Transcranial Magnetic Stimulation in Alzheimer's disease-Data

Abstract

Neural hyper-excitability and network dysfunction are neurophysiological hallmarks of Alzheimer’s disease (AD) in animal studies, but their presence and clinical relevance in humans remain poorly understood. We introduce a novel perturbation-based approach combining transcranial magnetic stimulation and electroencephalography (TMS-EEG), alongside resting-state EEG (rsEEG), to investigate neurophysiological basis of default mode network (DMN) dysfunction in early AD. While rsEEG revealed global neural slowing and disrupted synchrony, these measures reflected widespread changes in brain neurophysiology without network-specific insights. In contrast, TMS-EEG identified network-specific local hyper-excitability in the parietal DMN and disrupted connectivity with frontal DMN regions, which uniquely predicted distinct cognitive impairments and mediated the link between structural brain integrity and cognition. Our findings provide critical insights into how network-specific neurophysiological disruptions contribute to AD-related cognitive dysfunction. Perturbation-based assessments hold promise as novel markers of early detection, disease progression, and target engagement for disease-modifying therapies aiming to restore abnormal neurophysiology in AD.

Background

Alzheimer's disease (AD) is marked by the early deposition of amyloid-beta-42 (Aβ) plaques followed by the rapid accumulation of phosphorylated tau (p-tau) in neurofibrillary tangles (1), and is associated with  neurophysiological changes before the onset of clinical symptoms (2). In animal models, Aβ plaques disrupt synaptic activity, causing an imbalance in inhibition and excitation, with neurons surrounding these plaques becoming abnormally hyperactive (3). P-tau independently suppresses neural activity (4), leading to synaptic loss, and progressive cerebral atrophy in brain regions critical for memory function. Advanced imaging methods like positron emission tomography (PET) have transformed the monitoring of these pathological markers in AD patients, enhancing our understanding of their relevance to disease progression. However, the neurophysiological manifestations of AD pathology and their clinical significance, remain poorly understood in humans.

Neural activity and connectivity dynamics in AD have been largely investigated using resting-state functional MRI (rsfMRI), with the default mode network (DMN) showing early disruptions in disease progression (5,6). Early rsfMRI studies reported decreased activity in the posterior cingulate cortex (PCC) and inferior parietal lobule (IPL) (7,8), findings that align with PET studies showing increased amyloid deposition (9) and reduced metabolic activity in these regions (10).  Recent research has identified the posterior DMN as the first region to exhibit abnormal connectivity patterns that correlate with cognitive decline (11) and neuropathology (12) across the AD spectrum. While this research has greatly advanced our understanding of the functional network organization in AD pathology, rsfMRI primarily captures correlational changes in blood oxygenation across the brain, providing indirect estimates of neuronal activity and network connectivity with limited capacity to detect synaptic disruptions at the individual level (13). High temporal resolution methods such as electroencephalography (EEG) and magnetoencephalography (MEG) are therefore preferred for measuring synchronized synaptic activity across large-scale brain networks. In AD patients, these neurophysiological modalities have consistently shown a global slowing of spectral power dynamics, with increased power in slower and decreased power in faster neural oscillations (14–17). MEG connectivity analyses generally show decreased alpha (8-12 Hz) over visual cortices and increased delta-theta (1-7 Hz) neural synchrony across the cortex (18,19). One limitation of these resting state based EEG/MEG measures is that they do not directly reflect AD-related neural hyper-excitability and network dysconnectivity. Spontaneous neural oscillations may not be optimal for localizing network-specific neurophysiological dynamics, as neural activity during the resting state arises from multiple sources (20) with a diffuse and widespread distribution across overlapping networks. Additionally, without external stimuli, spectral power estimates can only serve as surrogates for neural excitability.

Combining transcranial magnetic stimulation (TMS) with EEG offers a promising approach to address some limitations of both rsfMRI and rsEEG. Single pulses of TMS (spTMS) initially excite the axonal terminals of cortical pyramidal and interneurons (21), particularly at gyral crowns of the cortex, and evoke a series of high-frequency synaptic activations that directly represent local cortical excitability at the site of stimulation (22,23). These early activations are followed by trans-synaptic network responses, reflecting the connectivity profile of the stimulated brain region (24). EEG effectively captures synchronous post-synaptic potentials in pyramidal and interneurons with precise temporal resolution and has the highest sensitivity to neural activity at the superficial layers of the cortex (25), making it an ideal neurophysiological modality to index the activity of neural populations targeted by TMS. In AD, TMS can directly stimulate well-defined nodes of the DMN to assess local activation and connectivity abnormalities. However, most TMS research in AD has focused on the motor cortex (26–33), with only a few recent studies examining other brain regions (34–38). These recent TMS-EEG studies have typically performed electrode-level analyses at local TMS sites, offering a limited understanding of the functional network dynamics implicated in AD. Recently, we developed an MRI-guided TMS-EEG method that personalizes the topography of fMRI-based functional networks on individual brains, identifies network-specific TMS targets, and generates local activation and causal network connectivity maps at the individual level (39). Using this approach, we demonstrated that EEG responses to TMS provide an accurate 'fingerprint' of individual brain activation patterns (40), preferentially propagate through the structural connectivity of the stimulated network (41,42), characterize causal network connectivity dynamics with high reproducibility (43,44), and identify brain-behavior relationships not observable through resting-state recordings in healthy individuals (39).

In this study, we aimed to characterize the neurophysiological signatures of the DMN dysfunction in AD by utilizing both TMS-EEG and rsEEG measures, and evaluated their relationships to cognition. Our study is the first to comprehensively integrate rsEEG and TMS-EEG within the same cohort, offering a unique opportunity to compare these modalities and uncover network-specific abnormalities in AD. Using spTMS, we targeted individually defined regions within the left IPL (IPL-TMS), a DMN node particularly vulnerable to AD pathology, in biomarker-confirmed AD patients and age-matched cognitively normal controls (Table 1). Additionally, we stimulated the left primary motor cortex (M1-TMS) as a control site and included a sham TMS condition (Sham-TMS). RsEEG was recorded for five minutes during an eyes-closed condition to assess spontaneous neural dynamics.  We further examined how these measures relate to specific cognitive impairments, accounting for participant demographics (age, education) and structural brain integrity (cortical thickness) at TMS targeted brain regions. While we expect spectral slowing and reduced alpha neural synchrony in AD patients, we hypothesize that rsEEG measures will reflect global brain changes but will not be specific to the DMN. In contrast, TMS-EEG measures would reveal DMN-specific dysfunction in AD, characterized by hyper-excitability in the IPL and abnormal hyper- and hypo-connectivity patterns with non-stimulated DMN regions, providing insight into the network-specific mechanisms underlying decline in distinct cognitive functions.

Methods

Forty-two biomarker-positive early AD (18 female, Agemean=70.95 ± 7.63 years) and 40 healthy older adult control (HC) participants (19 female, Agemean=70.63 ± 6.41 years) were included in this study. Baseline characteristics are shown in Table 1. Data were drawn from two parallel ongoing studies (cohorts) both include AD and HC participant. In Cohort-1, all stimulation conditions (IPL, M1, and Sham) were performed within a single visit, with the order of M1 and IPL stimulation counterbalanced across participants. In Cohort-2, data were collected across two separate visits, one for M1 stimulation and another for IPL stimulation, with visit order randomized. There was a minimum one-week interval between visits in Cohort-2. Throughout the session, participants were comfortably seated in an adjustable chair. At the beginning of the TMS visit rsEEG recordings were performed first in eyes closed (EC) condition for five minutes. Following rsEEG recordings, the motor hotspot for eliciting MEPs in the right FDI muscle was determined by delivering single TMS pulses and moving the TMS coil in small incremental steps after two to three stimulations in each spot, over the hand region of left motor cortex with 45° rotation in relation to the parasagittal plane. The hotspot was defined as the region where single-pulse TMS elicits larger and more consistent MEPs in the FDI muscle, as compared to the APB muscle, with the minimum stimulation intensity. Resting motor threshold (RMT) was determined on the FDI hotspot as the minimum stimulation intensity eliciting at least five MEPs (≥50 μV) out of 10 pulses in the relaxed FDI using biphasic current waveforms. In compliance with the International Federation of Clinical Neurophysiology safety recommendations (77), participants were asked to wear earplugs during hotspot and RMT trials to protect their hearing, and to minimize external noise. TMS was administered with a thin layer of foam placed under the coil to minimize somatosensory contamination of the TMS-evoked EEG potentials. To minimize AEPs related to the TMS click, auditory white noise masking was used throughout the TMS stimulation. The intensity of noise masking is determined as the highest noise level participants could tolerate below 90db. Following determination of RMT, a total of 150 single TMS pulses were delivered to each stimulation target (DMN node target in the left inferior parietal lobule IPL-TMS and motor hot spot in the left motor cortex M1-TMS) at an intensity of 120% RMT with randomly jittered (3,000 to 5,000ms) intervals. Details of TMS targeting for IPL were provided in supplementary materials.  Sham-TMS was administered on the motor hot spot of the FDI muscle over left M1. An active/sham TMS coil (Cool-B65 A/P, MagVenture A/S, Farum, Denmark) was flipped to the placebo side and stimulation intensity was kept identical to actual TMS, but with induced currents on the opposite vertical direction to the targeted gyri (Supplementary Fig. 11). A 3D printed 3 cm spacer was attached to the placebo side (MagVenture A/S, Farum, Denmark) of the coil to further ensure the elimination of residual currents on the placebo side of the coil. Small current pulses between 2 and 4 mA and proportional to the intensity of actual TMS pulse were delivered over the left forehead, over the frontalis muscle, using surface electrodes (Ambu Neuroline 715 12/Pouch) to approximate somatosensory sensations arising from skin mechanoreceptors and scalp muscles during active-TMS condition

Data Description

Data Description

Dataset Storage and Organization

This study’s de-identified dataset will be shared in a BIDS-compliant structure to facilitate reuse and reproducibility. The release includes raw for TMS-EEG, resting-state EEG, structural MRI, and cognition, along with complete metadata and processing provenance.

Study scope and cohorts.
Two cohorts are included. The TMS-EEG cohort comprises Alzheimer’s disease and healthy control participants used for perturbation-based analyses (group sizes reported in the manuscript Results). The rsEEG cohort includes Alzheimer’s disease and healthy control participants with resting-state EEG and structural MRI. Cognitive assessments include ADAS-Cog, RAVLT-Total, Digit-Span Backward, and Animal Fluency.

Repository and access.
The dataset is organized under the Brain Imaging Data Structure (BIDS) and distributed as a versioned archive containing /rawdata,  and /code directories. All identifiers are removed and MR images are defaced prior to release. 

File formats

• EEG raw in EEGlab set format ; event timing in <font face="monospace">EEG.urevent structure</font>

• MRI NIfTI-1 .nii.gz plus JSON; FreeSurfer outputs in standard directory form.

• Cognition data TSV and CSV with comprehensive Cognitive battery results

Usage Notes

How to Access the Data

Data access is provided via the Brain Data Science Platform (BDSP). Complete data access instructions and security protocols are available on bdsp.io.

Requirements:

• Signed Data Use Agreement with strict terms and conditions
• Proof of completed CITI Training certification

Access Methods: After application approval, data can be accessed through:

• AWS Command Line Interface using AWS Access Keys
• Directory listing and file downloads
• Bulk folder copying to local systems
• Cloud-based data processing

BDSP provides flexible options for both local download and cloud-based analysis workflows.

Privacy, QC, and integrity

• All PHI is removed. MRIs are defaced and EEG files contain only de-identified IDs.

Ethics

The study conformed to the Declaration of Helsinki and was approved by the Institutional Review Board of Beth Israel Deaconess Medical Center under the following protocols; Ahead ‘2021P000229’, AD Supplement ‘2018P-000569’, BrightFocus ‘2020P-000889’, and Lead ‘2019P000091’. If an AD participant was interested in participating in the study but was determined to be unable to consent by the study MD, a legally authorized representative provided informed consent. RsEEG data were collected from all participants.

Conflicts of Interest

The authors declare they have no competing interest.