A new report from Alan Evans of McGill University in Montreal, Quebec, puts a structural face on these functional studies. Just as functional correlations among brain regions weaken in AD, so do correlations of cortical thickness as measured by MRI, Evans and coauthors report in the April 30 Journal of Neuroscience. The work provides a different, anatomical measure of brain organization and its disruption in disease.

The cerebral cortex is the grey matter that forms the outermost layer of the cerebrum. It varies in thickness from about 1 mm to 7 mm in different parts of the brain, and grows thinner with age and with Alzheimer disease. Recent work from Evans and others has suggested that cortical thickness is correlated in regions of the brain that are known to be functionally connected. This has led to the idea that brain-wide maps of thickness correlations reveal the architecture of cortical networks (He et al., 2007; Chen et al., 2008).

In the current study, coauthors Yong He and Zhang Chen analyze global cortical thickness measures from MRI images of 97 healthy elderly people and 92 with AD, all from the Open Access Series of Imaging Studies database (OASIS, see Marcus et al., 2007). Their results indicate that interregional thickness correlations are greatly disrupted in the AD patients. This disruption shows up in the form of mismatches in thickness between specific cortical regions that are normally similar. The AD brains also reveal global changes in correlation characteristics, and a network analysis indicates the changes could result in a less efficient network architecture and lower resistance to further disruption.

Cortical thickness reflects the size, density, and arrangement of cells, and so it might not be surprising that structural correlations diminish under assault from AD pathology and associated cortical thinning. But do these structural alterations relate to the functional changes seen in AD patients? In support of this idea, the authors note that the structural networks they identify in healthy brain coincide with functional networks defined by fMRI, and are known to be disrupted in AD.

Interestingly, Evans and colleagues observed an increased positive correlation among several brain regions involved in the default network. That distributed network has gotten a lot of attention in AD, as it is involved in memory, is functionally disturbed early on in AD, and is an early site of amyloid deposition (see ARF related news story). Evans and coauthors speculate that the increased correlation of cortical thickness in these areas “may reflect a correlative cortical shrinking because of shared vulnerability to an insult, as preferentially affected by the pathological process of this disease.” In support of this idea, data tying amyloid deposition to cortical thinning in the default regions was presented recently by Keith Johnson of Massachusetts General Hospital, Boston, at the Human Amyloid Imaging conference in Chicago (see ARF related news story).

Beyond the current study, analysis of cortical thickness correlations represents a promising approach to look at distributed changes in the brain in AD, Evans told ARF. Rather than look at morphological changes in select regions of the brain, for example, by measuring hippocampal volume, Evans said, “Cortical thickness is a global index. We can extract it automatically over the entire cortex, and do exploratory analysis without making assumptions about which areas are important.” The program that Evans has developed to automate the analysis can be applied to large collections of MRI data. Using such collections, Evans has worked on large studies looking at cortical thickness during brain development in children (Evans et al., 2006, and see ARF related news story). With the advent of large, collaborative imaging studies such as the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and its counterparts in Japan and Europe, there will be a lot more data available to pursue with this approach. One goal will be to determine which changes in cortical thickness networks correlate best with the progression of AD, Evans said.—Pat McCaffrey.

Reference:
He Y, Chen Z, Evans A. Structural insights into aberrant topological patterns of large-scale cortical networks in Alzheimer's disease. J. Neurosci. 30 April 2008; 28(18):4756-4766. Abstract