Broad repetitive transcranial magnetic stimulation (rTMS) of the precuneus in Alzheimer's disease: A rationale and study design

Abstract

INTRODUCTION

Brain network dysfunction, particularly within the default mode network (DMN), is an increasingly apparent contributor to the clinical progression of Alzheimer's disease (AD). Repetitive transcranial magnetic stimulation (rTMS) can target key DMN hubs, maintain signaling function, and delay or improve clinical outcomes in AD. Here, we present the rationale and design of a study using off‐the‐shelf equipment and the latest clinical evidence to expand on prior rTMS work and reduce participant burden in the process.

METHODS

We will conduct a two‐stage trial of large‐coil rTMS targeting the precuneus (a key hub in the DMN affected by AD) in 54 participants with mild to moderate Alzheimer's Clinical Syndrome focused primarily on determining tolerability and feasibility and secondarily focused on determining short‐term efficacy for memory. The first stage will involve 5 to 10 participants receiving open‐label active treatment to refine the protocol. The following second stage will consist of a 1:1 randomized, double‐blind, sham‐controlled clinical trial to study feasibility and tolerability while exploring target engagement and short‐term efficacy for memory. Participants will undergo 16 total rTMS brain stimulation sessions over the course of 5 weeks. A full course of open‐label active treatment will be offered as an extension to the sham group after unblinding. Outcomes will focus on completion rates and adverse events to demonstrate feasibility and tolerability. Further exploratory outcomes will include neuropsychological assessments, electroencephalography, neuroimaging, and blood biomarkers to demonstrate the feasibility of collection and explore preliminary changes in these measures.

RESULTS

We anticipate this treatment is feasible and tolerable and may show evidence of target engagement and clinical improvement.

DISCUSSION

Should we achieve expected positive outcomes in feasibility and tolerability, this will justify future work focusing on clear demonstrations of clinical efficacy and biomarker engagement, as well as enhancement of generalizability and scalability.

Highlights

• Induction‐to‐maintenance repetitive transcranial magnetic stimulation (rTMS) of the precuneus is a promising treatment for Alzheimer's disease (AD), though recent methods require intensive personalization.

• We propose here a trial design of precuneus rTMS in mild‐to‐early‐moderate AD dementia using exclusively off‐the‐shelf equipment and protocol modifications to reduce participant burden.

• Our two novel modifications from prior work are (1) using a larger rTMS coil, and (2) consolidating the induction phase of treatment.

• This trial focuses primarily on tolerability and feasibility while exploring clinical measures of efficacy and biomarkers of target engagement.

• Our trial is registered at ClinicalTrials.gov NCT06597942.

1. INTRODUCTION

Alzheimer's disease (AD) affects 10% of individuals over the age of 65, with an estimated 6 million patients in the United States and an anticipated 13.8 million by 2060. ¹ With the anticipated growth of AD on the horizon, there is an urgent need for the development of novel therapeutic treatments to battle AD and delay the disease progression. Despite this need, there are fewer clinical trials for AD in 2024 than in 2023. ² There have been multiple randomized controlled trials examining repetitive transcranial magnetic stimulation (rTMS) as a treatment for AD. ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ Mounting evidence suggests synaptic dysfunction leading to brain network dysregulation is a significant contributor to AD pathology. ⁶ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ Theoretically, rTMS may act through increasing synaptic plasticity in regions where plasticity has been damaged by amyloid and tau pathology, thereby ameliorating the impacts of this dysfunction. ³ , ⁶ , ⁹ , ¹⁰ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ In early AD, fluorodeoxyglucose positron emission tomography and functional magnetic resonance imaging (fMRI) studies show the most heavily affected regions tend to be the posterior cingulate, the retrosplenial cortex, the lateral posterior parietal cortex, and the precuneus (a notable hub of the default mode network [DMN]). ¹⁴ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ Previously, rTMS studies have primarily focused on stimulating the dorsolateral prefrontal cortex (DLPFC). ⁷ , ⁹ , ²⁶ While this is effective for the treatment of major depressive disorder (MDD), the results in AD have been mixed, with recent larger studies suggesting a lack of efficacy for rTMS to DLPFC for AD. ⁹ , ²⁶ , ²⁷ Additionally, many studies performed brief treatments with little post‐treatment follow‐up and examined rTMS as a short‐term treatment, rather than as an ongoing treatment for a progressive neurodegenerative disease. ⁸ , ²⁸

However, further explorations of the massive parameter space in rTMS have yielded encouraging results targeting different cortical areas. Yao et al. examined bilateral cerebellar Crus II stimulation in AD patients and found significant improvement with active rTMS over sham in multiple cognitive domains that were maintained 8 weeks post‐treatment. ⁵ They additionally discovered corresponding increases in functional connectivity among the stimulation sites, DLPFC, and medial frontal cortex, which correlated with improvement in cognitive measures (including multiple memory and recall tests). Other promising sites include the precuneus, most notably studied by Koch et al., as well as other parietal sites as in Jung et al. ³ , ⁴ , ⁸ , ²⁹ , ³⁰ , ³¹

The double‐blind sham‐controlled randomized controlled trial (RCT) of Koch et al. examined 20 Hz 100% motor threshold stimulation to the midline/bilateral precuneus in AD patients, with an induction to maintenance treatment design. ⁴ Their method of placement over the target of interest involved searching for maximal engagement of the DMN within the precuneus region through TMS‐evoked potentials (TEPs). ³ , ⁴ , ³² Participants underwent 2 weeks of daily treatment during the induction phase, followed by once‐weekly stimulation for a total of 24 weeks of maintenance treatment. Their goal was to assess whether they could arrest or delay cognitive decline in the active stimulation group as opposed to the sham group over a period of 6 months. They found preservation of multiple cognitive and functional domains as well as signs of neurophysiologic preservation (using electroencephalography [EEG] measures of cortical excitability and brain imaging) in the active group compared to sham. While there was not a post‐treatment follow‐up period, the maintenance phase of treatment was designed to serve as a longitudinal marker of ongoing treatment efficacy in place of treatment completion and follow‐up. Increased engagement of the DMN as measured using their proprietary TEP marker (also used for targeting purposes) and fMRI was associated with better outcomes. ⁴ , ³³

Jung et al. ³¹ performed a 4‐week treatment protocol in which AD patients received 20 sessions of rTMS using the same stimulation parameters as Koch et al. ⁴ but using fMRI connectivity to target a specific region in the left superior parietal lobe connected to the hippocampus. Their primary outcome was a change in cognitive and functional performance as assessed by the Alzheimer's Disease Assessment Scale 13‐item Cognitive Subscale (ADAS‐Cog13) at 8 weeks from baseline. Like Koch et al., ⁴ they obtained encouraging results, showing improvement across functional and cognitive performance. ³¹ They also noted statistically significant improvement in measures of global disease status in the active group compared to the sham, and these improvements seemed correlated with the strengthening of DMN functional connectivity. ³¹ These structural and functional connectivity changes noted by multiple research groups are at the very least consistent with symptomatic improvement in AD and may represent disease modification at the level of brain networks distinct from monoclonal antibody therapies with molecular targets. ⁴ , ¹⁰ , ³³ , ³⁴

While personalized targeting techniques indicate engagement of the DMN is necessary for treatment effects, which appear to primarily be a “re‐synchronization” of brain oscillations (particularly in the gamma frequency band on EEG), it also appears stimulation of adjacent cortical areas (or “off‐target” areas) does not diminish these effects. ⁴ , ³³ Though these prior works discuss stimulation of the precuneus, it is important to note that rTMS has a limited penetration depth relative to the depth of the precuneus. Therefore, positive results from different targeting methods in the same hub of the same brain network (i.e., TMS‐EEG–based targeting around the precuneus vs. fMRI targeting more laterally on the left) suggest the benefits of stimulation focused in the precuneus may be achieved through stimulating multiple different sites over a broad cortical area near this crucial hub of the DMN. In fact, it is possible stimulation of a broader area may amplify benefits beyond those seen with narrowed targeting. Additionally, the spreading of the induction phase over weeks of daily treatment greatly unduly increases the burden of treatment. Ongoing work in rTMS for psychiatric indications supports the use of accelerated rTMS protocols with multiple sessions per day and a brief waiting period (or inter‐session interval) between sessions; acceleration has the potential to shift the burden of the induction phase of treatment onto days instead of weeks. ³⁵ , ³⁶ , ³⁷

The broad goal of our study, entitled Protocol for Maintaining and Improving Status in Alzheimer's Disease (PROMIS‐AD), is to validate and expand on existing work of rTMS of the precuneus (or rather, superficial to the precuneus) while reducing participant burden and using commercially available equipment. In place of a 70 mm figure of eight coil, we will use a larger coil (which can be referred to as “deep rTMS,” or “broad rTMS,” or most accurately “large‐coil rTMS”), leveraging rTMS physics to stimulate more cortical areas linked to the DMN without requiring complex targeting techniques, simultaneously simplifying and enhancing the procedure. This can be done using any number of coils, including (but not limited to) the MagVenture Cool D‐B80 (MagVenture), BrainsWay H7 coil (BrainsWay), or the BrainsWay H4 coil. We also plan to use an accelerated induction phase approach of four sessions per day for 3 days instead of spreading it over 2 weeks with daily treatment; this will drastically reduce the time and transport burdens of treatment. ³⁵ , ³⁷ Accelerated protocols are increasingly common in rTMS for depression, particularly now that the most well‐known accelerated protocol, the Stanford Neuromodulation Therapy protocol, is US Food and Drug Administration (FDA) approved. With a dire need for more strategies to approach AD, we believe it is crucial to ask feasibility and therapeutic questions centered around optimizing rTMS in a scalable manner by stimulating a broader cortical surface area with an accelerated induction phase.

2. METHODS

2.1. Study design

This pilot study is a single‐site, two‐stage clinical feasibility trial (1—open‐label active, 2—randomized sham controlled), designed to primarily examine feasibility and tolerability while secondarily exploring the efficacy of our PROMIS‐AD large‐coil rTMS protocol (Figure 1). The first stage consists of open‐label active PROMIS‐AD treatment of 5 to 10 participants for the purpose of protocol and workflow refinement. After refinement, we plan to enroll and randomize 44 participants to active or sham treatment with a permuted block design stratified by baseline Mini‐Mental State Examination (MMSE) score (strata of 18–21; 22–26). This will allow for the assessment of differences in feasibility, tolerability, and efficacy between groups. At the end of treatment, participants will be unblinded, and those in the sham group will be offered an open‐label extension of a full active treatment course. This will aid recruitment and retention while allowing for within‐participant active to sham comparison.

FIGURE 1.

Study flow diagram. Aβ, amyloid beta; AD, Alzheimer's disease; APOE, apolipoprotein E; EEG, electroencephalography; fMRI, functional magnetic resonance imaging; MRI, magnetic resonance imaging; PROMIS, Protocol for Maintaining and Improving Status; Screening + Baseline, pre‐randomization assessment that may impact eligibility; TMS, transcranial magnetic stimulation; W00, post‐randomization baseline assessment before induction treatment; W01, week 1 follow‐up visits between induction and maintenance treatment; W05, week 5 end‐of‐treatment visits

2.2. Study endpoints

2.2.1. Feasibility and tolerability

The primary goal of this study is to demonstrate the tolerability and feasibility of the PROMIS‐AD protocol. This is the primary focus because although prior work has demonstrated tolerability and feasibility of rTMS in AD over both brief periods (≈ 4 weeks) and longer periods (≥ 6 months), ³ , ⁴ , ⁵ , ³¹ our novel modifications to simplify the approach (particularly the change of coil size) have the potential to impact treatment feasibility and tolerability. Therefore, to develop a rTMS protocol with simplicity and scalability at its core, we first focus on confirming the ability to perform it. The main endpoint for this assessment will be dropout/completion rates. We hypothesize the protocol will be well tolerated and logistically practical as demonstrated by no significant difference in dropout rates between active versus sham treatment groups and ≥ 80% of overall participants completing the study. We know that dropout rates can be elevated in AD trials, particularly those with multiple visits; completion rates in recent similar rTMS in AD trials have been ≥ 80%. ³ , ⁴ , ⁵ , ²⁶ , ³⁸ Therefore, an overall treatment completion rate of ≥ 80% will be considered acceptable. At each treatment visit, we will additionally assess tolerability, adverse events, and unanticipated urgent medical visits occurring between study visits.

2.2.2. Clinical/neuropsychological

Though the primary focus of this study is safety and feasibility, the randomized phase is designed to also allow for preliminary assessments of efficacy, with the main clinical outcome measure being an improvement in memory and cognitive function as measured by the Repeatable Battery for the Assessment of Neuropsychological Status Update (RBANS Update). Additional clinical outcome measures assessing cognitive status, functional status, neuropsychiatric symptoms, and study partner burden are described further in section 2.5.1 and Table 1. Short‐term clinical efficacy will be assessed with between‐group and within‐group comparisons over time using linear mixed‐effects models.

TABLE 1.

Schedule of measures and study events.

Measure name/study event	Baseline	Decision point	Week 0 (start induction)	Week 1 (end induction)	Weekly during maintenance	Week 5 (end of treatment)
Trial logistics
Informed consent	X
Preliminary eligibility	X
Fully eligibility assessment		X
Randomization		X
Induction rTMS treatment, 4x sessions/visit			X
Maintenance rTMS treatment, 1x session/visit					X
Assess blind			X			X
Break blind						X
Sham→open‐label active extension						X
Neuropsychological assessment
MMSE	X			X		X
RBANS	X			X		X
CDR	X					X
Focused history	X
NPI	X					X
NIHTCB	X					X
ADL + IADL	X					X
Other clinical assessments
GDS	X		X	X		X
NPI‐Q			X	X		X
ZBI			X	X		X
Routine check‐in Q			X		X
Biomarkers
APOE	X
Aβ, tau	X					X
PLR			X			X
sMRI + fMRI	X					X
EEG + TMS‐EEG			X			X

Notes: Measure names and meanings of abbreviations can be found in Section 2.5. Column names refer to timing based on both week number and study step: Baseline column indicates measures obtained during pre‐screening before comprehensive assessment or through scheduled visits between pre‐screen and the start of the induction treatment phase, decision point refers to the review of assessments to ensure participant eligibility prior to initiation of treatment, Week 0 (start induction) column indicates measures that will be obtained at the first initiation/induction treatment visit (within 2 weeks of completing baseline assessments), Week 1 (end induction) column indicates measures that will be obtained at a separate visit after the completion of the induction treatment phase but before the start of maintenance treatment, Weekly during maintenance indicates measures that will be obtained on a weekly basis (at every treatment) during the maintenance phase (including the final treatment), and Week 5 (end of treatment) refers to measures obtained during or after completion of the final maintenance phase treatment visit.

Abbreviations: Aβ, amyloid beta; ADL, Activities of Daily Living; APOE, apolipoprotein E; CDR, Clinical Dementia Rating; EEG, electroencephalography; fMRI, functional magnetic resonance imaging; GDS, Geriatric Depression Scale; iADL, instrumental Activities of Daily Living; MMSE, Mini‐Mental State Examination; NIHTCB, National Institutes of Health Toolbox Cognition Battery; NPI, Neuropsychiatric Inventory; NPI‐Q, Neuropsychiatric Inventory Questionnaire; PLR, pupillary light reflex; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; rTMS, repetitive transcranial magnetic stimulation; sMRI, structural magnetic resonance imaging; ZBI, Zarit Burden Interview.

2.2.3. Biological/neurophysiological

Another aim of this planned study is to assess both the viability of collecting and exploring longitudinal differences in the progression of biomarkers of AD in the active versus sham conditions. Many of these markers, including imaging (structural MRI [sMRI] and fMRI), multiple EEG markers (gamma connectivity, gamma, and beta TMS‐evoked oscillatory activity), and plasma markers, have already been studied as markers of severity ²⁵ , ³² , ³⁹ , ⁴⁰ , ⁴¹ and rTMS target engagement ⁴ , ⁶ , ³¹ , ³³ in AD. Others, such as pupillary reactivity (a marker of autonomic dysfunction, which may subtly appear with AD) and other EEG markers are shown to differ in those with AD, ³⁹ , ⁴² though changes over time or in response to intervention in AD have not yet been studied. Measures are further detailed in section 2.5.2 and Table 1.

We will study these endpoints through a four‐step research plan, as shown in Figure 1. Potential participants with mild‐to‐early‐moderate AD (MMSE score 18–26) will undergo baseline cognitive and functional assessment to establish the level of impairment. They also will undergo EEG, sMRI and fMRI, serum disease marker testing, and pupillometry measures to characterize baseline neurophysiologic measures. First, five to ten participants will receive an initial 12 sessions (three sessions/day over 4 days) of active rTMS applied to the precuneus using a deep rTMS coil, followed by focused repeat measurement of cognitive and functional status. Second, participants will continue weekly maintenance treatment with active or sham rTMS for four sessions, followed by repeat neuropsychological and neurophysiological baseline assessments to assess tolerability and allow for examination of pre‐to‐post active stimulation effects. Third, after this first protocol refinement stage, we will enroll 44 participants and randomize them in a 1:1 fashion to undergo the above protocol with either active or sham rTMS to compare the effects of active and sham stimulation. Fourth, after completion of their initial sham course and repeat baseline testing, participants who receive sham stimulation will have the option to receive a full course of active stimulation as an open‐label extension. This study combines our group's expertise in rTMS, neurophysiology, and neurobehavior to provide preliminary data of primarily the feasibility and secondarily of the efficacy of PROMIS‐AD and serve as the basis for a larger longitudinal trial.

2.3. Study intervention—broad precuneus rTMS

2.3.1. rTMS coil selection

One of the largest considerations in the design of this study is the coil selection. Keeping in mind the burden of targeting based on EEG or fMRI, we have determined our coil options to maximize the use of off‐the‐shelf commercially available equipment. We propose this study, or a similar study, can be performed with a wide range of commercially available equipment. Induced electric field models for a 70 mm figure‐of‐eight coil (MagVenture Cool‐B70), double‐cone coil (MagVenture Cool D‐B80), and multiple H‐coils (BrainsWay 104 system with H4 and H7 coils) were simulated using SimNIBS 4.1.0 (Figures 2, 3, and 4). Simulation coils were positioned over the precuneus Montreal Neurocognitive Institute (MNI) coordinates (x = 0, y = −65, z = 45; ³ , ⁴ H4 coil orientation reversed to align coil windings with posterior cortex) and manually adjusted as needed to approximate placement. Coil models were obtained from Drakaki et al. publicly available files ⁴³ and H‐coil models were provided by BrainsWay; all simulations used dI/dt = 1.0 A/s (Figures 2, 3, and 4). A qualitative examination of field distribution demonstrates the significant expansion of the stimulated cortical area, in terms of both breadth and depth, through the use of a coil other than the figure‐of‐eight coil. It is plausible these fields provide clinically significant stimulation over all the personalized target cortical areas described in Koch et al. ⁴ and Jung et al. ³¹ at once through placement over the neuroanatomical precuneus or nearby scalp electrode targets. ⁴³ Maximum output intensity with the same input apparently increases as well; however, direct clinical translation of stimulation intensity from one coil to another is not an established process. It is also notable that the use of the H4 coil may stimulate the motor cortex (or even further) at the anterior field edge under the lateral portions of the winding (Figures 2, 3, and 4), as well as the angular gyrus. The main options presented here (double cone, H4, and H7) each have their potential risks and benefits, and it is not clear which will be tolerated by participants. These options should all be explored, and we are not yet committed to a specific coil for this procedure. While multiple options may be explored in the open‐label stage, the randomized stage will use a single coil, and we intend to make this choice known prior to the initiation of that stage. We anticipate this decision will be made based on logistics and preliminary qualitative assessments of feasibility, tolerability, and acceptability during the open‐label stage. Regardless of our ultimate choice, we encourage others to further explore best‐fit‐for‐purpose coil choices.

FIGURE 2.

Posterior surface cortical views with coils/casings of induced electric field models of 70 mm figure‐of‐eight (A), 80 mm double cone (B), reversed BrainsWay H4 (C), and BrainsWay H7 (D) rTMS coils. Models were generated using SimNIBS 4.1.0 with placement over MNI precuneus coordinates; H4 required a manual adjustment to approximate placement. MagnE, magnitude of induced electric field (V/m); rTMS, repetitive transcranial magnetic stimulation;

FIGURE 3.

Sagittal views of induced electric field models of 70 mm figure‐of‐eight (A), 80 mm double cone (B), reversed BrainsWay H4 (C), and BrainsWay H7 (D) rTMS coils. Models generated using SimNIBS 4.1.0 with placement over MNI precuneus coordinates; H4 required manual adjustment to approximate placement. Use of example head files for models resulted lack of midline volume preservation and partial visualization of E‐fields on opposite surface cortex, which is not representative of midline stimulation. MagnE, magnitude of induced electric field (V/m); MNI, Montreal Neurological Institute; rTMS, repetitive transcranial magnetic stimulation

FIGURE 4.

Transverse view of induced electric field models of 70 mm figure‐of‐eight (A), 80 mm double cone (B), reversed BrainsWay H4 (C), and BrainsWay H7 (D) rTMS coils. Models were generated using SimNIBS 4.1.0 with placement over MNI precuneus coordinates; H4 required manual adjustment to approximate placement. MagnE, magnitude of induced electric field (V/m); MNI, Montreal Neurological Institute; rTMS, repetitive transcranial magnetic stimulation

2.3.2. rTMS parameters

Each active treatment session will consist of 1600 pulses of 20 Hz rTMS delivered in 40 pulse trains to the precuneus at 100% of depth‐corrected motor threshold (MT) with a 28 second intertrain interval (roughly 20 minutes per session) using one of the coils discussed above. Sham treatment will consist of the same settings, though using a sham coil. Notably, MT with the treatment coil will be performed in an analogous manner as with a traditional figure‐of‐eight coil, with the elicitation of MEPs over M1. Depth correction from M1 to the surface cortex at the site of treatment will be performed by adapting the APEX MT as described by Caulfield et al., ⁴⁴ generating the following formula:

$$\begin{array}{ccc}& & Precuneus\%MachineOutput\hfill \\ & & =\frac{MTE-field}{PrecuneusE-Field}\ast MT\%MachineOutput\hfill \end{array}$$

This correction will be used in place of the commonly used Stokes ⁴⁵ correction as the field‐gradient effects of double‐cone and H‐coils have not been characterized.

Placement over the precuneus would vary by coil. Using the H7 coil, placement will be performed using a novel technique we refer to as “offline neuronavigation.” A BrainsWay 104 cap with a grid system will be placed, and MNI coordinates for the precuneus will be marked on the cap using an MRI neuronavigation system; the authors use Rogue Research's Brainsight system. The grid system will then be centered over this location on the cap, and the nasion to the front edge of the cap distance will also be marked for reproducibility of placement at each session. The placement with the H7 will then be performed in its standard fashion, with a posterior pitch to align it over the grid centered on the MNI precuneus target marked on the cap. Using the H4 coil, placement would be similar, though the coil would first be flipped 180 degrees such that the winding sits over the posterior aspect of the scalp rather than the anterior. Placement over the precuneus using the Cool D‐B80 coil will be performed using traditional structural MRI neuronavigation on Rogue Research's Brainsight system, with MNI coordinates used for placement (x = 0, y = −65, z = 45). Distance from the stimulation site to 10 to 20 EEG coordinates for Pz will be measured.

2.3.3. rTMS administration and monitoring

Treatment will be applied in an induction‐to‐maintenance manner. After randomization, participants will undergo three induction phase visits over the course of three consecutive days, followed by four once‐weekly maintenance phase visits. Induction phase visits will consist of four treatment sessions per day with 60 minutes between treatment sessions, while maintenance visits will consist of a single session once per week, for a total of 16 rTMS treatment sessions (12 induction sessions over three visits, four maintenance sessions over four visits). Every day during the induction phase and every week during the maintenance phase, each participant's primary study partner will complete a brief survey identifying any changes in the participant's medication, any emergency department (ED)/urgent care visits, any hospitalizations, and any other unexpected or adverse events. Study staff will contact a participant's study partner should they not attend a visit unexpectedly to assess the reason for non‐attendance. Standard rTMS monitoring and safety precautions will be taken. ⁴⁶

2.4. Population, setting, pre‐screening, and informed consent

2.4.1. Recruitment

A total of 54 participants aged > 60 years with Alzheimer's Clinical Syndrome (including confirmed AD dementia) per the 2018 National Institute on Aging–Alzheimer's Association framework and mild‐to‐moderate severity based on Clinical Dementia Rating (CDR) scale score will be included. ⁴⁷ , ⁴⁸ Participants will be recruited as a local convenience sample through fliers, newspaper ads, web advertising, University of California Los Angeles (UCLA) psychiatry and neurology clinics, other UCLA‐affiliated clinics, and local community clinics.

The first study visit will consist of a review and completion of pre‐screening materials and informed consent. Those who do not meet the inclusion criteria assessed in this visit or meet exclusion criteria will not proceed with further visits. Pre‐screening will include MMSE administration (including screening score range 18–26), Geriatric Depression Scale (GDS) administration, and preliminary evaluation of inclusion and exclusion criteria (Table 2).

TABLE 2.

Summary of study inclusion and exclusion criteria.

Inclusion criteria
Alzheimer's Clinical Syndrome (including AD dementia) per 2018 NIA‐AA framework
Age 60–100
Screening MMSE 18–26
Screening GDS <6
Can complete informed consent process with surrogate
Have a known alternate surrogate decision maker
Have at least one routine caregiver
No recent or planned cognitive‐enhancing medication changes for 2 months prior to enrollment
No changes in use of psychotropic medication for 2 weeks prior to enrollment

Exclusion criteria
Unable to consent
Pregnant or potentially pregnant
Dementia is known not to be due to AD
SEVERE dementia (CDR > 2.0)
Substance use disorder not in sustained remission
Substance misuse within the past 6 months (excluding nicotine or caffeine)
Major neurologic disorder other than AD (especially seizure disorders)
Uncontrolled cardio/cerebrovascular disease
Any other major illness affecting cognition or ability to safely and meaningfully participate in the study
Non‐fluent English
Standard contraindication to TMS or MRI (metallic implants, severe claustrophobia, etc.)
Not TMS‐naive
Enrolled in another memory‐enhancement study
Cognitive‐enhancement medication dose adjustment within the past 2 months
Received monoclonal antibody treatment for AD
Taking any of the following medication classes within past 2 weeks prior to enrollment: Anticholinergics (e.g., tolterodine, benztropine) Sedating antihistamines (e.g., diphenhydramine) Drugs with significant anticholinergic or antihistaminic side effects (e.g., TCAs, mirtazapine). Benzodiazepines. While not a strict rule‐out, will be decided on a case‐by‐case basis Antiepileptics. While not a strict rule‐out, will be decided on a case‐by‐case basis Antipsychotics. While not a strict rule‐out, will be decided on a case‐by‐case basis

Note: See supporting information for detailed inclusion/exclusion criteria.

Abbreviations: AD, Alzheimer's disease; CDR, Clinical Dementia Rating; GDS, Geriatric Depression Scale; MMSE, Mini‐Mental State Examination; MRI, magnetic resonance imaging; NIA‐AA, National Institute on Aging–Alzheimer's Association; TCA, tricyclic antidepressant; TMS, transcranial magnetic stimulation.

We will exclude participants who have a history of major medical, neurologic, or psychiatric comorbidities; are on medications that may negatively alter cognition; or have had recent changes to any medication that may alter cognition (e.g., a recently adjusted dose of a cholinesterase inhibitor). We will not include pregnant women, institutionalized mentally disabled, prisoners, patients with primary psychotic disorders, depressed individuals (GDS ≥ 6), or others whose ability to give voluntary informed consent may be affected by a condition other than a form of mild cognitive impairment (MCI) or dementia. We will also exclude those who have a diagnosis of severe dementia, standard contraindications to rTMS or MRI, not TMS‐naïve, are enrolled in other memory‐enhancement studies, are not fluent in English, or have been treated with a monoclonal antibody for AD (e.g., aducanumab, lecanemab, and donanemab). See Table 2 and the supporting information for additional inclusion and exclusion criteria.

2.4.2. Informed consent

All participants will be required to attend the first visit with an identified surrogate decision maker/legally authorized representative (spouse, adult child, other family member, or other qualified individual), and enrollment will require the agreement of both the potential participant and their legally authorized representative to allow the potential participant's enrollment, as well as informed consent of one of these parties.

2.5. Study assessments

2.5.1. Clinical assessments

Clinical assessments are chosen to allow for multi‐domain assessment of cognition, functional assessment, neuropsychiatric assessment, study partner burden assessment, and global disease assessment. Cognitive assessments will include the RBANS Update, National Institutes of Health Toolbox Cognition Battery (NIHTCB), and MMSE, with the RBANS Update and MMSE comprising the abbreviated cognitive assessment administered between the induction and maintenance phases. While practice effects with repeated administration are a concern, we anticipate the use of the multiple available RBANS Update versions, focus on a dementia population rather than MCI, and stratification by baseline MMSE score will mitigate this risk. ⁴⁹ , ⁵⁰ , ⁵¹ The Katz Activities of Daily Living (ADL) and Lawton instrumental Activities of Daily Living (iADL) scales will assess the functional domain. The GDS, Neuropsychiatric Inventory (NPI), and NPI Questionnaire (NPI‐Q) will assess neuropsychiatric symptoms, and the Zarit Burden Index (ZBI) will assess the study partner burden. The CDR will evaluate global severity.

Blinding of participants, study partners, and study staff (including rationale for and confidence in the group assignment guess) will be assessed at the end of the first treatment visit and before unblinding during the final study visit. Blinding assessment early in treatment will be performed in addition to at the end of the study to detect changes in guesses consistent with functional unblinding (i.e., correctly guessing due to clinical changes or lack thereof). For every treatment visit, each participant and their study partner will complete a routine check‐in questionnaire (RCQ) identifying any changes in the participant's medication, any ED/urgent care visits, any hospitalizations, and any other unexpected or adverse events. The schedule of administration is outlined in Table 1.

2.5.2. Biomarker assessment

EEG, TMS‐EEG, and pupillometry assessment

Neurophysiology baseline measurements will include EEG, pupillometry, and TMS‐evoked EEG markers over the precuneus. First, pupillary light reflex (PLR) measurements will be collected using the Neuroptics PLR‐200 (Neuroptics) for each eye separately with an 800 ms pulse duration and 30 Hz sampling rate. The PLR protocol will last for 6 seconds per measurement. Resting EEG and TMS‐EEG (during treatment) metrics will be collected as well using a TMS‐compatible EEG system. Additional details regarding EEG and PLR metrics, acquisition, and processing can be found in the supporting information. The schedule of administration is outlined in Table 1.

MRI and blood‐based assessment

We will acquire T1‐weighted, T2‐weighted, and diffusion‐weighted structural images, as well as resting‐state fMRI (rs‐fMRI). The sMRI data will be used to identify T1‐based degeneration patterns consistent with AD and the precuneus location for each participant based on MNI coordinates: x = 0, y = −65, z = 45. ⁴ Imaging will be used to evaluate gray matter volume and structural tract connectivity (with diffusion tensor imaging) on sMRI in the DMN (primarily in the precuneus and other parietal regions) and temporal lobe from baseline to week 5. ⁴ , ²⁵ , ³³ rs‐fMRI will allow for evaluation of rs‐fMRI connectivity between the bilateral precuneus and posterior cingulate cortex, medial temporal lobe, entorhinal cortex, and hippocampus relative to sham from baseline to week 5. ⁶ , ¹⁹ , ³³ During these visits, in addition to imaging assessment, a peripheral blood draw will be performed for apolipoprotein E (APOE) testing and measurement of amyloid beta (Aβ)_42/40 ratio, plasma phosphorylated tau (p‐tau)217, and plasma p‐tau181. The schedule of administration is outlined in Table 1, and additional details are included in Supporting information.

2.6. Sham to active open‐label extension

At the end of the final blinded study visit, after the assessment of blinding, participants and study treatment staff will be unblinded to participant group assignment. After unblinding, a participant assigned to the sham group will be offered an open‐label extension with active treatment and repeat assessments.

2.7. Data management and sharing

Participants are given number codes for identification to maintain confidentiality. All data are maintained in confidentiality. The data may be used for teaching purposes or publication in journals. No identification of patients is made on any of these materials. A pre‐specified study staff member will assign codes to each patient with select research team members having access to the identifiers. Data and specimens will be labeled with the code the research team can link to identifying information. Data and specimens will be used for this study and possible future research. Health Insurance Portability and Accountability Act regulations will be carefully adhered to in all cases.

After completion of the study, de‐identified data may be made available to other institutions or secure online databases on reasonable request (e.g., Clinical Study Data Request, https://www.clinicalstudydatarequest.com).

3. STATISTICAL CONSIDERATIONS

3.1. Sample definitions

Analysis of this study will include four different samples. The first will be the intention to treat (ITT) sample, which will include all participants in both study arms of the randomized phase who are randomized. The second will be the per protocol blinded (PPB) sample, which will include all participants in both study arms of the randomized phase who complete the entirety of their blinded treatment course. The third will be the crossover sample, which will include all participants in the sham group who complete the blinded phase of treatment and go on to complete the entire open‐label active protocol. The fourth will be the completer active (CA) sample, which will include all participants who receive active treatment (blinded or open label, either study phase) and complete the entirety of the study.

3.2. Sample size and power

We primarily examine the safety and feasibility of PROMIS‐AD in a clinical population. Summary statistics (e.g., relative frequency) will be used to describe the completion rates and adverse events within and across the active and sham conditions. Confidence intervals for proportions will be used to understand the likely rates of study completion within a larger sample. With a proposed randomized sample size of 44 patients, and assuming a 90% completion rate, we would be able to estimate the completion rate with a precision (half‐width of 95% confidence interval) of ± 0.08. This would allow us to demonstrate that the population completion rate is not < 82%. This sample size is in roughly the same range as (or even larger than) prior studies focused on efficacy. ³ , ⁴ , ⁵ , ³¹

Additionally, Yao et al. ⁵ examined between‐condition differences in the MMSE over time showing means with standard deviations at baseline and after 4 weeks of: Active mean pre = 19.87 ± 4.31, post = 23.30 ± 3.57; Sham mean pre = 18.40 ± 5.06, post = 18.80 ± 4.80. Using these estimates and a complete case analytic sample size of 40 (20 per condition), an empirical power simulation was conducted assuming within‐person correlation over time of ρ = 0.7. Based on the results of 1000 simulations using a two‐sided alpha level of 0.05, we would have 81% power to minimally detect the between‐condition differences in cognitive change found in Yao et al. (≈ Cohen d = 0.81). For examining changes between condition in the crossover sample, we would require a 40% increase in the difference‐in‐difference effect size observed in Yao et al. to have 80% power, suggesting our crossover subsample will only have adequate power for large effect sizes (Cohen d = 1.2). Similarly, for physiologic changes, we expect we will have 80% power to minimally detect a standardized difference (Cohen d) of d = 0.56 for our between‐patient between‐condition analysis and d = 1.2 for our within‐patient between‐condition analysis in the crossover sample. The former estimate (d = 0.56) is consistent with findings from Koch et al., ³ , ⁴ which showed between‐condition differences in event‐related spectral perturbation of ≈ d = 0.76. Calculations shown above do not account for multiple comparisons, as our primary outcome in this study is feasibility, and the estimation of efficacy is exploratory. Should effect sizes in this study be as large as seen in prior work, we expect to maintain sufficient power to detect these meaningful differences in the planned 5‐week duration.

3.3. Inferential analyses

For the primary endpoint of completion rates, the statistical design includes both open‐label components and a two‐arm treatment design. This analysis will include descriptive examination and Fisher exact testing of between‐group differences will be used. For descriptive examination, the randomized phase will be examined for completion rate ≥ 80%. This analysis will be performed in the PPB and ITT samples. The Fisher exact method will be applied to the PPB and ITT samples to examine differences in completion rates between groups.

For secondary and exploratory endpoints (including cognitive assessment scores and biomarker data), continuous outcomes will be examined in a two‐arm treatment design, and analyses of these endpoints will use a mixed‐effects linear model to compare the endpoints over time between the active treatment arm and the sham arm. The within–between group differences will serve as the features of interest. The models will also include random patient effects accounting for repeated measurement. The treatment by time interaction term will give an overall estimate and inferences for the effects of therapy/group assignment. Additionally, these models will allow us to include other clinical/demographic factors (e.g., age, sex, APOE status, education level), medications, baseline disease measures, and process variables as appropriate. Consideration will be given for the use of a mild to moderate strength statistical correction, such as the false discovery rate. Categorical outcomes will be analyzed using the Fisher exact method to examine between‐group differences. They will be performed in the PPB and ITT subsets, and some may be performed in the crossover and CA subsets as well.

4. ETHICAL AND REGULATORY CONSIDERATIONS

This study is regulated by the FDA due to medical device use. This study is approved by UCLA Medical Institutional Review Board (IRB) 3 and conducted in accordance with the IRB and Declaration of Helsinki. This study will involve informed consent of parties anticipated to potentially lack decision‐making capacity, and appropriate steps to assess informed consent by an authorized party and full assent by all parties will be taken. This study is registered with ClinicalTrials.gov, NCT06597942. This article is reflective of protocol version 3.5, last updated on October 2, 2024.

5. DISCUSSION

An ever‐growing body of literature supports the use of non‐invasive brain stimulation (NIBS) techniques to treat a wide range of neurologic and psychiatric disorders. The general concept is that many disorders either originate as disorders of brain network function or are (at least in part) mediated by the development of brain network dysfunction. ⁵² , ⁵³ While the core pathology at play in AD appears to be our often‐discussed molecular targets of amyloid and tau, elements of the pathophysiology appear capitulated through the development of oscillatory dysfunction. ¹¹ , ¹³ This is what makes rTMS an appealing treatment modality—it acts at the level of networks with known input and results in changes at that network level, as well as upstream and downstream of the network. ¹⁵ , ¹⁶ , ⁵³ The evidence of this in rTMS for AD in brain structural volume and functional connectivity changes that are correlated with clinical outcomes is encouraging thus far, and the successes of these prior trials are why we anticipate we will demonstrate the feasibility and tolerability of our intervention in this study. ⁴ , ³¹ , ³³ We also hypothesize we will observe clinical benefits (primarily in memory) and biomarker changes with between‐group differences favoring the active treatment group should effect sizes rise to previously published levels. ³ , ⁴ , ⁵ , ³¹ Prior work suggests biomarkers demonstrating engagement of the DMN will correlate with clinical outcomes, and while we will study this, we also know this study may not be able to detect correlations between biomarkers and clinical outcomes.

This study is a rigorously designed randomized double‐blind sham‐controlled clinical trial with a clear plan to demonstrate feasibility and tolerability. Its novelty lies in its relative simplicity—it relies on the physics of rTMS and the latest clinical evidence to leverage off‐the‐shelf equipment for AD in a manner never before attempted.

This study design is not without anticipated limitations. This is designed to primarily demonstrate feasibility and tolerability, and a positive outcome is defined on the demonstration of these; it is plausible we will not detect significant clinical or biomarker effects. This study is also designed to use a single‐site convenience sample and requires study partners during treatment, both of which limit the ability to extrapolate results to other sites or broader populations. This study is designed to have unblinding occur for each individual participant upon their completion of the study, and this increases the risk of creating cues for functional unblinding of study staff. We anticipate this will at least in part be mitigated by (or at least detected by) the assessment of blinding at multiple time points, though some creation of cues over the course of the study may not be detectable. The design choice to unblind each participant along the way is a practical one (particularly for recruitment and retention), though not without its trade‐offs. To further mitigate this risk, we will also use different staff to perform stimulation in the blinded and unblinded stages of the study. Our analyses are focused on the ITT and PPB samples, and we know our crossover and CA samples will be more vulnerable to placebo response and effects. This study does not include biomarkers of AD as a requirement for inclusion as plasma biomarker samples will be run in batches instead of as collected, and therefore prospective AD etiologic specificity may be limited. Study treatment is also limited to a 5‐week duration; while periods as short as 2 weeks of rTMS have been shown sufficient for both clinical and biomarker benefit in prior work, ³ , ⁵ , ³¹ we recognize a longer duration of months or years is of greater interest.

Therefore, should this study show positive outcomes in feasibility and tolerability as anticipated, future work will focus on four key areas: (1) clear demonstration of clinical efficacy in a larger clinical trial; (2) clear evidence of biomarker engagement; (3) engagement of broader populations beyond convenience samples to enhance generalizability, diversity, justice, and equity; and (4) using multiple trial sites to further diversify the study population and demonstrate the scalability of this intervention.

AUTHOR CONTRIBUTIONS

Concept and design: Michael K. Leuchter, Hanadi A. Oughli, Kelly A. Durbin, Nicholas J. Jackson, David Elashoff, Timothy S. Chang, Juliana Corlier, Doan Ngo, Cole Matthews, Darice Wong, Brent L. Fogel, Gal Bitan, Andrew F. Leuchter, Keith Vossel, and Nanthia Suthana. Drafting of the manuscript: Michael K. Leuchter, Hanadi A. Oughli, and Kelly A. Durbin. Critical revision of the manuscript for important intellectual content: Michael K. Leuchter, Hanadi A. Oughli, Kelly A. Durbin, Nicholas J. Jackson, David Elashoff, Timothy S. Chang, Juliana Corlier, Doan Ngo, Cole Matthews, Darice Wong, Brent L. Fogel, Gal Bitan, Andrew F. Leuchter, Keith Vossel, and Nanthia Suthana. Statistical analysis planning: Michael K. Leuchter, Nicholas J. Jackson, and David Elashoff. Obtained funding: Michael K. Leuchter, Andrew F. Leuchter, and Nanthia Suthana. Administrative, technical, or material support: Nicholas J. Jackson, David Elashoff, Timothy S. Chang, Doan Ngo, Cole Matthews, Darice Wong, Brent L. Fogel, Gal Bitan, Andrew F. Leuchter, Keith Vossel, and Nanthia Suthana. Supervision: Nicholas J. Jackson, David Elashoff, Keith Vossel, and Nanthia Suthana.

CONFLICT OF INTEREST STATEMENT

A.F.L. received research support from the NIH, Neuroptics, Brainsway, Kernel, and MagVenture. He has served as a consultant to eFovea, Options MD, and Elevance Health. All other authors have no disclosures to report. Author disclosures are available in the supporting information.

CONSENT STATEMENT

All participants will be required to attend the informed consent visit with an identified surrogate decision maker, and enrollment will require agreement/assent of both the potential participant and their surrogate to allow the potential participant's enrollment, as well as informed consent of one of these parties.

Supporting information

Supporting information

Supporting information

Leuchter MK, Oughli HA, Durbin KA, et al. Broad repetitive transcranial magnetic stimulation (rTMS) of the precuneus in Alzheimer's disease: A rationale and study design. Alzheimer's Dement. 2025;11:e70043. 10.1002/trc2.70043