Alzheimer’s disease (AD) is a progressive neurodegenerative condition characterized by the development of cerebral amyloid plaques, neurofibrillary tangles, and disruption of largescale brain networks (LSBNs) that affects nearly seven million Americans (2024) Alzheimer’s Disease Facts and Figures (2024). Until recently, the mainstay treatments for this condition have included cholinesterase inhibitors and memantine, neither of which have been found to impact the progression of Alzheimer’s disease. The development of anti-amyloid monoclonal antibodies (AMAs) has resulted in novel treatments that remove cerebral amyloid while reducing progression in cognitive and functional decline (Dyck et al., 2023; Sims et al., 2023). While these treatments represent a significant milestone in AD management, they are limited by infusion related reactions in up to 26% of patients (Dyck et al., 2023) and amyloid-related imaging abnormalities ranging from 21.5–40% (Dyck et al., 2023; Salloway et al., 2022). Furthermore, these drugs are primarily designed to impact disease progression (Cummings, 2023) rather than directly restore dysfunctional large scale brain networks (LSBNs).

An-FDA cleared treatment for refractory depression (Janicak and Dokucu, 2015), repetitive transcranial magnetic stimulation (rTMS) has significant potential as a non-invasive therapeutic for Alzheimer’s disease and related disorder (ADRD) and presents minimal risk in terms of adverse side effects. Currently, TMS is also FDA-cleared for the treatment of migraine with aura (Dodick et al., 2010), obsessive compulsive disorder (Carmi et al., 2019), and nicotine use disorder (Zangen et al., 2021) and is being explored as treatment for other neurological disorders such as stroke (Zhu et al., 2024), and Parkinson’s disease (Yang et al., 2024). A proposed mechanism of TMS action is that persistent enhancement of cortical excitability results in long-term potentiation (Huang et al., 2005) and increased expression of brain derived neurotrophic factors (Phillips et al., 1991). Repetitive TMS at 10 or 20 Hz applied to the dorsolateral prefrontal cortex (DLPFC), which is typically targeted for depression has resulted in cognitive benefits, such as everyday memory, or global cognition (Marra et al., 2015; Li et al., 2024). Furthermore, clinical trials have been performed in mild–moderate AD, demonstrating that intensive and maintenance to the precuneus with rTMS slows cognitive and functional decline as measured by the Clinical Dementia Rating- Sum of Boxes (CDR-SB) over a 6-month period (Koch et al., 2022) or with iTBS to the left dorsolateral prefrontal cortex (DLPFC) over 1 year as measured by the MOCA (Wu et al., 2024).

One potential therapeutic approach to ADRD is to leverage the role of functional connectivity using resting state functional magnetic resonance imaging (fc-rs-fMRI) with TMS to provide a precision-medicine approach to dysfunctional LSBNs. Through prior fc-rs-fMRI studies, we determined that temporal area G dorsal (TGd), a region defined by the Human Connectome Project (Baker et al., 2018; Glasser et al., 2016), frequently showed functional connectivity anomalies (high functional anomaly burden) in both AD and DLB (Rosenbloom et al., 2021). In addition, this region has been shown to be implicated in language processing and semantic dementia (Briggs et al., 2018b). Thus, our prior work identified this region as a particular TMS target. To allow for a precision-medicine approach to each patient, we also were interested in applying stimulation to other anomalous parcellations identified with fc-rs-fMRI.

We performed a pilot study evaluating the effect of combining an fc-rs-fMRI guided approach with intermittent theta burst (iTBS) TMS in early-stage AD subjects. Our primary aim was to estimate the effect of fc-rs-fMRI -guided iTBS TMS on connectivity dysfunction within the right temporal area G dorsal (RTGd) in persons with early-stage AD. We were also interested in evaluating the effect of fc-rs-fMRI -guided TBS on two additional dysfunctional parcellations within the default modal network (DMN), central executive network (CEN) and salience network (SN). Finally, we evaluated whether this approach impacted cognitive performance on the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) and the Geriatric Depression Score (GDS) in persons with early-stage AD.

We performed a pilot study of fc-rs-fMRI -guided TMS to treat specific parcellations within dysfunctional LSBNs in 10 early-stage biomarker-confirmed AD subjects. Our clinical trial methodology was unique in that we utilized an fc-rs-fMRI -guided, personalized theta burst TMS treatment in a neurodegenerative disease process. This therapeutic approach has been successfully used to treat patients with refractory depression (Cole et al., 2020) but has not been adequately explored in persons with dementia. Furthermore, we identified novel HCP-defined parcellations uniquely impacted by AD that have not been previously described in the literature. The precuneus, a cortical hub within the DMN, is most consistently associated with dysfunctional DMN connectivity in AD as demonstrated both by brain FDG-PET (Strom et al., 2022) and fc-rs-fMRI imaging (Yokoi et al., 2018) and has served as a target in other TMS treatments studies in AD (Koch et al., 2022). Our study has revealed AD is a disease that extends beyond the precuneus, involving area right TGd, left 8Av and left 55b within the middle frontal gyrus, area 8BL, and right SFL. Specifically, area 8Av is considered to have effective connectivity with memory related posterior cingulate cortex regions (Leech and Sharp, 2014), a key node within DMN (Buckner et al., 2008) along with connections with anterior cingulate and parietal cortex. An understanding of AD at the level of the HCP parcellation may provide valuable insights into the clinical phenotype and future treatment of this condition (Ren et al., 2020).

This investigation of 10 early-stage AD participants failed to meet the primary outcome measure of demonstrating significant changes in LSBNs between baseline and 6 weeks. However, there were clinically significant improvements compared to baseline on the RBANS attention index score. TMS did not result in any improvements in other areas of cognition or in the total RBANs score, unlike the recent rTMS study which showing that targeted hippocampal-cortex connectivity improved memory domain for ADAS-Cog (Jung et al., 2024) or with iTBS in terms of delayed recall (Wu et al., 2020) in actively treated patients with AD. Due to experimental limitations, we did not have a comparison sham placebo group, which does not rule out the possibility that there may have been efficacy relative to a placebo (e.g., sham coil). For instance, findings of LSBN stability in the treatment group could be a favorable finding when compared to a placebo group in which LSBN functional connectivity is worsening.

Potential explanations for the improvement on the attention index score without any effect on LSBNs may be due to the stimulation of our targeted regions resulting in remote stimulation of other areas or LSBNS (e.g., dorsal and ventral attention networks) that were not included in our analysis (Cole et al., 2020). Certain treatment targets such as area 8Av with its effective connections with posterior cingulate cortex may have had an impact on improving attention (Hahn et al., 2007). In addition, we did not assess immediate post- TMS treatment effects on cognition or on fMRI in this study, which likely limited our ability to capture immediate effects of TMS treatments (Wu et al., 2020). Future studies are needed to better understand whether maintenance dosing is necessary to maintain favorable cognitive benefits in early-stage AD (Koch et al., 2022; Wu et al., 2024). It is crucial to account for learning effects associated with cognitive testing in any longitudinal study. In our study, to minimize this practice or learning effects we utilized different versions of the RBANS.

In the 10 early AD subjects evaluated, there were no significant adverse events associated with TMS, supporting this treatment as relatively low risk compared to the AMAs. Overall, our findings are consistent with what has been described in the TMS literature (Blumberger et al., 2018; Cole et al., 2020). The most common side effect involved jaw twitching when the TMS was targeted to area TGd, an adverse effect that limited our ability to fully dose certain subjects. Because of this frequent unpleasant effect, we will avoid this region in future treatment trials.

Strengths of our study included the enrollment of AD-biomarker confirmed subjects from a community neurology clinic. In addition, we piloted an innovative non-invasive neurostimulation treatment approach combining fc-rs-fMRI targeted TMS treatment to dysfunctional LSBNs in a progressive neurodegenerative disorder. To our knowledge, this is one of the first studies to utilize multimodal imaging and functional connectivity for personalized TMS in AD. This enhances the precision and personalization of TMS treatment offering greater accuracy than traditional scalp measurement targeting. Additionally, the study leveraged the Human Connectome Project (HCP), which provides high granularity in defining regions of interest (Glasser et al., 2016). For instance, the dorsolateral prefrontal cortex (DLPFC), a common TMS target, includes 13 distinct functional regions linked to multiple large-scale brain networks (Rosen et al., 2021). Furthermore, the imaging protocol can be easily added to a clinical MRI session, taking only an additional 15 min (Rosenbloom et al., 2021).

Study limitations relate to the pilot nature of this clinical trial and the small number of subjects with high education who also lack diversity, the absence of a sham-control group, and a relatively short treatment interval of 6 weeks, which may not have been adequate to capture immediate or delayed treatment responses. As mentioned previously, the jaw twitching related to the stimulation of area RTGd may have led to certain subjects receiving a subtherapeutic dose of this treatment. The TMS intervention in this study involved 2 research assistants per subject and took greater than 5 h to administer, which may impact translation to clinical practice. Finally, the precision-medicine approach to choosing parcellations based on associated functional connectivity anomalies resulted in an inconsistency of target regions treated between experimental subjects and reduced our power to detect differences for these regions. Moreover, there may have been instances where stimulation of one region may have activated an LSBN whereas stimulation of a second area may have led to an inhibitory response in that given network. For future trials, we hope to recruit a larger number of subjects from multiple sites, evaluate both an immediate and delayed neuropsychological outcome, and stimulate a single area of interest rather than targeting three different areas at a given time.

Fc-rs-fMRI guided TMS may have potential therapeutic benefits on cognitive performance in early-stage AD through the modulation of LSBNs. Relative to other AD therapeutics, this treatment was safe and well-tolerated in all participating subjects. Future randomized, sham-controlled clinical trials are necessary to effectively investigate the role of non-invasive neurostimulation as a potential adjunctive therapy in early-stage AD.