The Alzheimer's Disease Neuroimaging Initiative 3: Continued innovation for clinical trial improvement

Efficient segmentation of fetal brain MRI based on the physical resolution

BACKGROUND

The image resolution of fetal brain magnetic resonance imaging (MRI) is a critical factor in brain development measures, which is mainly determined by the physical resolution configured in the MRI sequence. However, fetal brain MRI are commonly reconstructed to 3D images with a higher apparent resolution, compared to the original physical resolution.

PURPOSE

This work is to demonstrate that accurate segmentation can be achieved based on the MRI physical resolution, and the high apparent resolution segmentation can be achieved by a simple deep learning module.

METHODS

This retrospective study included 150 adult and 80 fetal brain MRIs. The adult brain MRIs were acquired at a high physical resolution, which were downsampled to visualize and quantify its impacts on the segmentation accuracy. The physical resolution of fetal images was estimated based on MRI acquisition settings and the images were downsampled accordingly before segmentation and restored using multiple upsampling strategies. Segmentation accuracy of ConvNet models were evaluated on the original and downsampled images. Dice coefficients were calculated, and compared to the original data.

RESULTS

When the apparent resolution was higher than the physical resolution, the accuracy of fetal brain segmentation had negligible degradation (accuracy reduced by 0.26%, 1.1%, and 1.8% with downsampling factors of 4/3, 2, and 4 in each dimension, without significant differences from the original data). Using a downsampling factor of 4 in each dimension, the proposed method provided 7× smaller and 10× faster models.

CONCLUSION

Efficient and accurate fetal brain segmentation models can be developed based on the physical resolution of MRI acquisitions.

A review of deep learning-based information fusion techniques for multimodal medical image classification

Multimodal medical imaging plays a pivotal role in clinical diagnosis and research, as it combines information from various imaging modalities to provide a more comprehensive understanding of the underlying pathology. Recently, deep learning-based multimodal fusion techniques have emerged as powerful tools for improving medical image classification. This review offers a thorough analysis of the developments in deep learning-based multimodal fusion for medical classification tasks. We explore the complementary relationships among prevalent clinical modalities and outline three main fusion schemes for multimodal classification networks: input fusion, intermediate fusion (encompassing single-level fusion, hierarchical fusion, and attention-based fusion), and output fusion. By evaluating the performance of these fusion techniques, we provide insight into the suitability of different network architectures for various multimodal fusion scenarios and application domains. Furthermore, we delve into challenges related to network architecture selection, handling incomplete multimodal data management, and the potential limitations of multimodal fusion. Finally, we spotlight the promising future of Transformer-based multimodal fusion techniques and give recommendations for future research in this rapidly evolving field.

The hippocampus as a structural and functional network epicentre for distant cortical thinning in neurocognitive aging

Alterations in grey matter (GM) and white matter (WM) are associated with memory impairment across the neurocognitive aging spectrum and theorised to spread throughout brain networks. Functional and structural connectivity (FC,SC) may explain widespread atrophy. We tested the effect of SC and FC to the hippocampus on cortical thickness (CT) of connected areas. In 419 (223 F) participants (agemean=73 ± 8) from the Alzheimer's Disease Neuroimaging Initiative, cortical regions associated with memory (Rey Auditory Verbal Learning Test) were identified using Lasso regression. Two structural equation models (SEM), for SC and resting-state FC, were fitted including CT areas, and SC and FC to the left and right hippocampus (LHIP,RHIP). LHIP (β=-0.150,p=<.001) and RHIP (β=-0.139,p=<.001) SC predicted left temporopolar/rhinal CT; RHIP SC predicted right temporopolar/rhinal CT (β=-0.191,p=<.001). LHIP SC predicted right fusiform/parahippocampal (β=-0.104,p=.011) and intraparietal sulcus/superior parietal CT (β=0.101,p=.028). Increased RHIP FC predicted higher left inferior parietal CT (β=0.132,p=.042) while increased LHIP FC predicted lower right fusiform/parahippocampal CT (β=-0.97; p=.023). The hippocampi may be epicentres for cortical thinning through disrupted connectivity.

The association between rs6859 in NECTIN2 gene and Alzheimer's disease is partly mediated by pTau

Introduction

Emerging evidence suggests a connection between vulnerability to infections and Alzheimer's disease (AD). The nectin cell adhesion molecule 2 (NECTIN2) gene coding for a membrane component of adherens junctions is involved in response to infection, and its single nucleotide polymorphism (SNP) rs6859 was significantly associated with AD risk in several human cohorts. It is unclear, however, how exactly rs6859 influences the development of AD pathology. The aggregation of hyperphosphorylated tau protein (pTau) is a key pathological feature of neurodegeneration in AD, which may be induced by infections, among other factors, and potentially influenced by genes involved in both AD and vulnerability to infections, such as NECTIN2.

Materials and methods

We conducted a causal mediation analysis (CMA) on a sample of 708 participants in the Alzheimer's Disease Neuroimaging Initiative (ADNI). The relationship between rs6859 and Alzheimer's disease (AD), with AD (yes/no) as the outcome and pTau-181 levels in the cerebrospinal fluid (CSF) acting as a mediator in this association, was assessed. Adjusted estimates from the probit and linear regression models were used in the CMA model, where an additive model considered an increase in dosage of the rs6859 A allele (AD risk factor).

Results

The increase in dose of allele A of the SNP rs6859 resulted in about 0.144 increase per standard deviation (SD) of pTau-181 (95% CI: 0.041, 0.248, p<0.01). When included together in the probit model, the change in A allele dose and each standard deviation change in pTau-181 predicted 6.84% and 9.79% higher probabilities for AD, respectively. In the CMA, the proportion of the average mediated effect was 17.05% and was higher for the risk allele homozygotes (AA), at 19.40% (95% CI: 6.20%, 43.00%, p<0.01). The sensitivity analysis confirmed the evidence of a robust mediation effect.

Conclusion

This study reported a new causal relationship between pTau-181 and AD. We found that the association between rs6859 in the NECTIN2 gene and AD is partly mediated by pTau-181 levels in CSF. The rest of this association may be mediated by other factors. Further research, using other biomarkers, is needed to uncover the remaining mechanisms of the association between the NECTIN2 gene and AD.

An interpretable generative multimodal neuroimaging-genomics framework for decoding Alzheimer's disease

Alzheimer's disease (AD) is the most prevalent form of dementia, affecting millions worldwide with a progressive decline in cognitive abilities. The AD continuum encompasses a prodormal stage known as Mild Cognitive Impairment (MCI), where patients may either progress to AD (MCIc) or remain stable (MCInc). Understanding the underlying mechanisms of AD requires complementary analysis derived from different data sources, leading to the development of multimodal deep learning models. In this study, we leveraged structural and functional Magnetic Resonance Imaging (sMRI/fMRI) to investigate the disease-induced grey matter and functional network connectivity changes. Moreover, considering AD's strong genetic component, we introduce Single Nucleotide Polymorphisms (SNPs) as a third channel. Given such diverse inputs, missing one or more modalities is a typical concern of multimodal methods. We hence propose a novel deep learning based classification framework where generative module employing Cycle Generative Adversarial Networks (cGAN) was adopted to impute missing data within the latent space. Additionally, we adopted an Explainable Artificial Intelligence (XAI) method, Integrated Gradients (IG), to extract input features relevance, enhancing our understanding of the learned representations. Two critical tasks were addressed: AD detection and MCI conversion prediction. Experimental results showed that our framework was able to reach the state-of-the-art in the classification of CN vs AD reaching an average test accuracy of 0.926 ± 0.02. For the MCInc vs MCIc task, we achieved an average prediction accuracy of 0.711 ± 0.01 using the pre-trained model for CN and AD. The interpretability analysis revealed that the classification performance was led by significant grey matter modulations in cortical and subcortical brain areas well known for their association with AD. Moreover, impairments in sensory-motor and visual resting state network connectivity along the disease continuum, as well as mutations in SNPs defining biological processes linked to amyloid-beta and cholesterol formation clearance and regulation, were identified as contributors to the achieved performance. Overall, our integrative deep learning approach shows promise for AD detection and MCI prediction, while shading light on important biological insights.

Assistive tools for classifying neurological disorders using fMRI and deep learning: A guide and example

BACKGROUND

Deep-learning (DL) methods are rapidly changing the way researchers classify neurological disorders. For example, combining functional magnetic resonance imaging (fMRI) and DL has helped researchers identify functional biomarkers of neurological disorders (e.g., brain activation and connectivity) and pilot innovative diagnostic models. However, the knowledge required to perform DL analyses is often domain-specific and is not widely taught in the brain sciences (e.g., psychology, neuroscience, and cognitive science). Conversely, neurological diagnoses and neuroimaging training (e.g., fMRI) are largely restricted to the brain and medical sciences. In turn, these disciplinary knowledge barriers and distinct specializations can act as hurdles that prevent the combination of fMRI and DL pipelines. The complexity of fMRI and DL methods also hinders their clinical adoption and generalization to real-world diagnoses. For example, most current models are not designed for clinical settings or use by nonspecialized populations such as students, clinicians, and healthcare workers. Accordingly, there is a growing area of assistive tools (e.g., software and programming packages) that aim to streamline and increase the accessibility of fMRI and DL pipelines for the diagnoses of neurological disorders.

OBJECTIVES AND METHODS

In this study, we present an introductory guide to some popular DL and fMRI assistive tools. We also create an example autism spectrum disorder (ASD) classification model using assistive tools (e.g., Optuna, GIFT, and the ABIDE preprocessed repository), fMRI, and a convolutional neural network.

RESULTS

In turn, we provide researchers with a guide to assistive tools and give an example of a streamlined fMRI and DL pipeline.

CONCLUSIONS

We are confident that this study can help more researchers enter the field and create accessible fMRI and deep-learning diagnostic models for neurological disorders.

Utility of cerebrovascular imaging biomarkers to detect cerebral amyloidosis

INTRODUCTION

The relationship between cerebrovascular disease (CVD) and amyloid-β (Aβ) in Alzheimer disease (AD) is understudied. We hypothesized that magnetic resonance imaging (MRI)-based CVD biomarkers, including cerebral microbleeds (CMBs), ischemic infarction, and white matter hyperintensities (WMH), would correlate with Aβ positivity on positron emission tomography (Aβ-PET).

METHODS

We cross-sectionally analyzed data from the Alzheimer's Disease Neuroimaging Initiative (ADNI, N=1,352). Logistic regression was used to calculate odds ratios (ORs), with Aβ-PET positivity as the standard-of-truth.

RESULTS

Following adjustment, WMH (OR=1.25) and superficial CMBs (OR=1.45) remained positively associated with Aβ-PET positivity (p<.001). Deep CMBs and infarcts exhibited a varied relationship with Aβ-PET in cognitive subgroups. The combined diagnostic model, which included CVD biomarkers and other accessible measures, significantly predicted Aβ-PET (pseudo-R 2 =.41).

DISCUSSION

The study highlights the translational value of CVD biomarkers in diagnosing AD, and underscores the need for more research on their inclusion in diagnostic criteria. ClinicalTrials.gov: ADNI-2 ( NCT01231971 ), ADNI-3 ( NCT02854033 ).

Unsupervised multimodal modeling of cognitive and brain health trajectories for early dementia prediction

Predicting the course of neurodegenerative disorders early has potential to greatly improve clinical management and patient outcomes. A key challenge for early prediction in real-world clinical settings is the lack of labeled data (i.e., clinical diagnosis). In contrast to supervised classification approaches that require labeled data, we propose an unsupervised multimodal trajectory modeling (MTM) approach based on a mixture of state space models that captures changes in longitudinal data (i.e., trajectories) and stratifies individuals without using clinical diagnosis for model training. MTM learns the relationship between states comprising expensive, invasive biomarkers (β-amyloid, grey matter density) and readily obtainable cognitive observations. MTM training on trajectories stratifies individuals into clinically meaningful clusters more reliably than MTM training on baseline data alone and is robust to missing data (i.e., cognitive data alone or single assessments). Extracting an individualized cognitive health index (i.e., MTM-derived cluster membership index) allows us to predict progression to AD more precisely than standard clinical assessments (i.e., cognitive tests or MRI scans alone). Importantly, MTM generalizes successfully from research cohort to real-world clinical data from memory clinic patients with missing data, enhancing the clinical utility of our approach. Thus, our multimodal trajectory modeling approach provides a cost-effective and non-invasive tool for early dementia prediction without labeled data (i.e., clinical diagnosis) with strong potential for translation to clinical practice.

APOE4 homozygozity represents a distinct genetic form of Alzheimer's disease

This study aimed to evaluate the impact of APOE4 homozygosity on Alzheimer's disease (AD) by examining its clinical, pathological and biomarker changes to see whether APOE4 homozygotes constitute a distinct, genetically determined form of AD. Data from the National Alzheimer's Coordinating Center and five large cohorts with AD biomarkers were analyzed. The analysis included 3,297 individuals for the pathological study and 10,039 for the clinical study. Findings revealed that almost all APOE4 homozygotes exhibited AD pathology and had significantly higher levels of AD biomarkers from age 55 compared to APOE3 homozygotes. By age 65, nearly all had abnormal amyloid levels in cerebrospinal fluid, and 75% had positive amyloid scans, with the prevalence of these markers increasing with age, indicating near-full penetrance of AD biology in APOE4 homozygotes. The age of symptom onset was earlier in APOE4 homozygotes at 65.1, with a narrower 95% prediction interval than APOE3 homozygotes. The predictability of symptom onset and the sequence of biomarker changes in APOE4 homozygotes mirrored those in autosomal dominant AD and Down syndrome. However, in the dementia stage, there were no differences in amyloid or tau positron emission tomography across haplotypes, despite earlier clinical and biomarker changes. The study concludes that APOE4 homozygotes represent a genetic form of AD, suggesting the need for individualized prevention strategies, clinical trials and treatments.

Mahmoud L, Sadek I. AI-based tool for early detection of Alzheimer's disease

In the context of Alzheimer's disease (AD), timely identification is paramount for effective management, acknowledging its chronic and irreversible nature, where medications can only impede its progression. Our study introduces a holistic solution, leveraging the hippocampus and the VGG16 model with transfer learning for early AD detection. The hippocampus, a pivotal early affected region linked to memory, plays a central role in classifying patients into three categories: cognitively normal (CN), representing individuals without cognitive impairment; mild cognitive impairment (MCI), indicative of a subtle decline in cognitive abilities; and AD, denoting Alzheimer's disease. Employing the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset, our model undergoes training enriched by advanced image preprocessing techniques, achieving outstanding accuracy (testing 98.17 %, validation 97.52 %, training 99.62 %). The strategic use of transfer learning fortifies our competitive edge, incorporating the hippocampus approach and, notably, a progressive data augmentation technique. This innovative augmentation strategy gradually introduces augmentation factors during training, significantly elevating accuracy and enhancing the model's generalization ability. The study emphasizes practical application with a user-friendly website, empowering radiologists to predict class probabilities, track disease progression, and visualize patient images in both 2D and 3D formats, contributing significantly to the advancement of early AD detection.

Impact of APOE-ε alleles on brain structure and cognitive function in healthy older adults: A VBM and DTI replication study

BACKGROUND

The Apolipoprotein E (APOE) gene has been established in the Alzheimer's disease (AD) literature to impact brain structure and function and may also show congruent effects in healthy older adults, although findings in this population are much less consistent. The current study aimed to replicate and expand the multimodal approach employed by Honea et al. Structural magnetic resonance imaging (MRI), diffusion tensor imaging (DTI), and neuropsychological measures were used to investigate the impact of APOE-ε status on grey matter structure, white matter integrity, and cognitive functioning.

METHODS

Data were obtained from the Alzheimer's Disease Initiative Phase 3 (ADNI3) database. Baseline MRI, DTI and cognitive composite scores for memory (ADNI-Mem) and executive function (ADNI-EF) were acquired from 116 healthy controls. Participants were grouped according to APOE allele presence (APOE-ε2+ N = 17, APOE-ε3ε3 N = 64, APOE-ε4+ N = 35). Voxel-based morphometry (VBM) and tract-based spatial statistics (TBSS) were used to compare grey matter volume (GMV) and white matter integrity, respectively, between APOE-ε2+ and APOE-ε3ε3 controls, and again between APOE-ε4+ and APOE-ε3ε3 controls. Multivariate analysis of covariance (MANCOVA) was used to examine the effects of APOE polymorphism on memory and EF across all APOE groups with age, sex and education as regressors of no interest. Cognitive scores were correlated (Pearson r) with imaging metrics within groups.

RESULTS

No significant differences were seen across groups, within groups in MRI metrics, or cognitive performance (p>0.05, corrected for multiple comparisons).

CONCLUSIONS

The current study partially replicated and extended previous findings from an earlier multimodal study (Honea 2009). Future studies should clarify APOE mechanisms in healthy ageing by adding other imaging, cognitive, and lifestyle metrics and longitudinal design in larger sample sizes.

Predictability of intelligence and age from structural connectomes

In this study, structural images of 1048 healthy subjects from the Human Connectome Project Young Adult study and 94 from ADNI-3 study were processed by an in-house tractography pipeline and analyzed together with pre-processed data of the same subjects from braingraph.org. Whole brain structural connectome features were used to build a simple correlation-based regression machine learning model to predict intelligence and age of healthy subjects. Our results showed that different forms of intelligence as well as age are predictable to a certain degree from diffusion tensor imaging detecting anatomical fiber tracts in the living human brain. Though we did not identify significant differences in the prediction capability for the investigated features depending on the imaging feature extraction method, we did find that crystallized intelligence was consistently better predictable than fluid intelligence from structural connectivity data through all datasets. Our findings suggest a practical and scalable processing and analysis framework to explore broader research topics employing brain MR imaging.

Sarcopenia in subjects with Alzheimer's disease: prevalence and comparison of agreement between EGWSOP1, EGWSOP2, and FNIH criteria

BACKGROUND

Sarcopenia is an age-related clinical syndrome characterized by the progressive loss of muscle mass and muscle strength. It appears to be closely linked to dementia, particularly Alzheimer's disease (AD); however, its prevalence among AD patients remains unclear. In this study, we assessed differences in sarcopenia prevalence between non-demented individuals and AD patients. Moreover, we assessed sex-specific differences in sarcopenia prevalence and explored the diagnostic value of the Muscle Quality Index (MQI) for diagnosing sarcopenia among AD patients.

METHOD

Cross-sectional study including 145 patients with probable AD and 51 older adults with normal cognition. Sarcopenia was diagnosed according to the criteria of the European Working Group on Sarcopenia in Older People (EWGSOP1 and EWGSOP2) and of the Foundation for the National Institutes of Health (FNIH). The MQI was computed as the ratio of handgrip strength to skeletal muscle mass.

RESULTS

No significant difference in sarcopenia prevalence was observed between AD patients and controls. Prevalence ranged from 3.4 to 23.4% in AD patients and from 2 to 11.8% in controls, depending on diagnostic criteria. Prevalence was higher using EWGSOP1 and decreased using EWGSOP2 and FNIH. Prevalence was higher in males than in females with AD. The MQI was lower in AD patients than in controls (95%CI: - 0.23, - 0.05, p < 0.001), but displayed poor diagnostic accuracy in identifying sarcopenia cases.

CONCLUSIONS

AD patients and controls show comparable sarcopenia prevalence. Sarcopenia prevalence is higher in males than females among AD patients and higher when using EWGSOP1 compared to FNIH and EWGSOP2 criteria.

Lifespan reference curves for harmonizing multi-site regional brain white matter metrics from diffusion MRI

Age-related white matter (WM) microstructure maturation and decline occur throughout the human lifespan, complementing the process of gray matter development and degeneration. Here, we create normative lifespan reference curves for global and regional WM microstructure by harmonizing diffusion MRI (dMRI)-derived data from ten public datasets (N = 40,898 subjects; age: 3-95 years; 47.6% male). We tested three harmonization methods on regional diffusion tensor imaging (DTI) based fractional anisotropy (FA), a metric of WM microstructure, extracted using the ENIGMA-DTI pipeline. ComBat-GAM harmonization provided multi-study trajectories most consistent with known WM maturation peaks. Lifespan FA reference curves were validated with test-retest data and used to assess the effect of the ApoE4 risk factor for dementia in WM across the lifespan. We found significant associations between ApoE4 and FA in WM regions associated with neurodegenerative disease even in healthy individuals across the lifespan, with regional age-by-genotype interactions. Our lifespan reference curves and tools to harmonize new dMRI data to the curves are publicly available as eHarmonize (https://github.com/ahzhu/eharmonize).

Florbetapir PET-assessed demyelination is associated with faster tau accumulation in an APOE ε4-dependent manner

PURPOSE

The main objectives were to test whether (1) a decrease in myelin is associated with enhanced rate of fibrillar tau accumulation and cognitive decline in Alzheimer's disease, and (2) whether apolipoprotein E (APOE) ε4 genotype is associated with worse myelin decrease and thus tau accumulation.

METHODS

To address our objectives, we repurposed florbetapir-PET as a marker of myelin in the white matter (WM) based on previous validation studies showing that beta-amyloid (Aβ) PET tracers bind to WM myelin. We assessed 43 Aβ-biomarker negative (Aβ-) cognitively normal participants and 108 Aβ+ participants within the AD spectrum with florbetapir-PET at baseline and longitudinal flortaucipir-PET as a measure of fibrillar tau (tau-PET) over ~ 2 years. In linear regression analyses, we tested florbetapir-PET in the whole WM and major fiber tracts as predictors of tau-PET accumulation in a priori defined regions of interest (ROIs) and fiber-tract projection areas. In mediation analyses we tested whether tau-PET accumulation mediates the effect of florbetapir-PET in the whole WM on cognition. Finally, we assessed the role of myelin alteration on the association between APOE and tau-PET accumulation.

RESULTS

Lower florbetapir-PET in the whole WM or at a given fiber tract was predictive of faster tau-PET accumulation in Braak stages or the connected grey matter areas in Aβ+ participants. Faster tau-PET accumulation in higher cortical brain areas mediated the association between a decrease in florbetapir-PET in the WM and a faster rate of decline in global cognition and episodic memory. APOE ε4 genotype was associated with a worse decrease in the whole WM florbetapir-PET and thus enhanced tau-PET accumulation.

CONCLUSION

Myelin alterations are associated in an APOE ε4 dependent manner with faster tau progression and cognitive decline, and may therefore play a role in the etiology of AD.

Deep learning for unsupervised domain adaptation in medical imaging: Recent advancements and future perspectives

Deep learning has demonstrated remarkable performance across various tasks in medical imaging. However, these approaches primarily focus on supervised learning, assuming that the training and testing data are drawn from the same distribution. Unfortunately, this assumption may not always hold true in practice. To address these issues, unsupervised domain adaptation (UDA) techniques have been developed to transfer knowledge from a labeled domain to a related but unlabeled domain. In recent years, significant advancements have been made in UDA, resulting in a wide range of methodologies, including feature alignment, image translation, self-supervision, and disentangled representation methods, among others. In this paper, we provide a comprehensive literature review of recent deep UDA approaches in medical imaging from a technical perspective. Specifically, we categorize current UDA research in medical imaging into six groups and further divide them into finer subcategories based on the different tasks they perform. We also discuss the respective datasets used in the studies to assess the divergence between the different domains. Finally, we discuss emerging areas and provide insights and discussions on future research directions to conclude this survey.

Individuals with Alzheimer's disease and low tau burden: Characteristics and implications

INTRODUCTION

Abnormal amyloid-beta (Aβ) and tau deposition define Alzheimer's Disease (AD), but non-elevated tau is relatively frequent in patients on the AD pathway.

METHODS

We examined characteristics and regional patterns of 397 Aβ+ unimpaired and impaired individuals with low tau (A+T-) in relation to their higher tau counterparts (A+T+).

RESULTS

Seventy-one percent of Aβ+ unimpaired and 42% of impaired Aβ+ individuals were categorized as A+T- based on global tau. In impaired individuals only, A+T- status was associated with older age, male sex, and greater cardiovascular risk. α-synuclein was linked to poorer cognition, particularly when tau was low. Tau burden was most frequently elevated in a common set of temporal regions regardless of T+/T- status.

DISCUSSION

Low tau is relatively common in patients on the AD pathway and is linked to comorbidities that contribute to impairment. These findings have implications for the selection of individuals for Aβ- and tau-modifying therapies.

Cognitively defined Alzheimer's dementia subgroups have distinct atrophy patterns

INTRODUCTION

We sought to determine structural magnetic resonance imaging (MRI) characteristics across subgroups defined based on relative cognitive domain impairments using data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and to compare cognitively defined to imaging-defined subgroups.

METHODS

We used data from 584 people with Alzheimer's disease (AD) (461 amyloid positive, 123 unknown amyloid status) and 118 amyloid-negative controls. We used voxel-based morphometry to compare gray matter volume (GMV) for each group compared to controls and to AD-Memory.

RESULTS

There was pronounced bilateral lower medial temporal lobe atrophy with relative cortical sparing for AD-Memory, lower left hemisphere GMV for AD-Language, anterior lower GMV for AD-Executive, and posterior lower GMV for AD-Visuospatial. Formal asymmetry comparisons showed substantially more asymmetry in the AD-Language group than any other group (p = 1.15 × 10-10 ). For overlap between imaging-defined and cognitively defined subgroups, AD-Memory matched up with an imaging-defined limbic predominant group.

DISCUSSION

MRI findings differ across cognitively defined AD subgroups.

Spatial transcriptomic patterns underlying amyloid-β and tau pathology are associated with cognitive dysfunction in Alzheimer's disease

Amyloid-β (Aβ) and tau proteins accumulate within distinct neuronal systems in Alzheimer's disease (AD). Although it is not clear why certain brain regions are more vulnerable to Aβ and tau pathologies than others, gene expression may play a role. We study the association between brain-wide gene expression profiles and regional vulnerability to Aβ (gene-to-Aβ associations) and tau (gene-to-tau associations) pathologies by leveraging two large independent AD cohorts. We identify AD susceptibility genes and gene modules in a gene co-expression network with expression profiles specifically related to regional vulnerability to Aβ and tau pathologies in AD. In addition, we identify distinct biochemical pathways associated with the gene-to-Aβ and the gene-to-tau associations. These findings may explain the discordance between regional Aβ and tau pathologies. Finally, we propose an analytic framework, linking the identified gene-to-pathology associations to cognitive dysfunction in AD at the individual level, suggesting potential clinical implication of the gene-to-pathology associations.

Sparse Deep Neural Network for Encoding and Decoding the Structural Connectome

Brain state classification by applying deep learning techniques on neuroimaging data has become a recent topic of research. However, unlike domains where the data is low dimensional or there are large number of available training samples, neuroimaging data is high dimensional and has few training samples. To tackle these issues, we present a sparse feedforward deep neural architecture for encoding and decoding the structural connectome of the human brain. We use a sparsely connected element-wise multiplication as the first hidden layer and a fixed transform layer as the output layer. The number of trainable parameters and the training time is significantly reduced compared to feedforward networks. We demonstrate superior performance of this architecture in encoding the structural connectome implicated in Alzheimer's disease (AD) and Parkinson's disease (PD) from DTI brain scans. For decoding, we propose recursive feature elimination (RFE) algorithm based on DeepLIFT, layer-wise relevance propagation (LRP), and Integrated Gradients (IG) algorithms to remove irrelevant features and thereby identify key biomarkers associated with AD and PD. We show that the proposed architecture reduces 45.1% and 47.1% of the trainable parameters compared to a feedforward DNN with an increase in accuracy by 2.6 % and 3.1% for cognitively normal (CN) vs AD and CN vs PD classification, respectively. We also show that the proposed RFE method leads to a further increase in accuracy by 2.1% and 4% for CN vs AD and CN vs PD classification, while removing approximately 90% to 95% irrelevant features. Furthermore, we argue that the biomarkers (i.e., key brain regions and connections) identified are consistent with previous literature. We show that relevancy score-based methods can yield high discriminative power and are suitable for brain decoding. We also show that the proposed approach led to a reduction in the number of trainable network parameters, an increase in classification accuracy, and a detection of brain connections and regions that were consistent with earlier studies.

Tau accumulation and its spatial progression across the Alzheimer's disease spectrum

The accumulation of tau abnormality in sporadic Alzheimer's disease is believed typically to follow neuropathologically defined Braak staging. Recent in-vivo PET evidence challenges this belief, however, as accumulation patterns for tau appear heterogeneous among individuals with varying clinical expressions of Alzheimer's disease. We, therefore, sought a better understanding of the spatial distribution of tau in the preclinical and clinical phases of sporadic Alzheimer's disease and its association with cognitive decline. Longitudinal tau-PET data (1370 scans) from 832 participants (463 cognitively unimpaired, 277 with mild cognitive impairment and 92 with Alzheimer's disease dementia) were obtained from the Alzheimer's Disease Neuroimaging Initiative. Among these, we defined thresholds of abnormal tau deposition in 70 brain regions from the Desikan atlas, and for each group of regions characteristic of Braak staging. We summed each scan's number of regions with abnormal tau deposition to form a spatial extent index. We then examined patterns of tau pathology cross-sectionally and longitudinally and assessed their heterogeneity. Finally, we compared our spatial extent index of tau uptake with a temporal meta-region of interest-a commonly used proxy of tau burden-assessing their association with cognitive scores and clinical progression. More than 80% of amyloid-beta positive participants across diagnostic groups followed typical Braak staging, both cross-sectionally and longitudinally. Within each Braak stage, however, the pattern of abnormality demonstrated significant heterogeneity such that the overlap of abnormal regions across participants averaged less than 50%, particularly in persons with mild cognitive impairment. Accumulation of tau progressed more rapidly among cognitively unimpaired and participants with mild cognitive impairment (1.2 newly abnormal regions per year) compared to participants with Alzheimer's disease dementia (less than 1 newly abnormal region per year). Comparing the association of tau pathology and cognitive performance our spatial extent index was superior to the temporal meta-region of interest for identifying associations with memory in cognitively unimpaired individuals and explained more variance for measures of executive function in patients with mild cognitive impairments and Alzheimer's disease dementia. Thus, while participants broadly followed Braak stages, significant individual regional heterogeneity of tau binding was observed at each clinical stage. Progression of the spatial extent of tau pathology appears to be fastest in cognitively unimpaired and persons with mild cognitive impairment. Exploring the spatial distribution of tau deposits throughout the entire brain may uncover further pathological variations and their correlation with cognitive impairments.

Deep Normative Tractometry for Identifying Joint White Matter Macro- and Micro-structural Abnormalities in Alzheimer's Disease

This study introduces the Deep Normative Tractometry (DNT) framework, that encodes the joint distribution of both macrostructural and microstructural profiles of the brain white matter tracts through a variational autoencoder (VAE). By training on data from healthy controls, DNT learns the normative distribution of tract data, and can delineate along-tract micro-and macro-structural abnormalities. Leveraging a large sample size via generative pre-training, we assess DNT's generalizability using transfer learning on data from an independent cohort acquired in India. Our findings demonstrate DNT's capacity to detect widespread diffusivity abnormalities along tracts in mild cognitive impairment and Alzheimer's disease, aligning closely with results from the Bundle Analytics (BUAN) tractometry pipeline. By incorporating tract geometry information, DNT may be able to distinguish disease-related abnormalities in anisotropy from tract macrostructure, and shows promise in enhancing fine-scale mapping and detection of white matter alterations in neurodegenerative conditions.

Combined Connectomics, MAPT Gene Expression, and Amyloid Deposition to Explain Regional Tau Deposition in Alzheimer Disease

OBJECTIVE

We aimed to test whether region-specific factors, including spatial expression patterns of the tau-encoding gene MAPT and regional levels of amyloid positron emission tomography (PET), enhance connectivity-based modeling of the spatial variability in tau-PET deposition in the Alzheimer disease (AD) spectrum.

METHODS

We included 685 participants (395 amyloid-positive participants within AD spectrum and 290 amyloid-negative controls) with tau-PET and amyloid-PET from 3 studies (Alzheimer's Disease Neuroimaging Initiative, 18 F-AV-1451-A05, and BioFINDER-1). Resting-state functional magnetic resonance imaging was obtained in healthy controls (n = 1,000) from the Human Connectome Project, and MAPT gene expression from the Allen Human Brain Atlas. Based on a brain-parcellation atlas superimposed onto all modalities, we obtained region of interest (ROI)-to-ROI functional connectivity, ROI-level PET values, and MAPT gene expression. In stepwise regression analyses, we tested connectivity, MAPT gene expression, and amyloid-PET as predictors of group-averaged and individual tau-PET ROI values in amyloid-positive participants.

RESULTS

Connectivity alone explained 21.8 to 39.2% (range across 3 studies) of the variance in tau-PET ROI values averaged across amyloid-positive participants. Stepwise addition of MAPT gene expression and amyloid-PET increased the proportion of explained variance to 30.2 to 46.0% and 45.0 to 49.9%, respectively. Similarly, for the prediction of patient-level tau-PET ROI values, combining all 3 predictors significantly improved the variability explained (mean adjusted R2 range across studies = 0.118-0.148, 0.156-0.196, and 0.251-0.333 for connectivity alone, connectivity plus MAPT expression, and all 3 modalities combined, respectively).

INTERPRETATION

Across 3 study samples, combining the functional connectome and molecular properties substantially enhanced the explanatory power compared to single modalities, providing a valuable tool to explain regional susceptibility to tau deposition in AD. ANN NEUROL 2024;95:274-287.

Growing importance of brain morphometry analysis in the clinical routine: The hidden impact of MR sequence parameters

Volumetric assessment based on structural MRI is increasingly recognized as an auxiliary tool to visual reading, also in examinations acquired in the clinical routine. However, MRI acquisition parameters can significantly influence these measures, which must be considered when interpreting the results on an individual patient level. This Technical Note shall demonstrate the problem. Using data from a dedicated experiment, we show the influence of two crucial sequence parameters on the GM/WM contrast and their impact on the measured volumes. A simulated contrast derived from acquisition parameters TI/TR may serve as surrogate and is highly correlated (r=0.96) with the measured contrast.

Cerebrospinal fluid level of proNGF as potential diagnostic biomarker in patients with frontotemporal dementia

Introduction

Frontotemporal dementia (FTD) is an extremely heterogeneous and complex neurodegenerative disease, exhibiting different phenotypes, genetic backgrounds, and pathological states. Due to these characteristics, and to the fact that clinical symptoms overlap with those of other neurodegenerative diseases or psychiatric disorders, the diagnosis based only on the clinical evaluation is very difficult. The currently used biomarkers help in the clinical diagnosis, but are insufficient and do not cover all the clinical needs.

Methods

By the means of a new immunoassay, we have measured and analyzed the proNGF levels in 43 cerebrospinal fluids (CSF) from FTD patients, and compared the results to those obtained in CSF from 84 Alzheimer's disease (AD), 15 subjective memory complaints (SMC) and 13 control subjects.

Results

A statistically significant difference between proNGF levels in FTD compared to AD, SMC and controls subjects was found. The statistical models reveal that proNGF determination increases the accuracy of FTD diagnosis, if added to the clinically validated CSF biomarkers.

Discussion

These results suggest that proNGF could be included in a panel of biomarkers to improve the FTD diagnosis.

The Alzheimer's Disease Neuroimaging Initiative in the era of Alzheimer's disease treatment: A review of ADNI studies from 2021 to 2022

The Alzheimer's Disease Neuroimaging Initiative (ADNI) aims to improve Alzheimer's disease (AD) clinical trials. Since 2006, ADNI has shared clinical, neuroimaging, and cognitive data, and biofluid samples. We used conventional search methods to identify 1459 publications from 2021 to 2022 using ADNI data/samples and reviewed 291 impactful studies. This review details how ADNI studies improved disease progression understanding and clinical trial efficiency. Advances in subject selection, detection of treatment effects, harmonization, and modeling improved clinical trials and plasma biomarkers like phosphorylated tau showed promise for clinical use. Biomarkers of amyloid beta, tau, neurodegeneration, inflammation, and others were prognostic with individualized prediction algorithms available online. Studies supported the amyloid cascade, emphasized the importance of neuroinflammation, and detailed widespread heterogeneity in disease, linked to genetic and vascular risk, co-pathologies, sex, and resilience. Biological subtypes were consistently observed. Generalizability of ADNI results is limited by lack of cohort diversity, an issue ADNI-4 aims to address by enrolling a diverse cohort.

A Pragmatic, Investigator-Driven Process for Disclosure of Amyloid PET Scan Results to ADNI-4 Research Participants

BACKGROUND

Prior studies of Alzheimer's disease (AD) biomarker disclosure have answered important questions about individuals' safety after learning and comprehending their amyloid PET results; however, these studies have typically employed highly structured disclosure protocols and focused on the psychological impact of disclosure (e.g., anxiety, depression, and suicidality) in homogeneous populations. More work is needed to develop flexible disclosure protocols and study outcomes in ethnoculturally representative samples.

METHODS

The Alzheimer's Disease Neuroimaging Initiative (ADNI) is formally incorporating amyloid PET disclosure into the newest protocol (ADNI-4). Participants across the cognitive spectrum who wish to know their amyloid PET results may learn them. The pragmatic disclosure process spans four timepoints: (1) a pre-disclosure visit, (2) the PET scan and its read, (3) a disclosure visit, and (4) a post-disclosure check-in. This process applies to all participants, with slight modifications to account for their cognitive status. In designing this process, special emphasis was placed on utilizing investigator discretion. Participant measures include perceived risk of dementia, purpose in life, and disclosure satisfaction. Investigator assessment of the disclosure visit (e.g., challenges encountered, topics discussed, etc.) is also included.

RESULTS

Data collection is ongoing. Results will allow for more robust characterization of the impact of learning amyloid PET results on individuals and describe the perspectives of investigators.

CONCLUSION

The pragmatic design of the disclosure process in ADNI-4 coupled with the novel participant and investigator data will inform future disclosure practices. This is especially important as disclosure of biomarker results expands in research and care.

Identifying individuals with non-Alzheimer's disease co-pathologies: A precision medicine approach to clinical trials in sporadic Alzheimer's disease

INTRODUCTION

Biomarkers remain mostly unavailable for non-Alzheimer's disease neuropathological changes (non-ADNC) such as transactive response DNA-binding protein 43 (TDP-43) proteinopathy, Lewy body disease (LBD), and cerebral amyloid angiopathy (CAA).

METHODS

A multilabel non-ADNC classifier using magnetic resonance imaging (MRI) signatures was developed for TDP-43, LBD, and CAA in an autopsy-confirmed cohort (N = 214).

RESULTS

A model using demographic, genetic, clinical, MRI, and ADNC variables (amyloid positive [Aβ+] and tau+) in autopsy-confirmed participants showed accuracies of 84% for TDP-43, 81% for LBD, and 81% to 93% for CAA, outperforming reference models without MRI and ADNC biomarkers. In an ADNI cohort (296 cognitively unimpaired, 401 mild cognitive impairment, 188 dementia), Aβ and tau explained 33% to 43% of variance in cognitive decline; imputed non-ADNC explained an additional 16% to 26%. Accounting for non-ADNC decreased the required sample size to detect a 30% effect on cognitive decline by up to 28%.

DISCUSSION

Our results lead to a better understanding of the factors that influence cognitive decline and may lead to improvements in AD clinical trial design.

Identifying the joint signature of brain atrophy and gene variant scores in Alzheimer's Disease

The joint modeling of genetic data and brain imaging information allows for determining the pathophysiological pathways of neurodegenerative diseases such as Alzheimer's disease (AD). This task has typically been approached using mass-univariate methods that rely on a complete set of Single Nucleotide Polymorphisms (SNPs) to assess their association with selected image-derived phenotypes (IDPs). However, such methods are prone to multiple comparisons bias and, most importantly, fail to account for potential cross-feature interactions, resulting in insufficient detection of significant associations. Ways to overcome these limitations while reducing the number of traits aim at conveying genetic information at the gene level and capturing the integrated genetic effects of a set of genetic variants, rather than looking at each SNP individually. Their associations with brain IDPs are still largely unexplored in the current literature, though they can uncover new potential genetic determinants for brain modulations in the AD continuum. In this work, we explored an explainable multivariate model to analyze the genetic basis of the grey matter modulations, relying on the AD Neuroimaging Initiative (ADNI) phase 3 dataset. Cortical thicknesses and subcortical volumes derived from T1-weighted Magnetic Resonance were considered to describe the imaging phenotypes. At the same time the genetic counterpart was represented by gene variant scores extracted by the Sequence Kernel Association Test (SKAT) filtering model. Moreover, transcriptomic analysis was carried on to assess the expression of the resulting genes in the main brain structures as a form of validation. Results highlighted meaningful genotype-phenotype interactionsas defined by three latent components showing a significant difference in the projection scores between patients and controls. Among the significant associations, the model highlighted EPHX1 and BCAS1 gene variant scores involved in neurodegenerative and myelination processes, hence relevant for AD. In particular, the first was associated with decreased subcortical volumes and the second with decreasedtemporal lobe thickness. Noteworthy, BCAS1 is particularly expressed in the dentate gyrus. Overall, the proposed approach allowed capturing genotype-phenotype interactions in a restricted study cohort that was confirmed by transcriptomic analysis, offering insights into the underlying mechanisms of neurodegeneration in AD in line with previous findings and suggesting new potential disease biomarkers.

Personalized Computational Causal Modeling of the Alzheimer Disease Biomarker Cascade

BACKGROUND

Mathematical models of complex diseases, such as Alzheimer's disease, have the potential to play a significant role in personalized medicine. Specifically, models can be personalized by fitting parameters with individual data for the purpose of discovering primary underlying disease drivers, predicting natural history, and assessing the effects of theoretical interventions. Previous work in causal/mechanistic modeling of Alzheimer's Disease progression has modeled the disease at the cellular level and on a short time scale, such as minutes to hours. No previous studies have addressed mechanistic modeling on a personalized level using clinically validated biomarkers in individual subjects.

OBJECTIVES

This study aimed to investigate the feasibility of personalizing a causal model of Alzheimer's Disease progression using longitudinal biomarker data.

DESIGN/SETTING/PARTICIPANTS/MEASUREMENTS

We chose the Alzheimer Disease Biomarker Cascade model, a widely-referenced hypothetical model of Alzheimer's Disease based on the amyloid cascade hypothesis, which we had previously implemented mathematically as a mechanistic model. We used available longitudinal demographic and serial biomarker data in over 800 subjects across the cognitive spectrum from the Alzheimer's Disease Neuroimaging Initiative. The data included participants that were cognitively normal, had mild cognitive impairment, or were diagnosed with dementia (probable Alzheimer's Disease). The model consisted of a sparse system of differential equations involving four measurable biomarkers based on cerebrospinal fluid proteins, imaging, and cognitive testing data.

RESULTS

Personalization of the Alzheimer Disease Biomarker Cascade model with individual serial biomarker data yielded fourteen personalized parameters in each subject reflecting physiologically meaningful characteristics. These included growth rates, latency values, and carrying capacities of the various biomarkers, most of which demonstrated significant differences across clinical diagnostic groups. The model fits to training data across the entire cohort had a root mean squared error (RMSE) of 0.09 (SD 0.081) on a variable scale between zero and one, and were robust, with over 90% of subjects showing an RMSE of < 0.2. Similarly, in a subset of subjects with data on all four biomarkers in at least one test set, performance was high on the test sets, with a mean RMSE of 0.15 (SD 0.117), with 80% of subjects demonstrating an RMSE < 0.2 in the estimation of future biomarker points. Cluster analysis of parameters revealed two distinct endophenotypic groups, with distinct biomarker profiles and disease trajectories.

CONCLUSION

Results support the feasibility of personalizing mechanistic models based on individual biomarker trajectories and suggest that this approach may be useful for reclassifying subjects on the Alzheimer's clinical spectrum. This computational modeling approach is not limited to the Alzheimer Disease Biomarker Cascade hypothesis, and can be applied to any mechanistic hypothesis of disease progression in the Alzheimer's field that can be monitored with biomarkers. Thus, it offers a computational platform to compare and validate various disease hypotheses, personalize individual biomarker trajectories and predict individual response to theoretical prevention and therapeutic intervention strategies.

Combining pathological and cognitive tests scores: A novel data analytics process to improve dementia prediction models1

BACKGROUND

The term 'dementia' covers a range of progressive brain diseases from which many elderly people suffer. Traditional cognitive and pathological tests are currently used to detect dementia, however, applications using Artificial Intelligence (AI) methods have recently shown improved results from improved detection accuracy and efficiency.

OBJECTIVE

This research paper investigates the efficacy of one type of data analytics called supervised learning to detect Alzheimer's disease (AD) - a common dementia condition.

METHODS

The aim is to evaluate cognitive tests and common biological markers (biomarkers) such as cerebrospinal fluid (CSF) to develop predictive classification systems for dementia detection.

RESULTS

A data analytics process has been proposed, implemented, and tested against real data obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI) repository.

CONCLUSION

The models showed good power in predicting AD levels, notably from specified cognitive tests' scores and tauopathy related features.

Precision Brain Morphometry Using Cluster Scanning

Measurement error limits the statistical power to detect group differences and longitudinal change in structural MRI morphometric measures (e.g., hippocampal volume, prefrontal thickness). Recent advances in scan acceleration enable extremely fast T1-weighted scans (~1 minute) to achieve morphometric errors that are close to the errors in longer traditional scans. As acceleration allows multiple scans to be acquired in rapid succession, it becomes possible to pool estimates to increase measurement precision, a strategy known as "cluster scanning." Here we explored brain morphometry using cluster scanning in a test-retest study of 40 individuals (12 younger adults, 18 cognitively unimpaired older adults, and 10 adults diagnosed with mild cognitive impairment or Alzheimer's Dementia). Morphometric errors from a single compressed sensing (CS) 1.0mm scan with 6x acceleration (CSx6) were, on average, 12% larger than a traditional scan using the Alzheimer's Disease Neuroimaging Initiative (ADNI) protocol. Pooled estimates from four clustered CSx6 acquisitions led to errors that were 34% smaller than ADNI despite having a shorter total acquisition time. Given a fixed amount of time, a gain in measurement precision can thus be achieved by acquiring multiple rapid scans instead of a single traditional scan. Errors were further reduced when estimates were pooled from eight CSx6 scans (51% smaller than ADNI). Neither pooling across a break nor pooling across multiple scan resolutions boosted this benefit. We discuss the potential of cluster scanning to improve morphometric precision, boost statistical power, and produce more sensitive disease progression biomarkers.

Causal functional connectivity in Alzheimer's disease computed from time series fMRI data

Functional connectivity between brain regions is known to be altered in Alzheimer's disease and promises to be a biomarker for early diagnosis. Several approaches for functional connectivity obtain an un-directed network representing stochastic associations (correlations) between brain regions. However, association does not necessarily imply causation. In contrast, Causal Functional Connectivity (CFC) is more informative, providing a directed network representing causal relationships between brain regions. In this paper, we obtained the causal functional connectome for the whole brain from resting-state functional magnetic resonance imaging (rs-fMRI) recordings of subjects from three clinical groups: cognitively normal, mild cognitive impairment, and Alzheimer's disease. We applied the recently developed Time-aware PC (TPC) algorithm to infer the causal functional connectome for the whole brain. TPC supports model-free estimation of whole brain CFC based on directed graphical modeling in a time series setting. We compared the CFC outcome of TPC with that of other related approaches in the literature. Then, we used the CFC outcomes of TPC and performed an exploratory analysis of the difference in strengths of CFC edges between Alzheimer's and cognitively normal groups, based on edge-wise p-values obtained by Welch's t-test. The brain regions thus identified are found to be in agreement with literature on brain regions impacted by Alzheimer's disease, published by researchers from clinical/medical institutions.

Atrophy of hippocampal subfields relates to memory decline during the pathological progression of Alzheimer's disease

Background

It has been well documented that atrophy of hippocampus and hippocampal subfields is closely linked to cognitive decline in normal aging and patients with mild cognitive impairment (MCI) and Alzheimer's disease (AD). However, evidence is still sparce regarding the atrophy of hippocampus and hippocampal subfields in normal aging adults who later developed MCI or AD.

Objective

To examine whether atrophy of hippocampus and hippocampal subfields has occurred in normal aging before a diagnosis of MCI or AD.

Methods

We analyzed structural magnetic resonance imaging (MRI) data of cognitively normal (CN, n = 144), MCI (n = 90), and AD (n = 145) participants obtained from the Alzheimer's Disease Neuroimaging Initiative. The CN participants were categorized into early dementia converters (CN-C) and non-converters (CN-NC) based on their scores of clinical dementia rating after an average of 36.2 months (range: 6-105 months). We extracted the whole hippocampus and hippocampal subfields for each participant using FreeSurfer, and analyzed the differences in volumes of hippocampus and hippocampal subfields between groups. We then examined the associations between volume of hippocampal subfields and delayed recall scores in each group separately.

Results

Hippocampus and most of the hippocampal subfields demonstrated significant atrophy during the progression of AD. The CN-C and CN-NC groups differed in the left hippocampus-amygdala transition area (HATA). Furthermore, the volume of presubiculum was significantly correlated with delayed recall scores in the CN-NC and AD groups, but not in the CN-C and MCI groups.

Conclusion

Hippocampal subfield atrophy (i.e., left HATA) had occurred in cognitively normal elderly individuals before clinical symptoms were recognized. Significant associations of presubiculum with delayed recall scores in the CN-NC and AD groups highlight the essential role of the hippocampal subfields in both early dementia detection and AD progression.

Functional network collapse in neurodegenerative disease

Cognitive and behavioral deficits in Alzheimer's disease (AD) and frontotemporal dementia (FTD) result from brain atrophy and altered functional connectivity. However, it is unclear how atrophy relates to functional connectivity disruptions across dementia subtypes and stages. We addressed this question using structural and functional MRI from 221 patients with AD (n=82), behavioral variant FTD (n=41), corticobasal syndrome (n=27), nonfluent (n=34) and semantic (n=37) variant primary progressive aphasia, and 100 cognitively normal individuals. Using partial least squares regression, we identified three principal structure-function components. The first component showed overall atrophy correlating with primary cortical hypo-connectivity and subcortical/association cortical hyper-connectivity. Components two and three linked focal syndrome-specific atrophy to peri-lesional hypo-connectivity and distal hyper-connectivity. Structural and functional component scores predicted global and domain-specific cognitive deficits. Anatomically, functional connectivity changes reflected alterations in specific brain activity gradients. Eigenmode analysis identified temporal phase and amplitude collapse as an explanation for atrophy-driven functional connectivity changes.

Genetic basis of anatomical asymmetry and aberrant dynamic functional networks in Alzheimer's disease

Genetic associations with macroscopic brain networks can provide insights into healthy and aberrant cortical connectivity in disease. However, associations specific to dynamic functional connectivity in Alzheimer's disease are still largely unexplored. Understanding the association between gene expression in the brain and functional networks may provide useful information about the molecular processes underlying variations in impaired brain function. Given the potential of dynamic functional connectivity to uncover brain states associated with Alzheimer's disease, it is interesting to ask: How does gene expression associated with Alzheimer's disease map onto the dynamic functional brain connectivity? If genetic variants associated with neurodegenerative processes involved in Alzheimer's disease are to be correlated with brain function, it is essential to generate such a map. Here, we investigate how the relation between gene expression in the brain and dynamic functional connectivity arises from nodal interactions, quantified by their role in network centrality (i.e. the drivers of the metastability), and the principal component of genetic co-expression across the brain. Our analyses include genetic variations associated with Alzheimer's disease and also genetic variants expressed within the cholinergic brain pathways. Our findings show that contrasts in metastability of functional networks between Alzheimer's and healthy individuals can in part be explained by the two combinations of genetic co-variations in the brain with the confidence interval between 72% and 92%. The highly central nodes, driving the brain aberrant metastable dynamics in Alzheimer's disease, highly correlate with the magnitude of variations from two combinations of genes expressed in the brain. These nodes include mainly the white matter, parietal and occipital brain regions, each of which (or their combinations) are involved in impaired cognitive function in Alzheimer's disease. In addition, our results provide evidence of the role of genetic associations across brain regions in asymmetric changes in ageing. We validated our findings on the same cohort using alternative brain parcellation methods. This work demonstrates how genetic variations underpin aberrant dynamic functional connectivity in Alzheimer's disease.

Profiling and predicting distinct tau progression patterns: An unsupervised data-driven approach to flortaucipir positron emission tomography

INTRODUCTION

How to detect patterns of greater tau burden and accumulation is still an open question.

METHODS

An unsupervised data-driven whole-brain pattern analysis of longitudinal tau positron emission tomography (PET) was used first to identify distinct tau accumulation profiles and then to build baseline models predictive of tau-accumulation type.

RESULTS

The data-driven analysis of longitudinal flortaucipir PET from studies done by the Alzheimer's Disease Neuroimaging Initiative, Avid Pharmaceuticals, and Harvard Aging Brain Study (N = 348 cognitively unimpaired, N = 188 mild cognitive impairment, N = 77 dementia), yielded three distinct flortaucipir-progression profiles: stable, moderate accumulator, and fast accumulator. Baseline flortaucipir levels, amyloid beta (Aβ) positivity, and clinical variables, identified moderate and fast accumulators with 81% and 95% positive predictive values, respectively. Screening for fast tau accumulation and Aβ positivity in early Alzheimer's disease, compared to Aβ positivity with variable tau progression profiles, required 46% to 77% lower sample size to achieve 80% power for 30% slowing of clinical decline.

DISCUSSION

Predicting tau progression with baseline imaging and clinical markers could allow screening of high-risk individuals most likely to benefit from a specific treatment regimen.

ApoE4 and Connectivity-Mediated Spreading of Tau Pathology at Lower Amyloid Levels

Importance

For the Alzheimer disease (AD) therapies to effectively attenuate clinical progression, it may be critical to intervene before the onset of amyloid-associated tau spreading, which drives neurodegeneration and cognitive decline. Time points at which amyloid-associated tau spreading accelerates may depend on individual risk factors, such as apolipoprotein E ε4 (ApoE4) carriership, which is linked to faster disease progression; however, the association of ApoE4 with amyloid-related tau spreading is unclear.

Objective

To assess if ApoE4 carriers show accelerated amyloid-related tau spreading and propose amyloid positron emission tomography (PET) thresholds at which tau spreading accelerates in ApoE4 carriers vs noncarriers.

Design, Setting, and Participants

This cohort study including combined ApoE genotyping, amyloid PET, and longitudinal tau PET from 2 independent samples: the Alzheimer's Disease Neuroimaging Initiative (ADNI; n = 237; collected from April 2015 to August 2022) and Avid-A05 (n = 130; collected from December 2013 to July 2017) with a mean (SD) tau PET follow-up time of 1.9 (0.96) years in ADNI and 1.4 (0.23) years in Avid-A05. ADNI is an observational multicenter Alzheimer disease neuroimaging initiative and Avid-A05 an observational clinical trial. Participants classified as cognitively normal (152 in ADNI and 77 in Avid-A05) or mildly cognitively impaired (107 in ADNI and 53 in Avid-A05) were selected based on ApoE genotyping, amyloid-PET, and longitudinal tau PET data availability. Participants with ApoE ε2/ε4 genotype or classified as having dementia were excluded. Resting-state functional magnetic resonance imaging connectivity templates were based on 42 healthy participants in ADNI.

Main Outcomes and Measures

Mediation of amyloid PET on the association between ApoE4 status and subsequent tau PET increase through Braak stage regions and interaction between ApoE4 status and amyloid PET with annual tau PET increase through Braak stage regions and connectivity-based spreading stages (tau epicenter connectivity ranked regions).

Results

The mean (SD) age was 73.9 (7.35) years among the 237 ADNI participants and 70.2 (9.7) years among the 130 Avid-A05 participants. A total of 107 individuals in ADNI (45.1%) and 45 in Avid-A05 (34.6%) were ApoE4 carriers. Across both samples, we found that higher amyloid PET-mediated ApoE4-related tau PET increased globally (ADNI b, 0.15; 95% CI, 0.05-0.28; P = .001 and Avid-A05 b, 0.33; 95% CI, 0.14-0.54; P < .001) and in earlier Braak regions. Further, we found a significant association between ApoE4 status by amyloid PET interaction and annual tau PET increases consistently through early Braak- and connectivity-based stages where amyloid-related tau accumulation was accelerated in ApoE4carriers vs noncarriers at lower centiloid thresholds, corrected for age and sex.

Conclusions and Relevance

The findings in this study indicate that amyloid-related tau accumulation was accelerated in ApoE4 carriers at lower amyloid levels, suggesting that ApoE4 may facilitate earlier amyloid-driven tau spreading across connected brain regions. Possible therapeutic implications might be further investigated to determine when best to prevent tau spreading and thus cognitive decline depending on ApoE4 status.

MERGE: A model for multi-input biomedical federated learning

Driven by the deep learning (DL) revolution, artificial intelligence (AI) has become a fundamental tool for many biomedical tasks, including analyzing and classifying diagnostic images. Imaging, however, is not the only source of information. Tabular data, such as personal and genomic data and blood test results, are routinely collected but rarely considered in DL pipelines. Nevertheless, DL requires large datasets that often must be pooled from different institutions, raising non-trivial privacy concerns. Federated learning (FL) is a cooperative learning paradigm that aims to address these issues by moving models instead of data across different institutions. Here, we present a federated multi-input architecture using images and tabular data as a methodology to enhance model performance while preserving data privacy. We evaluated it on two showcases: the prognosis of COVID-19 and patients' stratification in Alzheimer's disease, providing evidence of enhanced accuracy and F1 scores against single-input models and improved generalizability against non-federated models.

Biostatistical Estimation of Tau Threshold Hallmarks (BETTH) Algorithm for Human Tau PET Imaging Studies

A methodology for determining tau PET thresholds is needed to confidently detect early tau deposition. We compared multiple threshold-determining methods in participants who underwent either 18F-flortaucipir or 18F-MK-6240 PET scans. Methods: 18F-flortaucipir (n = 798) and 18F-MK-6240 (n = 216) scans were processed and sampled to obtain regional SUV ratios. Subsamples of the cohorts were based on participant diagnosis, age, amyloid-β status (positive or negative), and neurodegeneration status (positive or negative), creating older-adult (age ≥ 55 y) cognitively unimpaired (amyloid-β-negative, neurodegeneration-negative) and cognitively impaired (mild cognitive impairment/Alzheimer disease, amyloid-β-positive, neurodegeneration-positive) groups, and then were further subsampled via matching to reduce significant differences in diagnostic prevalence, age, and Mini-Mental State Examination score. We used the biostatistical estimation of tau threshold hallmarks (BETTH) algorithm to determine sensitivity and specificity in 6 composite regions. Results: Parametric double receiver operating characteristic analysis yielded the greatest joint sensitivity in 5 of the 6 regions, whereas hierarchic clustering, gaussian mixture modeling, and k-means clustering all yielded perfect joint specificity (2.00) in all regions. Conclusion: When 18F-flortaucipir and 18F-MK-6240 are used, Alzheimer disease-related tau status is best assessed using 2 thresholds, a sensitivity one based on parametric double receiver operating characteristic analysis and a specificity one based on gaussian mixture modeling, delimiting an uncertainty zone indicating participants who may require further evaluation.

MRAβ: A multimodal MRI-derived amyloid-β biomarker for Alzheimer's disease

Florbetapir 18 F (AV45), a highly sensitive and specific positron emission tomographic (PET) molecular biomarker binding to the amyloid-β of Alzheimer's disease (AD), is constrained by radiation and cost. We sought to combat it by combining multimodal magnetic resonance imaging (MRI) images and a collaborative generative adversarial networks model (CollaGAN) to develop a multimodal MRI-derived Amyloid-β (MRAβ) biomarker. We collected multimodal MRI and PET AV45 data of 380 qualified participants from the ADNI dataset and 64 subjects from OASIS3 dataset. A five-fold cross-validation CollaGAN were applied to generate MRAβ. In the ADNI dataset, we found MRAβ could characterize the subject-level AV45 spatial variations in both AD and mild cognitive impairment (MCI). Voxel-wise two-sample t-tests demonstrated amyloid-β depositions identified by MRAβ in AD and MCI were significantly higher than healthy controls (HCs) in widespread cortices (p < .05, corrected) and were much similar to those by AV45 (r > .92, p < .001). Moreover, a 3D ResNet classifier demonstrated that MRAβ was comparable to AV45 in discriminating AD from HC in both the ADNI and OASIS3 datasets, and in discriminate MCI from HC in ADNI. Finally, we found MRAβ could mimic cortical hyper-AV45 in HCs who later converted to MCI (r = .79, p < .001) and was comparable to AV45 in discriminating them from stable HC (p > .05). In summary, our work illustrates that MRAβ synthesized by multimodal MRI could mimic the cerebral amyloid-β depositions like AV45 and lends credence to the feasibility of advancing MRI toward molecular-explainable biomarkers.

Basal forebrain activity predicts functional degeneration in the entorhinal cortex in Alzheimer's disease

Recent models of Alzheimer's disease suggest the nucleus basalis of Meynert (NbM) as an early origin of structural degeneration followed by the entorhinal cortex (EC). However, the functional properties of NbM and EC regarding amyloid-β and hyperphosphorylated tau remain unclear. We analysed resting-state functional fMRI data with CSF assays from the Alzheimer's Disease Neuroimaging Initiative (n = 71) at baseline and 2 years later. At baseline, local activity, as quantified by fractional amplitude of low-frequency fluctuations, differentiated between normal and abnormal CSF groups in the NbM but not EC. Further, NbM activity linearly decreased as a function of CSF ratio, resembling the disease status. Finally, NbM activity predicted the annual percentage signal change in EC, but not the reverse, independent from CSF ratio. Our findings give novel insights into the pathogenesis of Alzheimer's disease by showing that local activity in NbM is affected by proteinopathology and predicts functional degeneration within the EC.

Decision making in clinical trials: Interim analyses, innovative design, and biomarkers

The efficient and accurate execution of clinical trials testing novel treatments for Alzheimer's disease (AD) is a critical component of the field's collective efforts to develop effective disease-modifying treatments for AD. The lengthy and heterogeneous nature of clinical progression in AD contributes to the challenges inherent in demonstrating a clinically meaningful benefit of any potential new AD therapy. The failure of many large and expensive clinical trials to date has prompted a focus on optimizing all aspects of decision making, to not only expedite the development of new treatments, but also maximize the value of the information that each clinical trial yields, so that all future clinical trials (including those that are negative) will contribute toward advancing the field. To address this important topic the Alzheimer's Association Research Roundtable convened December 1-2, 2020. The goals focused around identifying new directions and actionable steps to enhance clinical trial decision making in planned future studies.

Distinct neurophysiology during nonword repetition in logopenic and non-fluent variants of primary progressive aphasia

Overlapping clinical presentations in primary progressive aphasia (PPA) variants present challenges for diagnosis and understanding pathophysiology, particularly in the early stages of the disease when behavioral (speech) symptoms are not clearly evident. Divergent atrophy patterns (temporoparietal degeneration in logopenic variant lvPPA, frontal degeneration in nonfluent variant nfvPPA) can partially account for differential speech production errors in the two groups in the later stages of the disease. While the existing dogma states that neurodegeneration is the root cause of compromised behavior and cortical activity in PPA, the extent to which neurophysiological signatures of speech dysfunction manifest independent of their divergent atrophy patterns remain unknown. We test the hypothesis that nonword deficits in lvPPA and nfvPPA arise from distinct patterns of neural oscillations that are unrelated to atrophy. We use a novel structure-function imaging approach integrating magnetoencephalographic imaging of neural oscillations during a non-word repetition task with voxel-based morphometry-derived measures of gray matter volume to isolate neural oscillation abnormalities independent of atrophy. We find reduced beta band neural activity in left temporal regions associated with the late stages of auditory encoding unique to patients with lvPPA and reduced high-gamma neural activity over left frontal regions associated with the early stages of motor preparation in patients with nfvPPA. Neither of these patterns of reduced cortical oscillations was explained by cortical atrophy in our statistical model. These findings highlight the importance of structure-function imaging in revealing neurophysiological sequelae in early stages of dementia when neither structural atrophy nor behavioral deficits are clinically distinct.

Inflating 2D convolution weights for efficient generation of 3D medical images

BACKGROUND AND OBJECTIVE

The generation of three-dimensional (3D) medical images has great application potential since it takes into account the 3D anatomical structure. Two problems prevent effective training of a 3D medical generative model: (1) 3D medical images are expensive to acquire and annotate, resulting in an insufficient number of training images, and (2) a large number of parameters are involved in 3D convolution.

METHODS

We propose a novel GAN model called 3D Split&Shuffle-GAN. To address the 3D data scarcity issue, we first pre-train a two-dimensional (2D) GAN model using abundant image slices and inflate the 2D convolution weights to improve the initialization of the 3D GAN. Novel 3D network architectures are proposed for both the generator and discriminator of the GAN model to significantly reduce the number of parameters while maintaining the quality of image generation. Several weight inflation strategies and parameter-efficient 3D architectures are investigated.

RESULTS

Experiments on both heart (Stanford AIMI Coronary Calcium) and brain (Alzheimer's Disease Neuroimaging Initiative) datasets show that our method leads to improved 3D image generation quality (14.7 improvements on Frchet inception distance) with significantly fewer parameters (only 48.5% of the baseline method).

CONCLUSIONS

We built a parameter-efficient 3D medical image generation model. Due to the efficiency and effectiveness, it has the potential to generate high-quality 3D brain and heart images for real use cases.

Quantifying brain connectivity signatures by means of polyconnectomic scoring

A broad range of neuropsychiatric disorders are associated with alterations in macroscale brain circuitry and connectivity. Identifying consistent brain patterns underlying these disorders by means of structural and functional MRI has proven challenging, partly due to the vast number of tests required to examine the entire brain, which can lead to an increase in missed findings. In this study, we propose polyconnectomic score (PCS) as a metric designed to quantify the presence of disease-related brain connectivity signatures in connectomes. PCS summarizes evidence of brain patterns related to a phenotype across the entire landscape of brain connectivity into a subject-level score. We evaluated PCS across four brain disorders (autism spectrum disorder, schizophrenia, attention deficit hyperactivity disorder, and Alzheimer's disease) and 14 studies encompassing ~35,000 individuals. Our findings consistently show that patients exhibit significantly higher PCS compared to controls, with effect sizes that go beyond other single MRI metrics ([min, max]: Cohen's d = [0.30, 0.87], AUC = [0.58, 0.73]). We further demonstrate that PCS serves as a valuable tool for stratifying individuals, for example within the psychosis continuum, distinguishing patients with schizophrenia from their first-degree relatives (d = 0.42, p = 4 × 10-3, FDR-corrected), and first-degree relatives from healthy controls (d = 0.34, p = 0.034, FDR-corrected). We also show that PCS is useful to uncover associations between brain connectivity patterns related to neuropsychiatric disorders and mental health, psychosocial factors, and body measurements.

Global brain activity and its coupling with cerebrospinal fluid flow is related to tau pathology

Amyloid-β (Aβ) and tau deposition constitute Alzheimer's disease (AD) neuropathology. Cortical tau deposits first in the entorhinal cortex and hippocampus and then propagates to neocortex in an Aβ-dependent manner. Tau also tends to accumulate earlier in higher-order association cortex than in lower-order primary sensory-motor cortex. While previous research has examined the production and spread of tau, little attention has been paid to its clearance. Low-frequency (<0.1 Hz) global brain activity during the resting state is coupled with cerebrospinal fluid (CSF) flow and potentially reflects glymphatic clearance. Here we report that tau deposition in subjects with evaluated Aβ, accompanied by cortical thinning and cognitive decline, is strongly associated with decreased coupling between CSF flow and global brain activity. Substantial modulation of global brain activity is also manifested as propagating waves of brain activation between higher- and lower-order regions, resembling tau spreading. Together, the findings suggest an important role of resting-state global brain activity in AD tau pathology.

The relationship between learning slopes and Alzheimer's Disease biomarkers in cognitively unimpaired participants with and without subjective memory concerns

OBJECTIVE

Learning slopes represent serial acquisition of information during list-learning tasks. Although several calculations for learning slopes exist, the Learning Ratio (LR) has recently demonstrated the highest sensitivity toward changes in cognition and Alzheimer's disease (AD) biomarkers. However, investigation of learning slopes in cognitively unimpaired individuals with subjective memory concerns (SMC) has been limited. The current study examines the association of learning slopes to SMC, and the role of SMC in the relationship between learning slopes and AD biomarkers in cognitively unimpaired individuals.

METHOD

Data from 950 cognitively unimpaired participants from the Alzheimer's Disease Neuroimaging Initiative (aged 55 to 89) were used to calculate learning slope metrics. Learning slopes among those with and without SMC were compared with demographic correction, and the relationships of learning slopes with AD biomarkers of bilateral hippocampal volume and β-amyloid pathology were determined.

RESULTS

Learning slopes were consistently predictive of hippocampal atrophy and β-amyloid deposition. Results were heightened for LR relative to the other learning slopes. Additionally, interaction analyses revealed different associations between learning slopes and hippocampal volume as a function of SMC status.

CONCLUSIONS

Learning slopes appear to be sensitive to SMC and AD biomarkers, with SMC status influencing the relationship in cognitively unimpaired participants. These findings advance our knowledge of SMC, and suggest that LR - in particular - can be an important tool for the detection of AD pathology in both SMC and in AD clinical trials.

Sensitivity of memory subtests and learning slopes from the ADAS-Cog to distinguish along the continuum of the NIA-AA Research Framework for Alzheimer's Disease

Despite extensive use of the Alzheimer's Disease (AD) Assessment Scale - Cognitive Subscale (ADAS-Cog) in AD research, exploration of memory subtests or process scores from the measure has been limited. The current study sought to establish validity for the ADAS-Cog Word Recall Immediate and Delayed Memory subtests and learning slope scores by showing that they are sensitive to AD biomarker status. Word Recall subtest and learning slope scores were calculated for 441 participants from the Alzheimer's Disease Neuroimaging Initiative (aged 55 to 90). All participants were categorized using the NIA-AA Research Framework - based on PET-imaging of β-amyloid (A) and tau (T) deposition - as Normal AD Biomarkers (A-T-), Alzheimer's Pathologic Change (A + T-), or Alzheimer's disease (A + T+). Memory subtest and learning slope performances were compared between biomarker status groups, and with regard to how well they discriminated samples with (A + T+) and without (A-T-) biomarkers. Lower Word Recall memory subtest scores - and scores for a particular learning slope calculation, the Learning Ratio - were observed for the AD (A + T+) group than the other biomarker groups. Memory subtest and Learning Ratio scores further displayed fair to good receiver operator characteristics when differentiating those with and without AD biomarkers. When comparing across learning slopes, the Learning Ratio metric consistently outperformed others. ADAS-Cog memory subtests and the Learning Ratio score are sensitive to AD biomarker status along the continuum of the NIA-AA Research Framework, and the results offer criterion validity for use of these subtests and process scores as unique markers of memory capacity.