|Home | About | Journals | Submit | Contact Us | Français|
Recent studies have demonstrated alterations in cortical gray to white matter tissue contrast with nondemented aging and in individuals with Alzheimer’s disease (AD). However, little information exists about the clinical relevance of such changes. It is possible that changes in MRI tissue contrast occur via independent mechanisms from those traditionally used in the assessment of AD associated degeneration such as hippocampal degeneration measured by more traditional volumetric magnetic resonance imaging (MRI). We created cortical surface models of 95 cognitively healthy individuals and 98 individuals with AD to characterize changes in regional gray and white matter T1-weighted signal intensities in dementia, and in normal aging and to evaluate how such measures related to classically described hippocampal and cortical atrophy. We found a reduction in gray matter to white matter tissue contrast throughout portions of medial and lateral temporal cortical regions as well as in anatomically associated regions including the posterior cingulate, precuneus, and medial frontal cortex. Decreases in tissue contrast were associated with hippocampal volume, however, the regional patterns of these associations differed for demented and nondemented individuals. In nondemented controls, lower hippocampal volume was associated with decreased gray/white matter tissue contrast globally across the cortical mantle. In contrast, in individuals with AD, selective associations were found between hippocampal volume and tissue contrast in temporal and limbic tissue. These results demonstrate that there are strong regional changes in neural tissue properties in AD which follow a spatial pattern including regions known to be affected from pathology studies. Such changes are associated with traditional imaging metrics of degeneration and may provide a unique biomarker of the tissue loss that occurs as a result of AD.
Prior studies have demonstrated the effect of AD on several neuroimaging parameters of tissue integrity, including alterations in cortical morphometry and volume (Fotenos et al., 2005; Jernigan et al., 1991b; Thompson et al., 1998) and cortical thickness (Dickerson et al., 2009; Lerch et al., 2008; Lerch et al., 2005), as well as regional changes in tissue properties such as tissue microstructure measured by diffusion tensor imaging (DTI) (Bozzali et al., 2002; Bozzali et al., 2001; Head et al., 2004; Rose et al., 2000; Salat et al. 2010). Less studied in AD are changes in tissue signal properties, such as T1 relaxation times and signal intensity which have been examined more extensively in nondemented aging (Cho et al., 1997; Davatzikos and Resnick, 2002; Ogg and Steen, 1998; Raz et al., 1990; Salat et al., 2009; Steen et al., 1995; Westlye et al., 2009) and to some extent in AD (Westlye et al., 2009). While MRI tissue signal changes in individuals with AD have been demonstrated (Westlye et al., 2009), the association between contrast changes and more traditional metrics of degeneration in AD has thus far been completely unexplored.
In order to characterize the unique changes in tissue characteristics in AD, we created cortical surface models in a large sample of nondemented individuals and individuals with a clinical diagnosis of AD. Surface models were utilized to map gray and white matter signal intensities from high-resolution 3D T1-weighted MPRAGE images to examine the regional patterns of AD-specific associated signal alterations. This is a unique measure that can only be calculated with the creation of explicit models of the gray matter/white matter and gray matter/cerebrospinal fluid borders for a regional assessment of true tissue contrast. We additionally examined the association between this novel metric of tissue changes and traditional metrics of AD pathology including hippocampal volume and cortical thickness, and Clinical Dementia Rating (CDR) as a measure of dementia severity. We specifically focused on the gray to white matter signal intensity ratio (GWR) at each point along the cortical surface to determine whether contrast properties were altered in a regionally specific manner throughout the brain. The GWR showed a considerable increase (towards a value of 1) with increasing severity of AD, demonstrating an overall decrease in the contrast between tissue classes. These findings demonstrate that tissue signal properties are altered in AD in a regionally specific manner. Such changes may be an indirect marker of changes in the histological properties of the tissue that have a significant impact on neural and cognitive processes. These findings may therefore have important clinical applications in the monitoring of degenerative effects resultant from AD. Alterations in tissue contrast remained when controlling for changes in morphometric properties, suggesting that the novel GWR metric may provide a unique in vivo biomarker of degenerative changes in AD.
Images were obtained for 193 participants (Table 1). All older non-demented adults and ADs were recruited and clinically evaluated through the Washington University Alzheimer’s Disease Research Center (ADRC) as reported previously (Berg et al., 1998; Morris, 1993). These data are publically accessible as part of the Open Access Series of Imaging Studies (OASIS: http://www.oasis-brains.org/). Non-demented individuals were all clinically-screened to ensure no signs of even mild cognitive impairment (all CDR 0). Fotenos et al. (Fotenos et al., 2005) describe the recruitment characteristics of this sample in detail. Participants consented in accordance with guidelines of the Washington University Human Studies Committee.
Signal properties were examined from high-resolution 3D MPRAGE and cortical reconstruction procedures similar to descriptions in our prior work using cortical reconstruction (Dickerson et al., 2008; Han et al., 2006; Rosas et al., 2008; Salat et al., 2004; Salat et al., 2009) and intensity analysis (Salat et al., 2009). Two to five T1 weighted MP-RAGE scans per participant were acquired on a single scanner (Siemens 1.5T Vision System, resolution 1 × 1 × 1.25 mm, TR = 9.7 ms, TI = 20 ms, TE = 4.0 ms) motion corrected, and averaged to create high signal/contrast to noise volumes.
Cortical reconstruction was performed using the FreeSurfer image analysis suite, which is documented and freely available for download online (http://surfer.nmr.mgh.harvard.edu/). The technical details of these procedures are described in prior publications (Dale et al., 1999; Dale and Sereno, 1993; Fischl and Dale, 2000; Fischl et al., 2001; Fischl et al., 2002; Fischl et al., 2004a; Fischl et al., 1999a; Fischl et al., 1999b; Fischl et al., 2004b; Han et al., 2006; Jovicich et al., 2006; Segonne et al., 2004). Briefly, this processing includes removal of non-brain tissue using a hybrid watershed/surface deformation procedure (Segonne et al., 2004), automated Talairach transformation, segmentation of the subcortical white matter and deep gray matter volumetric structures (including hippocampus, amygdala, caudate, putamen, ventricles) (Fischl et al., 2002; Fischl et al., 2004a) intensity normalization (Sled et al., 1998), tessellation of the gray matter white matter boundary, automated topology correction (Fischl et al., 2001; Segonne et al., 2007), and surface deformation following intensity gradients to optimally place the gray/white and gray/CSF borders at the locations where the greatest shifts in intensity defines the transition to the other tissue class (Dale et al., 1999; Dale and Sereno, 1993; Fischl and Dale, 2000). Once the cortical models are complete, a number of deformable procedures are performed for further data processing and analysis including surface inflation (Fischl et al., 1999a), registration to a spherical atlas which utilizes individual cortical folding patterns to match cortical geometry across subjects (Fischl et al., 1999b), parcellation of the cerebral cortex into units based on gyral and sulcal structure (Desikan et al., 2006; Fischl et al., 2004b), and creation of a variety of surface based data including maps of curvature and sulcal depth. These procedures have been demonstrated to align histological properties such as cytoarchitectonic borders with greater accuracy than volumetric registration (Fischl et al., 2008). This method uses both intensity and continuity information from the entire three dimensional MR volume in segmentation and deformation procedures to produce representations of cortical thickness, calculated as the closest distance from the gray/white boundary to the gray/CSF boundary at each vertex on the tessellated surface (Fischl and Dale, 2000). The maps are created using spatial intensity gradients across tissue classes and are therefore not simply reliant on absolute signal intensity. The maps produced are not restricted to the voxel resolution of the original data and are thus capable of detecting submillimeter differences in thickness between groups. Procedures for the measurement of cortical thickness have been validated against histological analysis (Rosas et al., 2002) and manual measurements (Kuperberg et al., 2003; Salat et al., 2004). Freesurfer morphometric procedures have been demonstrated to show excellent test-retest reliability across scanner manufacturers and across field strengths and other imaging parameters (Han et al., 2006; Jovicich et al., 2009; Wonderlick et al., 2008). Images were corrected for intensity nonuniformity prior to sampling tissue intensities (Sled et al., 1998) to reduce the influence of inhomogeneity caused by B1 and coil profiles. In addition to this nonuniformtiy correction, examination of the GWR provided a highly localized (most voxels within ~2mm) normalization of tissue values because the gray and white matter intensity from closely neighboring voxels would be expected to be similarly influenced by any imaging parameters due to the smoothness of the nuisance parameter maps (e.g. field B0/B1 inhomogeneities) which can be assumed to be smooth relative to the local normalization operation. Thus, presenting the gray matter values as a ratio to bordering white matter values provides a unit which is normalized for the local imaging environment.
Gray matter tissue intensities were measured at a depth of 35% through the thickness of the cortical ribbon, starting from the gray matter/white matter border and projecting towards the gray matter/cerebrospinal fluid border (Figure 1). White matter signal intensities were sampled at 1mm subjacent to the gray matter/white matter border. The 35% sampling procedure was utilized to be conservatively close to the gray/white border and the white matter sampling voxel (in order to minimize potential spurious effects that could arise in sampling from more remote locations and from sampling in regions of low cortical thickness). After calculation of the GWR, values were smoothed with a Gaussian kernel of FWHM 30 mm for analysis utilizing a surface based smoothing procedure that averages data across neighboring cortical locations. In contrast to 3D volumetric smoothing, the use of surface based smoothing limited the integration of information to regions limited by the area along the cortical mantle, preventing combining of data across sulcal space and cerebrospinal fluid.
The relationships among surface measures (GWR, and thickness) and AD were examined with a vertex by vertex general linear model (GLM). A separate GLM was evaluated for each vertex to assess regional variation in each measure. The significance of contrasts of the regression parameters were computed using t-tests. Additional models examined the association between hippocampal volume and tissue contrast.
Gray and white matter intensities were first normalized to the mean cortical intensity and compared between the demented and nondemented groups (Figure 2). There was a regionally specific decrease in gray matter signal in cingulate and parahippocampal areas in AD compared to nondemented individuals. White matter signal was affected to a greater degree with decreased intensity throughout a large portion of the cortical mantle. The gray/white ratio (GWR) was significantly increased throughout portions of medial and lateral temporal cortex as well in regions including the cingulate and precuneus in individuals with AD compared to nondemented control participants.
Effects were strong when comparing the demented individuals to the nondemented control participants (Figure 2b, panel A-B). In contrast, only limited regions in medial temporal cortex differed in CDR 1 compared to CDR 0.5, suggesting that most tissue changes are present early in the disease. However, small patches of inferior lateral and medial temporal regions demonstrated progressive changes in the GWR with increasing dementia severity (Figure 2b, panel C).
There were strong associations between hippocampal volume and GWR in both healthy nondemented individuals and in patients with AD; however, the regional patterns of these associations differed. In nondemented individuals, more generalized regional increases in GWR were associated with reduced hippocampal volumes (Figure 3b); in contrast, in demented individuals these associations were limited to medial and lateral temporal cortex (Figure 3a). There was an interaction between demented and nondemented individuals when examining the association between hippocampal volume and GWR, with nondemented individuals showing a stronger association in most areas throughout the cortical mantle except for medial temporal and right lateral temporal areas (Figure 4).
Figure 5 demonstrates the effect of AD on GWR across all participants (CDR 1 and CDR 0.5 combined), and the effect of AD in the same individuals on cortical thickness. There was substantial regional overlap of the changes in these two metrics of tissue integrity with AD. Figure 5C demonstrates the effect of AD on GWR while controlling for the cortical thickness at each vertex along the cortical mantle. These results demonstrate that the effects of AD on GWR are beyond the variance accounted for by cortical thickness alone.
We performed a split sample analysis to examine the reliability of the AD associated effects in GWR. For this analysis, we sorted the demented group and the nondemented group individually by sex and then by age and assigned the odd numbered participants to group 1 and the even numbered participants to group 2 to create two equal age and sex matched samples. Differences in the GWR between demented and nondemented individuals were reliable across the two samples (Fig. 6), and were in accord with the overall group maps presented in Figures 2 and and55.
It is possible that the effects measured here could be influenced by gyral location (e.g. crown of a gyrus versus the fundus of a sulcus) for biological as well as for technical reasons. We therefore performed a secondary analysis to determine whether the effects of AD on GWR were influenced by gyral location by extracting unsmoothed GWR data from two ROIs; one ROI limited to the gyral portion of a region affected by AD, and the other ROI limited to the sulcal portion of the same region. The effects were similar across the two ROIs (Figure 7). Thus, although these results do no completely exclude the potential influence of curvature on the GWR measurements, they do demonstrate that the effects are not strictly limited to gyral or sulcal regions.
The current results demonstrate that there are regionally selective alterations in MR image properties that occur uniquely with AD that are distinct from the effects from nondemented healthy aging. Individuals with AD exhibited reduced gray matter to white matter tissue contrast in several regions throughout the cortical mantle with particularly strong effects in temporal and limbic areas. In contrast, associations with hippocampal volume were only found in limited regions of temporal and parietal cortex, and the effects were statistically beyond what could be explained by cortical thinning. These results thus importantly suggest that tissue contrast measures may provide a more sensitive metric, or reflect a distinct pathology from the traditional measures of volume and thickness. These changes were found in individuals who were in early stages of AD (CDR 0.5), and may provide an indirect and sensitive metric of early histological or pathological processes.
Although prior studies have demonstrated changes in brain morphometry due to AD with various MRI analysis procedures, few studies have examined the effects of AD on tissue signal properties. Westlye and colleagues recently demonstrate that tissue properties could be used to enhance the effects of cortical thickness measurements in the study of AD (Westlye et al., 2009). In addition to this, the current data suggest that the tissue properties are an interesting parameter to examine on their own; and that this metric correlates to some extent with more classical measures of pathology such as hippocampal volume. Although the effects of AD on GWR remained when considering effects on cortical thickness, there was substantial regional overlap between these metrics. Thus, GWR effects could be a microstructural marker of pathologic mechanisms that occur prior to cortical atrophy, and changes may occur with a distinct temporal pattern in the disease process; however it is currently unknown whether changes in GWR and thickness are due to similar or different biological mechanisms. Similarly, the current data do not address the relative strengths of GWR compared to traditional measures such as cortical thickness with regard to prediction of dementia. However, given the unique properties of the GWR measure, it is likely to be useful for understanding neural decline as well as to serve as biomarkers of early disease processes.
Primary attention has been put towards gray matter degeneration in AD given the known early and primary pathology noted through histopathological analysis, particularly within the entorhinal cortex (Ball, 1978; Hyman et al., 1984). However, histological work (Hyman et al., 1984) as well as recent neuroimaging studies (Choo et al., 2008; Rose et al., 2006; Salat et al., 2008a; Salat et al., 2008b; Zhang et al., 2007) demonstrate that white matter within the parahippocampal gyrus is substantially affected in AD. It is possible that these white matter changes contribute to the change in tissue contrast, and that signal properties are influenced by tissue changes on a microstructural scale that presage the macrostructural effect of cortical atrophy or white matter tissue degradation. The tissue specific analyses suggested that changes in white matter signal intensity may contribute substantially to the changes in GWR observed. These results should be viewed with caution because the signal intensities examined were not from quantitative acquisitions and the specifics of the normalization procedures will influence the outcome. Overall, these findings suggest that intensity measures may be an important biomarker of aging and age-associated disease. It is therefore possible that intensity and contrast measures may be useful in diagnostic and other clinical procedures. We note that the effects of AD on GWR were both spatially and statistically greatly reduced compared to the effects of age on GWR described in prior work (Salat et al., 2009). It is therefore possible that further development of the procedures for data acquisition and analysis will be necessary to fully exploit this potential biomarker of disease associated processes.
These data contribute to prior imaging studies describing a reduction in gray/white matter tissue contrast with aging (Davatzikos and Resnick, 2002; Jernigan et al., 1991a; Magnaldi et al., 1993; Raz et al., 1990; Salat et al., 2009; Westlye et al., 2009). There is a strong dependence of T1 relaxation times on regional iron concentration in the brain (Ogg and Steen, 1998), and subtle changes of this sort could contribute to the current results. It was somewhat surprising that hippocampal volume was strongly related to GWR generally throughout the brain in the nondemented older adults, and that the effect was more selective in the AD group. It is possible that hippocampal volume is a general marker of brain health in this population. When an aggressive process such as AD is present, the more generalized associations break down, and the hippocampus is instead tied more selectively to linked degenerative changes in temporal and limbic cortex. The effects measured were in regions where pathology has been reported in both gray and white matter. The spatial colocalization of these effects demonstrate that the effect measured are of pathologic significance associated with the primary disease process and not a more generic or global effect on tissue.
It is important to consider any potential technical contributions of changes in one measure, such as signal properties, to changes in another measure such as cortical thickness. It is possible that changes in cortical thickness influence contrast measures, or alternatively, changes in contrast influence cortical thickness measures. The specifics of the sampling do affect the results (e.g. how far through the cortex the gray matter signal is sampled), however, the major findings of this study are constant across a range of methods. There is also potential for bias of these procedures in regions of thinner or thicker cortex, and it is likely that the thickness of the cortex affects the measures somewhat. We demonstrated that thickness had some influence on the data, but that effects remained in GWR when thickness was considered in the analyses. Also, we demonstrated that there are similar effects in neighboring gyral (thicker) and sulcal (thinner) regions in tissue showing an effect of AD on GWR suggesting that thickness may have some influence, but is not a major component of the effects reported. It is important to note that the current results are cross-sectional, and ongoing work will examine the longitudinal effects of AD on the GWR. Additionally, the current study utilized a 3D MPRAGE sequence to define tissue intensities. This is a limitation as this sequence is not quantitative and the signal intensities can be affected by technical aspects of the scanning procedures. Indeed the parameters of the MPRAGE sequence were selected to maximize the sensitivity to differences between gray matter, white matter and CSF intensity per unit of acquisition time. Future work will examine changes in the GWR utilizing more optimized acquisition parameters for this type of analysis.
In summary, these results demonstrate changes in neural tissue signal properties with AD. These changes are regionally selective to areas known to show AD pathology, and were related to, yet statistically exceeding those of changes in morphometry. Thus, the GWR may be an important biomarker of pathologic changes with aging and disease and may reflect important histopathologic properties that will be of interest to examine in a range of studies to understand degenerative mechanisms and inform therapeutic intervention in AD.
This work was supported by the National Institutes of Health [K01AG024898, R01NR010827, P41RR14075, BIRN002, U24RR021382, R01EB001550, R01EB006758, R01NS052585, U54EB005149]; a postdoctoral fellowship from the Canadian Institutes of Health Research; the Athinoula A. Martinos Center for Biomedical Imaging, the Mental Illness and Neuroscience Discovery (MIND) Institute, and The Autism & Dyslexia Project funded by the Ellison Medical Foundation. We are grateful to Dr. Randy Buckner for his comments on this work, and for his support in the use of the OASIS dataset.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.