The main findings of this study are: (i) FTD is associated with FA reduction in the anterior corpus callosum, bilateral anterior and descending cingulum and uncinate fibre tracts. A voxel-by-voxel analysis showed even more widespread FA reduction in FTD which involved frontal and temporal white matter regions, expanded to parietal, and spared occipital white matter. (ii) In contrast, Alzheimer's disease is associated with FA reduction in bilateral descending cingulum, left posterior and anterior cingulum, and left uncinate fibre tracts, which is in agreement with previous DTI studies. (iii) A direct comparison between FTD and Alzheimer's disease reveals that FTD is associated with lower FA values than Alzheimer's disease in frontal fibres including the anterior corpus callosum, and bilateral uncinate fibres. By contrast, Alzheimer's disease shows no significant regional FA reductions relative to FTD. Furthermore, most FA reductions are related to increased Dra and some to increased Dax values. Taken together, the different regional patterns of DTI in FTD and Alzheimer's disease compared to CN suggest a characteristic distribution of white matter degradation for each disease. FTD is shown to be associated with more extensive white matter degradation than Alzheimer's disease.
The findings suggest that frontal and temporal white matter regions are particularly vulnerable in individuals diagnosed with FTD. These regions include: the anterior corpus callosum, which contains fibres connecting to frontal cortex; bilateral anterior and the descending cingulum, which connect to the major limbic cortices including anterior cingulate gyrus and hippocampus and bilateral uncinate fibres, which connect the frontal and inferior temporal cortices. This regional pattern of FA reduction overlaps with the pattern of white matter atrophy that we have previously reported with structural MRI studies (Cardenas, et al
; Chao et al
). Patterns of grey matter atrophy and hypoperfusion have also been reported in FTD (Rosen et al
; Jeong et al
; McMurtray et al
; Seeley et al
). Taken together, these findings suggest that degradation of the white matter fibre bundles occurs in association with white matter loss and cortical abnormalities in FTD. Therefore, white matter changes are a major pathological characteristic of FTD (Larsson et al
). The DTI findings of white matter vulnerability in FTD are most likely due to the axonal degeneration associated with injury/death of grey matter neuronal cell bodies. However, these findings are also consistent with histopathological evidence of tau protein deposition in white matter of FTD (Schofield et al
). Notably, we found substantial FA reduction in the bilateral descending cingulum, which is a major limbic fibre loop in the medial temporal lobe. The finding is consistent with reports from structural MRI studies of neurodegeneration of the paralimbic network in FTD (Rabinovici et al
; Seeley et al
), but has not been observed in DTI studies before.
Our findings of FA reductions in the parietal and temporal regions in Alzheimer's disease are in agreement with several other DTI studies (Takahashi et al
; Fellgiebel et al
; Naggara et al
; Taoka et al
; Xie et al
; Stahl et al
; Teipel et al
; Salat et al
., in press). DTI abnormalities in fibres located deep in the posterior white matter, such as the left posterior cingulum and the posterior corpus callosum have consistently been reported for Alzheimer's disease and even for mild cognitive impairment (MCI) (Fellgiebel et al
; Medina et al
; Stahl et al
; Zhang et al
; Parente et al
), which is a diagnostic category that will include many subjects with a preclinical stage of Alzheimer's disease. In addition, we found abnormalities in the descending branch of cingulum (i.e. the parahippocampal fibre), consistent with histological, structural and metabolic studies suggesting that the medial temporal region is particularly vulnerable to Alzheimer's pathology (see review papers: Anderson et al
; Drzezga, 2008
; Ries et al
). Moreover, the observation of FA reduction in the descending cingulum replicates our previous DTI results in an entirely different population of Alzheimer's disease and MCI patients (Zhang et al
). More recently, altered diffusion properties of parahippocampal white matter in Alzheimer's disease were found even in absence of significant grey matter degeneration as measured by hippocampal atrophy (Salat et al
., in press). These findings imply that disintegration of the parahippocampal fibres is a systematic and potentially early marker of incipient Alzheimer's disease. However, the pathological underpinning of FA changes in white matter in Alzheimer's disease remains elusive with several explanations possible. One possibility is that the degeneration of white matter is secondary to grey matter pathology. Alzheimer's disease is thought to begin in grey matter with deposition of β-amyloid around neuronal cell bodies. The disease is also associated with accumulation of tau protein in neuronal cells (Hashimoto et al
). These cellular processes lead to neurodegeneration including loss of axons within associated white matter tracks. This white matter degradation can then be detected via FA reductions measured by DTI. An alternative explanation is that white matter degeneration is a direct consequence of Alzheimer's pathology. Braak et al
) hypothesized that as brain development takes place, regions that become myelinated later in the process, such as cortical association areas, have fewer oligodendrocytes supporting greater numbers of axons compared to regions that become myelinated earlier (primary sensorimotor regions). Oligodendrocytes have particularly high metabolic demands in the cortical association areas to maintain the distributed network of their axons, which makes these axons more susceptible to pathological processes such as oxidative stress (Bartzokis, 2004
). These and other changes may lead to the diminished white matter integrity in the association cortical regions. Longitudinal studies, using DTI and cortical mapping together, are needed to investigate the relationship between white matter and grey matter changes in more detail.
Relative to Alzheimer's disease, FTD was associated with markedly reduced FA in the frontal brain regions. This finding is consistent with several structural and functional imaging studies which reported greater grey matter loss and reduced cerebral blood flow in frontal brain regions in FTD when compared to Alzheimer's disease (Julin et al
; Kitagaki et al
; Varma et al
; Varrone et al
). Taking these DTI, structural and functional findings in FTD together, the results emphasize an association between white matter degradation and cortical abnormalities in FTD. In contrast, Alzheimer's disease showed no significant FA reductions relative to FTD. This observation is surprising because several structural and functional imaging studies of Alzheimer's disease reported substantial abnormalities in posterior brain regions in Alzheimer's disease relative to FTD (Varrone et al
; Du et al
). One possible explanation is that white matter and grey matter are affected differently in Alzheimer's disease and FTD throughout the course of the disease. Our ongoing studies hope to better elucidate the relationship between grey matter and white matter changes in Alzheimer's disease and FTD. Another possibility, which cannot be ruled out is that the patients diagnosed with Alzheimer's disease in this study were at an earlier stage of the disease and less impaired than the FTD patients. It is difficult to compare disease severity in two different disorders which have quite different symptomatology. It is possible that these patients with Alzheimer's disease will eventually exhibit regions of reduced FA relative to FTD as the disease progresses. Prospective studies of FTD and Alzheimer's disease are needed to resolve this issue.
Examination of regional alterations in various diffusion parameters, including radial and Dax
in addition to anisotropy, showed distinct zones of alterations that potentially stemmed from different underlying pathological processes. For example, radial and Dax
of the uncinate fibres was substantially higher in FTD than in Alzheimer's disease where most regions of reduced FA values showed also increased Dra
. On the other hand, differences in diffusion anisotropy of the anterior corpus callosum fibres between FTD and Alzheimer's disease disappeared when analysed in terms of Dra
. These results suggest that some disease specificity may be obtained by further examination of different DTI eigenvalues. Diffusion studies in animals suggest that increased Dra
can be associated with myelin damage (Song et al
; Harsan et al
). Accordingly, our findings of increased Dra
in some regions in FTD and Alzheimer's disease, including the uncinate fibres, could reflect processes of demyelination. In contrast, demyelination would not explain the FA differences in anterior corpus callosum fibres between FTD and Alzheimer's disease, since Dra
was not significantly different. However, the interpretation of DTI indices with respect to pathology is notoriously ambiguous and more validation studies to the underlying causes of DTI changes are necessary.
Several limitations of our study ought to be mentioned. First, the clinical diagnoses were not confirmed by autopsy and, therefore, the pathological processes underlying the DTI alterations may not be associated with FTD and Alzheimer's disease. In particular, FTD, which is one of several variants of frontotemporal lobar degeneration that have different aetiologies, suffers from diagnostic uncertainty. Consequently, the pattern of DTI changes may vary among FTD patients with different aetiologies. Second, although the groups were matched for WMSH severity and regions including WMSH were avoided in analysis, WMSH due to cerebrovascular disease may have affected FA and the other DTI measures. The relation between fibre integrity and cerebrovascular disease needs to be explored further. Another concern is that large variations of brain shape in older subjects limited the ability to achieve spatial normalization of white matter regions across the subjects. This is particularly critical as cortical and ventricular shapes may have driven image registration. Some FA changes, especially in regions surrounding the ventricles as seen in and , could be a result of artefacts due to image mis-registration and underlying variations of brain shapes. The use of more sophisticated algorithms for DTI registration such as Tract-Based Spatial Statistics (TBSS) (Smith et al
) or dense feature deformation morphometry (Studholme, 2008
) could lead to different results. Further studies involving a larger number of patients and autopsied confirmation of the pathology are needed to support our results.
In summary, the regional DTI alterations suggest that FTD and Alzheimer's disease are each associated with a characteristic distribution of white matter degradation. The results also suggest a greater vulnerability of white matter in FTD than in Alzheimer's disease. DTI may provide additional diagnostic information to distinguish Alzheimer's disease from FTD.