In our previous work diffuse deposition patterns of neocorti

In our previous work, diffuse deposition patterns of neocortical Aβ and a hierarchical upward spreading pattern of tau were determined with in vivo 18F-florbetaben and 18F-flortaucipir positron emission tomography (PET) studies, based on the regional frequency of involvement by tau and Aβ (Cho et al., 2016a, Cho et al., 2016b). Interestingly, the frequency of tau deposition in the entorhinal qx 314 was approximately 10% greater than that of Aβ deposition in the middle and inferior temporal cortices. Although we may suspect that tau deposition in the entorhinal cortex precedes neocortical Aβ deposition, this method based on the involvement frequency does not provide information about the sequential order of tau and Aβ deposition. Conditional probability (CP) is a method for determining the likelihood of the occurrence of an event, given that a different event has occurred. It is thought to be theoretically more accurate for predicting temporal sequences of events with large cross-sectional data. A recent pathological study successfully applied this CP method to determine the spreading pattern of TAR DNA-binding protein 43 in AD (Josephs et al., 2016).
Methods
Results
Discussion Hyperphosphorylated tau pathology already appears in the subcortical nuclei including the locus ceruleus, raphe nucleus, and magnocellular nuclei of the basal forebrain even in the second decade and propagates to the transentorhinal and entorhinal cortices (Braak and Del Tredici, 2011, Braak and Del Tredici, 2015), and then the NFT pathology begins in the transentorhinal and entorhinal cortices and spreads toward the neighboring medial temporal regions and later to the distant association and primary cortices (Braak and Braak, 1991). In transgenic mouse models, tau spread via the downstream anatomical connections across the synapses inside and outside of the medial temporal cortex (Liu et al., 2012, de Calignon et al., 2012) and even to the contralateral regions strongly connected with the origin rather than to the spatially close regions (Ahmed et al., 2014). These support that tau spreads hierarchically through the anatomically and functionally connected network. In contrast to this upward hierarchically spreading pattern of tau, Aβ appears in the diffuse neocortex and spreads downward to the medial temporal cortex (Thal et al., 2002). In vivo imaging studies also replicated distinct spreading patterns of tau and Aβ. In our previous 18F-flortaupir PET study (Cho et al., 2016a, Cho et al., 2016b), entorhinal cortex was most frequently involved by tau deposition and followed by neighboring medial and basal temporal regions and the association cortices distant from the medial temporal regions. Primary cortices were least affected by tau. This sequence was almost similar to the regional order found by CP analysis in the present study with larger sample size. Unlike this pattern, there was a variability in the predicted regional order of Aβ deposition in the neocortex. In a 18F-florbetapir PET study, Aβ deposition was most frequently observed in the basal temporal, anterior cingulate, and parietal operculum (Grothe et al., 2017), while most frequently involved areas were the basal temporal, frontal and precuneus cortices in our previous 18F-florbetaben study (Cho et al., 2016a, Cho et al., 2016b). One longitudinal 18F-florbetapir PET study found that the early change of Aβ deposition occurs in the posterior cingulate, precuneus, and medial orbitofrontal cortices (Palmqvist et al., 2017). In our present study, based on CP model, ventral frontotemporal association cortices showed highest conditional probabilities over other areas. Variable regional order of Aβ deposition may be attributable to the pattern of diffuse Aβ deposition or the methodological difference. However, the medial temporal cortex is least affected by Aβ deposition in the cerebral cortex.