 (2006).

Uchida RR, Del-Ben CM, Araújo D 93 et al. Correlation between voxel based morphometry and manual volumetry in magnetic resonance images of the human brain. An. Acad. Bras. Cienc. 80(1), 149–156 (2008).

Ashburner J. A fast diffeomorphic image 94

registration algorithm. Neuroimage 38(1), 95–113 (2007).

Ashburner J, Friston KJ. Unified 95

segmentation. Neuroimage 26(3), 839–851 (2005).

Schaufelberger MS, Duran FL, Lappin JM 96

et al. Grey matter abnormalities in Brazilians with first-episode psychosis. Br. J. Psychiatry 51 (Suppl.), S117–S122 (2007).

Ewers M, Teipel SJ, Dietrich O 97 et al. Multicenter assessment of reliability of cranial MRI. Neurobiol. Aging 27(8), 1051–1059 (2006).

Schnack HG, van Haren NE, Hulshoff Pol 98

HE et al. Reliability of brain volumes from multicenter MRI acquisition: a calibration study. Hum. Brain Mapp. 22(4), 312–320 (2004).

Teipel S, Ewers M, Dietrich O 99 et al. Reliability of multicenter magnetic resonance imaging. Results of a phantom test and in vivo measurements by the German Dementia Competence Network. Nervenarzt 77(9), 1086–1092, 1094–1095 (2006).

Chaim TM, Duran FL, Uchida RR, Périco 100

CA, de Castro CC, Busatto GF. Volumetric reduction of the corpus callosum in Alzheimer’s disease in vivo as assessed with voxel-based morphometry. Psychiatry Res. 154(1), 59–68 (2007).

Li S, Pu F, Shi F, Xie S, Wang Y, Jiang T. 101

Regional white matter decreases in Alzheimer’s disease using optimized voxel-based morphometry. Acta Radiol. 49(1), 84–90 (2008).

Villain N, Desgranges B, Viader F 102 et al. Relationships between hippocampal atrophy, white matter disruption, and gray matter hypometabolism in Alzheimer’s disease. J. Neurosci. 28(24), 6174–6181 (2008).

Lao Z, Shen D, Xue Z, Karacali B, Resnick 103

SM, Davatzikos C. Morphological classification of brains via high-dimensional shape transformations and machine learning methods. Neuroimage 21(1), 46–57 (2004).

Fan Y, Shen D, Davatzikos C. Classification 104

of structural images via high-dimensional image warping, robust feature extraction, and SVM. Med. Image Comput. Comput. Assist. Interv. Int. Conf. Med. Image Comput. Comput. Assist. Interv. 8(Pt 1), 1–8 (2005).

Davatzikos C, Fan Y, Wu X, Shen D, 105

Resnick SM. Detection of prodromal Alzheimer’s disease via pattern classification of magnetic resonance imaging. Neurobiol. Aging 29(4), 514–523 (2008).

Klöppel S, Stonnington CM, Chu C 106 et al. Automatic classification of MR scans in Alzheimer’s disease. Brain 131(Pt 3), 681–689 (2008).

Vemuri P, Gunter JL, Senjem ML 107 et al. Alzheimer’s disease diagnosis in individual subjects using structural MR images: validation studies. Neuroimage 39(3), 1186–1197 (2008).

Mueller SG, Weiner MW, Thal LJ 108 et al. Ways toward an early diagnosis in Alzheimer’s disease: the Alzheimer’s Disease Neuroimaging Initiative (ADNI). Alzheimers Dement. 1(1), 55–66 (2005).

Hampel H, Bürger K, Teipel SJ, Bokde AL, 109

Zetterberg H, Blennow K. Core candidate neurochemical and imaging biomarkers of Alzheimer’s disease. Alzheimers Dement. 4(1), 38–48 (2008).

Frisoni GB, Whitwell JL. How fast will it 110

go, doc? New tools for an old question from patients with Alzheimer disease. Neurology 70(23), 2194–2195 (2008).

Frisoni GB, Henneman WJ, Weiner MW 111

et al. Alzheimer’s Disease Neuroimaging Initiative. The pilot European Alzheimer’s Disease Neuroimaging Initiative of the European Alzheimer’s Disease Consortium. Alzheimers Dement. 4(4), 255–264 (2008).

Jack CR Jr, Bernstein MA, Fox NC 112 et al. The Alzheimer’s Disease Neuroimaging Initiative (ADNI): MRI methods. J. Magn. Reson. Imaging 27(4), 685–691 (2008).

Dubois B, Feldman HH, Jacova C 113 et al. Research criteria for the diagnosis of Alzheimer’s disease: revising the NINCDS–ADRDA criteria. Lancet Neurol. 6, 734–746 (2007).

Feldman HH, Ferris S, Winblad B 114 et al. Effect of rivastigmine on delay to diagnosis of Alzheimer’s disease from mild cognitive impairment: the InDDEx study. Lancet Neurol. 6, 501–512 (2007).

Jack Jr CR, Petersen RC, Grundman M 115

et al. Longitudinal MRI findings from the vitamin E and donepezil treatment study for MCI. Neurobiol. Aging 29, 1285–1295 (2008).

Website

Alzheimer’s Disease Neuroimaging Initiative 201

www.adni-info.org

AffiliationsGeraldo F Busatto, MD, PhD •Laboratory of Psychiatric Neuroimaging (LIM-21), Department and Institute of Psychiatry, University of São Paulo Medical School, São Paulo, SP, Brazil Centro de Medicina Nuclear, 3°andar, LIM-21, Rua Dr. Ovídio Pires de Campos, s/n 05403–010, São Paulo, SP, Brazil Tel.: +55 113 069 8193 Fax: +55 113 082 1015 [email protected]

Breno S Diniz, MD •Laboratory of Neurosciences (LIM-27), Department and Institute of Psychiatry, University of São Paulo Medical School, São Paulo, SP, Brazil Instituto de Psiquiatria, 3°andar, Laboratório de Neurociências, Rua Dr. Ovídio Pires de Campos, 785, 05403–010, São Paulo, SP, Brazil Tel.: +55 113 069 7267 Fax: +55 113 069 7924 [email protected]

Marcus V Zanetti, MD •Laboratory of Psychiatric Neuroimaging (LIM-21), Department and Institute of Psychiatry, University of São Paulo Medical School, São Paulo, SP, Brazil Centro de Medicina Nuclear, 3°andar, LIM-21, Rua Dr. Ovídio Pires de Campos, s/n 05403–010, São Paulo, SP, Brazil Tel.: +55 113 069 8193 Fax: +55 113 082 1015 [email protected]