Aims

Non-invasive high-resolution imaging of amyloid-beta (Aβ) deposits across the entire rodent brain can greatly advance our understanding on the underlying pathophysiology of Alzheimer's disease. At present, a large gap exists between sub-millimeter scale highly invasive intravital microscopy and whole-brain imaging modalities lacking the resolution or molecular specificity for accurate characterization of Aβ pathologies. Here we introduce a multi-scale optical molecular imaging platform from transcranial observations of single plaques to 3D whoe brain Aβ mapping in animal models.

Methods

The in vitro binding between amyloid probes AOI987 and luminescent conjugated oligothiophene HS-169 and recombinant Aβ1-42 fibrils were assessed. High precision in vivo assessment of Aβ deposits was performed in transgenic APP/PS1, arcAβ mice and wild-type littermates (12-24 months-old) with newly devised large-field multifocal illumination fluorescence microscopy using luminacient conjugated oligothiophene HS-169 and multi-view volumetric multispectral optoacoustic tomography using oxazine-derivative AOI987 probe. Ex vivo light-sheet microscopy and immunohistochemistry was performed at ex vivo to assess the specificity of the probes binding to Aβ.

Results

We achieved transcranial cortex-wide Aβ imaging with 10 μm (single plaque) resolution, scaling into 100 μm resolution imaging at the whole brain level, including deeply embedded areas such as hippocampus and thalamus inaccessible by conventional intravital microscopy. Ex vivo light-sheet microscopy and immunohistochemistry confirmed the specificity and regional distributions unveiled by in vivo Aβ imaging.

Conclusions

We developed a novel in vivo Aβ imaging pipeline for multi-scale high precision assessment, which facilitates region-specific studies of Aβ spread and accumulation, and the monitoring of putative treatments targeting Alzheimer’s disease.

Aims

To assess the feasibility of Aβ42/Aβ40 ratio, measured by a novel HPLC-MS/MS method, as a useful and cost-effective biomarker of PET status in a large Korean cohort from the DPUK Study.

Methods

580 participants were included in six groups: ADD (Alzheimer´s Disease Dementia), AMCI (Alzheimer´s Mild Cognitive Impairment), OC (Old Controls), SVCI (Subcortical Vascular Cognitive Impairment), and CAA (Cerebral Amyloid Angiopathy). Aβ40 and Aβ42 plasma levels were quantitated using a new antibody-free, liquid chromatography-mass spectrometry method (HPLC-MS/MS), which drastically reduces sample preparation time and cost. Receiver Operating Characteristic (ROC) curves were built according to Centiloid-based positivity. Several statistical models were evaluated.

Results

Aβ42/Aβ40 ratios were statistically lower for PET positive individuals in ADD, AMCI, OC (p<0.001) and SVCI groups (p=0.034) but not in CAA group (p=0.544). ROC curves are presented for the total group (n=562, age over 45), which includes ADD (n=131), AMCI (n=212), OC (n=149), SVCI (n=58), and CAA (n=12). Model 1 (Aβ42/Aβ40 ratio) and Model 2 (Aβ42/Aβ40 ratio + ApoE4 + age + diagnosis) are compared. Area Under the ROC Curve (AUC) values for PET positivity based on Centiloid cut-off were 0.808 and 0.871 for Model 1 and Model 2, respectively (p between models=0.002).

Conclusions

Aβ42/Aβ40 ratio in human plasma, measured by this novel HPLC-MS/MS method, showed good discriminating performance based on PET positivity, as demonstrated in this study that contains the largest series of patients being analyzed with this methodology.

Aims

Astrogliosis in response to amyloid-beta (Aβ) plaques is an early feature of Alzheimer's disease (AD). Glial fibrillary acidic protein (GFAP) is expressed in astrocytes and is increased in CSF in AD. Studies on plasma GFAP as AD biomarker are few and not longitudinal. Our aim was to evaluate plasma GFAP as potential biomarker for Aβ status and for future development of AD dementia.

Methods

161 subjects with a baseline clinical diagnosis of mild cognitive impairment (MCI) were included, genotyped for APOE, followed for 4.7 years (average) and assessed for conversion to AD at follow-up. Plasma was collected at baseline and follow-up. GFAP was measured with Simoa GFAP Discovery kit for SR-X (Quanterix). Aß positivity (Aß+) was defined as CSF Aβ42/40 <0.07 (cut-off calculated with Youden index within the cohort).

Results

Baseline GFAP was increased in Aβ+ MCI patients (p<0.0001). Plasma GFAP could predict Aβ+ status (p<0.0001, AIC=184.3, AUC=0.787, sensitivity=73%, specificity=75%). Accuracy was increased by combining plasma GFAP and APOE genotype (p<0.0001, AIC 154.7, AUC=0.859). Plasma GFAP could also predict subsequent development of AD dementia (p<0.0001, AIC=154.4, AUC=0.836, sensitivity=72%, specificity=85%). Predictive accuracy of future AD dementia was improved by combining plasma GFAP with APOE genotype and age (p<0.0001, AIC=140, AUC=0.864). Longitudinal slopes showed a significant increase of plasma GFAP over time in Aβ+ MCI compared to Aβ- (p<0.0001) and in subjects later diagnosed with AD compared to those that remained clinically stable (stable Aβ-:p<0.0001; stable Aβ+:p=0.049).

Conclusions

Plasma GFAP is strongly associated to Aβ status and is a good predictor of clinical evolution to AD.

Aims

To identify causal pathway of AD and PD risk and drug targets by generating and analyzing multi-tissue proteomic and genetic data from a large cohort.

Methods

We generated a genomic atlas of protein levels in multiple neurologically relevant tissues (380 brain, 835 cerebrospinal fluid (CSF) and 529 plasma), by profiling thousands of proteins in a large and well-characterized cohort. We used Mendelian Randomization (MR) and colocalization methods to identify proteins in the causal pathway of two neurological diseases and drug targets for repurposing.

Results

Combining both MR and colocalization results, we found that one CSF, 13 plasma and six brain proteins were likely to be in the causal pathways for AD risk. Among these proteins, plasma CD33 was a risk factor towards AD and had been used as a drug target for other diseases, such as prostate cancer. As for PD risk, 13 CSF, 12 plasma and 23 brain proteins were likely to be the cause. Among these proteins, plasma IDUA was prioritized as it was encoded by a risk locus for PD and as a drug target for chondroitin sulfate, reported to treat osteoarthritis. IDUA is required for the lysosomal degradation of glycosaminoglycans, dermatan sulfate and heparan sulfate.

Conclusions

Our results prioritized several proteins likely to be in the causal pathways leading to AD and PD risk. These nominated proteins can facilitate mapping the disease GWAS results into biological mechanisms, and further leading to precision medicine in neurological and/or psychiatric traits.

Aims

To compare cerebrospinal fluid (CSF) levels of the synaptic protein, VAMP-2, in Lewy body dementia (LBD) and Alzheimer’s disease (AD) patients.

Methods

We quantified VAMP-2 using a Single Molecular Array (ADx NeuroSciences) in aged cognitively normal controls (n=68) and patients from the Sant Pau Initiative for Neurodegeneration cohort clinically diagnosed with mild cognitive impairment/dementia due to LBD (n=47) or AD (n=119). LBD with AD co-pathology (LBD+AD n=28) was distinguished from pure LBD (n=19) using our validated CSF p-tau/Aβ₄₂ cut-off. Regression analyses were controlled for age and APOE E4 status.

Results

Compared to controls, mean CSF VAMP-2 levels were lower in pure LBD (0.86-fold, p=.005) but elevated in LBD+AD (1.58-fold, p=.009) and AD (1.25-fold, p=.04).The CSF p-tau/Aβ₄₂ ratio*LBD diagnosis interaction term contributed more to VAMP-2 levels (t=7.0, p<.0001) than either variable alone (t=5.6, p<.0001 and t=-4.4, p<.0001, respectively).

In ROC analyses, VAMP-2 showed good accuracy (area under the curve) to discriminate LBD+AD (.796, 95%CI .67-.90) and AD (.715, 95%CI .64-.79) from controls but poor accuracy to discriminate pure LBD (.579, 95%CI .42-.74) from controls.

VAMP-2 was associated with CSF Aβ₄₂/Aβ₄₀ in LBD+AD (r²=.40, p=.0002) and AD (r²=.17, p<.0001), with CSF tau markers in all groups (r²=.42 to .67, all p<.002) and with CSF Nf-L in controls (r²=.25, p=.01).

Conclusions

CSF VAMP-2 is not a surrogate marker of neurodegeneration. Low CSF VAMP-2 levels may reflect reduced synapse density in LBD patients, an effect that may be masked by AD pathology. Synuclein and AD pathologies may have a synergistic effect on CSF VAMP-2 levels.

Aims

A large body of research has shown cerebral blood hypoperfusion across the Alzheimer’s disease (AD) continuum. However, the relationship between the primary AD pathologies and cerebral blood flow (CBF) remains unclear. Here, we examined the link between amyloid-beta (Aβ) and tau with CBF.

Methods

Baseline CBF was measured by arterial spin labeling at a 3T MRI scanner in 94 Aβ- cognitively unimpaired, 43 Aβ+ cognitively unimpaired, and 119 Aβ+ cognitively impaired participants i.e. those with mild cognitive impairment (MCI) or AD dementia. Aβ and tau burden was measured using [18F] flutemetamol and [18F] RO948 positron emission tomography (PET), respectively. Additionally, cerebrospinal fluid (CSF) was analyzed for Aβ42 and Aβ40. Voxel-wise and linear regression analyses were used to assess the interrelation between CBF with Aβ and tau load.

Results

CBF was not associated with amyloid PET or CSF Aβ42/40 in cognitively unimpaired individuals. However, tau PET was inversely related to CBF in lateral temporal, parietal and superior occipital cortices in individuals on the AD continuum i.e. in Aβ+ individuals. Event-based modeling suggested that the observed CBF alterations occurred after neocortical Aβ pathology, temporal tau pathology, and memory deficits, but before widespread neocortical tau pathology.

Conclusions

These findings provide in vivo evidence for an association between tau aggregation and CBF reduction in the AD spectrum and indicate that CBF changes after the occurrence of tau pathology in temporal areas.

Aims

The connectomes obtained from different neuroimaging modalities are often analyzed separately, despite growing evidence showing they are not independent and often interact with each other. The aim of this study is to assess the multilayer connectome across different stages of Alzheimer’s disease (AD) using amyloid and cortical thickness data.

Methods

One-hundred ninety-nine amyloid-negative controls (CN-), 98 amyloid-positive controls (CN+), 234 amyloid-positive patients with mild cognitive impairment (MCI+), and 166 amyloid-positive AD patients with T1-weighted MRI and 18F-Florbetapir PET data were included from the Alzheimer’s Disease Neuroimaging Initiative. We integrated the networks built using amyloid and cortical thickness data into a multiplex network using BRAPH 2.0 (http://braph.org/). This network was binarized with different densities and the multiplex participation, overlapping degree, and degree overlap were compared between groups.

Results

We found significant multiplex participation decreases in the right entorhinal cortex and increases in the left entorhinal cortex in MCI+ compared to CN-. In AD+, in addition to the entorhinal changes, significant decreases in the temporal pole and caudal anterior cingulate were also found compared to CN-. There were significant overlapping degree decreases in the parahippocampal, precentral, and entorhinal cortices in addition to increases in the caudal anterior cingulate in CN+ compared to CN-. Finally, the degree overlap was lower in widespread areas in AD+ and in temporal areas in MCI+ and CN+ compared to CN-.

Conclusions

These findings suggest that the multilayer brain connectome can detect widespread changes in the interaction between amyloid pathology and gray matter atrophy across different stages of AD.