Structural and functional imaging studies searching for

Structural and functional imaging studies searching for specific cortical areas related to each cognitive function domain have provided clues to the spreading patterns of cognitive dysfunction (Domoto-Reilly et al., 2012, Firbank et al., 2016, Li et al., 2012, Machulda et al., 2003, Mandal et al., 2012, McDonald et al., 2012, Thiyagesh et al., 2009, Tomaszewki Farias et al., 2005, Wolk and Dickerson, 2011). In addition, there have been several reports showing the relationship between tau deposition and cognitive dysfunction. The severity of memory dysfunction was closely related to tau deposition in the medial temporal Topiroxostat sale in in vivo tau PET studies (Cho et al., 2016a, Cho et al., 2016b, Johnson et al., 2016, Scholl et al., 2016). Deterioration in visuospatial function was related to tau deposition in the right parietal cortex, particularly in early-onset AD and MCI (Cho et al., 2017). Therefore, the hypothetical model of progression of cognitive dysfunction may be simplified according to the pathological progression model in AD. Memory dysfunction can be attributed to the involvement of the medial temporal cortex, language dysfunction to the involvement of the lateral temporal cortex, and attention/executive and visuospatial functions to the involvement of the fronto-parietal association cortices. In this study, a modification of a previously reported method for selecting a reference data set and Z-score-based determination of regional involvement was used (Schwarz et al., 2016). However, it remains uncertain whether the Z-score of 2.5 to minimize false positives was higher or even lower than the true cutoff. To overcome this limitation, different cutoffs around the value 2.5 were tested, and we obtained an almost similar spreading pattern and sequential orders for tau and Aβ. Nevertheless, the question remains on whether to use the Z-score cutoff of 1.5 to determine the involvement of neuropsychological domains. In addition, it seems to be inadequate to use cutoff 18F-flortaucipir SUVR value of entorhinal cortex for selecting the reference controls. This logic for selecting the reference controls least affected by tau deposition raises a concern of pre-existing assumption for tau deposition and thereby artificially higher CPs for tau deposition in the entorhinal cortex over other cortical regions. To test if the regional order can be affected by different set of reference controls, we also tested same CP analysis by using the 78 reference controls screened by cutoff 18F-flortaucipir SUVR value 1.2 of the global cortex. Although statistical significance for the CP for tau deposition in the entorhinal cortex markedly reduced and thereby entorhinal cortex no longer has significantly higher CPs for tau over those for Aβ, the regional order was largely unaffected by different reference controls (Fig. S7). Although the CP model is appropriate for explaining the precedence of 2 separate events with cross-sectional data, an absence of evidence with longitudinal data is another limitation. In addition, it does not prove a causal relationship between those events. Therefore, it may be possible that the propagation of tau and Aβ pathologies are mutually independent processes, and the results were over construed due to the opposite direction and different mode of propagation; stepwise spreading of tau pathology from the medial temporal cortex versus rapid spreading of Aβ throughout the widespread neocortical areas and toward the medial temporal cortex. However, previous animal studies have shown interactions between the 2 pathological proteins during the development and progression of AD (Frautschy et al., 1991, Gotz et al., 2001, Hurtado et al., 2010, Lewis et al., 2001). In addition, we have to consider the possibility that tau restricted to medial temporal cortex promotes Aβ deposition in the distant areas because the CP model does not provide an evidence for causal relationship.