Prevent-AD

Description¶

The PREVENT-AD research group is now releasing data openly with the intention to contribute to the community’s growing understanding of AD pathogenesis.

The PREVENT-Alzheimer program wants to recruit 500 participants. Their contributions will be the key to finding strategies that can slow or reverse brain changes that may occur in older people who do not suffer from dementia. Recruitment is currently closed.

PREVENT-Alzheimer means PRe-symptomatic EValuation of Experimental or Novel Treatments for Alzheimer’s Disease. The PREVENT-AD program is the principal clinical research activity of the Centre for Studies on Prevention of Alzheimer’s Disease, or StoP-AD Centre. The goal of this program is to study memory and brain changes in healthy people over the age of 55. While some people in this age group begin to develop apparent memory problems, many more may have undetected brain changes that mark the very beginning of the disease long before the onset of symptoms.

With rapid advances in technology, such as high-resolution brain scanning methods or the precise examination of brain chemicals, we can now detect early evidence of brain changes. What’s more, we can likely track these changes over time. These brain changes are more likely to occur in individuals who have had either a parent or sibling affected by AD. This is why we require that our participants have a close family member affected by the disease.

The Dataset¶

The cohort selection process for the PREVENT-AD study, which was established in 2011, aimed to identify a population enriched for the risk of developing Alzheimer’s disease (AD).

The goal of the StoP-AD Centre, which developed the cohort, was to pursue innovative studies in the preclinical, or asymptomatic, phase of AD.

Target Population and Screening¶

The PREVENT-AD cohort included cognitively normal older individuals who were at-risk due to a strong family history of sporadic AD. This family history is known to increase the risk of AD dementia by 2- to 3-fold.

Between 2011 and 2017, the screening process involved the following steps and numbers:

• Screening: More than 1700 persons from the greater Montreal area were screened.

• On-site Eligibility Visit (EL00): 692 screened persons completed an on-site eligibility visit.

• Baseline Visit (BL00): 426 participants completed a baseline visit.

• Final Inclusion: 387 participants were included in the cohort for sustained investigation, or 386 participants met criteria for sustained investigation.

Inclusion Criteria¶

Participants had to meet the following criteria to be included in the cohort:

1. Family History: Self-reported parental or multiple-sibling (two or more) history of Alzheimer-like dementia.

   • For the diagnosis of “AD-like dementia,” investigators used either a compelling AD diagnosis from an experienced clinician or a structured questionnaire to establish memory or concentration issues severe enough to cause disability, with insidious onset and gradual progression.

   • A list notes that 8 participants had only 1 sibling affected with AD-like dementia.

2. Age: They needed to be 60 years of age or older. However, individuals aged 55–59 years were eligible if their age was within 15 years of the age of symptom onset of their youngest-affected first-degree relative.

3. Education and Study Partner: A minimum of 6 years of formal education was required, and a study partner needed to be available to provide information on cognitive status.

4. Protocol Compliance: Participants needed the ability and intention to participate in regular visits, sufficient fluency in French and/or English, and agreement for periodic donation of biofluids (blood and urine) and periodic multimodal assessments via MRI and Lumbar Puncture (LP). LP was initially optional but became mandatory in 2017 for participation of new individuals.

Exclusion Criteria and Cognitive Screening¶

The primary objective of the screening process was to ensure participants were cognitively unimpaired (CU).

1. Initial Screening: At the eligibility visit, the Montreal Cognitive Assessment (MoCA) and the Clinical Dementia Rating (CDR) were administered to rule out cognitive impairment.

2. Follow-up Assessment for Impairment: If an individual scored MoCA $\le 26$ or CDR $> 0$, an exhaustive neuropsychological evaluation was performed by a clinician. A very small fraction of these individuals were still enrolled if the evaluation confirmed normal cognition.

3. Baseline Confirmation: At the baseline visit, the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) was performed. Participants whose scores fell below normative values also needed to undergo an exhaustive neuropsychological evaluation to confirm normal cognition, otherwise they were excluded.

4. Exclusions During Phase 1 Follow-up (2011–2017): Generally, participants who developed Mild Cognitive Impairment (MCI) during follow-up visits were excluded from the study, with the exception of those enrolled in the naproxen clinical trial (INTREPAD). Ten participants were excluded between 2012 and 2017 due to cognitive decline. (Note: This exclusion practice changed in Phase 2, starting around 2016, where participants who developed MCI or dementia were no longer excluded).

5. Medical/Pharmacological Exclusions: Exclusion criteria also included known or identified cognitive disorders, current alcohol/barbiturate/benzodiazepine abuse, and use of certain medications such as acetyl-cholinesterase inhibitors, memantine, certain doses of Vitamin E or aspirin, opiates, NSAIDs, or certain anti-coagulants. Clinically significant liver/kidney disease, anemia, and uncontrolled hypertension were also exclusionary.

Final determination of eligibility was made by clinical consensus among study physicians, a research nurse, and a neuropsychologist.

Retrospective Data Sharing Consent¶

Of the 387 participants included in the cohort, 348 retrospectively agreed to have their data shared openly with the neuroscience community following a restructuring of the StoP-AD Centre operations in 2017. A total of 373 participants were reached during this re-consent process, and 93% of those reached agreed to the sharing.

INTREPID¶

The cohort divided between individuals enrolled in the observational cohort of the INTREPAD trial. The INTREPAD trial was interested in screening for a genomic familial link. This criteria changed the timing of some of the longitudinal observations.

The eligibility sessions collected T1w and and fMRI data, while the follow up enrollment session collected DWI. The observational cohorts had followups scheduled at 12, 24, 36, and 48 months. Additionally, the INTREPAD cohort had much more rapid follow ups from their baseline scan, at 3, 12, and 24 months as well as the 36 and 48 month observations.

Inclusion Criteria¶

Taken from Table 1.

• Inclusion criteria

• • Self-reported parental or multiple-sibling (2 or more*) history of Alzheimer-like dementia

• • Age 60 years or older (persons aged 55–59 years and < 15 years younger than their affected index relative were also eligible)

• • Minimum of 6 years of formal education

• • Study partner available to provide information on cognitive status

• • Sufficient fluency in spoken and written French and/or English

• • Ability and intention to participate in regular visits

• • Agreement for periodic donation of blood and urine samples

• • Agreement to participate in periodic multimodal assessments via MRI and LP for CSF collection (LP optional at first, then mandatory (in 2017) for participation)

• • Agreement to limit use of medicines as required by clinical trial protocols, if applicable

• • Provision of informed consent of the different protocols

• Exclusion criteria

• • Cognitive disorders - Known or identified during eligibility assessments (MoCA and CDR or exhaustive neuropsychological evaluation when needed)

• • Use of acetyl-cholinesterase inhibitors including tacrine, donepezil, rivastigmine, galantamine

• • Use of memantine or other approved prescription cognitive enhancer

• • Use of vitamin E at>600 i.u. / day or aspirin at > 325 mg / day

• • Use of opiates (oxycodone, hydrocodone, tramadol, meperidine, hydromorphone)

• • Use of NSAIDs or regular use of systemic or inhalation corticosteroids

• • Clinically significant hypertension (accepted if controlled medically), anemia, significant liver or kidney disease

• • Concurrent use of warfarin, ticlopidine, clopidrogel, or similar anti-coagulant

• • Current plasma Creatinine > 1.5 mg/dl (132 mmol/l)

• • Current alcohol, barbiturate or benzodiazepine abuse/dependence

Imaging Data¶

Stage one of the data consists of 349 subjects with up to 4 years of follow-up scans. The imaging data collected for the participants include:

• T1w Anatomical

• Resting State fMRI (rsfMRI)

• Task fMRI (tfMRI) - Memory Encoding and Retrieval

• Diffusion Weighted imaging (DWI)

• Arterial Spin Labeling (ASL)

dMRI Data Features¶

The Diffusion Magnetic Resonance Imaging (dMRI) data collection for the PREVENT-AD cohort evolved across two major phases of data acquisition (Phase 1: 2011–2017; Phase 2: Post-2019).

The changes in the dMRI protocol primarily reflect an upgrade in the MRI scanner.

Phase 1 dMRI Parameters (2011–2017, Stage 1)¶

During Phase 1 of data acquisition (Stage 1), the multi-modal MRI sequences, including dMRI, were harmonized with the Alzheimer’s Disease Neuroimaging Initiative (ADNI) protocol.

The specific dMRI sequence parameters available for this phase are detailed below:

Phase 2 dMRI Parameters (Post-2019)¶

MRI acquisition resumed in January 2019 after the original Siemens Tim Trio scanner was upgraded to a Prisma Fit. The MRI protocol was subsequently updated.

The key changes for diffusion MRI acquisition in Phase 2 included:

1. Scanner Upgrade: The scanner used was the upgraded Prisma Fit 3T.

2. Coil Usage: The 32-channel head coil was used exclusively.

3. Protocol Change: The diffusion MRI shifted from a single shell to a multi-shell protocol.

4. Data Type: The acquired data is described as multi-shell diffusion imaging.

While the sources confirm the use of multi-shell diffusion imaging in Phase 2, the exact numerical acquisition parameters (TR, TE, b-values, directions for the multi-shell sequence) are referenced to supplementary Table 4 in the sources, the details of which are not explicitly provided in the available text.

Derivative Measures¶

For both phases, the dMRI data was processed to derive quantitative measures of white matter properties. These analytic measures include diffusion tensor imaging (DTI) metrics derived from the multi-shell dMRI data, specifically:

• Diffusion properties of white matter tracts quantified using TractoFlow, RBXFlow, and Tractometry Flow.

• DTI measures in 52 white matter bundles reconstructed from whole-brain tractograms computed using the SCIL Tractoflow pipelines.

• Common dMRI analytical measures listed in the available data repository include Fractional Anisotropy (FA), Mean Diffusivity (MD), Axial Diffusivity (AD), and Radial Diffusivity (RD).

Clinical and Behavioral Data¶

The Repeatable Battery for Assessment of Neuropsychological Status (RBANS) was used to evaluate cognitive ability by estimating 5 Index scores (immediate memory, delayed memory, language, attention and visuospatial capacities) and a total score, consists of 12 subtests:

• list learning

• story learning

• figure copy

• line orientation

• picture naming

• semantic fluency

• digit span

• coding

• list recall

• list recognition

• story recall

• figure recall

The RBANS is a comprehensive, approximately 30-minute test designed to assess cognitive performance over time. It yields not only a total score (Global cognition), but also five specific Index scores that categorize cognitive function:

RBANS Subtests and Corresponding Cognitive Domains¶

The 12 subtests listed contribute to one of the five specific cognitive domains (Index scores) as follows:

Clinical Metrics¶

The PREVENT-AD cohort, which enrolls cognitively unimpaired older adults at risk for sporadic Alzheimer’s Disease (AD), collects a vast array of clinically relevant metrics and scales across cognitive function, clinical status, physical health, and psycho-affective measures.

Here are the key clinically relevant metrics and scales observed, along with the time points of their collection:

Cognitive Assessment Metrics¶

The primary cognitive data is collected longitudinally using standardized, detailed neuropsychological batteries:

Clinical Progression and Diagnosis¶

Clinical status classification, crucial for defining the preclinical cohort, is tracked:

• Clinical Progression/Status: The study tracks the transition of participants from Cognitively Unimpaired (CU) to Mild Cognitive Impairment (MCI) or Dementia. This classification is based on multidisciplinary consensus meetings, reviewing all available longitudinal cognitive information, specifically for participants who performed below norms on RBANS or RAVLT.

  • Time Points: Classification reviews are based on all available time points up to the latest follow-up (as of December 2023, 105 participants had been classified as having MCI at their latest follow-up time point). An additional CSV file reports the CDR score and latest clinical status (including dementia diagnosis reported by a physician or study partner).

Behavioral, Lifestyle, and Psycho-Affective Measures¶

Questionnaires and objective measures were introduced, primarily starting in 2016, to evaluate modifiable risk factors:

Clinical Health and Risk Factors¶

Other health and risk metrics are collected:

• CAIDE score (Cardiovascular Risk Factors, Aging, and Incidence of Dementia): Calculated at the Eligibility visit (EL00). The score uses baseline age, sex, education, blood pressure, BMI, cholesterol, physical activity, and $APOE \text{ } \epsilon 4$ status.

• McGill Pain Questionnaire - Short Version: Assesses Pain duration and intensity. Collected at multiple time points (Phase 2).

• Physical Health Metrics: Blood Pressure, pulse, and weight were updated annually. Blood test results (including glycosylated hemoglobin, thyroid stimulating hormone, Vitamin B12, and cholesterol levels) were obtained annually.

• Medication Use: Self-reported medication intake information and medical history were updated annually.

Context and Scoring in PREVENT-AD¶

• Longitudinal Assessment: RBANS was performed annually at baseline (BL00) and at subsequent follow-up visits (FU12, FU24, FU36, FU48, etc.).

• Version Rotation: To reduce practice effects during longitudinal assessment, the RBANS is available in four equivalent versions (A, B, C, D) in both French and English.

• Scoring Metrics: The RBANS data are scored following the manual. Index scores are typically age-adjusted with a mean of 100 and a standard deviation (SD) of 15, while subtest scores have a mean of 10 and a SD of 3.

• Age-Specific Norms: In PREVENT-AD, scores were calculated using both the standard age-adjusted method AND using norms specified exclusively for individuals aged 60–69 years. Both scoring methods are available in the registered data repository, allowing for the detection of potential decline with advancing age.

• Clinical Screening: Low performance on the RBANS (e.g., index score > 1 SD below the mean in two different cognitive domains) during follow-up visits could trigger a complete cognitive evaluation by a certified neuropsychologist to determine if the individual developed Mild Cognitive Impairment (MCI).

• Data Availability: A total of 2662 RBANS assessments were collected between 2012 and 2023, with 2493 assessments from the 348 participants who agreed to data sharing available to researchers. However, RBANS data collected during the COVID-19 pandemic (March 2020 and July 2021) were obtained virtually and are not shared because not all tasks were completed, and index scores could not be computed reliably. Additionally, standard Clinical measures like the MOCA and UPDRS are also collected.

Other scores and demographics are available as well.

Longitudinal Features¶

The PREVENT-AD dataset’s longitudinal neuroimaging acquisitions are divided into two distinct periods, Phase 1 (referred to as Stage 1) and Phase 2 (referred to as Stage 2), reflecting changes in funding, protocol, and available technologies.

Here are the details regarding the characteristics of the longitudinal neuroimaging acquisitions and the differences between Phase 1 and Phase 2:

Overview of Phases¶

Phase 1 (Stage 1) Neuroimaging Acquisitions (2011–2017)¶

Phase 1 neuroimaging primarily focused on Magnetic Resonance Imaging (MRI). Most MRI sequences (3T) were harmonized with the Alzheimer’s Disease Neuroimaging Initiative (ADNI) protocol to facilitate interoperability between cohorts.

Acquired Modalities (Standard Protocol): The acquisitions were performed using a Siemens Tim Trio 3T scanner, typically with a standard 12-channel coil. The sequences included:

• T1-weighted (T1w): MPRAGE.

• T2-weighted*.

• FLAIR (Fluid-attenuated inversion recovery) images.

• Diffusion MRI (DWI).

• Arterial Spin Labeling (ASL).

• Resting-state functional MRI (fMRI) (Resting State BOLD).

• Task fMRI: An episodic memory fMRI task was performed for most participants.

Protocol Change (Starting June 2016): For participants enrolled between June 2016 and November 2017 (n=48), the protocol changed slightly, using a 32-channel coil:

• The task fMRI was removed/replaced.

• MP2RAGE was added for quantitative T1 maps.

• A multi-echo gradient echo was added for T2* maps.

• A high in-plane resolution T2-weighted image (T2-weighted SPACE) was added to assess brain microstructure and segment hippocampal subfields.

Phase 2 (Stage 2) Neuroimaging Acquisitions (Post-2017)¶

The shift to Phase 2 operations in late 2017 necessitated changes, although core cognitive data collection continued. MRI scanning resumed in January 2019, utilizing the upgraded hardware and revised protocols.

MRI Protocol Updates: When MRI scanning resumed in 2019, the 3T Siemens Tim Trio scanner had been upgraded to a Prisma Fit, and the 32-channel head coil was used exclusively. Key differences included changes to existing sequences:

• fMRI changed from single- to multi-echo.

• Diffusion MRI changed from single shell to multi-shell.

• Sequences acquired include 1mm$^3$ 3D T1-weighted MPRAGE, 0.6mm$^3$ 3D T2-weighted SPACE, 3mm$^3$ resting-state multi-echo functional MRI, 1x1x3mm FLAIR, and multi-shell diffusion imaging.

• FLAIR data were only acquired in 2019-2020 after the upgrade.

New Modalities Introduced: Phase 2 expanded the neuroimaging scope significantly by adding Positron Emission Tomography (PET) and Magnetoencephalography (MEG).

1. Positron Emission Tomography (PET):

   • Amyloid-beta (A$\beta$) PET scans were collected (n=232 shared) using $[^{18}\text{F}]\text{NAV}4694$.

   • Tau PET scans were collected (n=229 shared) using $[^{18}\text{F}]\text{Flortaucipir}$.

   • Scans were acquired on a high-resolution research tomograph (HRRT) at the McConnell Brain Imaging Centre.

2. Magnetoencephalography (MEG):

   • Resting-state MEG scans were collected (n=114 shared), usually on the same day as an A$\beta$ or tau PET scan.

   • MEG data were collected using a whole-head CTF MEG system.

Improvements Over Time¶

The second phase (Phase 2) of data acquisition in the PREVENT-AD cohort, which began after November 2017, introduced several substantial methodological improvements and new modalities aimed at enhancing the quality, depth, and predictive value of the collected data.

The expected improvements to the quality of the data in PREVENT-AD Phase 2 stem from three main areas: enhanced neuroimaging acquisition techniques, the introduction of novel high-sensitivity molecular biomarkers, and extended longitudinal follow-up/clinical tracking.

I. Enhanced Neuroimaging Acquisition Quality¶

The greatest expected quality improvements come from hardware upgrades and modified imaging protocols, especially for MRI and functional imaging:

• MRI Scanner Upgrade and Coil Usage: The Siemens Tim Trio MRI scanner used in Phase 1 was upgraded to a Prisma Fit scanner when scanning resumed in 2019. Furthermore, the higher-channel 32-channel head coil was used exclusively, replacing the mixed use of 12-channel and 32-channel coils in Phase 1. This hardware upgrade generally yields superior signal-to-noise ratio (SNR) and image quality.

• Diffusion MRI (dMRI) Improvement: The dMRI protocol was updated from a single-shell sequence (used in Phase 1) to a multi-shell diffusion MRI sequence. Multi-shell data allows for fitting increasingly complex models that better summarize the orientation of water movement in tissue, leading to more advanced microstructure estimates beyond standard Diffusion Tensor Imaging (DTI), such as Neurite Orientation Dispersion and Density Imaging (NODDI) measures (Neurite Density Index [NDI], Orientation Dispersion Index [ODI], and Isotropic Volume Fraction [ISOVF]). The new multi-shell data has already led to key insights into brain microstructural alterations.

• Functional MRI (fMRI) Improvement: The resting-state fMRI acquisition was changed from a single-echo to a multi-echo sequence. Multi-echo fMRI is known to help improve image quality by facilitating the removal of non-BOLD noise components, leading to cleaner functional connectivity estimates.

• Introduction of PET Imaging (Amyloid and Tau): Phase 2 introduced standardized in-vivo measures of core Alzheimer’s pathologies through Positron Emission Tomography (PET) scanning.

  • High Resolution: The PET scans (A$\beta$ and tau) were acquired on a brain-dedicated Siemens/CTI high-resolution research tomograph (HRRT). This scanner provides a high spatial resolution of $2.4 \text{mm}$ at the center of the field of view, which is superior to most other scanners.

  • Standardized Quantification: The analytic measures provided include Standardized Uptake Value Ratios (SUVRs) across Desikan-Killiany (DK) atlas regions and, specifically for A$\beta$ PET, the standardized Global A$\beta$ PET Centiloid values.

• Inclusion of Neuroimaging Analytic Measures: Instead of only sharing raw data, the Phase 2 release innovated by including analytic neuroimaging measures (data derivatives), such as connectome matrices and volumes, computed by imaging experts. This eases data usage for researchers who may not specialize in neuroimaging processing. A concrete example of this is the sharing of five dominant patterns of brain atrophy quantified by an independent group and shared back to the repository.

II. Introduction of Novel High-Sensitivity Biomarkers¶

Phase 2 significantly expanded the range and sensitivity of fluid biomarkers collected, enhancing the ability to track the preclinical stage of the disease:

• Advanced Plasma Biomarkers: The new release incorporates novel longitudinal AD plasma biomarkers. This includes highly specific and sensitive markers quantified using in-house Single molecule array (Simoa) assays in collaboration with the University of Gothenburg:

  • Amyloid-beta (A$\beta$1-40, A$\beta$1-42).

  • Neurodegeneration and inflammation markers (GFAP, NfL).

  • Highly predictive Phosphorylated tau species ($\text{p-tau}181$, $\text{p-tau}231$, and $\text{p-tau}217$). Plasma $\text{p-tau}217$ has been shown to be highly predictive, with most PREVENT-AD participants having abnormal values progressing from cognitively unimpaired (CU) to Mild Cognitive Impairment (MCI) within 10 years.

• CSF Biomarker Expansion: New CSF data includes A$\beta$1-40 and A$\beta$1-42 measured using automated immunoassay and $\text{p-tau}217$ measured using an in-house Simoa assay.

• Comprehensive Proteomics: High-throughput aptamer-based proteomics assay data (SomaScan 7K panel) measuring approximately 6,600 unique protein targets in baseline CSF samples is now shared.

III. Extended Longitudinal Follow-up and Clinical Depth¶

Phase 2 data enhances quality by addressing the crucial need for longer follow-up in preclinical AD studies, which affects the interpretation of results:

• Extended Follow-up: Phase 2 includes six years of additional cognitive and MRI follow-ups, extending the data set beyond the Phase 1 period (2011–2017). The median follow-up is now over 8 years, with observations available for more than 12 years in total. The length of follow-up has been shown to “dramatically change the interpretation of the results,” for instance, increasing the percentage of A$\beta$ positive/tau negative ($\text{A+T-}$ ) individuals who progressed to MCI from 9% to 42% after 2.4 years of additional follow-up.

• Improved Clinical Progression Tracking: The systematic exclusion of participants who developed MCI was discontinued in Phase 2, meaning data is now available for participants who developed MCI or dementia (105 participants were classified as having MCI as of December 2023). This is critical for characterizing the trajectory from preclinical status to cognitive impairment.

• New Modalities (MEG and Actigraphy): The inclusion of Magnetoencephalography (MEG) is a novelty, as this neuroimaging modality is rarely acquired in large longitudinal AD studies. MEG data provides unique insights into neurophysiological alterations (e.g., spectral power density estimates across six frequency bands). Additionally, objective sleep measures collected via wrist Actiwatches were introduced, providing high-quality objective behavioral data on sleep characteristics.

Neuroimaging Analytic Measures¶

Phase 2 also emphasized the sharing of analytic neuroimaging measures (data derivatives) computed by imaging experts, provided in CSV files to ease data usage.

These analytical measures derived from all MRI modalities in Phase 2 include:

• Morphometric measurements (cortical thickness, volume, surface area) using FreeSurfer.

• Machine learning derived measures (brain parcellations, indices of aging and AD) using NiChart.

• Brain-wide functional connectivity matrices from resting-state fMRI.

• Diffusion tensor imaging (DTI) measures (diffusion properties of white matter tracts) using TractoFlow and related pipelines.

• White matter hyperintensity (WMH) volumes derived from T1w, T2w, and FLAIR images.

• Regional Cerebral Blood Flow (CBF) quantification from ASL sequences.

• Standardized Uptake Value Ratio (SUVR) images and values for A$\beta$ and tau PET scans.

• Relative spectral power density estimates across cortical regions from MEG data.

The overall transition from Phase 1 to Phase 2 represents a move from standardized annual multi-modal MRI and core clinical measures to a more flexible, investigator-driven data collection strategy that dramatically expanded the breadth of high-end biomarkers (PET, MEG, multi-shell dMRI) collected.

Possible Data Products¶

The PREVENT-AD cohort collects rich multimodal data, including extensive neuroimaging, genetics, and pathology biomarkers, enabling the construction of various multimodal connectomes beyond the standard functional connectivity derived from resting-state functional magnetic resonance imaging (fMRI).

The data available supports the construction of connectomes across three primary domains: Structural Connectivity, Neurophysiological Connectivity, and Pathology-based Networks.

1. Structural Connectomes (using Diffusion MRI)¶

The raw Diffusion MRI (dMRI) data, including multi-shell sequences collected in Phase 2, is crucial for modeling white matter pathways and building structural connectomes (often referred to as the structural “wiring diagram” of the brain).

Structural connectomes can be built by running tractography algorithms (such as those used in TractoFlow, which has already processed PREVENT-AD dMRI data) between parcellated gray matter regions. Possible resulting structural connectomes include:

• Streamline Count/Density Connectomes: These networks quantify the strength of the anatomical connection (edge) between two brain regions (nodes) based on the log$_{10}$ Streamline Count or Log$_{10}$ Streamline Density derived from tractography.

• Streamline Length Connectomes: These networks quantify connections based on the Log$_{10}$ Streamline Length between regions.

• Validated Structural Connectomes using LiFE: Connectome edges can be weighted using the Linear Fascicle Evaluation (LiFE) method. This method estimates a connectome by testing the “Strength of Evidence” for each edge based on how well the tractography streamlines fit the underlying diffusion data (using a virtual lesion method).

2. Multimodal Structural Connectomes (Combining dMRI Measures)¶

The edges of the structural connectome derived from dMRI can be mapped with microstructural properties to create multimodal networks that reflect tissue integrity, rather than just anatomical presence.

• DTI-derived Microstructural Connectomes: By averaging quantitative metrics derived from Diffusion Tensor Imaging (DTI) along reconstructed white matter pathways (tracts), connectomes can be built based on:

  • Fractional Anisotropy (FA): Connectome weights could reflect the mean or standard deviation of FA along the tract profile connecting two regions.

  • Axial Diffusivity (AD), Radial Diffusivity (RD), and Mean Diffusivity (MD): Although not explicitly listed as connectome outputs, DTI measures are already extracted for white matter tracts in PREVENT-AD.

• Advanced Microstructural Connectomes (e.g., NODDI): Given the multi-shell diffusion data available from Phase 2, more complex models like Neurite Orientation Dispersion and Density Imaging (NODDI) can be applied. Connectomes could then be built based on measures such as:

  • Neurite Density Index (NDI).

  • Orientation Dispersion Index (ODI).

  • Isotropic Volume Fraction (ISOVF).

3. Neurophysiological Connectomes (using MEG)¶

The PREVENT-AD dataset includes resting-state Magnetoencephalography (MEG) data, a modality that measures brain electrical activity directly.

• Frequency-Band Specific Connectomes: Unlike fMRI, which measures slow hemodynamic changes, MEG data can be used to construct connectomes reflecting instantaneous connectivity (e.g., phase synchrony or power coupling) across specific neural oscillation bands:

  • Delta (2–4 Hz).

  • Theta (5–7 Hz).

  • Alpha (8–12 Hz).

  • Beta (13–30 Hz).

  • Gamma1 (30–60 Hz).

  • Gamma2 (60–90 Hz).

4. Morphometric and Molecular Networks (Multimodal Integration)¶

Multimodal networks can be derived by correlating non-connectivity measures across anatomical regions defined by atlases like the Desikan-Killiany (DK) atlas.

• Morphometric Similarity Networks: Based on the volumetric measures already calculated (e.g., cortical thickness, cortical volume, and subcortical volume derived from T1w using FreeSurfer), researchers could build networks where the connectivity between two nodes is defined by the statistical similarity (covariance or correlation) of their morphological properties across participants.

• Pathology-Pathology Covariance Networks: The availability of regional standardized uptake value ratios (SUVRs) for both Amyloid-$A\beta$ PET and tau PET across DK atlas regions, allows for the construction of networks describing how $A\beta$ deposition in one region correlates with $A\beta$ or tau deposition in another region across subjects. This maps patterns of pathology accumulation and spread, which is highly relevant to Alzheimer’s disease progression.

CNS Projects¶

The goal is to create common processing derivatives for multiple projects to utilize.

Access¶

Access the dataset here.

Villeneuve, S., Poirier, J., Breitner, J. C., Tremblay‐Mercier, J., Remz, J., Raoult, J. M., ... & PREVENT‐AD Research Group. (2025). The PREVENT‐AD cohort: Accelerating Alzheimer’s disease research and treatment in Canada and beyond. Alzheimer’s & Dementia, 21(10), e70653.

Tremblay-Mercier, J., Madjar, C., Das, S., Binette, A. P., Dyke, S. O., Étienne, P., ... & PREVENT-AD Research Group. (2021). Open science datasets from PREVENT-AD, a longitudinal cohort of pre-symptomatic Alzheimer’s disease. NeuroImage: Clinical, 31, 102733.