The major findings of this study are: (i) Alzheimer’s disease is associated with cortical thinning primarily in the frontal, parietal, temporal and occipital lobes, while a different regional pattern of cortical thinning is found in FTD, involving primarily the frontal and temporal lobes. The pattern of cortical thinning in each disease is consistent with previous MRI studies using volumetric measurements of brain atrophy and also with histopathological findings of the brain’s selective vulnerability to Alzheimer’s disease and FTD; (ii) dementia severity is negatively correlated with cortical thickness in Alzheimer’s disease, while comparable correlations in FTD were not significant; (iii) measurement of cortical thickness provided similar accuracy as that of cortical volume for differentiating between CN, Alzheimer’s disease and FTD and significantly improved the classification between Alzheimer’s disease and FTD based on neuropsychological scores alone.
The finding of characteristic patterns of cortical thinning in Alzheimer’s disease replicates the previous study (Lerch et al., 2005
) and is consistent with the pattern of tissue loss reported by histopathological and volumetric MRI studies (Braak and Braak, 1995
; Baron et al., 2001
). Furthermore, we found that FTD is associated with a characteristic regional pattern of cortical thinning in frontal, temporal regions and parietal lobe, and also with greatest cortical atrophy in prefrontal regions, and less cortical atrophy in parietal regions. The findings in FTD are consistent with previous pathological reports and volumetric MRI studies (Rosen et al., 2002
; Broe et al., 2003
; Grossman et al., 2004
; Whitwell et al., 2005
). In particular, prominent thinning of frontal and temporal cortex in FTD is in agreement with autopsy findings from macroscopic examinations of FTD brains showing consistently frontal and temporal and, less commonly, parietal atrophy (Dickson, 2001
; Kersaitis et al., 2004
). Furthermore, this study showed that FTD was associated with the cortical thinning in both orbital and medial frontal cortices with a similar severity, which is also consistent with the previous pathological study that FTD is associated with the orbital and medial frontal cortices in the early stage (Broe et al., 2003
). However, the previous VBM studies (Rosen et al., 2005
; Williams et al., 2005
) have shown that in FTD, the behaviour change is related to grey matter loss in the medial frontal cortex other than the orbital frontal cortex. More studies may be needed to elucidate the relationship between the behaviour change and cortical atrophy with cortical thickness measurement in FTD. While patients with Alzheimer’s disease had thinner cortices on MRI than FTD patients, specifically in the precuneus and the parietal lobe regions, no cortical region in FTD was significantly thinner than in Alzheimer’s disease. The finding that Alzheimer’s disease and FTD patients showed similar levels of cortical thinning in the frontal lobe is surprising, given that greater brain atrophy in frontal regions in FTD than Alzheimer’s disease has previously been reported (Grossman et al., 2004
). Different demographics of patients including stage and severity of cognitive impairment may explain the different findings. For example, in the previous MRI study (Grossman et al., 2004
), patients with Alzheimer’s disease or FTD were matched for dementia severity based solely on MMSE scores, which assesses predominantly memory impairments but less executive functions. In fact, a previous neuropsychological study demonstrated that FTD patients could be more impaired in judgement and problem solving than Alzheimer’s disease patients when MMSE scores of the patients are matched (Rosen et al., 2004
). It is therefore conceivable that previous findings of more frontal lobe atrophy in FTD than in Alzheimer’s disease can be explained at least in part by greater executive dysfunction in some of FTD patients than in Alzheimer’s disease patients. It is also possible that Alzheimer’s disease heterogeneity may lead to different results, depending whether the frontal cortex is involved in the disease process or not. Although there was no difference in cortical thinning in the frontal regions between Alzheimer’s disease and FTD in this study, we found that Alzheimer’s disease was associated with a thinner cortex in the dorsolateral frontal cortex than the orbital and medial frontal cortices (), while FTD was associated with a similar cortical thinning in the dorsolateral, orbital and medial frontal cortices (). Taken together, these results suggest that while there are areas of overlap between the atrophic patterns of the two dementias, their patterns are dissociable and measurements of cortical thickness may be a useful surrogate marker for Alzheimer’s disease and FTD.
In this study, we found that FTD was associated with cortical thinning in inferior parietal lobes and the posterior cingulate cortex, which are prominent regions affected by Alzheimer’s disease pathology. Other MRI studies using voxel-based morphometry found no significant structural abnormalities in these regions in FTD (Rosen et al., 2002
; Gee et al., 2003
; Whitwell et al., 2005
). There are several possible explanations for the discrepancy. First, cortical thickness measurements may be more sensitive than voxel-based morphometry in detecting cortical alterations. Second, older patients with FTD may have concomitant Alzheimer’s disease pathology or could be false negatively classified as Alzheimer’s disease patients. However, we did not find systematic differences in brain atrophy between younger and older FTD patients; nor did we discover differences in the neurocognitive characteristics between FTD patients with more parietal atrophy and less parietal atrophy. Nonetheless, only autopsy can provide conclusive results for comorbidity of Alzheimer’s disease pathology in some patients with FTD symptoms.
Another finding is that regional cortical thinning correlated with dementia severity in Alzheimer’s disease but no significant correlations were seen in FTD. It is unclear why correlations show a disease selective effect. One reason could be that severity of FTD, which presents with predominantly behavioural problems, is not accurately reflected in MMSE and CDR tests, which assess primarily cognitive functions, though other studies have used MMSE and CDR together to compare dementia severity between Alzheimer’s disease and FTD (Likeman et al., 2005
). Previous pathological studies showed that FTD impacts also white matter and several subcortical nuclei such as the thalamus in addition to cortical involvement (Mann and South, 1993
; Broe et al., 2003
; Schofield et al., 2003
). Thus, damage of white matter and subcortical nuclei may play a role in cognitive impairment in FTD. However, both dementias are neuropathologically defined by cortical neuronal changes, thus involvement of white matter is not surprising, because neuronal damage implies damage to the white matter as the white matter reflects the neuronal axons and dendrites that are affected in Alzheimer’s disease as well as in FTD. More studies involving novel MRI techniques, such as diffusion imaging that is more sensitive to white matter changes than volumetric MRI are needed to evaluate the difference of white matter and subcortical nuclei between Alzheimer’s disease and FTD. In addition, the negative finding between cortical thickness and cognitive function may be due to the small range of MMSE and CDR scores in FTD patients, which makes detecting a relationship between cortical thickness and cognitive functions unreliable. More studies of FTD patients with a broader spread of dementia severity are needed to evaluate a potential relationship between cortical thickness and cognitive functions.
The best discriminator between Alzheimer’s disease and FTD was parietal lobe atrophy in Alzheimer’s disease. This is in agreement with a recent diagnostic MRI study, which reported for a range of pathology confirmed dementia cases that only posterior greater than anterior gradient of atrophy was highly specific for Alzheimer’s disease when compared to other dementias, including FTD (Likeman et al., 2005
). Overall, however, we found no significant improvement in correctly classifying CN, Alzheimer’s disease and FTD from cortical thickness measurement as compared to volumetric measurements of the cortex. We expected that measurements of cortical thinning would be less confounded by underlying white matter atrophy than measurements of cortical volume. Our argument was based on the assumption that the inner surface area of grey matter, which is interfaced with the surface area of white matter, might shrink as a consequence of white matter atrophy or white matter lesions, hence impacting computations of volume measurements more than computations of cortical thickness. The negative outcome could be due to the fact that thickness measurements are intrinsically limited by the finite resolution of MRI, diminishing the advantage of assessing cortical thickness versus cortical volume. Studies at higher magnetic fields that can afford higher image resolution may overcome this limitation. Although cortical thickness and volume measurements provided similar divisions between Alzheimer’s disease and FTD, cortical thickness measurement significantly improved the classification between the two diseases based on neuropsychological scores alone. In addition, measurement of cortical thickness provided a similar distinction between Alzheimer’s disease and FTD patients than biomarkers such as CSF based tau and isoprostane and better classifications than CSF amyloid beta (1–42), as reported in the previous study (Grossman et al., 2005
). Taken together, this suggests that cortical thickness is a useful marker for differentiating the illnesses.
A major limitation of this study is that the diagnosis of dementia and its type was made clinically without autopsy confirmation. Therefore, it is possible that some of the patients with FTD had also Alzheimer’s disease, vice versa or had other causes of dementia. The inclusion of patients at a more advanced stage of disease may reduce the relevance of our findings to the earliest detection of dementia and differentiation between Alzheimer’s disease and FTD. Studies including early stage patients may be more useful in evaluating the clinical diagnostic value of cortical thickness. The relatively young age of patients with Alzheimer’s disease, used to match the average age of the patients with FTD, may limit the projection of the findings to an older Alzheimer’s disease population.