Non-linear associations of amyloid-β with resting-state functional networks and their cognitive relevance in a large community-based cohort of cognitively normal older adults

Basic Information

• Paper ID: 2510.12751
• Title: Non-linear associations of amyloid-β with resting-state functional networks and their cognitive relevance in a large community-based cohort of cognitively normal older adults
• Authors: Junjie Wu, Benjamin B Risk, Taylor A James, Nicholas Seyfried, David W Loring, Felicia C Goldstein, Allan I Levey, James J Lah, Deqiang Qiu
• Classification: q-bio.NC (Neuroscience)
• Affiliated Institution: Emory University
• Paper Type: Neuroimaging Research

Abstract

This study systematically investigated the relationship between 14 large-scale functional brain networks and Alzheimer's disease (AD) pathology in 968 cognitively normal community-dwelling older adults using resting-state functional magnetic resonance imaging (rs-fMRI), cerebrospinal fluid (CSF) biomarkers, and neuropsychological assessments. The study revealed inverted U-shaped non-linear associations between cerebrospinal fluid amyloid-β (Aβ) and functional connectivity in the precuneus network and ventral default mode network (DMN), but this pattern was not observed in the dorsal DMN. Higher connectivity in Aβ-related networks was associated with better performance in visual memory, visuospatial, and executive functions. This study provides important functional network biomarkers for early detection of AD.

Research Background and Motivation

Research Questions

Pathological changes in Alzheimer's disease can begin decades before cognitive symptoms appear. How to identify sensitive biomarkers during the asymptomatic phase is a key question in current research.

Importance of the Problem

1. Early intervention window: Therapeutic interventions are most effective before extensive neurodegeneration occurs
2. Sensitivity of functional connectivity: Functional MRI can capture subtle synaptic and network abnormalities before structural atrophy and cognitive decline
3. Clinical translational value: Provides neuroimaging biomarkers for early detection and prognostic assessment of AD

Limitations of Existing Research

1. Sample bias: Most studies recruit patients from memory clinics, often accompanied by subjective memory concerns, limiting the generalizability of results
2. Limited network scope: Primarily focus on single networks (e.g., DMN), with sample sizes typically less than 100 subjects
3. Linear model limitations: Insufficient consideration of biphasic trajectories between Aβ and functional connectivity (early high connectivity → later low connectivity)

Research Innovation

This study addresses the above research gaps through a large-scale community cohort, multi-network systematic analysis, and non-linear modeling.

Core Contributions

1. First systematic study in a large-scale community cohort of non-linear associations between 14 large-scale brain networks and AD pathology biomarkers
2. Discovery of network-specific susceptibility: The precuneus network and ventral DMN are more sensitive to early Aβ pathology
3. Establishment of functional-cognitive associations: Confirms the relationship between Aβ-related network connectivity and cognitive performance
4. Provision of early biomarkers: Provides important evidence for functional network biomarkers during the asymptomatic phase of AD

Methods

Study Design

• Study Type: Cross-sectional observational study
• Study Cohort: Emory Healthy Brain Study
• Sample Size: 968 cognitively normal older adults
• Age: Median 63.8 years (58.1-69.2 years)
• Gender Distribution: 66.2% female

Data Acquisition Methods

MRI Data Acquisition

• Equipment: Siemens Magnetom Prisma 3T scanner
• Structural imaging: MPRAGE sequence (TR/TE=2300/2.96ms, 1×1×1mm³)
• Functional imaging: Multiband accelerated EPI sequence (TR/TE=1890/30ms, 1.5×1.5×1.5mm³, 10 minutes)

CSF Biomarker Detection

• Collection method: Standardized lumbar puncture with 24G atraumatic Sprotte needle
• Detection platform: Roche Cobas e601 analyzer, Elecsys immunoassay platform
• Biomarkers: Aβ1-42, total tau (T-tau), phosphorylated tau181 (P-tau)

Neuropsychological Assessment

Includes 7 standardized tests covering memory, language, visuospatial, and executive function domains:

• Rey Complex Figure Test (RCFT)
• Judgment of Line Orientation (JoLO)
• Rey Auditory Verbal Learning Test (RAVLT)
• Letter and animal fluency tests
• Trail Making Tests A and B (TMTA/TMTB)

Data Analysis Methods

Functional Network Identification

1. Preprocessing: Motion correction, spatial normalization, smoothing (8mm FWHM), and band-pass filtering (0.01-0.1Hz) using the CONN toolbox
2. Network extraction: Spatially constrained independent component analysis (ICA) using the GIFT toolbox
3. Network definition: Identification of 14 large-scale networks based on prior templates

Statistical Analysis Strategy

1. Non-linear modeling: Multiple regression including quadratic terms of CSF biomarkers, controlling for age and sex
2. Multiple comparison correction: Benjamini-Hochberg false discovery rate (FDR) correction
3. Functional-cognitive associations: Multiple regression analysis of the relationship between functional connectivity and cognitive performance

Experimental Setup

Study Subject Inclusion Criteria

• Cognitively normal community-dwelling older adults
• Completed MRI scanning, CSF collection, and neuropsychological assessment
• No significant history of neuropsychiatric disease

Functional Network Definition

The 14 large-scale networks identified in the study include:

• Sensorimotor networks: Auditory, visual (primary and higher-order), sensorimotor networks
• Cognitive control networks: Left and right executive control networks, salience network (anterior and posterior)
• Default mode network: Dorsal and ventral DMN, precuneus network
• Other networks: Language, visuospatial, basal ganglia networks

Evaluation Metrics

• Functional connectivity strength: Average strength of within-network connectivity
• Cognitive performance: Standardized scores on neuropsychological tests
• Biomarker levels: CSF concentrations of Aβ, T-tau, P-tau

Experimental Results

Main Findings

Non-linear Associations Between Aβ and Functional Connectivity

1. Precuneus Network:
   • Linear term: β=0.126 (95% CI: 0.058-0.193, P=0.004)
   • Quadratic term: β=-0.071 (95% CI: -0.122 to -0.020, P=0.045)
   • Demonstrates significant inverted U-shaped association
2. Ventral DMN:
   • Linear term: β=0.079 (95% CI: 0.010-0.148, P=0.119)
   • Quadratic term: β=-0.089 (95% CI: -0.141 to -0.037, P=0.012)
   • Primarily manifested as significant quadratic term
3. Dorsal DMN: No significant linear or non-linear associations observed

Tau Protein and Functional Connectivity

• No significant associations between CSF tau (T-tau or P-tau) and functional connectivity in any of the 14 networks
• Suggests that tau pathology has limited impact on network connectivity during the cognitively normal phase

Associations Between Functional Connectivity and Cognitive Performance

Visual Memory (RCFT Immediate Recall)

Significantly positively correlated networks:

• Precuneus: β=0.168 (P<0.001)
• Dorsal DMN: β=0.168 (P<0.001)
• Posterior salience: β=0.141 (P=0.001)
• Ventral DMN: β=0.100 (P=0.043)

Visuospatial Ability (JoLO)

• Precuneus: β=0.113 (P=0.021)
• Dorsal DMN: β=0.099 (P=0.043)

Executive Function (TMTB)

• Precuneus: β=-0.100 (P=0.032)
• Ventral DMN: β=-0.091 (P=0.043)

Network-Specific Patterns

The study found that Aβ-related functional changes are primarily concentrated in:

• Precuneus network
• Ventral DMN
• Posterior salience network
• Dorsal DMN (significant only in cognitive associations)

Related Work

DMN and AD Pathology Research

Previous studies primarily focused on the DMN as a whole network. This study is the first to differentiate between dorsal and ventral DMN subsystems, revealing their differential susceptibility to Aβ pathology.

Biphasic Trajectories of Functional Connectivity

Previous small-sample studies proposed the hypothesis of early high connectivity → later low connectivity. This study validates this non-linear pattern in a large-scale community cohort.

Biomarker Specificity

Compared to tau protein, Aβ has a more prominent effect on functional networks during the cognitively normal phase, consistent with the AD pathological cascade hypothesis.

Conclusions and Discussion

Main Conclusions

1. Network-specific susceptibility: The precuneus network and ventral DMN are most sensitive to early Aβ pathology
2. Non-linear pathological trajectory: Aβ and functional connectivity show an inverted U-shaped association, supporting the compensation-decompensation hypothesis
3. Cognitive relevance: Aβ-related network connectivity is closely associated with visual memory, visuospatial, and executive functions
4. Early biomarker value: Functional networks can serve as sensitive neuroimaging biomarkers for early detection of AD

Clinical Significance

• Provides new neuroimaging biomarkers for early identification of AD
• Facilitates understanding of pathology-function relationships during the cognitively normal phase
• Provides reference for timing of therapeutic interventions

Limitations

1. Unbalanced gender distribution: Higher proportion of female participants (66.2%) may affect the representativeness of results
2. Cross-sectional design: Unable to track dynamic changes in functional connectivity over time
3. Tau biomarker limitations: CSF P-tau181 has higher correlation with amyloid PET, which may mask its independent effects

Future Directions

1. Longitudinal follow-up studies: Ongoing follow-up data from the Emory Healthy Brain Study will provide information on dynamic changes
2. Tau PET imaging: Combined tau PET imaging can more accurately assess the network effects of tau pathology
3. Gender-balanced studies: Future research should strive for more balanced gender distribution

In-Depth Evaluation

Strengths

1. Sample size advantage: 968 subjects represent the largest community cohort study in this field to date
2. Methodological rigor: Multimodal data integration, non-linear modeling, strict multiple comparison correction
3. Network systematicity: First systematic study of 14 large-scale networks, avoiding limitations of single-network studies
4. Clinical relevance: Establishes direct associations between functional connectivity and cognitive performance
5. Community representativeness: Avoids selection bias inherent in memory clinic cohorts

Limitations

1. Limited mechanistic explanation: Insufficient exploration of biological mechanisms underlying network-specific susceptibility
2. Missing genetic factors: Does not account for genetic risk factors such as APOE genotype
3. Limited cognitive test scope: Assessment of certain cognitive domains (e.g., attention) is relatively limited

Impact Assessment

1. Academic contribution: Provides important benchmarks for functional network research on early AD pathology
2. Clinical value: Establishes foundation for developing early diagnostic tools based on functional MRI
3. Reproducibility: Detailed methodological description and large-sample results demonstrate good reproducibility

Applicable Scenarios

• Early screening of high-risk populations for AD
• Functional endpoint measures in clinical trials
• Mechanistic research on cognitive aging
• Research on neural network plasticity

References

This study cites important literature on functional connectivity research in AD, including seminal work by Schultz et al. on biphasic trajectories of the DMN, and research by Sheline et al. on resting-state functional connectivity in the preclinical phase of AD. These works provide important theoretical foundation and methodological support for this study.

Overall Assessment: This is a high-quality neuroimaging study that provides important evidence for understanding early functional network changes in AD through a large-scale community cohort and rigorous multimodal analysis. The findings have significant clinical implications for early diagnosis and intervention in AD, representing an important advance in the field.