Received: 31 July 2020 Revised: 16 December 2020 Accepted: 11 January 2021 DOI: 10.1002/hbm.25350 RESEARCH ARTICLE Network interdigitations of Tau and amyloid-beta deposits define cognitive levels in aging Chan-Mi Kim1,2 | Victor Montal1,3,4 | Ibai Diez1,2 | William Orwig1,2 | Alzheimer's Disease Neuroimaging Initiative (ADNI) | Jorge Sepulcre1,2 1 Gordon Center for Medical Imaging, Department of Radiology, Massachusetts Abstract General Hospital, Harvard Medical School, Amyloid-beta (Aβ) plaques and tau neurofibrillary tangles are pathological hallmarks Boston, Massachusetts 2 of Alzheimer's disease (AD); their contribution to neurodegeneration and clinical Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, manifestations are critical in understanding preclinical AD. At present, the mecha- Massachusetts General Hospital, Harvard nisms related to Aβ and tau pathogenesis leading to cognitive decline in older adults Medical School, Charlestown, Massachusetts 3 Memory Unit, Department of Neurology, remain largely unknown. Here, we examined graph theory-based positron emission Hospital de la Santa Creu i Sant Pau, tomography (PET) analytical approaches, within and between tau and Aβ PET modali- Biomedical Research Institute Sant Pau, Universitat Autonoma de Barcelona, ties, and tested the effects on cognitive changes in cognitively normal older adults Barcelona, Spain (CN). Particularly, we focused on the network interdigitations of Aβ and tau deposits, 4 Centro de Investigacón Biomédica en Red de along with cognitive test scores in CN at both baseline and 2-year follow-up (FU). Enfermedades Neurodegenerativas (CIBERNED), Spain We found highly significant Aβ-tau network integrations in AD vulnerable areas, as well as significant associations between those Aβ-tau interdigitations and general Correspondence Jorge Sepulcre, Gordon Center for Medical cognitive impairment in CN at baseline and FU. Our findings suggest a distinctive Imaging, Department of Radiology, contribution of interlinking network relationships between Aβ and tau deposits in Massachusetts General Hospital, Harvard Medical School, 13th Street, Charlestown, MA heteromodal areas of the human brain. They support a network-based interaction 02129. between Aβ and tau accumulations as a key factor for cognitive deterioration in CN Email: sepulcre@nmr.mgh.harvard.edu, jsepulcre-bernad@mgh.harvard.edu prior to dementia. Funding information KEYWORDS National Institute on Aging, Grant/Award aging, Alzheimer's disease, amyloid PET, cognition, connectivity analysis, Tau PET Numbers: R01AG061445, R01AG061811 1 | I N T RO DU CT I O N contribution to neurodegeneration and cognitive decline is presumed to be a key to understand preclinical AD (Ballard et al., 2011; Johnson Alzheimer's disease (AD) is the most common type of dementia char- et al., 2013; Pike et al., 2007). However, at present, it remains largely acterized by progressive memory loss and the loss of independence in unknown how Aβ and tau accumulation spatially intersect in brain cir- daily activities. Aggregation of amyloid-beta (Aβ) peptides and accu- cuits and whether these factors may explain the emergence of cogni- mulation of aggregated forms of tau proteins are pathological hall- tive decline in older adults. marks of AD, appearing in initial stages of AD prior to onset of Increased Aβ plaques and tau neurofibrillary tangles are com- symptoms (Fleisher et al., 2015; Jack Jr et al., 2010; Jansen monly found in the neocortex of AD patients in postmortem autopsies et al., 2015). Two pathological hallmarks, Aβ plaques and the neurofi- (Braak & Braak, 1991a, 1991b, 1995) as well as in positron emission brillary tangles, are considered sensitive markers for AD, and their tomography (PET) imaging studies (Barthel et al., 2011; Chien This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. © 2021 The Authors. Human Brain Mapping published by Wiley Periodicals LLC. 2990 wileyonlinelibrary.com/journal/hbm Hum Brain Mapp. 2021;42:2990–3004. KIM ET AL. 2991 et al., 2013; Clark et al., 2011; Clark et al., 2012; Johnson et al., 2016; potential network dependency between Aβ and tau accumulations in Klunk et al., 2004; Marquié et al., 2015; Pike et al., 2007; Van- older adults or mild AD (Iaccarino et al., 2018; Sepulcre et al., 2016). denberghe et al., 2010; Wong et al., 2010). The coexistence of both Previously, a strong association between the spatial distribution pathological hallmarks is associated with synaptic dysfunction and of tau accumulation and cognitive impairment was observed, primarily neuronal loss that mediate memory and cognition (Iqbal & Grundke- in the medial temporal lobe (MTL), in older adults and AD (Ghoshal Iqbal, 2002; Selkoe, 2002; Sperling et al., 2019). Several studies have et al., 2002; Mitchell et al., 2002; Nelson et al., 2012; Quiroz shown a strong association between Aβ and greater tau accumula- et al., 2018; Rolstad et al., 2013; van Rossum et al., 2012). Although tions at the cellular and molecular levels (Bennett et al., 2017; Götz, tau is more closely associated with cognitive impairment than Aβ Chen, Van Dorpe, & Nitsch, 2001; He et al., 2018; Hurtado (Ghoshal et al., 2002; Mitchell et al., 2002; Nelson et al., 2012; Quiroz et al., 2010; Lewis et al., 2001; Williamson, Usardi, Hanger, & et al., 2018; Rolstad et al., 2013; van Rossum et al., 2012), a significant Anderton, 2008). The spatial association between Aβ and tau accumu- association between the longitudinal trajectories of Aβ and the pro- lations is somewhat inconsistent or minimally overlapping in specific gressive cognitive decline, particularly memory function, has been also regions of the brain in older adults prior to dementia (Schöll observed in aging or in the early stages of AD (Hanseeuw et al., 2019; et al., 2016; Whitwell et al., 2018). Aβ plaques first arise in the neo- Landau et al., 2012; Resnick et al., 2010; Villemagne et al., 2013). cortex and spread to deep subcortical regions, while tau neurofibril- Other studies have suggested that neocortical Aβ accumulation lary tangles occur in entorhinal and limbic areas first, then spread to potentially contributes to the association of tau with cognitive impair- the neocortical areas (Arnold, Hyman, Flory, Damasio, & Van ment in older adults (Hanseeuw et al., 2019; Jacobs et al., 2018; Schöll Hoesen, 1991; Braak & Braak, 1991b, 1995). Thus, the lack of spatial et al., 2016; Sperling et al., 2019) and early stages of AD (Fagan consistency between Aβ and tau accumulations has resulted from the et al., 2007; Johnson et al., 2016). However, the mechanism of the distinctive spreading patterns of Aβ pathology and tau pathology in association between Aβ and tau pathogenesis that links to cognitive AD progression (Arnold et al., 1991; Braak & Braak, 1991b, 1995). decline in older adults prior to AD remains largely unknown. Although there is not a perfect alignment of spatial distributions, a Given that the network physical dependencies and overlapping potential contribution of global Aβ accumulation to increasing accu- distributions of Aβ and tau in the neocortex -also called Aβ-tau inter- mulation of tau was suggested by several imaging studies (Jacobs digitations in this study- have been suggested recently (Kim et al., 2018; Pontecorvo et al., 2017; Quiroz et al., 2018; Wang et al., 2019; Sepulcre et al., 2016; Sepulcre et al., 2018), it is critical to et al., 2016), particularly if distributed network information is taken examine whether Aβ accumulation and tau accumulation affect subtle into consideration. In fact, it is now known that regions of Aβ and tau cognitive impairment independently, or combination of Aβ and tau accumulation show network bonds, even without spatial overlapping accumulations have synergetic effects in older adults with clinically between them (Brier et al., 2016;Iaccarino et al., 2018; Sepulcre normal cognition prior to symptom onset. To date, little is known et al., 2016). Therefore, understanding the spatial and network inter- about the role of the association between Aβ and tau accumulations actions between Aβ and tau accumulations in cognitively healthy par- in longitudinal cognitive changes in aging. Therefore, investigating the ticipants may elucidate the underlying role of both pathologies in the network-based PET correlation between Aβ and tau deposits, and development of cognitive decline. their association with the cognitive variability in older adults may pro- Given the increasing evidence of the transcellular propagation of vide insights into how preclinical AD progresses into a more patent AD pathogenesis along neuronal circuits as well as in functionally cognitive failure. linked distributed regions (Braak & Del Tredici, 2011; Clavaguera In this study, we explore the network-based relationships et al., 2009; De Calignon et al., 2012; Guo & Lee, 2014; Iba between Aβ and tau accumulations in cognitively healthy older adults et al., 2013; Stöhr et al., 2012; Thal, Rub, Orantes, & Braak, 2002; measured by PET images. Next, we examine the association between Walker & Jucker, 2011), graph theory-based network approaches PET network profiles and neuropsychological (NP) test scores in order have been applied to understand how Aβ and tau propagate in the to understand the effects of the Aβ and tau network relationships at in vivo human brain (Kim et al., 2019; Sepulcre et al., 2016; Sepulcre baseline and 2-year follow-up (FU). Our findings support the distinc- et al., 2018). The network-based spatial spreading patterns of Aβ and tive contribution of Aβ-tau pathological connectivity crosstalk in the tau accumulations have shown distinctive pathways, such as tau prop- appearance of the initial cognitive changes in putative preclinical agation from medial/inferior temporal lobe to orbitofrontal cortex, stages of AD in older adults. and Aβ propagation from posterior cingulate cortex (PCC) to lateral parietal lobe, in older adults (Sepulcre et al., 2016), mild AD (Iaccarino et al., 2018), and AD spectrum patients (Kim et al., 2019). However, 2 | M A T E R I A L S A N D M ET H O D S some hubs of both Aβ and tau pathology networks overlapped in regions which are partly associated with AD vulnerability (Iaccarino 2.1 | Participants et al., 2018; Kim et al., 2019; Sepulcre et al., 2016; Sepulcre et al., 2018). Additionally, local tau accumulation was significantly A total of 193 cognitively normal older adults (CN) from Alzheimer's associated with distributed Aβ accumulation, which suggested a Disease Neuroimaging Initiative (ADNI) projects (http://www.loni. 2992 KIM ET AL. ucla.edu/ADNI/) who completed cognitive assessment, T1-weighted angle = 9.0 , image matrix = 256 × 256, 210 slices, voxel magnetic resonance imaging (MRI), 18F-Florbetapir (FBP) PET, and size = 1.0 × 1.0 mm, slice thickness = 1.0 mm) or on 3-T Phillips Sys- 18F-AV-1451 (tau) PET at baseline visits were investigated. We tems (repetition time = 6.5 ms, echo time = 2.9 ms, flip angle = 9.0 , excluded nine subjects due to technical limitations of image processing, image matrix = 256 × 256, 210 slices, voxel size = 1.0 × 1.0 mm, slice such as a failure of coregistration between T1-MRI and PET or the fail- thickness = 1.0 mm). Some T1-weighted MRIs were acquired using an ure of the partial volume correction (PVC) process in PET images. All accelerated sagittal inversion recovery-prepared fast spoiled gradient- data used in this study was acquired from ADNI-3 project between echo sequence on 3-T GE systems (repetition time = 7.7 ms, echo August 2016 and December 2019 (Weiner et al., 2017). Cognitive time = 3.1 ms, flip angle = 11.0 , image matrix = 256 × 256, 196 slices, assessment was performed using NP tests, such as Clinical Dementia voxel size = 1.0 × 1.0 mm, slice thickness = 1.0 mm). Detailed infor- Rating, Mini Mental State Examination (MMSE) (Folstein, Folstein, & mation of the MRI acquisition protocol is available from ADNI website McHugh, 1975), Geriatric Depression Scale, Rey Auditory Verbal Learn- (http://www.loni.ucla.edu/ADNI/). ing Test (RAVALT) (Rey, 1958), Alzheimer's Disease Assessment Scale- Cognitive Subscale (ADAS-Cog) (Mohs et al., 1997a, 1997b), ADNI composite score for memory (ADNI-MEM) (Crane et al., 2012), and 2.3 | PET acquisition ADNI composite score for executive function (ADNI-EF) (Gibbons et al., 2012). We additionally investigated the FU data in twenty CN, FBP (Aβ) PET scan was performed for 20 min in duration by 4-by-5 min who completed the cognitive assessment, T1-weighted MRI, FBP PET frames and was acquired starting 50–70 min after injection of and tau PET at the 2-year FU time point, for the validation analysis in 10.0 ± 1.0 mCi F18-Florbetapir. Tau PET scan was performed for this study. Two FU data were excluded due to the failure of the PET- 30 min in duration by 6-by-5 min frames and was acquired starting preprocessing. Finally, we defined three groups of CN, with 184 CN at 75–105 min after injection of 10.0 ± 1.0 mCi F18-AV-1451. Both FBP baseline, 18 CN at 2-year FU, and FU-matched 18 CN at baseline, for PET and tau PET images were acquired using one of several PET scan- the longitudinal analysis. FU-matched CN at baseline was subselected ners, including GE, Siemens, or Phillips Medical Systems. The detailed from all CN at baseline. Detailed demographic characteristics and NP methods for FBP PET and tau PET acquisitions are described elsewhere test scores of all participants are described in Table 1. (Weiner et al., 2017). We used an averaged single FBP PET and the averaged single tau PET from ADNI database, which were generated by averaging all frames of FBP PET or tau PET image set. All PET 2.2 | MRI acquisition images were reconstructed with 1.02 × 1.02 mm pixel size and 2.0 mm slice thickness; or 1.33 × 1.33 mm pixel size and 3.27 mm slice thick- T1-weighted MRIs were acquired using an accelerated sagittal ness; or 2.0 × 2.0 mm pixel size and 3.27 mm slice thickness depending magnetization-prepared rapid gradient echo sequence on 3-T SIE- on their scanner types. Detailed PET acquisition information is available MENS systems (repetition time = 2.3 ms, echo time = 2.9 ms, flip from ADNI website (http://www.loni.ucla.edu/ADNI/). TABLE 1 Demographics and clinical characteristics CN at baseline (n = 184) FU-matched CN at baseline (n = 18) CN at 2-year FU (n = 18) Gender F = 106/M = 78 F = 11/M = 7 F = 11/M = 7 Age (years) 74.92 ± 9.50 77.13 ± 6.31 79.22 ± 7.22 Education (years) 16.25 ± 3.44 15.30 ± 4.32 15.30 ± 4.32 MMSE 28.80 ± 2.51 29.06 ± 1.21 28.40 ± 1.64 CDR (IQR) 0.04 (0.0–0.0) 0.11 (0.0–0.0) 0.25 (0.0–0.5) MoCA 24.73 ± 3.41 24.72 ± 3.20 23.35 ± 3.25 GDS 0.92 ± 1.57 0.89 ± 1.49 1.35 ± 1.87 RAVALT 45.24 ± 10.63 44.00 ± 8.22 36.00 ± 14.59 ADAS-Cog 12.97 ± 5.22 13.04 ± 4.40 13.53 ± 6.35 ADNI-MEM 0.93 ± 0.56 0.83 ± 0.40 0.66 ± 0.86 ADNI-EF 1.02 ± 0.79 1.18 ± 0.80 0.92 ± 0.74 Amyloid positivity (global FBP-SUVR ≥ 1.10) 30.77% 44.44% tau positivity (composite ROIs ≥ 1.25) 2.73% 5.56% Abbreviations: ADAS-Cog, Alzheimer's Disease Assessment Scale-Cognitive Subscale; ADNI-EF, Alzheimer's Disease Neuroimaging Initiative composite score for executive function; ADNI-MEM, Alzheimer's Disease Neuroimaging Initiative composite score for memory; CDR, Clinical Dementia Rating; CN, cognitively normal older adults; FBP, 18F-Florbetapir; FU, follow-up; GDS, Geriatric Depression Scale; IQR, interquartile range; MMSE, Mini Mental State Examination; MoCA, Montreal Cognitive Assessment; RAVALT, Rey Auditory Verbal Learning Test; SUVR, standardized uptake value ratio. KIM ET AL. 2993 2.4 | MRI image processing cingulate, and posterior cingulate cortices, to whole cerebellum refer- ence region without performing PVC. The tau-positivity was defined All T1-weighted MRIs at baseline and 2-year FU were automatically by using the tau cut-off threshold of composite ROIs ≥1.25 in each processed by FreeSurfer recon-all procedure (FreeSurfer version individual (Mishra et al., 2017). The mean tau-SUVR in the composite 6.0.0; http://surfer.nmr.mgh.harvard.edu/) to reconstruct cortical sur- ROIs, including the entorhinal, lateral occipital, inferior temporal corti- faces and to segment region-of-interests (ROI) volumes (Dale, Fischl, & ces and amygdala, was divided by a mean value in the cerebellar gray Sereno, 1999; Desikan et al., 2006; Fischl, Sereno, Tootell, & reference region to decide tau-positivity in CN at baseline (Mishra Dale, 1999). The ROI volumes were defined by the gyral-based et al., 2017). Individual FBP-SUVR and tau-SUVR images were Desikan–Killiany atlas through an automated FreeSurfer process, corrected for partial volume effects by using an extended Müller– which used the depth from one sulcus to another to parcellate the Gärtner (MG) method, which estimates a true radioactivity concentra- cerebral cortex into standard neuroanatomical regions, for each indi- tion in human brain GM by considering a heterogeneity of GM activity vidual MRI (Desikan et al., 2006). All technical details of these proce- via four-compartment model, within PetSurfer procedure (Meltzer dures are described in prior publications (Dale et al., 1999; Desikan et al., 1996; Muller-Gartner et al., 1992; Rousset, Ma, & Evans, 1998). et al., 2006; Fischl et al., 2002; Fischl, van der Kouwe, et al., 2004; Fis- A GM threshold for PVC was set at 0.1 and the point spread function chl, Salat, et al., 2004; Fischl, Liu, & Dale, 2001). The brain tissues, for PVC was estimated at 8 mm. Detailed methodological explanation including gray matter (GM), white matter (WM), and cerebrospinal of MG PVC is described in a prior publication (Greve et al., 2016). fluid, were segmented from T1-weighted MRI and volume registered After that, individual partial volume corrected FBP-SUVR and tau- into Montreal Neurological Institute/International Consortium for SUVR images were coregistered to the corresponding MRI and then Brain Mapping (MNI/ICBM) space using watershed/surface deforma- registered to MNI/ICBM template by using the transformation tion procedure (Ségonne et al., 2004). The registered MRIs automati- matrixes obtained from the previous step. All partial volume corrected cally mapped into a common surface template using a surface-based PET data were down sampled from the standard space to 6 mm iso- averaging technique by considering cortical folding patterns. Then, voxel to avoid computational limitations of the high-dimensional data. predefined ROIs in the standard space were inversely mapped to each We used 6 mm isotropic MNI/ICBM template to define the structural native MRI using a high-dimensional spherical morphing procedure to information in all PET data. segment the ROIs in each individual (Fischl et al., 2002; Fischl, Salat, et al., 2004; Fischl, van der Kouwe, et al., 2004). In order to calculate the standard uptake value ratio (SUVR) and to perform the PVC in 2.6 | Group averaged FBP-SUVR and tau-SUVR PET image processing, we used ROIs to define the reference regions of PET images for each individual. To examine overall distributed patterns of both Aβ deposition and tau deposition in CN, the group averaged FBP-SUVR map and tau-SUVR map at baseline and 2-year FU were calculated in voxel-based MNI 2.5 | PET image processing space. Then, the voxel-based group averaged PET-SUVR maps were mapped into an averaged cortical surface by mapping a value of the All FBP PET and tau PET at both baseline and 2-year FU were middle point between inner and outer surfaces in each vertex point processed by FMRIB Software Library (FSL; http://fsl.fmrib.ox.ac.uk) via FreeSurfer recon-all procedure (FreeSurfer version 6.0.0; http:// and FreeSurfer PetSurfer procedure (FreeSurfer version 6.0.0; http:// surfer.nmr.mgh.harvard.edu/). Longitudinal changes in FBP-SUVR and surfer.nmr.mgh.harvard.edu/fswikiPetSurfer/) to perform cor- tau-SUVR between CN at 2-year FU and FU-matched baseline were egistration, calculation of SUVR, and to perform PVC (Greve investigated using a paired t-test in each voxel, and then mapped into et al., 2016). Each individual FBP PET or tau PET image was cor- an averaged cortical surface via FreeSurfer recon-all procedure egistered to the corresponding native T1-weighted MRI using a rigid- (FreeSurfer version 6.0.0; http://surfer.nmr.mgh.harvard.edu/). To body registration with mutual information cost function. All ROIs that determine significant change, false discovery rate (FDR) set at segmented from individual MRI were inversely registered to each indi- q < 0.05 was performed for multiple comparisons (Benjamini & vidual FBP PET or tau PET by using an inverse transformation of each Hochberg, 1995). coregistration matrix. Then, each individual FBP PET or tau PET was scaled by a mean value in the cerebellar gray reference region to cal- culate the SUVR (Sepulcre et al., 2018). To examine the pathological 2.7 | Graph theory-based PET correlation within status of our population, we additionally investigated amyloid- single PET modality and between PET modalities positivity and tau-positivity in CN at baseline. The amyloid-positivity was defined by considering the recommended threshold of global We measured voxel-level correlations within single PET modality and FBP-SUVR ≥1.10 in each individual (Joshi et al., 2012). The global different PET modalities. We created graph theory-based correlation FBP-SUVR was calculated as the ratio of the mean FBP-SUVR of six matrices, which defined a node as a voxel in GM and an edge as a par- cortical ROIs, including frontal, temporal, precuneus, parietal, anterior tial correlation coefficient between a pair of voxels of PET SUVR data 2994 KIM ET AL. in GM across all CN. To remove confounding effects, we adjusted all voxel b have significantly large number of correlations to the rest of partial correlations by age, sex, as well as the PET-SUVR values of the paired-voxels in comparison with other voxels. Thus, a voxel with high PET modality not included in that specific analysis (see details in the WD can be described as a hub voxel, which is strongly correlated with following sections). All voxel-based partial correlations were calcu- the rest of the paired-voxels. The WD of all 6,848 voxels was calcu- lated by Statistics and Machine Learning Toolbox within MATLAB. lated in each correlation matrix D and then resampled to the corresponding GM voxels in the MNI template. Finally, we obtained the WD map of Aβ-to-Aβ, Tau-to-Tau, Aβ-to-Tau, and Tau-to-Aβ 2.7.1 | Partial PET correlation within single PET correlation matrixes in group level at baseline. modality Each Aβ-to-Aβ correlation between different voxels was measured 2.8 | Relationship between the information of the by the partial correlation between a FBP-SUVR in a start voxel graph theory-based correlation matrix and cognitive b and a FBP-SUVR in the paired voxel d across all possible pairs of variability voxels within GM in CN while controlling for age, sex, and tau- SUVR in the paired voxel d. Similarly, each Tau-to-Tau correlation Flowchart for identifying the relationship between the averaged PET- was measured by the partial correlation between a tau-SUVR in a SUVR and cognitive variability in the PET correlation between differ- start voxel b and a tau-SUVR in the paired voxel d across all possi- ent PET modalities is shown in Figure 1. We examined correlation ble pairs of voxels within GM in CN while controlling for age, sex, between averaged PET-SUVRs of each paired-voxels and cognitive and FBP-SUVR in the paired voxel d. We obtained 6,848-by- scores in each PET-to-PET correlation combination. The linear associ- 6,848 matrix DwithinPET in each correlation within single PET ation analysis was performed to determine whether the information modality in group level. To determine significant correlation, we of the graph theory-based PET correlations is associated with cogni- corrected for multiple comparisons in each correlation matrix tive variability or not, as well as to identify which PET correlation, D withinPET by using a FDR set at q < 0.05 (Benjamini & such as the correlation within single PET modality or the PET correla- Hochberg, 1995). tion between different PET modalities, is better associated with the cognitive variability in CN. First, we calculated the averaged PET- SUVR between a start voxel and the paired voxel, which was defined 2.7.2 | Partial PET correlation between different by a significant correlation in the PET correlation matrix D, across all PET modalities possible voxel pairs in each correlation matrix D, in each individual level at baseline. Next, we investigated relationships between the Each Aβ-to-Tau correlation between different voxels was measured averaged PET-SUVR and NP scores, including MMSE, RAVALT, by the partial correlation between a FBP-SUVR in a start voxel b and ADAS-Cog, ADNI-MEM, and ADNI-EF, in each type of PET correla- a tau-SUVR in the paired voxel d across all possible pairs of voxels tion matrix in CN at baseline by using Pearson's partial correlation within GM in CN while controlling for age, sex, and FBP-SUVR in the analysis. Third, we constructed a linear plot between the averaged paired voxel d (Figure 1a). Each Tau-to-Aβ correlation was measured PET-SUVR in each paired-voxels and each NP score to examine a lin- by the partial correlation between a tau-SUVR in a start voxel b and a ear trend between the averaged PET-SUVR in each paired-voxels and FBP-SUVR in the paired voxel d across all possible pairs of voxels each NP score in CN at baseline (Figure 1b). within GM in CN while controlling for age, sex, and tau-SUVR in the Our correlation approach between PET-SUVRs and NP scores is paired voxel d (Figure 1a). We obtained 6,848-by-6,848 matrix sensitive to detect direct association between the network profiles DbetweenPET in each correlation between different PET modalities in of Aβ and tau accumulation and the cognitive scores of our samples. group level. The FDR correction was performed at q as 0.05 in each Thus, they indicate whether a voxel displays an overall significant correlation matrix DbetweenPET to determine the significant correlation association between network pathology and cognitive variability. (Benjamini & Hochberg, 1995). However, these correlations values do not provide information about how initial or late are these associations along the pathologi- cal process. To obtain this additional information, we used a 2.7.3 | Weighted degree of the correlation matrix centroid-based strategy. We calculated the centroid values of voxel-level correlations in order to determinate whether an Aβ, tau We calculated the weighted degree (WD) of each voxel in each corre- and NP correlation was predominantly populated by values in spe- lation matrix D to identify which voxel's PET-SUVR were highly corre- cific areas of the dispersion graph. For instance, a low centroid value lated with PET-SUVRs of the rest of GM voxels. The WD of a voxel would reflect a combination of low pathology and normal cognition b was calculated as the sum of the significant correlation coefficients (or initial states), while a high centroid value would reflect a combi- between voxel b and all possible paired voxel d, excluding a self- nation of high pathology and impaired cognition (late states) connection. High WD in voxel b indicates that the PET-SUVR of the (Figure 1b). We calculated centroid values of the cognition-related KIM ET AL. 2995 F I G U R E 1 Flowchart for measurement of sum of centroid values of averaged positron emission tomography (PET)-SUVR between different modalities by using PET-to-NP correlation. (a) An averaged PET-SUVR (Avr(FBPb,taud)) was calculated by averaging between FBP-SUVR of a voxel b (FBPb) and tau-SUVR of the paired voxel d (taud) that is significantly associated with the voxel b in FBP-to-Tau correlation matrix (DbetweenPET) (FDR corrected q < 0.05) across all CN at baseline and 2-year FU. Then, PET-to-NP correlation matrix was calculated by Pearson's partial correlation between the averaged PET-SUVR (Avr(FBPb,taud)) and ADAS-Cog scores while controlling for age, sex, years of education in CN at baseline only (FDR corrected q < 0.05). (b) A centroid of the averaged PET-SUVRs was calculated by considering the scattered plots between the averaged PET-SUVRs of each paired voxels (b,d) and ADAS-Cog scores. The centroid value was only calculated if the FBP-SUVR was significantly correlated with tau-SUVR in the corresponding link between voxel b and voxel d from the FBP-to-Tau correlation matrix (FDR corrected q < 0.05). (c) A sum of the centroid values in each voxel was calculated by summation of the centroid values across all possible correlations between the voxel b and the rest of the gray matter voxels from the centroid matrix to make a spatial map of the cognition-related PET uptakes in different PET modalities. The map of the sum of the centroid values was calculated in CN at baseline. Red color of (a) indicates high PET uptakes. Black circle of (b) indicates each individual in CN at baseline and red cross marker of (b) indicates the centroid of the averaged PET- SUVRs in plot distribution. Color bar of (c) indicates the sum of the centroid values in CN at baseline. ADAS-Cog, Alzheimer's Disease Assessment Scale-Cognitive Subscale; Avr, averaged; CN, cognitively normal older adults; FBP, 18F-Florbetapir; FDR, false discovery rate; NP, neuropsychological; SUVR, standardized uptake value ratio; tau, F18-AV-1451 averaged PET-SUVRs and then made centroid maps using the CN 2.8.1 | Measurement of averaged PET-SUVR sample at baseline (Figure 1b,c). We only calculated the centroid if the averaged PET-SUVRs in each paired-voxels was significantly We calculated the averaged PET-SUVR between a PET-SUVR of a start correlated with NP score. Next, we made spatial maps of the cen- voxel and a PET-SUVR of the paired voxel, which had a significant cor- troid by summation of the centroid values across all possible con- relation in the PET correlation matrix D after performing FDR correc- nections related to each voxel in group level to visualize the effects tion, in each PET correlation matrix D (Figure 1a, left side and right of the graph theory-based single PET uptakes or multimodal PET bottom). The averaged PET-SUVR between a PET-SUVR of a start uptakes in cognitive variability in CN at baseline. The centroid map voxel b (PETb) and a PET-SUVR of the paired voxel d (PETd) that linked was used to examine whether the centroid value of the graph by ith edge (Ei) of each PET correlation matrix D in each subject k was theory-based PET uptake is reliably and constantly associated with defined as follows: the NP test scores in CN at both baseline and FU point for further validation analysis. μDk ðiÞ = AvrðPET b , PET d ÞjEi of matrix D ðFDR corrected q < 0:05Þ 2996 KIM ET AL. According to this equation, We obtained four types of the aver- centroid values across all links connected to the voxel b in the cen- aged PET-SUVRs (μ), such as Avr(FBPb, FBPd), Avr(TAUb, TAUd), Avr troid matrix DCentroid, excluding self-connection (Figure 1c). The sum (FBPb, TAUd), and Avr(TAUb, FBPd), across all CN at baseline by consid- of the centroid in all 6,848 voxels were calculated in each centroid ering each type of PET correlation matrix D. Finally, we obtained matrix DCentroid, then resampled to the corresponding GM voxels in 6,848-by-6,848 matrix, which consisted of the averaged PET-SUVRs, the MNI template. Finally, we obtained the volume-based map of the in each type of PET correlation matrix D in each individual. All aver- sum of the centroid values derived from each centroid matrix DCentroid aged PET-SUVRs were adjusted for age and sex via a general regres- in group level. We resampled each volume-based map of the sum of sion model by Statistics and Machine Learning Toolbox within the centroid values onto the group averaged cortical surface by map- MATLAB. ping a value of middle point between inner and outer surfaces in each vertex point via FreeSurfer recon-all procedure (FreeSurfer version 6.0.0; http://surfer.nmr.mgh.harvard.edu/) (Figure 1c). 2.8.2 | Partial correlation between the averaged PET-SUVRs and cognitive scores 2.9 | Longitudinal analysis The correlations between the averaged PET-SUVRs in each paired- voxels and NP scores, including MMSE, RAVALT, ADAS-Cog, We performed a similarity approach based on the centroid map to ADNI-MEM, and ADNI-EF, were performed by partial correlation examine whether the PET uptakes were reliably associated with the analysis in each type of PET correlation matrix D in CN at baseline. cognitive impairment in CN at both baseline and 2-year FU. We per- We constructed four types of 6,848-by-6,848 correlation matrices formed spatial correlations between the centroid map and the individ- DPETNP, which consisted of the correlation coefficients between ual averaged PET-SUVR images to determine the topological similarity the averaged PET-SUVRs and each NP score in group level between them, across all CN at both baseline and 2-year FU. Then, (Figure 1a, right upper matrix). All correlations between the we used these spatial similarity scores from all individuals to investi- averaged PET-SUVRs and NP scores were in each correlation gate whether significant associations exist between PET patterns and matrix DPETNP, were adjusted for age, sex, and years of education cognitive scores in CN at baseline, and in an independent longitudinal and performed FDR correction at q as 0.05 (Benjamini & Hochberg, sample. 1995). Additionally, Bonferroni correction was performed in the WD map (Bonferroni corrected adjusted p < .01) to confirm statisti- cal robustness (Bonferroni, 1936). 2.9.1 | Measurement of similarity score We measured the spatial similarity between the centroid map, which 2.8.3 | Measurement of centroid of cognition- was defined by CN at baseline only, and a local corresponding voxel- related PET uptakes based averaged PET-SUVR map of each individual across all CN at baseline and 2-year FU by using Pearson correlation analysis. First, We constructed a linear plot between the averaged PET-SUVR in we constructed a local voxel-based averaged PET-SUVR map by each paired-voxels, which is defined as a link in each correlation averaging a local voxel's PET-SUVR and the same voxel's PET-SUVR matrix DPETNP, and each NP across all CN at baseline to examine a lin- across all voxels in each individual in CN at baseline and 2-year ear trend between two variables at group level (Figure 1b). Then, we FU. For example, if the centroid map is based on different PET calculated a centroid of the averaged PET-SUVRs from the plot modalities, each local voxel-based averaged PET-SUVR was calcu- between the averaged PET-SUVRs in each paired-voxels and each NP lated by averaging between a local PET-SUVR and the corresponding score by considering the distribution of the plot in group level. The voxel's another PET-SUVR in each voxel in each individual level. If centroid was calculated by assessing cluster centroid locations in the the centroid map is based on the single PET modality, we just used a plot distribution between the averaged PET-SUVRs and NP scores via single PET-SUVR map of each individual instead of the calculation of k-means clustering analysis within MATLAB toolbox. The centroid the local voxel-based averaged PET-SUVRs. Because each voxel's value in each link was only calculated if the averaged PET-SUVRs in value of the centroid map was calculated by summing the centroid this link had significantly correlated with NP score in the correlation values across all possible associations between the one voxel and matrix DPETNP (Figure 1a,b). This centroid calculation was iterated for the rest of the GM voxels from the centroid matrix, the centroid map the number of all links of the correlation matrix DPETNP. We obtained was based on the local-to-distributed voxel's PET correlations. 6,848-by-6,848 centroid matrix DCentroid in each NP score in each type Therefore, we used the local voxel-based averaged PET-SUVR maps, of PET correlation. which were based on the local-to-local PET-SUVRs, to avoid circular In order to visualize the effects of the graph theory-based mea- dependency in the final analysis. Consequently, we obtained a simi- surements, the centroid values, in cognitive variability in CN, we cal- larity score for each individual across all CN at baseline and culated the sum of the centroid values in each voxel b by adding the 2-year FU. KIM ET AL. 2997 2.9.2 | Linear association between the similarity cortex, and part of parietal cortex in CN at 2-year FU (Figure 2b). A scores and cognitive scores slightly higher group averaged tau-SUVR was observed mainly in wide areas of the temporal cortex, MTL, medial prefrontal cortex, and PCC To better visualize whether the high similarity scores are significantly in CN at baseline (Figure 2c). The group averaged tau-SUVR was more associated with the worsening of the cognitive test performances across increased in CN at 2-year FU, partly in temporal cortex (Figure 2d). individuals, we performed linear regression analysis between the similar- We observed significant longitudinal changes between CN at ity scores and cognitive test scores. All linear regression analyses were 2-year FU and FU-matched baseline in tau-SUVR, but not obviously performed with SurfStat MATLAB toolbox (http://www.math.mcgill.ca/ noted in FBP-SUVR (Supplementary Figure 3). We observed a signifi- keith/surfstat). The group difference in slopes of the linear regression cantly increased FBP-SUVR in only a few spots of cortical areas lines between the FU-matched CN at baseline and CN at 2-year FU data (Supplementary Figure 3a), while, a significantly increased tau-SUVR were tested by analysis of variance (ANOVA) within MATLAB toolbox. was noted in wide areas of the temporal cortex, prefrontal cortex, parietal cortex, and PCC in CN at 2-year FU compared with the FU- matched baseline (Supplementary Figure 3b). 3 | RESULTS 3.1 | Group demographics and clinical 3.3 | Graph theory-based PET correlation within characteristics single PET modality and between different PET modalities All demographic characteristics and NP tests scores are described in Table 1. We found that most of our sample of CN displayed low levels We observed significant PET correlation within single PET modality, of amyloid and tau at baseline (threshold of global FBP-SUVR ≥ 1.10 such as Aβ-to-Aβ and Tau-to-Tau correlations, as well as the PET cor- and mean tau-SUVR in composite ROIs > 1.25). There was no signifi- relation between different PET modalities, such as Aβ-to-Tau and cant group difference in NP scores between 2-year FU data and FU- Tau-to-Aβ correlation, in CN at baseline (FDR corrected q < 0.05, matched baseline data. Figure 3). Spatial distributions of WD of the PET correlation within single PET modality, which was calculated by summation of all correla- tion coefficients in all possible links between each voxel and the rest 3.2 | Group averaged FBP-SUVR and tau-SUVR of voxels, were observed across almost all cortical areas, excluding motor cortex, in CN at baseline (Figure 3a,b). A voxel with the high A high group averaged FBP-SUVR was observed in CN at baseline, WD indicates that this voxel have the large number of connections to mainly in medial frontal cortex, PCC/precuneus, inferior temporal cor- the rest of voxels, that means the PET-SUVR of this voxel is strongly tex, and lateral occipital cortex (Figure 2a). Compared to baseline, rela- correlated with the PET-SUVR of the rest of the paired voxels, tively higher group averaged FBP-SUVR was observed in temporal broadly. A high WD of the PET correlation was noted mainly in both F I G U R E 2 Surface-mapped group averaged (a) FBP-SUVR and (b) tau-SUVR in CN at all baseline and group averaged (c) FBP-SUVR and (d) tau-SUVR in CN at 2-year FU after performing PVC. Red color indicates high positron emission tomography (PET)-SUVR. CN, cognitively normal older adults; FBP, 18F-Florbetapir; FU, follow-up; PVC, partial volume correction; SUVR, standardized uptake value ratio; tau, F18-AV-1451 2998 KIM ET AL. F I G U R E 3 Weighted degree maps of positron emission tomography (PET) correlations in CN at baseline. Cortical distributions of FBP-to-FBP (a), Tau-to-Tau (b), FBP-to-Tau (c), and Tau-to-FBP (d) weighted degree maps in CN at baseline while adjusting for age, sex, and FBP-SUVR or tau-SUVR when appropriate in all paired-voxels. Only multiple comparison corrected significant correlation coefficients were included in the weighted degree maps (FDR corrected q < 0.05). Color bar indicates the weighted degree value; a high score indicates that the PET-SUVR of this surfaced-mapped local voxel is strongly correlated with the rest of voxels, broadly. CN, cognitively normal older adults; FBP, 18F-Florbetapir; FDR, False discovery rate; SUVR, standardized uptake value ratio; tau, F18-AV-1451 PCC/precuneus, medial frontal cortex, parietal cortex, and broad areas the type of Aβ-to-Tau correlation after controlling for years of educa- of temporal cortex in Aβ-to-Aβ correlation in CN at baseline tion in CN at baseline (FDR corrected q < 0.05). In contrast, no signifi- (Figure 3a). Compared to Aβ-to-Aβ correlation, Tau-to-Tau correlation cant correlation between the averaged PET-SUVRs and any NP scores showed relatively low WD in the temporal cortex, precuneus, and was found in other types of PET correlations, such as Aβ-to-Aβ, Tau- frontal cortex in CN at baseline (Figure 3b). to-Tau, and Tau-to-Aβ correlations, after performing FDR correction A spatial distribution of WD of the PET correlations between dif- in CN at baseline. In this analysis, a voxel with high WD of the PET- ferent PET modalities, such as Aβ-to-Tau and Tau-to-Aβ correlations, to-NP correlation indicates that the averaged PET-SUVR from the dif- was shown in large areas of cortical regions in CN at baseline, but that ferent PET modalities between this voxel and the rest of the voxels is showed relatively lower PET correlations compared with the PET cor- strongly correlated with the NP scores, broadly. Thus, high WD relations within single PET modality (Figure 3c,d). High WD was reflects the strong effects of the averaged PET-SUVR between differ- observed mainly in the left lateral parietal cortex, left precuneus, left ent PET modalities between different voxels in cognitive scores. We orbitofrontal cortex, and both superior temporal cortices in Aβ-to-Tau found that the high averaged PET-SUVRs derived from Aβ-to-Tau correlation in CN at baseline (Figure 3c). A spatial distribution of WD connectivity was significantly correlated with high ADAS-Cog scores of Tau-to-Aβ correlation was less spread out than the other PET cor- in left superior temporal cortex, left inferior cortex, left superior parie- relations, but showed more intensive WD pattern in both inferior tal cortex, part of left PCC/precuneus, part of right superior temporal temporal cortices, both precuneus, right fusiform gyrus, and right lat- cortex, right middle parietal cortex, right precuneus, both middle fron- eral parietal cortex in CN at baseline (Figure 3d). tal cortices, and both cuneus (Figure 4a). Similar topology remains sig- nificant after performing post hoc Bonferroni correction (Figure 4b). 3.4 | Relationship between the information of the graph theory-based correlation matrix and cognitive 3.5 | Measurement of cognition-related PET variability uptakes at baseline and longitudinal FU There was a significant positive PET-to-NP correlation between aver- Of note, in this section, an individual with high similarity score indi- aged PET-SUVRs, the mean values between FBP-SUVR of a local cates that the uptake spatial pattern of the averaged Aβ and tau PET- voxel and tau-SUVR of the rest of voxels, and ADAS-Cog scores in SUVRs is well overlapped to the centroid map, which reflects the KIM ET AL. 2999 F I G U R E 4 Weighted degree maps of positron emission tomography (PET)-to-NP correlations between the averaged PET-SUVRs and ADAS- Cog scores in CN at baseline adjusted by age, sex, and years of education. (a) Only multiple comparison corrected significant correlation coefficients were included in the weighted degree map (FDR corrected q < 0.05). (b) Bonferroni correction was also performed in the weighted degree map (Bonferroni corrected adjusted p < .01) to confirm statistical robustness. Color bar indicates the weighted degree of PET-to-NP correlations; a high score indicates that the averaged PET-SUVRs between this voxel and the rest of the voxels are strongly correlated with the ADAS-Cog scores. ADAS-Cog, Alzheimer's Disease Assessment Scale-Cognitive Subscale; CN, cognitively normal older adults; FBP, 18F- Florbetapir; FDR, False discovery rate; NP, neuropsychological scores; SUVR, standardized uptake value ratio; tau, F18-AV-1451 F I G U R E 5 Linear regression associations between similarity scores and ADAS-Cog in CN at baseline and 2-year FU time point. (a) A significant positive association between the similarity score and ADAS-Cog was noted in CN at baseline (t(182) = 3.22, p = .0015). (b) Similar positive association was noted in both 2-year FU (t(16) = 2.41, p = .025), as well as FU-matched baseline (t(16) = 2.36, p = .031). ADAS-Cog, Alzheimer's Disease Assessment Scale-Cognitive Subscale; CN, cognitively normal older adults; FBP, 18F-Florbetapir; FU, follow-up; SUVR, standardized uptake value ratio; tau, F18-AV-1451 information of the spatial patterns of cognitive impairment-related FU, the slope was slightly higher in 2-year FU compared with the FU- PET uptakes. Thus, the high similarity score reflects the worsening of matched baseline (Figure 5b). the cognitive test performance. We found that increased spatial simi- larity scores were significantly associated with increased ADAS-Cog score in CN at baseline (t(182) = 3.22, p = .0015, Figure 5a). A similar 4 | DI SCU SSION linear association between the similarity score and ADAS-Cog was observed in CN at 2-year FU (t(16) = 2.41, p = .025, Figure 5b) as well AD is a neurodegenerative disorder in which the network-level accu- as in FU-matched baseline (t(16) = 2.36, p = .031, Figure 5b). There is mulation of Aβ and tau pathology seems to play a critical role in the no significant slope difference between FU-matched baseline and appearance of cognitive symptoms. Currently, we believe that close 2-year FU groups after performing ANOVA. Although there was no and interdigitated spatial interactions of Aβ and tau within neural cir- significant slope difference between FU-matched baseline and 2-year cuits might be a better indicator of early preclinical processes and 3000 KIM ET AL. cognitive decline of aging populations. Following this line of research, decline and tau accumulation in MTL in older adults; however, an in the present study, we used a novel graph-theory approach to inves- increased tau accumulation in the temporal cortex was commonly tigate brain network profiles of Aβ and tau PET imaging and cognitive observed in Braak stage of III or IV (Braak, Alafuzoff, Arzberger, scores. We found striking positive correlations between the Aβ-to- Kretzschmar, & Del Tredici, 2006; Braak & Braak, 1991b, 1995) and in Tau interactions and ADAS-Cog score, which is the most widely used tau PET image (Johnson et al., 2016). Thus, a relatively weaker tau-to- measure to assess multiple cognitive domains in AD (Mohs tau correlation in MTL may also explain the absence of the significant et al., 1997a, 1997b; Rosen, Mohs, & Davis, 1984) and was developed PET-to-NP correlation in our study. Finally, we did not observe signifi- to measure general cognition, including verbal memory, nonverbal cant PET-to-NP correlations in the type of Tau-to-Aβ. As we con- memory, planning an executive function, attention and concentration, trolled tau-SUVR effects in FBP-SUVRs of all possible paired-voxels praxis, and orientation (Mohs et al., 1997a, 1997b; Podhorna, when we measured the correlation between the averaged PET-SUVR Krahnke, Shear, Harrison, & Alzheimer's Disease Neuroimaging, 2016; and ADAS-Cog scores in types of Tau-to-Aβ correlation, one possible Skinner et al., 2012). Particularly, we observed a high correlation explanation is that tau-independent Aβ may not sufficiently contribute between Aβ-to-Tau interactions and ADAS-Cog in the temporal cor- to the initial cognitive changes in CN at baseline even though tau- tex. This is consistent with the previous findings that show not only a controlled FBP-SUVR showed a strong association with the local tau- significant association between tau and cognitive impairment in MTL SUVR in MTL. (Sepulcre et al., 2018), but also that these cognition-related tau accu- Currently, the major concern to investigate the relationship mulation in MTL are associated with an increased Aβ burden in older between Aβ and tau accumulations is the spatial inconsistencies adults (Schöll et al., 2016; Sperling et al., 2019) or mild-to-moderate between the two brain pathologies. The lack of spatial correlation or AD (Johnson et al., 2016). Other areas in the fronto-parieto-occipital overlap between a local Aβ and the corresponding local tau accumula- cortices displayed similar network-based ADAS-Cog associations. Pre- tions has been reported in imaging studies, which makes it difficult to vious literature has shown a strong relationship between local Aβ and investigate their potential synergistic effects on cognitive decline in local tau in probable AD (Iaccarino et al., 2018), and a strong associa- aging. The benefit of conducting a graph theory-based approach is tion between the inferior temporal tau accumulation and distributed that it provides a high-resolution strategy to relate these two Aβ accumulation in older adults (Sepulcre et al., 2016; Sperling unmatched anatomies. Our proposed graph theory-based approach et al., 2019). Thus, although a significant association between both measures the local-to-distributed correlations within and between Aβ pathology accumulations and cognitive decline was reported mainly in and tau PET modalities, thus solving the spatial inconsistency MTL in previous studies (Johnson et al., 2016; Schöll et al., 2016; between them. In this network-based study, we only observed signifi- Sperling et al., 2019), the cognition-related tau accumulation cant findings with ADAS-Cog scores. The absence of significance in (Ossenkoppele et al., 2019; Sepulcre et al., 2018) or Aβ accumulation the correlation analysis between the averaged PET-SUVRs and other (Schöll et al., 2016) in regions beyond MTL, such as the parietal cortex, NP scores included in the study is somewhat surprising. It is possible frontal cortex, or occipital cortex, was also reported in several studies. that ADAS-Cog is the most sensitive to test in this work due to its These selective associations of cognitive impairment or neurodegen- multi-domain nature. However, future studies are needed to examine erative changes with AD pathology accumulations depend on the AD other PET-to-NP correlations by assessing multiple cognitive mea- severity (Ossenkoppele et al., 2019) and Braak stages (Schöll surements in both cognitively normal and impaired older adults et al., 2016). Thus, we believe that high associations of Aβ-to-Tau through multiple longitudinal FUs. Moreover, we used multicenter interactions and ADAS-Cog scores in MTL-related regions, as well as ADNI data, which may have a potential bias in PET images caused by the fronto-parieto-occipital lobes, may indicate specific states of net- multiple scanner types or visit sites. We observed consistency in the work interactions, from local Aβ toward distributed tau, that potenti- cortical distribution of the mean tau-SUVRs and mean FBP-SUVRs ates neurodegenerative and cognitive changes via Aβ-triggered tau across three scanner types in CN at baseline (Supplementary (Duyckaerts, 2011; Ittner & Götz, 2011; Jacobs et al., 2018; Small & Figure 1). We performed one-way ANOVA to determine whether Duff, 2008). there are any statistically significant differences in the means of PET- Our findings show that the cross-talking of Aβ-to-Tau better SUVRs of GM between the three types of scanner, and we found no explains the initial cognitive changes of CN compared to brain net- significant difference in mean tau-SUVR or mean FBP-SUVR between work profiles derived from single PET modalities, such as Aβ-to-Aβ, or three types of scanners (Supplementary Figure 2). No meaningful Tau-to-Tau. The effects of Aβ pathology in cognitive decline generally trends of group difference between two groups in the mean PET- appeared in prodromal AD and dementia, rather than in cognitively SUVRs was noted (Supplementary Figure 2). Nevertheless, future healthy older adults (Ossenkoppele et al., 2019). Therefore, the studies should apply similar network PET-based strategies within uni- absence of the significant PET-to-NP correlation in type of Aβ-to-Aβ center data to avoid potential scanner biases. Our study has an addi- may be caused by the insufficient Aβ accumulation in CN at baseline, tional limitation: we mainly focused on a large CN sample at baseline, which seems to be a late phenomenon. Similarly, no significant PET- and a relatively small longitudinal sample from ADNI-3 study dataset. to-NP correlation in type of tau-to-tau was observed in CN at base- Therefore, larger number of FU individuals should be used and tested line. Sperling et al. (2019) showed the association between cognitive in future studies. KIM ET AL. 3001 5 | C O N CL U S I O N Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society. Series B (Methodological), 57, 289–300. We examined network interaction patterns within single PET modal- Bennett, R. E., DeVos, S. L., Dujardin, S., Corjuc, B., Gor, R., Gonzalez, J., … ity, such as Aβ-to-Aβ or Tau-to-Tau correlations, and between differ- Carlson, G. A. (2017). Enhanced tau aggregation in the presence of ent PET modalities, such as Aβ-to-Tau or Tau-to-Aβ, at high- amyloid β. The American Journal of Pathology, 187(7), 1601–1612. resolution (voxel-level) in CN, using a graph theory-based analysis. Bonferroni, C. (1936). Teoria statistica delle classi e calcolo delle probabilita. In Pubblicazioni del R Istituto Superiore di Scienze We observed that the PET uptakes derived from Aβ-to-Tau interdigi- Economiche e Commericiali di Firenze (Vol. 8, pp. 3–62). tations were significantly associated with ADAS-Cog in AD vulnerable Braak, H., Alafuzoff, I., Arzberger, T., Kretzschmar, H., & del Tredici, K. brain areas, a finding confirmed by our longitudinal investigation. (2006). Staging of Alzheimer disease-associated neurofibrillary pathol- Therefore, our work suggests the preceding contribution of network ogy using paraffin sections and immunocytochemistry. Acta Neu- ropathologica, 112(4), 389–404. interactions between Aβ and tau deposits to explain initial cognitive Braak, H., & Braak, E. (1991a). Demonstration of amyloid deposits and changes in CN prior to the conversion of dementia. neurofibrillary changes in whole brain sections. Brain Pathology, 1(3), 213–216. ACKNOWLEDGMENT Braak, H., & Braak, E. (1991b). Neuropathological stageing of Alzheimer- related changes. Acta Neuropathologica, 82(4), 239–259. This research was supported by grants from the National Institutes of Braak, H., & Braak, E. (1995). Staging of Alzheimer's disease-related neuro- Health (NIH) (R01AG061811, and R01AG061445 to J. S.). fibrillary changes. Neurobiology of Aging, 16(3), 271–278. Braak, H., & del Tredici, K. (2011). Alzheimer's pathogenesis: Is there CONF LICT OF IN TE RE ST neuron-to-neuron propagation? Acta Neuropathologica, 121(5), The authors declare no potential conflict of interest. 589–595. Brier, M. R., Gordon, B., Friedrichsen, K., McCarthy, J., Stern, A., Christensen, J., … Hassenstab, J. (2016). Tau and Aβ imaging, CSF mea- AUTHOR CONTRIBUTIONS sures, and cognition in Alzheimer's disease. Science Translational Medi- Chan-Mi Kim: Study concept and design, analysis of the data, inter- cine, 8(338), 338ra366–338ra366. pretation of the data, drafting of manuscript, revision of the manu- Chien, D. T., Bahri, S., Szardenings, A. K., Walsh, J. C., Mu, F., Su, M.-Y., … Kolb, H. C. (2013). Early clinical PET imaging results with the novel script. Victor Montal: Data quality check and analysis of the data. Ibai PHF-tau radioligand [F-18]-T807. Journal of Alzheimer's Disease, 34(2), Diez: Analysis of the data. William Orwig: Assistant of the revision of 457–468. the manuscript. Jorge Sepulcre: Study concept and design and revi- Clark, C. M., Pontecorvo, M. J., Beach, T. G., Bedell, B. J., Coleman, R. E., sion of the manuscript. Doraiswamy, P. M., … Sadowsky, C. H. (2012). Cerebral PET with florbetapir compared with neuropathology at autopsy for detection of neuritic amyloid-β plaques: A prospective cohort study. The Lancet DATA AVAI LAB ILITY S TATEMENT Neurology, 11(8), 669–678. Data used in preparation of this article were obtained from the Clark, C. M., Schneider, J. A., Bedell, B. J., Beach, T. G., Bilker, W. B., Alzheimer's Disease Neuroimaging Initiative (ADNI) database (adni. Mintun, M. A., … Flitter, M. L. (2011). Use of florbetapir-PET for imag- ing β-amyloid pathology. JAMA, 305(3), 275–283. loni.usc.edu). As such, the investigators within the ADNI contributed Clavaguera, F., Bolmont, T., Crowther, R. A., Abramowski, D., Frank, S., to the design and implementation of ADNI and/or provided data but Probst, A., … Staufenbiel, M. (2009). Transmission and spreading of did not participate in analysis or writing of this report. A complete tauopathy in transgenic mouse brain. Nature Cell Biology, 11(7), 909–913. listing of ADNI investigators can be found at: http://adni.loni.usc. Crane, P. K., Carle, A., Gibbons, L. E., Insel, P., Mackin, R. S., Gross, A., … Alzheimer's Disease Neuroimaging Initiative. (2012). Development and edu/wp-content/uploads/how_to_apply/ADNI_Acknowledgement_ assessment of a composite score for memory in the Alzheimer's Dis- List.pdf. ease Neuroimaging Initiative (ADNI). Brain Imaging and Behavior, 6(4), 502–516. https://doi.org/10.1007/s11682-012-9186-z ORCID Dale, A. M., Fischl, B., & Sereno, M. I. (1999). Cortical surface-based analy- Chan-Mi Kim https://orcid.org/0000-0001-5686-1668 sis. I. Segmentation and surface reconstruction. NeuroImage, 9(2), 179–194. https://doi.org/10.1006/nimg.1998.0395 de Calignon, A., Polydoro, M., Suárez-Calvet, M., William, C., RE FE R ENC E S Adamowicz, D. H., Kopeikina, K. J., … Carlson, G. A. (2012). Propaga- Arnold, S. E., Hyman, B. T., Flory, J., Damasio, A. R., & van Hoesen, G. W. tion of tau pathology in a model of early Alzheimer's disease. Neuron, (1991). The topographical and neuroanatomical distribution of neurofi- 73(4), 685–697. brillary tangles and neuritic plaques in the cerebral cortex of patients Desikan, R. S., Segonne, F., Fischl, B., Quinn, B. T., Dickerson, B. C., with Alzheimer's disease. Cerebral Cortex, 1(1), 103–116. Blacker, D., … Killiany, R. J. (2006). An automated labeling system for Ballard, C., Gauthier, S., Corbett, A., Brayne, C., Aarsland, D., & Jones, E. subdividing the human cerebral cortex on MRI scans into gyral based (2011). Alzheimer's disease. Lancet (London, England), 377(9770), regions of interest. NeuroImage, 31(3), 968–980. https://doi.org/10. 1019–1031. 1016/j.neuroimage.2006.01.021 Barthel, H., Gertz, H.-J., Dresel, S., Peters, O., Bartenstein, P., Buerger, K., Duyckaerts, C. (2011). Tau pathology in children and young adults: Can … Reininger, C. (2011). Cerebral amyloid-β PET with florbetaben (18F) you still be unconditionally Baptist?. Acta Neuropathologica. 121(2), in patients with Alzheimer's disease and healthy controls: A multi- 145–147. centre phase 2 diagnostic study. The Lancet Neurology, 10(5), Fagan, A. M., Roe, C. M., Xiong, C., Mintun, M. A., Morris, J. C., & 424–435. Holtzman, D. M. (2007). Cerebrospinal fluid tau/β-amyloid42 ratio as a 3002 KIM ET AL. prediction of cognitive decline in nondemented older adults. Archives between amyloid, tau and neurodegeneration in Alzheimer's Disease. of Neurology, 64(3), 343–349. NeuroImage: Clinical, 17, 452–464. Fischl, B., Liu, A., & Dale, A. M. (2001). Automated manifold surgery: Con- Iba, M., Guo, J. L., McBride, J. D., Zhang, B., Trojanowski, J. Q., & structing geometrically accurate and topologically correct models of Lee, V. M.-Y. (2013). Synthetic tau fibrils mediate transmission of neu- the human cerebral cortex. IEEE Transactions on Medical Imaging, 20(1), rofibrillary tangles in a transgenic mouse model of Alzheimer's-like 70–80. https://doi.org/10.1109/42.906426 tauopathy. Journal of Neuroscience, 33(3), 1024–1037. Fischl, B., Salat, D. H., Busa, E., Albert, M., Dieterich, M., Haselgrove, C., … Iqbal, K., & Grundke-Iqbal, I. (2002). Neurofibrillary pathology leads to syn- Dale, A. M. (2002). Whole brain segmentation: Automated labeling of aptic loss and not the other way around in Alzheimer disease. Journal neuroanatomical structures in the human brain. Neuron, 33(3), of Alzheimer's Disease, 4(3), 235–238. 341–355. Ittner, L. M., & Götz, J. (2011). Amyloid-β and tau—A toxic pas de deux in Fischl, B., Salat, D. H., van der Kouwe, A. J., Makris, N., Ségonne, F., Alzheimer's disease. Nature Reviews Neuroscience, 12(2), 67–72. Quinn, B. T., & Dale, A. M. (2004). Sequence-independent segmenta- Jack, C. R., Jr., Knopman, D. S., Jagust, W. J., Shaw, L. M., Aisen, P. S., tion of magnetic resonance images. NeuroImage, 23, S69–S84. Weiner, M. W., … Trojanowski, J. Q. (2010). Hypothetical model of Fischl, B., Sereno, M. I., Tootell, R. B., & Dale, A. M. (1999). High-resolution dynamic biomarkers of the Alzheimer's pathological cascade. The Lan- intersubject averaging and a coordinate system for the cortical surface. cet Neurology, 9(1), 119–128. Human Brain Mapping, 8(4), 272–284. Jacobs, H. I., Hedden, T., Schultz, A. P., Sepulcre, J., Perea, R. D., Fischl, B., van der Kouwe, A., Destrieux, C., Halgren, E., Segonne, F., Amariglio, R. E., … Johnson, K. A. (2018). Structural tract alterations Salat, D. H., … Dale, A. M. (2004). Automatically parcellating the predict downstream tau accumulation in amyloid-positive older indi- human cerebral cortex. Cerebral Cortex, 14(1), 11–22. viduals. Nature Neuroscience, 21(3), 424–431. Fleisher, A. S., Chen, K., Quiroz, Y. T., Jakimovich, L. J., Gomez, M. G., Jansen, W. J., Ossenkoppele, R., Knol, D. L., Tijms, B. M., Scheltens, P., Langois, C. M., … Lee, W. (2015). Associations between biomarkers Verhey, F. R., … Alcolea, D. (2015). Prevalence of cerebral amyloid and age in the presenilin 1 E280A autosomal dominant Alzheimer dis- pathology in persons without dementia: A meta-analysis. JAMA, 313 ease kindred: A cross-sectional study. JAMA Neurology, 72(3), (19), 1924–1938. 316–324. Johnson, K. A., Minoshima, S., Bohnen, N. I., Donohoe, K. J., Foster, N. L., Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). "Mini-mental state". Herscovitch, P., … Pappas, V. (2013). Update on appropriate use A practical method for grading the cognitive state of patients for the criteria for amyloid PET imaging: Dementia experts, mild cognitive clinician. Journal of Psychiatric Research, 12(3), 189–198. https://doi. impairment, and education. Journal of Nuclear Medicine, 54(7), org/10.1016/0022-3956(75)90026-6 1011–1013. Ghoshal, N., Garcı ́a-Sierra, F., Wuu, J., Leurgans, S., Bennett, D. A., Johnson, K. A., Schultz, A., Betensky, R. A., Becker, J. A., Sepulcre, J., Berry, R. W., & Binder, L. I. (2002). Tau conformational changes corre- Rentz, D., … Papp, K. (2016). Tau positron emission tomographic imag- spond to impairments of episodic memory in mild cognitive impair- ing in aging and early Alzheimer's disease. Annals of Neurology, 79(1), ment and Alzheimer's disease. Experimental Neurology, 177(2), 110–119. 475–493. Joshi, A. D., Pontecorvo, M. J., Clark, C. M., Carpenter, A. P., Gibbons, L. E., Carle, A. C., Mackin, R. S., Harvey, D., Mukherjee, S., Jennings, D. L., Sadowsky, C. H., … Arora, A. (2012). Performance char- Insel, P., … Alzheimer's Disease Neuroimaging Initiative. (2012). A acteristics of amyloid PET with florbetapir F 18 in patients with composite score for executive functioning, validated in Alzheimer's Alzheimer's disease and cognitively normal subjects. Journal of Nuclear Disease Neuroimaging Initiative (ADNI) participants with baseline mild Medicine, 53(3), 378–384. cognitive impairment. Brain Imaging and Behavior, 6(4), 517–527. Kim, H.-R., Lee, P., Seo, S. W., Roh, J. H., Oh, M., Oh, J. S., … Jeong, Y. https://doi.org/10.1007/s11682-012-9176-1 (2019). Comparison of amyloid β and Tau spread models in Alzheimer's Götz, J., Chen, F. V., van Dorpe, J., & Nitsch, R. (2001). Formation of neu- Disease. Cerebral Cortex, 29(10), 4291–4302. rofibrillary tangles in P301L tau transgenic mice induced by Aβ42 Klunk, W. E., Engler, H., Nordberg, A., Wang, Y., Blomqvist, G., Holt, D. P., fibrils. Science, 293(5534), 1491–1495. … Estrada, S. (2004). Imaging brain amyloid in Alzheimer's disease with Greve, D. N., Salat, D. H., Bowen, S. L., Izquierdo-Garcia, D., Schultz, A. P., Pittsburgh compound-B. Annals of Neurology, 55(3), 306–319. Catana, C., … Johnson, K. A. (2016). Different partial volume correc- Landau, S. M., Mintun, M. A., Joshi, A. D., Koeppe, R. A., Petersen, R. C., tion methods lead to different conclusions: An (18)F-FDG-PET study Aisen, P. S., … Alzheimer's Disease Neuroimaging Initiative. (2012). of aging. NeuroImage, 132, 334–343. https://doi.org/10.1016/j. Amyloid deposition, hypometabolism, and longitudinal cognitive neuroimage.2016.02.042 decline. Annals of Neurology, 72(4), 578–586. Guo, J. L., & Lee, V. M. (2014). Cell-to-cell transmission of pathogenic pro- Lewis, J., Dickson, D. W., Lin, W.-L., Chisholm, L., Corral, A., Jones, G., … teins in neurodegenerative diseases. Nature Medicine, 20(2), 130–138. Yager, D. (2001). Enhanced neurofibrillary degeneration in trans- https://doi.org/10.1038/nm.3457 genic mice expressing mutant tau and APP. Science, 293(5534), Hanseeuw, B. J., Betensky, R. A., Jacobs, H. I., Schultz, A. P., Sepulcre, J., 1487–1491. Becker, J. A., … Mormino, E. C. (2019). Association of amyloid and tau Marquié, M., Normandin, M. D., Vanderburg, C. R., Costantino, I. M., with cognition in preclinical Alzheimer disease: A longitudinal study. Bien, E. A., Rycyna, L. G., … Debnath, M. L. (2015). Validating novel tau JAMA Neurology, 76(8), 915–924. positron emission tomography tracer [F-18]-AV-1451 (T807) on post- He, Z., Guo, J. L., McBride, J. D., Narasimhan, S., Kim, H., Changolkar, L., … mortem brain tissue. Annals of Neurology, 78(5), 787–800. Dengler, C. (2018). Amyloid-β plaques enhance Alzheimer's brain tau- Meltzer, C. C., Zubieta, J. K., Links, J. M., Brakeman, P., Stumpf, M. J., & seeded pathologies by facilitating neuritic plaque tau aggregation. Frost, J. J. (1996). MR-based correction of brain PET measurements Nature Medicine, 24(1), 29–38. for heterogeneous gray matter radioactivity distribution. Journal of Hurtado, D. E., Molina-Porcel, L., Iba, M., Aboagye, A. K., Paul, S. M., Cerebral Blood Flow and Metabolism, 16(4), 650–658. https://doi.org/ Trojanowski, J. Q., & Lee, V. M.-Y. (2010). Aβ accelerates the spatio- 10.1097/00004647-199607000-00016 temporal progression of tau pathology and augments tau amyloidosis Mishra, S., Gordon, B. A., Su, Y., Christensen, J., Friedrichsen, K., in an Alzheimer mouse model. The American Journal of Pathology, 177 Jackson, K., … Benzinger, T. L. S. (2017). AV-1451 PET imaging of tau (4), 1977–1988. pathology in preclinical Alzheimer disease: Defining a summary mea- Iaccarino, L., Tammewar, G., Ayakta, N., Baker, S. L., Bejanin, A., sure. NeuroImage, 161, 171–178. https://doi.org/10.1016/j. Boxer, A. L., … Lazaris, A. (2018). Local and distant relationships neuroimage.2017.07.050 KIM ET AL. 3003 Mitchell, T. W., Mufson, E. J., Schneider, J. A., Cochran, E. J., Nissanov, J., Ségonne, F., Dale, A. M., Busa, E., Glessner, M., Salat, D., Hahn, H. K., & Han, L. Y., … Bennett, D. A. (2002). Parahippocampal tau pathology in Fischl, B. (2004). A hybrid approach to the skull stripping problem in healthy aging, mild cognitive impairment, and early Alzheimer's dis- MRI. NeuroImage, 22(3), 1060–1075. ease. Annals of Neurology, 51(2), 182–189. Selkoe, D. J. (2002). Alzheimer's disease is a synaptic failure. Science, 298 Mohs, R. C., Knopman, D., Petersen, R. C., Ferris, S. H., Ernesto, C., (5594), 789–791. Grundman, M., … Clark, C. (1997a). Development of cognitive instru- Sepulcre, J., Grothe, M. J., d'Oleire Uquillas, F., Ortiz-Teran, L., Diez, I., ments for use in clinical trials of antidementia drugs: Additions to the Yang, H. S., … Johnson, K. A. (2018). Neurogenetic contributions to Alzheimer's Disease Assessment Scale that broaden its scope. amyloid beta and tau spreading in the human cortex. Nature Medicine, Alzheimer Disease and Associated Disorders, 11, 13–21. 24(12), 1910–1918. https://doi.org/10.1038/s41591-018-0206-4 Mohs, R. C., Knopman, D., Petersen, R. C., Ferris, S. H., Ernesto, C., Sepulcre, J., Schultz, A. P., Sabuncu, M., Gomez-Isla, T., Chhatwal, J., Grundman, M., … Thal, L. J. (1997b). Development of cognitive instru- Becker, A., … Johnson, K. A. (2016). In vivo tau, amyloid, and gray mat- ments for use in clinical trials of antidementia drugs: Additions to the ter profiles in the aging brain. Journal of Neuroscience, 36(28), Alzheimer's Disease Assessment Scale that broaden its scope. The 7364–7374. Alzheimer's Disease Cooperative Study. Alzheimer Disease and Associ- Skinner, J., Carvalho, J. O., Potter, G. G., Thames, A., Zelinski, E., ated Disorders, 11(Suppl 2), S13–S21. Crane, P. K., … Alzheimer's Disease Neuroimaging Initiative. (2012). Muller-Gartner, H. W., Links, J. M., Prince, J. L., Bryan, R. N., McVeigh, E., The Alzheimer's Disease Assessment Scale-Cognitive-Plus (ADAS- Leal, J. P., … Frost, J. J. (1992). Measurement of radiotracer concentra- Cog-Plus): An expansion of the ADAS-Cog to improve responsiveness tion in brain gray matter using positron emission tomography: MRI- in MCI. Brain Imaging and Behavior, 6(4), 489–501. https://doi.org/10. based correction for partial volume effects. Journal of Cerebral Blood 1007/s11682-012-9166-3 Flow and Metabolism, 12(4), 571–583. https://doi.org/10.1038/jcbfm. Small, S. A., & Duff, K. (2008). Linking Aβ and tau in late-onset Alzheimer's 1992.81 disease: A dual pathway hypothesis. Neuron, 60(4), 534–542. Nelson, P. T., Alafuzoff, I., Bigio, E. H., Bouras, C., Braak, H., Cairns, N. J., … Sperling, R. A., Mormino, E. C., Schultz, A. P., Betensky, R. A., Papp, K. V., Tredici, K. D. (2012). Correlation of Alzheimer disease neuropathologic Amariglio, R. E., … Hedden, T. (2019). The impact of amyloid-beta and changes with cognitive status: A review of the literature. Journal of tau on prospective cognitive decline in older individuals. Annals of Neu- Neuropathology & Experimental Neurology, 71(5), 362–381. rology, 85(2), 181–193. Ossenkoppele, R., Smith, R., Ohlsson, T., Strandberg, O., Mattsson, N., Stöhr, J., Watts, J. C., Mensinger, Z. L., Oehler, A., Grillo, S. K., Insel, P. S., … Hansson, O. (2019). Associations between tau, Aβ, and DeArmond, S. J., … Giles, K. (2012). Purified and synthetic Alzheimer's cortical thickness with cognition in Alzheimer disease. Neurology, 92 amyloid beta (Aβ) prions. Proceedings of the National Academy of Sci- (6), e601–e612. ences of the United States of America, 109(27), 11025–11030. Pike, K. E., Savage, G., Villemagne, V. L., Ng, S., Moss, S. A., Maruff, P., … Thal, D. R., Rub, U., Orantes, M., & Braak, H. (2002). Phases of A beta- Rowe, C. C. (2007). β-Amyloid imaging and memory in non-demented deposition in the human brain and its relevance for the development individuals: Evidence for preclinical Alzheimer's disease. Brain, 130(11), of AD. Neurology, 58(12), 1791–1800. https://doi.org/10.1212/wnl. 2837–2844. 58.12.1791 Podhorna, J., Krahnke, T., Shear, M., Harrison, J. E., & Alzheimer's Disease van Rossum, I. A., Visser, P. J., Knol, D. L., van der Flier, W. M., Neuroimaging Initiative. (2016). Alzheimer's Disease Assessment Teunissen, C. E., Barkhof, F., … Scheltens, P. (2012). Injury markers but Scale-cognitive subscale variants in mild cognitive impairment and mild not amyloid markers are associated with rapid progression from mild Alzheimer's disease: Change over time and the effect of enrichment cognitive impairment to dementia in Alzheimer's disease. Journal of strategies. Alzheimer's Research & Therapy, 8, 8. https://doi.org/10. Alzheimer's Disease, 29(2), 319–327. 1186/s13195-016-0170-5 Vandenberghe, R., van Laere, K., Ivanoiu, A., Salmon, E., Bastin, C., Pontecorvo, M. J., Devous, M. D., Sr., Navitsky, M., Lu, M., Salloway, S., Triau, E., … Korner, A. (2010). 18F-flutemetamol amyloid imaging in Schaerf, F. W., … Lim, N. C. (2017). Relationships between flortaucipir Alzheimer disease and mild cognitive impairment: A phase 2 trial. PET tau binding and amyloid burden, clinical diagnosis, age and cogni- Annals of Neurology, 68(3), 319–329. tion. Brain, 140(3), 748–763. Villemagne, V. L., Burnham, S., Bourgeat, P., Brown, B., Ellis, K. A., Quiroz, Y. T., Sperling, R. A., Norton, D. J., Baena, A., Arboleda- Salvado, O., … Maruff, P. (2013). Amyloid β deposition, neu- Velasquez, J. F., Cosio, D., … Miller, J. B. (2018). Association between rodegeneration, and cognitive decline in sporadic Alzheimer's dis- amyloid and tau accumulation in young adults with autosomal domi- ease: A prospective cohort study. The Lancet Neurology, 12(4), nant Alzheimer disease. JAMA Neurology, 75(5), 548–556. 357–367. Resnick, S. M., Sojkova, J., Zhou, Y., An, Y., Ye, W., Holt, D. P., … Walker, L. C., & Jucker, M. (2011). Amyloid by default. Nature Neurosci- Wong, D. F. (2010). Longitudinal cognitive decline is associated with ence, 14(6), 669–670. fibrillar amyloid-beta measured by [11C]PiB. Neurology, 74(10), Wang, L., Benzinger, T. L., Su, Y., Christensen, J., Friedrichsen, K., Aldea, P., 807–815. https://doi.org/10.1212/WNL.0b013e3181d3e3e9 … Morris, J. C. (2016). Evaluation of tau imaging in staging Alzheimer Rey, A. (1958). L'examen clinique en psychologie. [The clinical examination disease and revealing interactions between β-amyloid and tauopathy. in psychology]. Presses Universitaries De France. JAMA Neurology, 73(9), 1070–1077. Rolstad, S., Berg, A. I., Bjerke, M., Johansson, B., Zetterberg, H., & Weiner, M. W., Veitch, D. P., Aisen, P. S., Beckett, L. A., Cairns, N. J., Wallin, A. (2013). Cerebrospinal fluid biomarkers mirror rate of cogni- Green, R. C., … Alzheimer's Disease Neuroimaging Initiative. (2017). tive decline. Journal of Alzheimer's Disease, 34(4), 949–956. The Alzheimer's Disease Neuroimaging Initiative 3: Continued innova- Rosen, W. G., Mohs, R. C., & Davis, K. L. (1984). A new rating scale for tion for clinical trial improvement. Alzheimer's & Dementia, 13(5), Alzheimer's disease. The American Journal of Psychiatry, 141(11), 561–571. https://doi.org/10.1016/j.jalz.2016.10.006 1356–1364. https://doi.org/10.1176/ajp.141.11.1356 Whitwell, J. L., Graff-Radford, J., Tosakulwong, N., Weigand, S. D., Rousset, O. G., Ma, Y., & Evans, A. C. (1998). Correction for partial volume Machulda, M. M., Senjem, M. L., … Drubach, D. A. (2018). Imaging cor- effects in PET: Principle and validation. Journal of Nuclear Medicine, 39 relations of tau, amyloid, metabolism, and atrophy in typical and atypi- (5), 904–911. cal Alzheimer's disease. Alzheimer's & Dementia, 14(8), 1005–1014. Schöll, M., Lockhart, S. N., Schonhaut, D. R., O'Neil, J. P., Janabi, M., Williamson, R., Usardi, A., Hanger, D. P., & Anderton, B. H. (2008). Mem- Ossenkoppele, R., … Schwimmer, H. D. (2016). PET imaging of tau brane-bound β-amyloid oligomers are recruited into lipid rafts by a deposition in the aging human brain. Neuron, 89(5), 971–982. fyn-dependent mechanism. The FASEB Journal, 22(5), 1552–1559. 3004 KIM ET AL. Wong, D. F., Rosenberg, P. B., Zhou, Y., Kumar, A., Raymont, V., Ravert, H. T., … Ye, W. (2010). In vivo imaging of amyloid deposition in How to cite this article: Kim C-M, Montal V, Diez I, Orwig W, Alzheimer disease using the radioligand 18F-AV-45 (flobetapir F 18). Alzheimer's Disease Neuroimaging Initiative (ADNI), Journal of Nuclear Medicine, 51(6), 913–920. Sepulcre J. Network interdigitations of Tau and amyloid-beta SUPPORTING INFORMATION deposits define cognitive levels in aging. Hum Brain Mapp. Additional supporting information may be found online in the 2021;42:2990–3004. https://doi.org/10.1002/hbm.25350 Supporting Information section at the end of this article.