Alzheimer's Disease Neuroimaging Initiative Research Articles

PARTIAL VOLUME CORRECTION TECHNIQUES FOR LONGITUDINAL TAU‐PET QUANTIFICATION: A COMPARISON STUDY

AbstractBackgroundThe selection of an appropriate Partial Volume Correction (PVC) is crucial for evaluating tau accumulation rates using 18F‐Flortaucipir (18F‐FTP). PVC can reduce the impact of off‐target binding, but it can also increase noise and variability. While these effects may be averaged on cross‐sectional studies, they might be of paramount importance when evaluating within‐subject longitudinal tau‐PET signal. The aim of this study is to evaluate the robustness of different PVCs for the evaluation of tau accumulation rates in longitudinal 18F‐FTP studies, as measured by their correlation with cognitive clinical progression.Method111 cognitively impaired subjects from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) with serial 18F‐FTP PET and T1 images, as well as cognitive performance evaluation available for a 2‐year follow‐up period, were included in the study. 18F‐FTP PET images were corrected by applying a traditional deconvolution PVC (Grey matter only), the reblurred Van‐Cittert (RVC) analytical PVC (grey + white matter), and 2 voxel‐based PVCs, the region‐based voxel‐wise correction (RBV) and the iterative Yang (iY). Longitudinal rates of change for the regional standard uptake value ratios (SUVR) of the temporal meta‐ROI were computed for the images corrected for the different PVCs, and their correlation with the rates of change of the ADAS‐Cog 11 cognitive performance metric was analyzed.ResultThe deconvolution PVC correction showed the strongest correlation between decreased cognitive performance and increased tau accumulation rates (p<0.001, ρ = 0.38), followed closely by the RBV (p<0.001, ρ = 0.36) and iY (p<0.001, ρ = 0.35) corrections (Fig. 1). In contrast, the RVC results (p = 0.03, ρ = 0.20) were comparable to not applying any PVC at all (p = 0.03, ρ = 0.21).ConclusionOur findings suggest that longitudinal FTP studies might benefit from the use of PVC techniques for the evaluation of tau accumulation rates. Tau uptake rates of change tracked the observed clinical progression more accurately after certain PVC corrections were applied. Moreover, a simple deconvolution PVC has been shown to perform similarly to more complex, voxel‐wise corrections such as the RBV and iY.

Discerning neurodegenerative and aging genetic influences on hippocampal subfields trajectories in the Alzheimer’s disease continuum

AbstractBackgroundHippocampal atrophy is one of the most characteristic biomarkers of Alzheimer’s disease (AD), and its genetic basis, as well as that of its distinct subfields, is of current interest. However, hippocampal subfields have so far only been assessed largely unrelated to brain atrophy, and the majority of studies were focused on general population samples, explicitly excluding diagnosed patients. In this study, we evaluated whether hippocampal subfields trajectories are a suitable endophenotype to discern between neurodegenerative and aging genetic influences in the Alzheimer’s continuum.MethodThe dataset includes AD patients (N = 113), mild cognitive impairment (N = 426), and control normal (CN) (N = 272) subjects from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset. Individuals were evaluated with 3T MRI at baseline and every 6 months for a total of 96 months. Hippocampal subfields were extracted using longitudinal processing within FreeSurfer(6.0). We used longitudinal mixed‐effect models with random intercept for individuals to investigated the association between hippocampal subfields volumetric changes and genetic predisposition to AD, Parkinson’s disease, frontotemporal dementia (neurodegeneration), as well as intrinsic and extrinsic epigenetic clock acceleration and telomere length (biological aging). Models were adjusted by age, sex, AD‐stage and total hippocampal volume. Disease status‐specific trajectories were also assessed.ResultGenetic predisposition to biological aging (driven by genetic predisposition to telomere length) was significantly associated with subfield changes in presubiculum (β = ‐11.2, p = 4.1e‐04), parasubiculum (β = ‐4.4, padj = 4.4e‐04), molecular layer (β = 5.9, padj = 2.0e‐04), dentate gyrus (DG) (β = 5.79, padj = 4.4e‐04), CA4 (β = 7.2, padj = 4.4e‐05), and HATA (β = ‐1.7, padj = 0.027). Genetic predisposition to neurodegeneration (driven by predisposition to AD) was associated with volumetric changes in CA3 (β = 4.4, p = 0.047). In stratified models, we observed significant volumetric changes due to genetic predisposition to neurodegeneration in controls (CA1, CA4, DG, parasubiculum, molecular layer). In MCI (subiculum regions and CA4) and AD (hippocampal fissure, molecular layer and DG), volumetric changes were mostly associated with genetic predisposition to biological aging processes.ConclusionWe showed differential genetic mechanisms associated with hippocampal subfield trajectories that were specific to Alzheimer’s disease stage. These results may help elucidate new biological mechanisms in the course of the disease as well as discern additional molecular mechanisms related to neurodegeneration and aging.

Association of creatine kinase with cognition, neuroimaging, and Alzheimer’s pathology

AbstractBackgroundCreatine kinase (CK) and its related metabolites are known to imply cardiovascular system disease, whereas the cardiovascular metabolic state is associated with the state of cognitive well‐being of the brain. However, the extent to which CK levels is related with cognitive impairment has not been revealed.MethodThe Alzheimer’s Disease Neuroimaging Initiative (ADNI) database were used for this study. We first investigated the intergroup difference of participants demographic, cognitive status, Alzheimer’s disease (AD) diagnosis, and cerebrospinal fluid (CSF) AD pathology biomarkers. Next, we conducted to use liner regression model to examine the association of CK with CSF AD pathology biomarkers, cognition, and brain structure. Last, mediation analyses were utilized to explore whether the association between CK and cognition were mediated by CSF AD pathology biomarkers.ResultA total of 1,607 participants (mean age were 73.82±7.23 and 55.76% were male) from the ADNI database were involved in this cross‐sectional study. We found a significant difference through age, gender, AD diagnosis, and cognitive status subgroups in intergroup difference analysis (P<0.05). Meanwhile, significant association of lower level of CK with decreased level of β‐amyloid1‐42 (Aβ; β = 0.069, P = 0.040) and increased level of phosphorylated tau (P‐tau; β = ‐0.084, P = 0.007), and total‐tau (T‐tau; β = ‐0.071, P = 0.021), respectively. Liner regression analysis also indicated that the lower level of CK was associated with worse cognition (Mini‐Mental State Examination: β = 0.059, P = 0.021; Alzheimer’s Disease Assessment Scale: β = ‐0.065, P = 0.012), however not to brain structure. Moreover, mediation analysis identified CSF AD pathology biomarkers mediated the association between level of CK and cognition.ConclusionOur findings offered detailed proof to advise that the lower level of CK was significantly associated with poor cognition and it might be mediated by CSF AD pathology biomarkers. For future research, it is necessary to identify the precise reason and process by which changed levels of CK impact cognition.

The limbic‐predominant hypometabolic subtype of AD is not enriched for comorbid TDP‐43 pathology or hippocampal sclerosis

AbstractBackgroundPrevious research has identified distinct subtypes of AD that differed with respect to the distribution of neurofibrillary tangles (NFT). One of them, a limbic‐predominant subtype, was enriched for comorbid TDP‐43 pathology. Our previous data‐driven analysis used FDG‐PET to identify a limbic‐predominant subtype of AD that may correspond to the limbic‐predominant NFT subtype. In the current study, we tested the expected associations between measures of neuropathology such as TDP‐43 pathology, hippocampal sclerosis, and Braak stages of NFT accumulation, and the limbic‐predominant hypometabolic subtype.MethodWe analyzed ante‐mortem FDG‐PET and neuropathological data of 74 individuals from the autopsy cohort of the Alzheimer’s Disease Neuroimaging Initiative (ADNI), including 7 cognitively normal participants, 12 participants with MCI, and 55 with dementia (as assessed at the last available clinical evaluation). Severity of TDP‐43 pathology was operationalized as the number of brain areas with TDP‐43 immunoreactive inclusions (including amygdala, hippocampus, entorhinal/inferior temporal cortex and neocortex). Hippocampal sclerosis was operationalized as a binary variable. We measured similarity of individual FDG‐PET profiles to the limbic‐predominant subtype. For that, we used Euclidean distances from each individual profile to the mean profile of the limbic‐predominant AD subtype identified previously in a larger ADNI sample using a data‐driven hierarchical clustering approach (Fig.1). Lower Euclidean distances indicated higher similarity to the subtype. Bayesian linear regressions in JASP were used to assess associations between neuropathological markers and similarity to the limbic‐predominant FDG‐PET pattern, controlling for Thal phase of amyloid accumulation, age, sex, and time delay between the FDG‐PET scan and death in the null model. We assessed Bayes factors representing evidence for a given model in comparison to the null model (BF10).ResultOur results suggest that similarity of individual FDG‐PET profiles to the limbic‐predominant FDG‐PET pattern was not related to severity of TDP‐43 pathology (BF10 = 0.449), presence of hippocampal sclerosis (BF10 = 0.544), or Braak NFT stages (BF10 = 0.525).ConclusionContrary to the expected effects, our current analysis indicates a lack of reliable associations between neuropathological markers such as TDP‐43 pathology, hippocampal sclerosis, or global NFT accumulation and the limbic‐predominant hypometabolic subtype. Future analyses will assess potential alternative contributors, such as region‐specific patterns of NFT distribution.

Sexual dimorphisms in longitudinal associations of the blood lipidome with cognitive decline in Alzheimer’s disease

AbstractBackgroundLate‐onset Alzheimer’s disease (AD) is the leading cause of dementia with neither cure nor a clearly understood disease trajectory. Previous studies have shown cross‐sectional and longitudinal associations of the blood lipidome and AD. We previously reported on significant sex differences in the associations of lipids with AD biomarkers and it has been suggested that such sex differences may result in distinct trajectories of AD progression. We performed an association study of sex‐stratified longitudinal lipidomics data with cognitive function to further characterize sexual dimorphisms in lipidomic manifestations of AD.MethodWe used a subset of the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset: 1349 individuals (745 males and 604 females) with blood samples and clinical data collected at 10 time points with intervals of 6 to12 months (4145 samples total). We analyzed 749 lipid species measured by quantitative LC‐MS lipidomics profiling (Baker Heart and Diabetes Institute, Melbourne, Australia). We calculated linear mixed models (LMMs) and generalized additive mixed models (GAMMs) to analyze time‐resolved associations between lipids and two composite cognitive scores (ADNI memory and executive function) while adjusting for age, sex (pooled analysis only), BMI, HDL‐C, total cholesterol, copies of APOE ε4, years of education as well as significant medication effects.ResultWe observed a significant difference in lipid concentrations between males and females in line with previous studies, however, these differences were not fully consistent across diagnostic groups. In longitudinal association analyses, we found a substantial fraction of lipid species to have significant sex‐specific associations with cognitive function, which were not observed in the pooled analysis. These lipids were distributed across several lipid classes with enrichment in phosphatidylcholines (PC), phosphatidylethanolamines (PE) and triglycerides (TG). Comparisons between LMMs and GAMMs further indicated that some lipids might have a nonlinear relationship with cognitive trajectories.ConclusionOur results suggest that there are substantial sex differences in longitudinal associations of the blood lipidome with cognitive decline resulting in complex and partially nonlinear relationships. The highlighted lipid pathways can serve as a starting point to better characterize sex‐specific molecular changes linked to cognitive decline in AD.

Discerning neurodegenerative and aging genetic influences on hippocampal subfields trajectories in the Alzheimer’s disease continuum

AbstractBackgroundHippocampal atrophy is one of the most characteristic biomarkers of Alzheimer’s disease (AD), and its genetic basis, as well as that of its distinct subfields, is of current interest. However, hippocampal subfields have so far only been assessed largely unrelated to brain atrophy, and the majority of studies were focused on general population samples, explicitly excluding diagnosed patients. In this study, we evaluated whether hippocampal subfields trajectories are a suitable endophenotype to discern between neurodegenerative and aging genetic influences in the Alzheimer’s continuum.MethodThe dataset includes AD patients (N = 113), mild cognitive impairment (N = 426), and control normal (CN) (N = 272) subjects from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset. Individuals were evaluated with 3T MRI at baseline and every 6 months for a total of 96 months. Hippocampal subfields were extracted using longitudinal processing within FreeSurfer(6.0). We used longitudinal mixed‐effect models with random intercept for individuals to investigated the association between hippocampal subfields volumetric changes and genetic predisposition to AD, Parkinson’s disease, frontotemporal dementia (neurodegeneration), as well as intrinsic and extrinsic epigenetic clock acceleration and telomere length (biological aging). Models were adjusted by age, sex, AD‐stage and total hippocampal volume. Disease status‐specific trajectories were also assessed.ResultGenetic predisposition to biological aging (driven by genetic predisposition to telomere length) was significantly associated with subfield changes in presubiculum (β = ‐11.2, p = 4.1e‐04), parasubiculum (β = ‐4.4, padj = 4.4e‐04), molecular layer (β = 5.9, padj = 2.0e‐04), dentate gyrus (DG) (β = 5.79, padj = 4.4e‐04), CA4 (β = 7.2, padj = 4.4e‐05), and HATA (β = ‐1.7, padj = 0.027). Genetic predisposition to neurodegeneration (driven by predisposition to AD) was associated with volumetric changes in CA3 (β = 4.4, p = 0.047). In stratified models, we observed significant volumetric changes due to genetic predisposition to neurodegeneration in controls (CA1, CA4, DG, parasubiculum, molecular layer). In MCI (subiculum regions and CA4) and AD (hippocampal fissure, molecular layer and DG), volumetric changes were mostly associated with genetic predisposition to biological aging processes.ConclusionWe showed differential genetic mechanisms associated with hippocampal subfield trajectories that were specific to Alzheimer’s disease stage. These results may help elucidate new biological mechanisms in the course of the disease as well as discern additional molecular mechanisms related to neurodegeneration and aging.

Structural brain network degeneration associated with agitation in dementia

AbstractBackgroundAgitation is a behavioral syndrome involving increased motor activity, restlessness, aggressiveness and emotional distress. Its has a 30% prevalence across multiple types of dementia and is associated with negative outcomes, including reduced quality of life, caregiver distress, and mortality. Pharmacological treatment risks serious side effects, including mortality. Understanding brain network topology could provide insights into novel treatment methods.MethodParticipants comprised 600 subjects from 3 existing databases: the Alzheimer’s Disease Neuroimaging Initiative (ADNI), Frontotemporal Lobar Degeneration Neuroimaging Initiative (NIFD) and the Mesulam Center for Cognitive Neurology and Alzheimer’s Disease at Northwestern University. Participants were clinically diagnosed with either behavioral variant frontotemporal dementia (bvFTD), mild cognitive impairment (MCI), dementia of the Alzeheimer type (DAT) or cognitively normal (CN). The Neuropsychiatric Inventory Questionnaire (NPI‐Q) caregiver rating was used to determine whether individuals were agitated or not. Morphometric similarity networks (MSNs) were generated from Freesurfer statistics calculated from T1‐weighted MRIs. 7 surface‐based cortical metrics (e.g. gray matter volume, surface area) were calculated for each of 360 parcels. Pairwise inter‐parcel Pearson correlations of feature vectors were calculated to produce a morphometric similarity matrix for each individual. We calculated sub‐matrices for salience (SN) cognitive control (CCN) and default mode networks (DMN). Brain Connectivity Toolbox calculated transitivity and global efficiency of each network. Metrics were calculated at different thresholds to ensure the results were not threshold‐dependent. We calculated 2 (Agitation present/absent) x 4 (Diagnosis) repeated measures ANCOVAS for transitivity and global efficiency for each network. Covariates were age, sex, race, database, education, ICV, CDR‐SB and days MRI‐NPI‐Q.ResultFor the SN, people with Agitation had significantly lower global efficiency than people without Agitation (Figure 1). There were no significant effects of diagnosis or interaction. For the CCN, people with agitation had significantly lower transitivity than people without Agitation (Figure 2). There were no significant effect of diagnosis or interaction.ConclusionAcross different forms of dementia, Agitation is associated with reduced integration (global efficiency) of the salience network, and reduced segregation (transitivity) of the cognitive control network. Interventions that alter these topological network features may be effective in reducing agitation in dementia, regardless of clinical diagnosis.

Neuropsychological assessment as a predictor of dementia conversion in Alzheimer’s disease

AbstractBackgroundProgressive cognitive decline is a characteristic of Alzheimer’s disease (AD). Identifying reliable predictors of conversion to dementia in older adults, particularly those with mild cognitive impairment (MCI), is critical for clinicians, patients, and patient families.Method1083 adult participants (Age: M = 73.20, SD = 6.92) were included in this study from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). The average follow up was over 5 years (M = 65.25, SD = 40.16). Cox proportional hazards regression models were implemented to estimate the risk of dementia. Models were built sequentially to examine changes in predictive contributions when including biomarkers and cognitive predictors.Result57% of participants were classified as MCI and 43% as cognitively normal (CN). By the end of study, there is an overall conversion rate of 24% from the participants with cognitive normal and MCI to dementia. Participants with MCI demonstrated significantly higher rates of conversion than those with CN (F(1,258) = 92.51, p < .001, η2 = .26). The initial model indicated greater age (HR = 1.03, p = .004), fewer years of education (HR = 0.95, p < .027), the presence of at least one APOE ε4 gene (HR = 3.23, p < .001), and higher levels of neurofilament light chain (NFL) (HR = 1.28, p < .001) predicted conversion to dementia. A second model indicated higher levels of Ab42 was predictive of lower risk (HR = 0.41, p < .001), whereas higher levels of P‐tau was predictive of greater risk (HR = 1.48, p < .001). APOE ε4 presence and NfL levels remained significant predictors. Finally, composites of memory and executive scores were added. Memory performance at baseline was highly associated with dementia risk (HR = 0.26, p < .001), with poorer performance at baseline predicting greater likelihood of conversion. Similarly, poorer performance on executive functioning tasks was predictive of dementia risk (HR = 0.57, p < .001).ConclusionComposite scores of memory and executive functions are strong predictors of conversion to dementia. Such composites are approximately as predictive as Ab42, and more predictive than demographics, APOE ε4 presence, baseline P‐tau and NFL.

Genetic predisposition to Alzheimer’s disease is differentially associated with joint volumetric variations in the disease continuum

AbstractBackgroundImaging genetics studies commonly assess the individual association between genetic information and brain features without considering joint brain modulations. The aim of this study was to explore the structural variation of the brain at different stages of Alzheimer’s disease (AD), depending on the genetic predisposition to AD.MethodA total of 351 cognitively unimpaired (CU) middle‐age participants from the ALFA (Alzheimer’s and FAmilies) study and 330 cognitively impaired participants from ADNI (Alzheimer’s Disease Neuroimaging Initiative) with magnetic resonance imaging scans, genetic information and CSF‐amyloid status were included. They were classified as CU Aβ‐, CU Aβ+, (mild cognitive impairment) MCI Aβ+ and AD Aβ+[Table 1]. Freesurfer was used to obtain cortical and subcortical parcellations using the Desikan‐Killiany atlas. A polygenic risk score estimating each participant’s genetic predisposition to AD (PRS‐AD) was calculated including genetic variants at the genome‐wide suggestive level (5×10‐6). Individuals were classified into high/low genetic predisposition groups. Statistical analysis was based on compositional data analysis. First, we computed the optimal multivariate brain structural variation signature (elastic net selection). Then, a logistic regression was performed to determine the association between the optimal signature and the PRS‐AD. Models were stratified by disease stage and adjusted for age and sex.ResultIndividuals at different stages on the AD continuum displayed different brain volumetric modulations associated with higher genetic predisposition to AD [Figure 1]. For instance, the optimal signature associated with an increased risk of AD was mainly characterized by increased hippocampal volumes in Aβ‐ individuals, but decreased volumes in MCI individuals along with the modulation of other temporal regions. Moreover, in AD patients at higher genetic predisposition of AD, the optimal signature was characterized by increased and decreased volumes in frontal regions (e.g. pars triangularis, lateral orbitofrontal vs. medial orbitofrontal, respectively) [Figure 2].ConclusionResults showed AD stage‐specific multivariate volumetric variation associated with an increased genetic risk of AD. Main results suggested that regions that modulated together, although differing according to AD stage, were implicated in the structural brain topology of specific memory networks. The analysis of the joint volumetric variation of brain subregions brings an innovative modeling perspective for neurogenetic studies of AD.

Epigenetic age from peripheral blood predicts progression to Alzheimer’s disease, white matter disease burden, and cortical atrophy

AbstractBackgroundCross‐sectional studies suggest a limited but promising relationship between epigenetic clocks, estimated using DNA methylation patterns, and Alzheimer’s disease (AD) pathophysiology and risk. However, prior analyses did not utilize longitudinal analyses or neuroimaging biomarkers of AD. Herein, we employed survival and structural neuroimaging analyses to test the hypothesis that DNA methylation age predicts AD progression.MethodUsing data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), we calculated two established epigenetic ages, ‘DNAmPhenoAge’ and ‘DNAmGrimAge’ which proxy biologic age and mortality risk, respectively. Using Cox proportional hazard regression analysis in normal controls (NC), we tested whether DNAmPhenoAge/DNAmGrimAge associated with progression to mild cognitive impairment (MCI) or AD. Structural magnetic resonance (MR) images were segmented using Freesurfer 5.1 and a previously described white matter hyperintensity (WMH) quantification pipeline. Multiple regression analysis was used to test whether DNAmPhenoAge/DNAmGrimAge predicted WMH volumes and cortical atrophy.ResultReplicating prior work, we found no significant differences in DNAmPhenoAge and DNAmGrimAge between NC (n = 128), MCI (n = 251), and AD (n = 50) diagnoses. Strikingly, however, in NC participants, both DNAmPhenoAge (p = 0.004) and DNAmGrimAge (p = 0.05) associated with progression to MCI or AD (Figure 1). Even after correction for chronologic age, DNAmPhenoAge remained significantly associated with progression to MCI or AD (p = 0.01). Utilizing samples from across the spectrum of normal aging to AD, DNAmPhenoAge and DNAmGrimAge strongly associated with WMH in all diagnoses (Table 1). Moreover, DNAmPhenoAge remained significantly associated with WMH in select groups after correction for chronologic age (Table 1). Finally, both DNAmPhenoAge and DNAmGrimAge predicted cortical thickness in multiple AD‐relevant regions (Figure 2).ConclusionIn contrast to earlier work suggesting limited applicability of blood‐based DNA methylation scores in the context of AD, our novel survival and neuroimaging analyses have revealed the broad utility of second‐generation epigenetic clocks not only in predicting progression to MCI and AD but also via their association with neuroimaging biomarkers of AD, including white matter disease burden and cortical atrophy in AD‐relevant regions. Easily accessible, blood‐based measures of biologic age therefore represent promising predictors of AD risk and pathophysiology.

Comparison of plasma and cerebrospinal fluid protein markers and their significance for the characterization of dementia disease status

AbstractBackgroundAlthough Aβ42 and phosphorylated tau (p‐tau) in cerebrospinal fluid (CSF) are valuable to diagnose Alzheimer’s disease (AD), many patients are reluctant to accept lumbar puncture. In addition, there is lack of biomarkers for the different stages of AD dementia. This study sought to compare the plasma and CSF protein markers to characterize dementia status.MethodThis cohort study included Alzheimer’s Disease Neuroimaging Initiative (ADNI) participants who had baseline levels of 73 proteins measured in plasma (566 total, 58 cognitively normal (CN), 396 mild cognitive impairment (MCI), and 112 AD) and in CSF (310 total, 92 CN, 149 MCI, and 69 AD). Univariate/multivariate logistic regression and least absolute shrinkage and selection operator (LASSO) were used to identify the protein candidates for the disease classification. The following three models were examined for CN vs. MCI and MCI vs. AD; model 1: proteins only, model 2: proteins adjusted for age, sex, education, and model 3: model 2 plus APOE4 status.ResultNine proteins in plasma and CSF were chosen based on their importance from LASSO. The area under the curve (AUC) for the final multivariate logistic regression including the 9 proteins was calculated for each disease group in plasma [CN vs. MCI: 0.895 (model 1), 0.899 (model 2), 0.921 (model 3), and MCI vs. AD: 0.828 (model 1), 0.842 (model 2), 0.856 (model 3)]. In addition, AUC in CSF [CN vs. MCI: 0.678 (model 1), 0.726 (model 2), 0.770 (model 3) and MCI vs. AD: 0.711 (model 1), 0.738 (model 2), 0.765 (model 3)] was also estimated. We observed that plasma proteins outperformed the CSF proteins to be associated with MCI and AD. Also, plasma proteins were significantly better than the established AD markers, namely CSF Aβ42 and p‐tau in CN vs. MCI [AUC: 0.836 (model 1), 0.842 (model 2), 0.862 (model 3)] and MCI vs. AD [AUC: 0.665 (model 1), 0.673 (model 2), 0.680 (model 3)].ConclusionThis study demonstrates that the combination of plasma proteins could be considered an effective marker for diagnosing AD and its early‐stage symptoms with better performance than the other markers.

DAT1 and BDNF polymorphisms interact to predict cross‐sectional and longitudinal amyloid and tau pathology in cognitively normal older adults

AbstractBackgroundDysfunctional dopamine signaling and disruptions in neurotrophic support are potential exacerbators of Alzheimer’s disease (AD) pathophysiology. A single nucleotide polymorphism (SNP) in the dopamine transporter gene (rs6347 in DAT1/SLC6A3) is associated with greater ventricular expansion and lower scores on the Mini Mental State Exam (MMSE; Roussotte et al., 2015). A polymorphism in the brain derived neurotrophic factor gene (Val66Met in BDNF) is related to greater hippocampal atrophy and more severe episodic memory decline for amyloid positive Met carriers (Lim et al., 2013). However, it is unclear whether and how these polymorphisms directly relate to AD pathology.MethodWe tested a moderated mediation model (Fig. 1) involving both BDNF and DAT1 polymorphisms in a sample of cognitively normal older adults (n = 321) from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). This mediation involved DAT1 (independent variable), b‐amyloid (Ab; mediator), and tau pathology (dependent variable) with BDNF as a moderator. Ab and tau were measured both cross‐sectionally and longitudinally with positron emission tomography (PET) tracers [18F]Florbetapir/[18F]Florbetaben and [18F]Flortaucipir, respectively.ResultA significant moderated mediation demonstrated that DAT1 and BDNF interacted to predict cross‐sectional Ab‐PET (p = .02) and tau‐PET (p = .04), such that carriers of both DAT1 C and BDNF Met alleles demonstrated greater pathology (Fig. 2), consistent with previous research associating these variants with AD symptoms. Longitudinal analyses replicated these cross‐sectional results, with DAT1 C and BDNF Met carriers exhibiting greater accumulation of tau pathology (p = .0002) and marginally increased Ab accumulation (p = .07).ConclusionOur analyses revealed novel interaction effects between genetic polymorphisms in the DAT1 and BDNF genes. These results suggest that carrying DAT1 CC and BDNF Met results in higher Ab, which then contributes to higher tau pathology. While our analyses do not demonstrate causality, these findings are consistent with current hypotheses of AD in which Ab drives increases in tau pathology (Karran et al., 2011).

Deep learning based normative models identify structures that become abnormal with Alzheimer’s disease

AbstractBackgroundImproving dementia diagnosis is a major global concern, given that up to 20% of subjects are misdiagnosed. Normative modelling, which quantifies how far individuals deviate from healthy controls (HC), may help to better situate subjects along the spectrum. To determine whether this is the case, we implemented a deep learning based normative model and tested it on structural neuroimaging data from subjects with HC, mild cognitive impairment (MCI), and Alzheimer’s disease (AD).MethodWe developed an adversarial autoencoder (AAE) (Figure 1) leveraging FreeSurfer features, namely cortical surface area of 68 cortical subregions (34 by hemisphere; Desikan‐Killiany atlas) and volumes of 33 neuroanatomical structures (Aseg atlas). Our model not only returns individual “deviation” scores—divergence from HC—but also identifies regions of interest that contribute to said score, facilitating interpretability. We retrieved T1‐weighted MRI scans (n = 204) from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (HC/MCI/AD 105/29/60; Table 1). We trained our AAE on 71 HC samples (n = 71) and tested it on the rest (n = 133). We used bootstrapping to compute confidence intervals (CI) for each point estimate and examined regions where scores differed for the considered groups [1].ResultDeviation scores were lowest for HC (mean 0.74 [95%‐CI 0.63–0.88]), highest for AD (mean 1.21 [95%‐CI 1.03–1.50]), and intermediate for MCI (mean 1.04 [95%‐CI 0.88–1.27]), implying subjects indeed fell along the AD spectrum (Figure 2A). Said scores also enabled classifying subjects into HC, MCI, and AD with a mean area under the curve (AUC) of 75% (Figure 2B‐C; HC vs. MCI: 0.75 [95%‐CI 0.62–0.85]; HC vs. AD: 0.76 [95%‐CI 0.64–0.82]). According to the AAE, the hippocampus, parahippocampal cortex, ventricles, and inferior parietal lobe contributed to such differences (Figure 2D‐E).ConclusionNormative models can identify structures that become abnormal as the brain undergoes ageing and AD, in line with the literature. Further testing on a larger and more heterogeneous sample will reveal the potential of this tool and its potential deployment in clinics.

Longitudinal cognitive change in older adults with elevated tau PET signal in the absence of amyloid

AbstractBackgroundMost theoretical frameworks for Alzheimer’s disease (AD) pathogenesis propose amyloid as the primary pathologic process, whereas tau in the absence of amyloid (A‐/T+) is considered a “non‐AD pathologic change” (Jack et al., 2018). Previous research has demonstrated AD‐like characteristics of the A‐/T+ biomarker profile, including poorer memory and language performance and entorhinal thinning (Weigand et al., 2022; Yoon et al., 2022). Assessment of domain‐specific longitudinal cognitive decline in this group, however, has not yet been addressed. Our study sought to evaluate 1‐year memory, language, and attention/executive function trajectories of A‐/T+ older adults.MethodWe included 756 older adults at baseline across the AD clinical continuum from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), of whom 359 had 1‐year follow‐up cognitive data. Four A/T groups were classified based on cortical amyloid‐beta and medial temporal tau positron emission tomography positivity. Linear mixed effects models examined 1‐year trajectories of each baseline A/T group in domains of memory, language, and attention/executive function. Both statistical significance and effect size estimates were evaluated.ResultThe following proportions of A/T groups were observed at baseline: 47.0% A‐/T‐, 13.2% A‐/T+, 13.4% A+/T‐, and 26.4% A+/T+. Only the A+/T+ group exhibited a statistically significant decline in language and attention/executive function domains relative to the A‐/T‐ group. Examination of effect sizes, however, revealed that the A‐/T+ group had a quantitatively similar degree of decline as the A+/T+ group in memory (A‐/T+ and A+/T+ bs = ‐0.13), and an intermediate degree of decline between the A‐/T‐ and A+/T+ groups for language (A‐/T+ b = ‐0.17) and attention/executive function (A‐/T+ b = ‐0.19). Similar patterns were observed when examining cognitively unimpaired participants only.ConclusionOlder adults with the A‐/T+ biomarker profile demonstrate a degree of cognitive decline across multiple domains consistent with a possible AD trajectory. The systematic exclusion of these individuals from AD clinical trials may hinder early intervention efforts to attenuate the cognitive decline observed in this biomarker group, and an expansion of our conceptualization of the AD continuum to include A‐/T+ individuals may offer improved insight into the pathologic and cognitive changes that occur in the context of preclinical AD.

The effect of BDNF Val66Met on memory, CSF total‐tau, CSF ptau181 and CSF Aβ42 in early sporadic Alzheimer’s disease

AbstractBackgroundAllelic variation in the brain‐derived neurotrophic factor (BDNF) Val66Met polymorphism has been shown to moderate rates of cognitive decline in preclinical sporadic Alzheimer’s disease (i.e., Aβ+ older adults), and pre‐symptomatic dominantly inherited Alzheimer’s disease (DIAD). In DIAD, Met66 was also associated with greater increases in CSF levels of total‐tau and phosphorylated tau (p‐tau181). This study sought to determine the extent to which BDNF Val66Met is also associated with changes in episodic memory and CSF total‐tau and p‐tau181 in Aβ+ older adults in early‐stage sporadic AD.MethodAβ+ Met66 carriers (n = 94) and Val66 homozygotes (n = 192) enrolled in the Alzheimer’s Disease Neuroimaging Initiative (ADNI) who did not meet criteria for AD dementia were included. Only participants with at least one follow‐up neuropsychological and CSF assessment were included. Episodic memory was measured by the ADNI Memory Composite and CSF samples were collected via lumbar puncture to determine levels of total‐tau, p‐tau181 and Aβ42. A series of linear mixed models were conducted to investigate changes in each outcome variable over 10 years, covarying for CSF Aβ42 (except when examining changes in CSF Aβ42), APOE ε4 status, sex, age, baseline diagnosis and years of education.ResultWhen followed over 10 years, Aβ+ Met66 carriers demonstrated significantly faster memory decline (d = 0.33) and significantly greater increases in CSF total‐tau (d = 0.30) and p‐tau181 (d = 0.29) compared to Val66 homozygotes, despite showing equivalent rates of change in CSF Aβ42 (Table 1; Figure 1).ConclusionIn non‐demented Aβ+ older adults, Met66 carriers showed greater memory decline and greater increases in both CSF total‐tau and CSF p‐tau181 over 10 years compared to Val66 homozygotes, despite showing equivalent changes in CSF Aβ42. This suggests that Met66 carriage, which is associated with lower BDNF availability and lower neurotrophic support, may confer greater vulnerability to Aβ‐related increases in tau compared to Val66 homozygosity in the early stages of sporadic AD.

Interleukin‐3 is associated with soluble TREM2 and mediates amyloid and tau pathology in Alzheimer’s disease

AbstractBackgroundDysfunction of glial cell communication is involved in Alzheimer’s disease (AD) pathogenesis. Recent studies reported that interleukin‐3 (IL‐3) participated in astrocyte‐microglia crosstalk and restricted pathological progression in AD mice. However, how IL‐3 relates to microglial activity and AD pathology in humans remains unclear.MethodWe included 311 participants with cerebrospinal fluid (CSF) IL‐3, sTREM2 (microglial biomarker), and AD biomarkers (Aβ42, p‐tau, and t‐tau) from the Alzheimer’s disease Neuroimaging Initiative (ADNI). We compared IL‐3 levels in different stages of AD and assessed the associations between IL‐3 with sTREM2 and AD biomarkers at baseline and with longitudinal cognitive change. Mediation models were used to explore the potential mechanism of how IL‐3 affects sTREM2 and AD pathology.ResultCSF IL‐3 levels decreased with increasing Aβ pathology and increased with increasing tau pathology, especially in cognitive impairment participants (Figure 1). CSF IL‐3 was significantly associated with CSF sTREM2 and CSF AD biomarkers (Figure 2). Mediation analysis showed that IL‐3 and sTREM2 mediated the effects of CSF Aβ on CSF tau pathology (βp‐tau = 0.053, βt‐tau = 0.055, all p < 0.001) (Figure 3).ConclusionOur findings suggest that IL‐3 is associated with the microglia biomarker and that they respond strongly to amyloid and tau pathologies. Hence, IL‐3 could be a key marker for clinical trial design. We encourage future studies on enhancing IL‐3 levels as a therapeutic intervention to slow the development of AD pathology.

Predicting of β‐amyloid Positivity Using Integer‐based Risk Scores with Blood‐based Biomarkers

AbstractBackgroundβ‐amyloid (Aβ), a pathologic hallmark of Alzheimer’s disease (AD) is the target of recently FDA‐approved drugs. In trials and in practice, identifying individuals likely to be β‐amyloid positive (Aβ+) with accessible and affordable tests will be increasingly important. Blood‐based biomarkers (BBMs) have shown early promise in determining amyloid status. We aimed to develop machine learning‐based integer risk scores using BBMs and other accessible measures to predict Aβ positivity by PET in older, non‐demented (cognitively unimpaired or mild cognitive impairment) adults.MethodsWe used data from 722 individuals (72.1 ± 7.0 years old) from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) to develop integer risk scores for predicting Aβ positivity. The sample was divided into training (75%) and test (25%) sets. Variables were categorized as demographics (D: age, sex, and education), genetic information (G: APOE4 status), cognitive measures (C: ADAS‐Cog‐13 and MMSE), and plasma biomarkers of p‐Tau181(T) and neurofilament light (N: NfL). Feature‐sets were formed as hierarchically constructed combinations of these categorized variables (see Figure 1). Using the AutoScore algorithm, variables were ordered according to their Random Forest Gini Index and were iteratively included in the construction of logistic regression‐derived risk score models. Variables were included based on increases in Area Under the ROC Curves (AUCs).ResultsFour models were developed based on four feature‐sets. All models utilized age and APOE4 status and, when considered, all models retained ADAS‐Cog‐13, p‐Tau, and NfL. The best performing model corresponded to the DGCTN feature‐set with performance on the dedicated test set of AUC = 0.810 (C.I.: 0.745‐0.876). Using the DGCTN model on the test set, 97% of individuals with scores in the top quartile of possible risk score points were correctly identified as Aβ+.ConclusionsWe developed risk score models utilizing accessible and affordable measures to predict Aβ positivity. Plasma biomarkers improve performance of risk scores indicating their value. Applications include screening in clinical settings for therapeutic care and enrichment of clinical trials. These data can be used to optimize the efficient use of PET in determining eligibility for clinical trials and amyloid targeted therapies in practice.

Plasma p‐tau181 and nfl as surrogates biomarkers in clinical trials targeting preclinical stage of Alzheimer’s disease

AbstractBackgroundThe preclinical stage of Alzheimer’s disease (AD) is one of the main focuses of clinical trials. Plasma and positron emission tomography (PET) biomarkers have already been proposed to monitor participants’ disease progression in clinical trials targeting cognitively unimpaired (CU) individuals. Although longitudinal changes in plasma phosphorylated tau 181 (p‐tau181) and neurofilament light chain(NfL) correlate with AD progression, it is unknown whether these changes can be used to monitor drug effects in preventive clinical trials. Here, we aimed to evaluate the utility of using changes in plasma p‐tau181 and NfL as surrogate biomarkers for clinical trials targeting preclinical AD.MethodWe access 257 CU older individuals that had available Aß‐PET at baseline, as well as the baseline, up to 24‐month plasma p‐tau181 and NfL measures from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). We calculate the estimated sample size needed to test a 25% drug effect with 80% of power at a 0.05 level on reducing changes in plasma markers.ResultWe demonstrated that therapeutical clinical trials of 24‐month follow‐up using p‐tau181 and NfL would require 78% and 63% smaller sample sizes compared with a 12‐month follow‐up(Figure 1). The use of intermediate levels of Aß(CL20‐40), rather than merely Aß positivity, as an enrichment strategy reduced the sample size by 73% for p‐tau181 and 59% for NfL over 24 months. As demonstrated in Figure 2A the estimated cost of a clinical trial using only plasma biomarkers is lower than using neuroimaging biomarkers for surrogacy(Figure 2A). However, as showed by Figure 2B a trial including all Aß positive individuals, the total estimated cost when considering surrogate biomarkers plus other related assessments are numerically higher using plasma than neuroimaging biomarkers; while trials including only individuals with intermediate Aß levels, the cost was similar using plasma and neuroimaging biomarkers for surrogacy.ConclusionOur results suggest that to monitor large scale large‐scale population interventions in CU Aß positive individuals’ plasma p‐tau181/ NfL could potentially be used. Furthermore, using a strategy that has been proposed in recent clinical trials, we demonstrated that the use of intermediate levels of Aß is more cost‐effective than trials using Aß‐positive individuals in clinical.

Genetic predisposition to Alzheimer’s disease is differentially associated with joint volumetric variations in the disease continuum

AbstractBackgroundImaging genetics studies commonly assess the individual association between genetic information and brain features without considering joint brain modulations. The aim of this study was to explore the structural variation of the brain at different stages of Alzheimer’s disease (AD), depending on the genetic predisposition to AD.MethodA total of 351 cognitively unimpaired (CU) middle‐age participants from the ALFA (Alzheimer’s and FAmilies) study and 330 cognitively impaired participants from ADNI (Alzheimer’s Disease Neuroimaging Initiative) with magnetic resonance imaging scans, genetic information and CSF‐amyloid status were included. They were classified as CU Aβ‐, CU Aβ+, (mild cognitive impairment) MCI Aβ+ and AD Aβ+[Table 1]. Freesurfer was used to obtain cortical and subcortical parcellations using the Desikan‐Killiany atlas. A polygenic risk score estimating each participant’s genetic predisposition to AD (PRS‐AD) was calculated including genetic variants at the genome‐wide suggestive level (5×10‐6). Individuals were classified into high/low genetic predisposition groups. Statistical analysis was based on compositional data analysis. First, we computed the optimal multivariate brain structural variation signature (elastic net selection). Then, a logistic regression was performed to determine the association between the optimal signature and the PRS‐AD. Models were stratified by disease stage and adjusted for age and sex.ResultIndividuals at different stages on the AD continuum displayed different brain volumetric modulations associated with higher genetic predisposition to AD [Figure 1]. For instance, the optimal signature associated with an increased risk of AD was mainly characterized by increased hippocampal volumes in Aβ‐ individuals, but decreased volumes in MCI individuals along with the modulation of other temporal regions. Moreover, in AD patients at higher genetic predisposition of AD, the optimal signature was characterized by increased and decreased volumes in frontal regions (e.g. pars triangularis, lateral orbitofrontal vs. medial orbitofrontal, respectively) [Figure 2].ConclusionResults showed AD stage‐specific multivariate volumetric variation associated with an increased genetic risk of AD. Main results suggested that regions that modulated together, although differing according to AD stage, were implicated in the structural brain topology of specific memory networks. The analysis of the joint volumetric variation of brain subregions brings an innovative modeling perspective for neurogenetic studies of AD.

Visualizing Braak stages with deformation‐based morphometry in super‐sampled MRI

AbstractBackgroundUnlike MRI data, histological data collected post‐mortem can’t be used to study the longitudinal formation of neurofibrillary tangles (NFT) and corresponding brain changes over time for individuals. Although structural MRI offers the possibility of non‐invasive longitudinal antemortem imaging, its resolution is a limiting factor. Using image processing and super‐sampling techniques, we examined the possibility of employing structural MRI to observe the NFT footprints in the brain.MethodData included baseline MRI scans of 1473 subjects from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset including ADNI1, 2, and GO. Voxel size was super‐sampled to 0.5×0.5×0.5mm (Manjón et al. 2010) before non‐linear registration to an ADNI‐based unbiased symmetric MRI template. The resulting deformation fields were used to compute the Jacobian determinant map for each subject as a proxy for local volume. For each subject, the sex‐ and age‐corrected z‐scored Jacobian maps were then calculated and averaged for each diagnostic group: normal subjects(CN), early MCI, late MCI, and AD. The average map for CN group was subtracted from the other average maps to show the AD‐stage differences.ResultThe Braak tau staging pattern (Braak & Braak, 1995) is seen in average Jacobian maps as the disease progresses. Subjects in the MCI groups showed on average more pronounced neurodegeneration in the transentorhinal and entorhinal region (Braak Stages 1‐2), followed by degeneration in the hippocampus (Stages 3‐4). Later on, this neurodegeneration progresses toward the neocortex (Stages 5‐6). Figure 1 illustrates these results: A) the template, B)EMCI vs CN, C) LMCI vs EMCI, D) LMCI vs CN, E) AD vs CN. Red color shows an increase and blue indicates a decrease in local volume.ConclusionUsing novel image processing techniques, we were able to verify the staging of AD‐related neurofibrillary pathology progression in MRI data. This shows that structural MRI data, although limited in resolution, can still offer in‐depth insight into the brain antemortem which will in return allow the longitudinal and possible lifespan studies with an accuracy close to histological data.