The findings from this study suggest that grey matter loss in the temporoparietal neocortex may be a sensitive and specific marker of AD pathology. Temporoparietal grey matter loss was a feature of AD pathology in subjects with both a typical and atypical clinical presentation. Importantly, however, it was also significantly more affected in atypical AD than in a clinically matched cohort of subjects with FTLD pathology. Volume measurements of the temporoparietal cortex allowed good discrimination between the AD and FTLD groups demonstrating potential clinical utility.
Temporoparietal atrophy is well documented to be a classic, and relatively early, feature in subjects with typical AD (Baron et al., 2001
; Chetelat et al., 2002
; Frisoni et al., 2005
; Whitwell et al., 2007
). Temporoparietal atrophy has also been previously associated with aphasia and AD pathology in which subjects were clinically diagnosed with aphasic dementia (Josephs et al., 2008b
). The term aphasic dementia has been used at our institution as a diagnosis whenever the clinical presentation was that of a prominent aphasia; however because of more widespread cognitive impairment, such as loss of episodic memory or poor calculations, the patient was not given a diagnosis of primary progressive aphasia. The diagnosis of aphasic dementia predates more recent terminology. As discussed in our previous manuscripts (Josephs et al., 2008b
), many of our subjects with aphasic dementia, but not all, would fulfill criteria for the more recently characterized logopenic progressive aphasia syndrome (Gorno-Tempini et al., 2004
) which is also characterized by temporoparietal and medial temporal atrophy.
It is less common, however, for subjects presenting with CBS or bvFTD to show underlying AD pathology (Johnson et al., 1999
; Alladi et al., 2007
; Hu et al., 2009
; Shelley et al., 2009
), and neither syndrome is typically associated with predominant temporoparietal atrophy. Corticobasal syndrome is characterized by insidious onset and a progressive course of asymmetric cortical dysfunction, for example, ideomotor apraxia, cortical sensory loss or myoclonus, and asymmetric extrapyramidal dysfunction reflected by at least levodopa unresponsive appendicular rigidity or dystonia (Boeve et al., 2003
; Litvan et al., 2003
). It is typically associated with frontoparietal atrophy (Boxer et al., 2006
; Josephs et al., 2008a
), although in a recent study we found some suggestion of temporoparietal hypoperfusion on SPECT in a subset of CBS subjects with AD pathology (Hu et al., 2009
). Subjects with bvFTD show behavioral and personality changes (Neary et al., 1998
) and are typically associated with frontotemporal atrophy (Neary et al., 1998
; Rosen et al., 2002
; Boccardi et al., 2005
). Our findings suggest that the presence of AD pathology is associated with a somewhat different pattern of atrophy than what is typically observed in CBS and bvFTD. Specifically, in bvFTD, the pattern of atrophy involved more parietal lobe atrophy than is typically observed, and in CBS, temporal lobe atrophy was observed in addition to the typical frontoparietal pattern. It therefore appears that a pattern of temporoparietal atrophy may suggest the presence of AD pathology even in subjects presenting with non-amnestic clinical syndromes.
Another important finding from this study was that the atypical AD subjects showed a relative sparing of the hippocampus and medial temporal lobes, which concurs with the fact that episodic memory impairment was not the dominant symptom in these subjects. Conversely, severe involvement of the hippocampus and medial temporal lobes was observed in the typical AD subjects, as has previously been demonstrated (Jack et al., 1992
; Fox et al., 1996
) and is expected, given the early and dominant episodic memory impairment observed in typical AD (McKhann et al., 1984
). Severe involvement of the hippocampus and medial temporal lobe was also found in the FTLD group, driven mainly by the bvFTD and aphasic dementia subjects, which concurs with findings from a previous study that investigated pathologically confirmed FTLD subjects (Barnes et al., 2006
). Some involvement of the left hippocampus was observed in the aphasic dementia subjects fitting with previous aphasia studies (Gorno-Tempini et al., 2004
; Josephs et al., 2008b
), although to a lesser degree than the typical AD and FTLD subjects. Although there was a slight trend for disease severity to be worse in the typical AD and FTLD subjects there was no significant difference across the three groups suggesting that greater medial temporal lobe atrophy is likely not a result of more severe disease. Similarly, there was no difference in the time from onset to scan across groups. The fact that the atypical AD subjects showed more temporoparietal atrophy than the other groups also argues against this confound. The ROC analysis showed that while hippocampal volumes provide good discrimination between the atypical AD and FTLD groups the temporoparietal volume provides a little better discrimination. However, the addition of hippocampal volume to the temporoparietal volume further improves the discrimination over temporoparietal volume alone. Therefore, consideration of both these regions would improve the ability of a clinician to predict underlying pathology. If a subject with a bvFTD, CBS or aphasic dementia clinical syndrome showed temporoparietal atrophy in the context of relative sparing of the hippocampus one could predict the presence of AD pathology.
Atrophy of medial parietal regions, including the posterior cingulate and precuneus, has also been associated with AD (Baron et al., 2001
; Jones et al., 2006
). While no significant differences were identified across disease groups for the medial parietal lobe in this study, there was a trend for greater involvement in the typical and atypical AD groups compared to the FTLD group. Therefore, although the temporoparietal cortex provides the better marker of AD pathology the medial parietal cortex may also be helpful. Another finding was that the atypical AD subjects were also associated with greater involvement of the putamen than the typical AD subjects. This finding appears to be driven by the presence of putamen volume loss in the CBS subjects which is not unexpected. In fact, we could speculate that parkinsonism in the CBS patients with AD pathology may be associated with atrophy of this region.
The strengths of this study are that all subjects had histological confirmation of diagnosis and that the subject groups were well matched. The sample sizes were also reasonable in the main disease groups given that they are pathologically confirmed, although were smaller for the atypical AD group when split by clinical diagnoses. It should also be noted that these ROC analyses demonstrate group discrimination for our cohort and will need to be validated. Furthermore, the degree of hippocampal atrophy observed in the FTLD group is likely to vary dependent on the specific FTLD pathologies (Whitwell et al., 2005
; Josephs et al., 2008a
; Whitwell et al., 2009
). Additional strengths are that both the VBM and atlas-based parcellation techniques are automated and unbiased. The presence of secondary pathologies was also assessed and we found no differences across the typical and atypical AD groups in either the presence of Lewy Bodies or vascular pathology. The atypical AD subjects had a lower proportion of cases with TDP-43 immunoreactive inclusions than the typical AD subjects, although this difference was not significant. The presence of these pathologies therefore does not appear to be driving the clinical findings in the atypical group. We have previously shown that TDP-43 in AD does not account for a frontotemporal like phenotype (Hu et al., 2007
). We have also demonstrated that hippocampal atrophy is a feature of AD in the absence of TDP-43 (Josephs et al., 2008c
), showing that the lack of TDP-43 immunoreactivity in the atypical AD subjects could not explain the lack of hippocampal atrophy. The box plot in also shows that the finding of temporoparietal loss in the atypical AD subjects is not being driven by one particular clinical group with small volumes observed in bvFTD, CBS and aphasic dementia subjects. This study therefore suggests that patterns of atrophy on MRI, particularly with reference to the temporoparietal cortex and hippocampus, could prove to be useful in predicting the presence of underlying AD pathology.