Background
Mild behavioral impairment (MBI) is a neurobehavioral syndrome characterized by the emergence of persistent non‐cognitive neuropsychiatric symptoms (NPS) in older adults, representing an ...at‐risk state for dementia and a potential marker of Alzheimer’s disease (AD). However, few studies have investigated associations between MBI and plasma biomarkers of AD pathophysiology, specifically tau hyperphosphorylation.
Method
Individuals across the AD spectrum (cognitively unimpaired (CU) and impaired (CI)) were selected from the Translational Biomarkers of Aging and Dementia (TRIAD) cohort. MBI was assessed using the MBI‐Checklist (MBI‐C), which assesses NPS severity in five subdomains: decreased motivation (apathy), emotional dysregulation (mood/anxiety symptoms), impulse dyscontrol (agitation, impulsivity, abnormal reward salience), social inappropriateness (impaired social cognition), and abnormal thoughts/perception (psychosis). Plasma levels of p‐tau181 were quantified in all individual, and log‐transformed due to right skew. Spearman correlation analysis investigated cross‐sectional associations between plasma p‐tau181 and MBI‐C scores, and between baseline plasma p‐tau181 and annual change in MBI‐C. Multivariable linear regression analyses assessed the ability of cross‐sectional and longitudinal MBI‐C scores to predict plasma p‐tau181, adjusting for age, sex, education, and diagnostic group (CU/CI, Aβ +/‐).
Result
Cross‐sectional MBI data were available for 211 individuals, with longitudinal data available for 128 individuals (). Significant correlations were found between plasma p‐tau181 and MBI‐C total and apathy scores in the cross‐sectional sample (Figure A). Further, significant correlations were found with plasma p‐tau181 and the annual change in MBI‐C total and apathy scores (Figure B). These associations were corroborated by significant standardized and adjusted regression coefficients from multivariate regression models, which showed that baseline and longitudinal MBI‐C total and apathy score predicted plasma level of p‐tau181 (Figure C).
Conclusion
Our study provides novel findings on the association between MBI and tau hyperphosphorylation assessed using plasma biomarkers in an AD cohort. These results add to the existing evidence for MBI as a clinical manifestation of AD pathology and support the use of the MBI‐C as an enrichment tool for clinical trial enrollment and therapeutic intervention.
Background
The presence of p‐tau in biofluids has previously been proposed to be a response to neurofibrillary tangle pathology, one of the hallmarks of Alzheimer’s disease (AD). However, the ...increase of p‐tau in cerebrospinal fluid (CSF) precedes detectable neurofibrillary tangle pathology, as indexed by tau Positron Emission Tomography (PET), by up to a decade, suggesting that soluble tau could be an indication of early tau pathology. With this study, we investigated the heterogeneity of p‐tau species in CSF to assess the disease status of participants of the Translational Biomarkers of Aging and Dementia (TRIAD) cohort.
Methods
Support vector machines were used to determine cutoff values of p‐tau181, p‐tau217, p‐tau231 and p‐tau235 in CSF, identifying a group of participants that were amyloid positive (58 from a total of 165 participants). Amyloid positivity was determined by using an 18FAZD4694 SUVR threshold value of 1.55 in the neocortex. Using these cutoff values, signatures were calculated on an individual level to identify the number of phosphorylated sites. Based on the number of phosphorylated sites, 18FMK6240 SUVR maps and 18FAZD4694 SUVR maps were calculated.
Results
When combining different CSF p‐tau species, the largest contribution in identifying amyloid positivity came from p‐tau217, followed by p‐tau231, p‐tau181 and p‐tau235. Achieving the cutoff for multiple p‐tau species was associated with more tau pathology particularly in the later Braak stages (fig 1) and increased amyloid‐β plaque accumulation (fig 2).
Conclusion
Our findings suggest that heterogeneity in p‐tau species carries predictive power in the identification of disease severity in incipient Alzheimer’s Disease.
Highlights • Chronic quinpirole led to increased locomotor activity and frequency of visits. • Acute exposure increased while chronic exposure to quinpirole decreased the total glucose consumption in ...the brain. • Acute administration increased the local cerebral glucose uptake in regions with poor direct connection to the dopaminergic system. • Strong reductions after chronic exposure were seen in the cortico-striato-thalamico-cortical circuit. • Alterations in (para)limbic regions indicates a more extensive disease mechanism.
Background
Neuroinflammation typically involves the activation of microglial cells in the brain and has been linked to Alzheimer’s disease (AD) pathology. However, how microglial activation ...influences brain neurodegeneration in individuals across the AD continuum is still poorly understood. Here, we aimed to investigate the influence of microglial activation in longitudinal brain atrophy in individuals across the AD continuum. We hypothesize that high levels of regional brain inflammation predict widespread brain atrophy.
Method
We assessed 95 individuals from the TRIAD cohort (60 cognitively unimpaired and 35 cognitively impaired) with available 11CPBR28‐PET, a measure of microglial activation, and a 2‐year longitudinal MRI (mean = 2.07 years). We generated grey matter voxel‐based morphometry (VBM) images using SPM12 and DARTEL, smoothed with a Gaussian kernel of full‐width half maximum of 8mm. We built the uncorrected (p < 0.05) association matrix between the 11CPBR28‐PET SUVR and longitudinal VBM ROIs (z‐score) with the β‐estimates from linear regressions accounting for age, sex, and diagnosis. We divided the 11CPBR28‐PET levels into terciles (low, intermediate, and high) to generate the averages of longitudinal VBM changes.
Result
Baseline ROI‐based 11CPBR28‐PET levels associate with longitudinal brain atrophy not only locally but also in distinct brain regions (Fig 1a). After FDR‐correction, the inferior temporal cortex was the region where 11CPBR28‐PET levels were better associated with widespread longitudinal brain atrophy (Fig. 1b). Brain atrophy was observed in AD‐related regions, including the amygdala, insula, and the superior temporal cortex, independently of global amyloid load and tau (Fig. 1b). Accordingly, individuals with higher 11CPBR28‐PET levels in the inferior temporal cortex presented increased longitudinal brain atrophy compared to individuals with lower 11CPBR28‐PET (Fig. 1c).
Conclusion
We identified increased baseline 11CPBR28‐PET levels in the inferior temporal cortex that were highly associated with longitudinal brain atrophy in individuals across the AD continuum. Our results demonstrated that higher levels of inflammation in key brain regions could predict widespread longitudinal brain atrophy, suggesting that microglial activation has a detrimental impact on AD‐related neurodegeneration progression.
Background
There is a growing body of evidence suggesting changes in blood flow and metabolism as a trigger of the cascade of events leading to Alzheimer’s disease (AD). Oxygen extraction fraction ...(OEF) MRI images provide valuable information about the brain’s metabolism and blood flow in stroke, brain injury, and recently neurodegenerative diseases. However, OEF is a research tool, and its clinical utility is still being evaluated. Here, we aimed to assess brain oxygenation in aging and AD using OEF images. We test the hypothesis that OEF reduction is associated with the clinical stages of AD.
Method
A cohort of 310 subjects with cognitively unimpaired (CU) (n = 182), mild cognitive impairment (MCI) (n = 80), and AD (n = 48), were recruited. All participants received 3D gradient‐recalled echo sequence MRI. All OEF images were constructed by QSM+qBOLD model with CCTV (Temporal clustering, tissue composition, and total variation) (Figure1). Eighty‐two various brain regions of interest (ROI) and 6 Braak ROI were investigated in the current study.
Result
Our results demonstrated that the OEF mean value in almost all of the brain ROI encompassed the lowest amount of OEF in AD compared with MCI and CU. In most parts of the brain, the OEF mean value was AD<MCI<CU including bilateral caudate. However, in other parts of the brain, the OEF mean value increased from CU to MCI and then dipped to lower than CU in AD including the bilateral hippocampus, which clearly shows that brain metabolism and blood flow are decreasing in AD (Table1).
Conclusion
OEF is an affordable and relevant non‐invasive MR approach to assess brain metabolism during disease progression or therapeutic interventions. OEF decreased in symptomatic cases. Further studies should explore the mechanisms underlying increased OEF in MCI. Our results support the potential of OEF to improve our understanding of metabolic changes associated with AD pathophysiology.
Background
Plasma markers of tau are currently being studied as proxies of cerebral neurofibrillary tangle (NFT) accumulation. Phosphorylated tau (pTau) and N‐terminal tau fragment (NTA) assays are ...associated with present and future tau‐PET load. Our aim was to investigate whether plasma markers could predict whether someone will be a slow or fast accumulator.
Method
We assessed 143 individuals 72 CU, 54 MCI, 17 AD from the TRIAD cohort, with two available 18FMK6240 tau‐PET scans and calculated the relative change (Δ18FMK6240) between baseline and follow‐up mean follow‐up time: 2.1 ± 0.7 years. We used tertiles to divide individuals as slow, medium and fast accumulators. Additionally, we measured baseline plasma pTau181, pTau217, pTau231 and NTA concentrations. We computed the effect size (Cohen’s d) and area under the curve (AUC) for each plasma marker for Δ18FMK6240 between slow and fast accumulators. Δ18FMK6240 was calculated in Braak stages I/II, III/IV and V/VI.
Result
We first observed that the highest effect size for Δ18FMK6240 in Braak I/II was depicted by pTau231. For Δ18FMK6240 in Braak III/IV and Braak V/VI, pTau217 presented the highest effect size (Figure 1). Moreover, AUC values were the highest, and highly similar, in ΔBraak I/II for pTau181, pTau217 and pTau231. For ΔBraak III/IV, pTau181 and pTau217 presented the highest values. Finally, AUC for ΔBraak V/VI, pTau231 and NTA had the highest values, which were also similar (Figure 2).
Conclusion
Plasma pTau biomarkers (181, 217 and 231) are great predictors of fast accumulation in early to middle Braak regions. For late Braak regions, fast accumulation was best predicted by pTau217 and NTA. Plasma markers are able to determine whether someone will be a fast accumulator in a stage‐specific manner. The currently available tau biofluid measures could be used in the clinical and clinical trial settings, as these are less invasive and cheaper than CSF or PET assessments. Especially in the recruitment phase, pTau217 could be used for screening individuals that are more likely to accumulate tau fast.
Abstract
Background
Deep learning models, particularly convolutional neural networks (CNNs), have shown promise in Alzheimer’s disease (AD) classification using tau PET data. However, limited sample ...sizes and unharmonized tau tracers present challenges to developing an agnostic tau tracer tool to predict AD using machine learning. Transfer learning, which leverages pre‐trained models for related tasks, may address these issues. Here we evaluate the effectiveness of transfer learning in optimizing 3D CNNs for AD classification with distinct cohorts and tau tracers.
Method
We used tau PET images from ADNI (
18
FFlortaucipir, n = 437) and TRIAD (
18
FMK‐6240, n = 423) cohorts, categorizing patients into CU (cognitively unimpaired) and CI (cognitively impaired). Standardized uptake value ratios (SUVR) were used for tau PET data. Separate 3D CNNs were trained for each tracer, with SUVR volumes as input and diagnosis as output. For transfer learning, we trained a model on
18
FFlortaucipir data with a reduced learning rate, using a pre‐trained model from
18
FMK‐6240. Models underwent 5‐fold cross‐validation, and metrics were computed as the average of validation metrics across folds. To avoid data leakage, images from the same subject were assigned to the same fold.
Result
The model trained on
18
FMK‐6240 tracer demonstrated higher classification performance than
18
FFlortaucipir (AUC = 0.84 vs 0.67;
Figure 1
. F1‐score = 79.77% vs 64.66%;
Table 1
). To enhance the classification performance of
18
FFlortaucipir model, we employed a transfer learning approach by leveraging the model pre‐trained with
18
FMK‐6240. With this approach, we observed a slight improvement in all classification metrics compared to the model trained solely on
18
FFlortaucipir data (AUC = 0.71 vs 0.67;
Figure 1
. Accuracy = 71.39% vs 67.72, F1‐score = 67.97% vs 64.66%;
Table 1
).
Conclusion
This finding highlights the value of transfer learning in optimizing deep learning models for Alzheimer’s disease classification, particularly when handling tau tracers with varying performance levels. Our results are consistent with previous on transfer learning’s effectiveness in this context. These preliminary findings indicate that applying this technique to larger datasets of tau tracers may further enhance model performance, potentially leading to the development of a tau tracer‐agnostic tool that overcomes the need of tracer harmonization for predicting dementia.
Background
The sequential model predicts synaptic depletion as a downstream of amyloid, tau and neuroinflammation. However, synaptic toxicity might be consequence of toxic forms of amyloid in the ...absence of tau. In this study we explore the role of synaptic depletion and neuroinflammation as determinants of tau pathology. GAP‐43 and neurogranin are pre‐ and post‐synaptic biomarkers known to be detected at elevated levels in cerebrospinal fluid (CSF) of patients with Alzheimer’s disease (AD).
Method
We included 126 individuals from TRIAD cohort. Brain inflammation, tau tangle and amyloid‐β (Aβ) deposition were assessed via 11CPBR28‐PET, 18FMK6240‐PET and 18FAZD4694‐PET, respectively. All patients had plasma GFAP quantified, and a subset of 78 individuals had CSF GFAP, CSF Neurogranin and CSF GAP‐43 quantifications available. Voxel and region of interest regression models evaluated the relationship between PET tracers and the fluid biomarkers. Models with 18FAZD4694‐PET as the outcome were adjusted for 18FMK6240‐PET voxel‐wise and vice‐versa. A linear regression interaction model evaluated the interaction between CSF GFAP and both synaptic biomarkers with Aβ‐ and tau‐PET as the outcome.
Result
Positive associations were found between CSF neurogranin and Aβ‐, tau‐ and brain inflammation‐PET in AD related regions; the strongest associations were found with tau‐pet and CSF neurogranin in the medial temporal lobe. GAP‐43 was also positively associated with tau‐ and TSPO‐PET, but no associations were found with Aβ‐PET. Both plasma and CSF GFAP were associated with Aβ‐, tau‐ and TSPO‐PET in AD related regions, with the strongest t‐values in the model including plasma GFAP and amyloid‐PET. A negative interaction was found between CSF GFAP and both synaptic biomarkers with amyloid‐ and tau‐PET as the outcomes. These associations were stronger with tau‐PET and were found in the temporal, occipital and parietal areas.
Conclusion
This study supports the role of synaptic dysfunction and neuroinflammation on tau load. As both synaptic biomarkers and GFAP increase as a function of tangles load, tau accumulation is linked to the relationship between synaptic abnormalities and inflammation in early stages of the disease. Our results suggest that synaptic depletion is a phenomenon that might start prior to tau tangles.
Background
Accumulation of tau neurofibrillary tangles in Alzheimer’s disease (AD) follows a stereotypical pattern, known as Braak staging. As individuals show different rates of tau accumulation, ...depending on their initial stage, this potentially biases drug effects on tau pathology over time. We hypothesized that amyloid would be a driving force behind tau deposition in later Braak regions for those that were at earlier stages of the disease.
Method
79 individuals (CU: 57, MCI: 21, AD: 1) were recruited from the Translational Biomarkers of Aging and Dementia (TRIAD) cohort. All individuals underwent baseline and 2‐year follow‐up amyloid (18FAZD4694) and tau (18FMK6240) PET imaging, as well as baseline plasma pTau181, pTau217, and pTau231 assessments. Differences in standardized uptake value ratios between the two timepoints in Braak I‐IV were used to separate fast accumulators of tau from slow accumulators, using the standard deviation of a Young group (n = 13).
Pearson R’s were calculated between tau deposition in Braak regions IV‐VI and the neocortical amyloid load at baseline, within both these groups. Furthermore, a voxelwise analysis was performed to identify specific regional differences in amyloid SUVR between slow and fast progressors, while correcting for age, sex, ApoE4 carriership, as well as the individual’s Braak stage. In addition, an independent t‐test was performed to compare the levels of each pTau species at baseline.
Result
Fast accumulators had stronger increases (p < 0.01) in tau accumulation in Braak region V and VI, compared to slower accumulators. Furthermore, the level of tau deposition in Braak region IV and V was positively associated (p < 0.01) with the amyloid load in the neocortex, but only in the fast accumulator group (figure 1). The voxelwise analysis confirmed that faster progressors had stronger associations with amyloid‐PET (figure 2). Furthermore, those who accumulated tau at a faster rate also had higher levels of plasma pTau at baseline (figure 3).
Conclusion
Tau deposition in later Braak regions is associated with a higher amyloid load and increased values of pTau181, pTau217 and pTau231 at baseline in individuals in early Braak stages that accumulate tau at a more rapid rate.
Background
The sequential model predicts synaptic depletion as a downstream of amyloid, tau and neuroinflammation. However, synaptic toxicity might be consequence of toxic forms of amyloid in the ...absence of tau. In this study we explore the role of synaptic depletion and neuroinflammation as determinants of tau pathology. GAP‐43 and neurogranin are pre‐ and post‐synaptic biomarkers known to be detected at elevated levels in cerebrospinal fluid (CSF) of patients with Alzheimer’s disease (AD).
Method
We included 126 individuals from TRIAD cohort. Brain inflammation, tau tangle and amyloid‐ß (Aß) deposition were assessed via 11CPBR28‐PET, 18FMK6240‐PET and 18FAZD4694‐PET, respectively. All patients had plasma GFAP quantified, and a subset of 78 individuals had CSF GFAP, CSF Neurogranin and CSF GAP‐43 quantifications available. Voxel and region of interest regression models evaluated the relationship between PET tracers and the fluid biomarkers. Models with 18FAZD4694‐PET as the outcome were adjusted for 18FMK6240‐PET voxel‐wise and vice‐versa. A linear regression interaction model evaluated the interaction between CSF GFAP and both synaptic biomarkers with Aß‐ and tau‐PET as the outcome.
Result
Positive associations were found between CSF neurogranin and Aß‐, tau‐ and brain inflammation‐PET in AD related regions; the strongest associations were found with tau‐pet and CSF neurogranin in the medial temporal lobe. GAP‐43 was also positively associated with tau‐ and TSPO‐PET, but no associations were found with Aß‐PET. Both plasma and CSF GFAP were associated with Aß‐, tau‐ and TSPO‐PET in AD related regions, with the strongest t‐values in the model including plasma GFAP and amyloid‐PET. A negative interaction was found between CSF GFAP and both synaptic biomarkers with amyloid‐ and tau‐PET as the outcomes. These associations were stronger with tau‐PET and were found in the temporal, occipital and parietal areas.
Conclusion
This study supports the role of synaptic dysfunction and neuroinflammation on tau load. As both synaptic biomarkers and GFAP increase as a function of tangles load, tau accumulation is linked to the relationship between synaptic abnormalities and inflammation in early stages of the disease. Our results suggest that synaptic depletion is a phenomenon that might start prior to tau tangles.