A phase I trial of accelerated intermittent theta burst rTMS for amnestic MCI

Abstract

Background:

Emerging evidence suggests that repetitive transcranial magnetic stimulation (rTMS) enhances cognition in mild cognitive impairment (MCI). Accelerated intermittent theta burst (iTBS) rTMS protocols are promising as they substantially reduce burden by shortening the treatment course, but the safety, feasibility, and acceptability has not been established in MCI.

Methods:

Twenty-four older adults with amnestic MCI (aMCI) due to possible Alzheimer’s disease enrolled in a phase I trial of open-label accelerated iTBS to the left dorsolateral prefrontal cortex (8 stimulation sessions of 600 pulses of iTBS/day for 3 days). Participants rated common side effects during and after each session and retrospectively (at post-treatment and 4-week follow-up). They completed brain MRI (for safety assessments and electric field modeling), neuropsychiatric evaluations, and neuropsychological testing before and after treatment; a subset of measures was administered at follow-up.

Results:

Retention was high (95%) and there were no adverse neuroradiological, neuropsychiatric, or neurocognitive effects of treatment. Participants reported high acceptability, minimal side effects, and low desire to quit despite some rating the treatment as tiring. Electric field modeling data suggest that all participants received safe and therapeutic cortical stimulation intensities. We observed a significant, large effect-size (d = 0.98) improvement in fluid cognition using the NIH Toolbox Cognition Battery from pre- to post-treatment.

Conclusions:

Our findings support the safety, feasibility, and acceptability of accelerated iTBS in aMCI. In addition, we provide evidence of target engagement in the form of improved cognition following treatment. These promising results directly inform future trials aimed at optimizing treatment parameters.

INTRODUCTION

Mild Cognitive Impairment (MCI) is the pre-dementia stage in which individuals demonstrate the early cognitive and behavioral changes of neurodegenerative disease yet remain functionally intact. Alzheimer’s disease (AD) is a common cause of MCI that typically presents with primary deficits in episodic memory [1], referred to as amnestic MCI (aMCI). While most current AD clinical trials focus on disease-modifying pharmacologic monotherapies [2], such approaches may not fully address the multifactorial pathological [3] and syndromal [4,5] contributors to dementia onset and progression which already manifest in the MCI stage. Thus, there is a need for alternative interventions for the secondary prevention of dementia that address the cognitive and neuropsychiatric symptoms that commonly co-occur in MCI [6]. Repetitive transcranial magnetic stimulation (rTMS) is a promising candidate as it can be used to modulate the function of brain regions that are implicated in both the cognitive and affective symptoms of MCI [7].

In rTMS treatment, a magnetic coil is placed against the scalp centered over the target brain region, generating an electromagnetic field that penetrates the skull and stimulates the cortex (i.e., depolarizes neurons). Therapeutic use of rTMS has been most extensively studied in treatment-refractory depression. For this application, high-frequency stimulation is delivered to the left dorsolateral prefrontal cortex (l-dlPFC), often defined as the F3 position of the EEG 10–20 system [8] approximating the left middle frontal gyrus, to stimulate the neural circuits that support prefrontal regulation over dysregulated limbic regions [9]. In addition to the this role in affect regulation, l-dlPFC is integral to higher-order cognition (i.e., executive function) [10,11], which modulates episodic memory in MCI by way of enhancing encoding and facilitating retrieval [12–14]. Regions of the lateral prefrontal cortex participate in multiple large-scale functional networks that are implicated in executive function [15], with the l-dlPFC (i.e., middle frontal gyrus) being a key node of the frontoparietal network that is less activated during cognitive tasks in MCI relative to healthy controls [16]. Thus, high frequency rTMS of the l-dlPFC may be a viable treatment to compensate for memory deficits in aMCI. Evidence from the modest accumulation of 12 randomized clinical trials suggests that high-frequency rTMS (5, 10, 15, 20 Hz) to l-dlPFC delivered at 80–120% resting motor threshold (rMT) for >10 stimulation sessions produces modest improvements in global cognition and memory in MCI [17]. The highest dose delivered to l-dlPFC in these trials was 3,000 pulses on each of 10 consecutive weekdays (30,000 total pulses) [18].

While these emerging findings are promising, completing a conventional course of rTMS, requiring 4–6 weeks of once daily treatment, may be burdensome and logistically difficult for those with cognitive impairment. These barriers could be addressed by two recent advances in clinical rTMS delivery: intermittent theta burst stimulation (iTBS) delivered in an accelerated treatment schedule. iTBS refers to a higher frequency pulse pattern of 3-pulse bursts at 50 Hz delivered every 200 ms, with each stimulation session lasting less than 3 minutes versus 20–30 minutes for conventional 10 Hz rTMS sessions. iTBS is particularly appealing as the shorter stimulation sessions improve the feasibility of delivering multiple same-day stimulation sessions (i.e., an accelerated rTMS protocol). Such an accelerated schedule is advantageous in that it shortens the overall duration of treatment, enabling the delivery of the same number of stimulation sessions and pulses as conventional rTMS treatment (daily single-session treatments for 4–6 weeks) in as little as 3–5 days [19]. Thus, accelerated iTBS may be a more feasible and accessible treatment option for people with MCI than conventional rTMS. Accelerated iTBS is now FDA approved for treatment-refractory major depression [20,21], with some indication of increased adherence and faster treatment response compared to conventional rTMS [22–25]. iTBS has shown promise in enhancing cognition in healthy adults [26], and evidence for accelerated iTBS for impaired cognition is emerging. While no sham-controlled trials of accelerated iTBS to l-dlPFC have been conducted in MCI, one randomized, double-blind, sham-controlled iTBS trial conducted in mild to moderate AD dementia found improved associative memory and global cognition after 14 days of 3 stimulation sessions each at 70% rMT (25,200 total pulses) [27].

With the lack of studies of accelerated iTBS for MCI, an important first step towards advancing this promising, low-burden neuromodulation treatment is to systematically evaluate its safety and to carefully consider factors that may impact its implementation. Thus, we conducted an open-label phase I trial of accelerated iTBS to l-dlPFC in individuals with aMCI due to possible AD. We implemented an accelerated iTBS protocol over just 3 days (i.e., 8 stimulation sessions/day; 14,400 total pulses) at a higher stimulation intensity (i.e., 120% rMT) than prior investigations in this population. The aims were to establish the safety, feasibility, tolerability, and acceptability of this treatment in MCI. We collected neuroradiological, neuropsychiatric, and neuropsychological measures to monitor safety and detailed self-report data regarding side effects and perceptions of treatment. Our secondary aim was to determine target engagement via assessing change in fluid cognition, which is a psychometrically robust composite of tests of higher-order cognitive functions that are subserved by functional networks of which the l-dlPFC is a hub [28,29].

METHODS

Participants

Participants were patients of outpatient neurology clinics from a single-site tertiary care university hospital, with clinics staffed by behavioral neurologists, neuropsychologists, and advanced practice providers. Participants were identified either prospectively or retrospectively via chart review using a secure web-based tool. Out of 67 potential participants screened, 24 individuals with aMCI were enrolled; 22 initiated treatment (see Figure 1). This sample was composed of older adults (range: 61.5 – 85.2 years, M = 74.1, SD = 5.71) who were predominantly White/non-Hispanic (n = 23; Black/non-Hispanic: n = 1), roughly half female (n = 13), with a college education on average (range: 12 – 20 years, M = 15.9, SD = 2.5).

Figure 1. CONSORT diagram.

Participants were ineligible at screening due to: a) not meeting inclusion criteria, mostly diagnostic or neuropsychological MCI study criteria (7 of 10) or b) meeting exclusion criteria, mostly MRI, TMS, or medical contraindications (10 of 15).

Inclusion/Exclusion Criteria.

Participants had to be between ages 60–85 years and report English as their first/primary language. All participants received a diagnosis of MCI due to possible AD from a healthcare provider (i.e., neurologist or neuropsychologist) within the past two years per NIA-AA criteria [1]. The primary suspected etiology had to be neurodegenerative, with competing differential diagnoses (e.g., psychiatric disorder, movement disorder, reversible causes, substance use) ruled out. Additionally, in the interest of diagnostic consistency and stability, participants needed to meet actuarial neuropsychological criteria for aMCI upon clinical eligibility assessment, which requires impairment (≤16^th percentile using demographically-corrected norms [30–32]) in ≥2 scores within one cognitive domain or ≥1 scores in ≥3 domains, one of which being memory. Consistent with these criteria, the profile of pre-treatment performance shows primarily amnestic impairment (Figure 2; Table 1), and all participants were independent per the Lawton instrumental activities of daily living scale [33]; all scored 8/8 (i.e., no difficulties), except for one participant who scored 6/8 due to choosing not to drive and receiving reminders to take their medications.

Figure 2. Baseline Neuropsychological Profile.

Data are shown for the 22 treatment initiators. Boxplots show medians and quartiles for each of the 10 standard neuropsychological measures, points represent individual participant’s scores. The horizontal black line indicates the normative mean (50^th percentile) and the horizontal gray line indicates −1 SD (16^th percentile), below which scores were considered impaired.

Table 1.

Scores on cognitive and neuropsychiatric measures for participants with complete data (N = 21).

	Pre-Treatment Assessment		Post-Treatment Assessment		Change Over Time (Pre- to Post-Treatment)
	Mean	SD	Mean	SD	Stat	p-value
Neuropsychiatric Symptoms
YMRS	0	[0,0]	0	[0,0]	-	-
C-SSRS (no and %)
Suicidal ideation	0	0%	0	0%	-	-
Suicidal behaviora	1	5%	1	5%	-	-
GDS	4.65	3.12	4.55	3.14	t(20) = 0.13	0.897
Ham-D	0.24	0.77	0.05	0.22	t(20) = 0.53	0.605
Global Cognition	Mean	SD	Mean	SD
MoCA (raw score)b	19.05	4.14	19.00	3.65	-	-
MoCA (normed z-score)	−1.25	1.38	−1.20	1.18	t(20) = 0.36	0.725
NINDS-CSN Battery (percentiles)	Median	[Q1,Q3]	Median	[Q1,Q3]
Memory
HVLT Immediate	3	[1,14]	6	[1,18]	U = 57.5	0.228
HVLT Delayed	0	[0,0]	0	[0,0]	U = 9.5	0.685
HVLT Retention	0	[0,0]	0	[0,0]	U = 3.0	1
HVLT Recognition	0	[0,3]	0	[0,3]	U = 26.0	0.562
Attention
Coding	31	[11,41]	20	[12,56]	U = 66.0	0.15
Trails A	12	[0,47]	21	[4,55]	U = 70.0	0.514
Executive Function
Trails B	2	[0,34]	19	[0,48]	U = 31.0	0.059
Digits Forward	22	[8,35]	23	[15,55]	U = 42.5	0.196
Digits Backward	24	[17,40]	21	[11,41]	U = 63.5	0.554
Language
Animal Fluency	8	[3,19]	8	[2,15]	U = 34.0	0.724
Letter Fluency	28	[13,52]	15	[9,54]	U = 102.5	0.778
NIHTB-CB	Mean	SD	Mean	SD
Fluid Cognition Composite (normed T-score)	40.25	8.07	43.67	7.85	t(20) = −4.48	< .001

^aOne participant endorsed a history of suicidal behavior occurring over a year prior to enrolling in the study.

^bThe MoCA was administered virtually at the 4-week follow-up assessment and did not include the trails subtest due to the remote nature of this visit. The sample obtained a mean score of 20.6 (SD = 4.2) (maximum possible score = 29). Due to this difference in measures, these scores are not directly comparable to the original version given at pre- and post-treatment, and demographically adjusted normative scores are not available. Therefore, we did not conduct statistical tests of changes in MoCA scores at 4-week follow-up.

Note: C-SSRS=Columbia Suicide Severity Rating Scale; GDS=Geriatric Depression Scale; Ham-D=Hamilton Depression Rating Scale; HVLT=Hopkins Verbal Learning Test; MoCA=Montreal Cognitive Assessment; YMRS=Young Mania Rating Scale.

Participants were excluded if they: had a dementia diagnosis per DSM-5 or NIA-AA criteria [34]; had a significant or unstable condition/s that impact cognition such as severe cardiac, cerebrovascular, or metabolic disease, severe mental illness (e.g. bipolar disorder, psychoses), alcohol or substance use disorder, developmental disorder, or other neurologic disease (e.g. severe brain injury, seizures); were enrolled in a clinical trial and/or had received an investigational medication within the last 30 days; had MRI and TMS contraindications; or had daily/weekly use of anticholinergics, neuroleptics, sedative-hypnotics, or bupropion. Stimulant medication use was permitted if deemed low risk by a physician co-investigator (which applied to only one participant enrolled while on a stable dose that was deemed safe). Other specified medications (i.e., cholinesterase inhibitors, NMDA receptor antagonists, and antidepressants) were permitted if on a stable regimen for 4 weeks prior to enrollment.

Procedures

The timeline of study procedures is depicted in Figure 3. This open-label phase I trial was registered on ClinicalTrials.gov (NCT04503096) and approved by the Medical University of South Carolina Institutional Review Board (Pro00100536). It was conducted from April 2021 to July 2022 (when recruitment goals were satisfied). Briefly, potential participants underwent a phone eligibility screening prior to providing written informed consent. They completed pre- and post-treatment assessments (neuropsychological testing, neuropsychiatric assessments, questionnaires, brain MRI), 3 accelerated iTBS treatment days, questionnaires on each of the 4 post-treatment weeks, and a 4-week post-treatment assessment (questionnaires and a virtual global cognitive screening).

Figure 3. Study Procedures.

Accelerated iTBS Treatment.

Treatment was delivered using a MagVenture R30 MagPro TMS System with Cool-B65 figure-8 coil placed at the F3 location of the International 10–20 EEG placement system to target l-dlPFC [8] (Figure 4A). We selected this common targeting method as it is highly translatable and scalable to clinical contexts and facilitates comparison of our findings to the bulk of the literature. The distribution of cortical stimulation for each participant is presented in the electric field models (Figure 4B), in which the center of the highest intensity (i.e., red) area approximates the coil location, with an “x” denoting the center of the TMS coil positioned over the F3 location for each participant. A Brainsight Neuronavigation system (Version 2.4.8) ensured accuracy, reliability, and reproducibility of coil placement both within and between sessions [35]. Stimulation intensity was 120% of the rMT (average in the 22 treatment initiators = 43.5 ± 6.2%; range: 35 – 60 maximal stimulator output). Treatment was conducted over 3 days within an 8-day span delivering 24 total stimulation sessions (8 sessions/day, approximately 2 hours per day) to approximate the cumulative stimulation sessions delivered in a conventional once/weekday 5-week rTMS treatment. Please see the Supplement for additional details regarding neuronavigation and dosing decisions. Stimulation parameters were: 50 Hz iTBS triplet bursts every 200 ms in 2 second trains repeated every 10 seconds (8 second inter-train interval) for 190 seconds, resulting in 600 pulses/session (14,400 total pulses). Each of the same-day sessions was separated by 10–15 minutes or more accounting for participant comfort. The inter-session intervals ranged from 9–27 minutes; the vast majority were in the 10–15-minute range (median = 10 min, Q1 = 10, Q3 = 11) and only one participant had an interval >20 minutes. To facilitate adherence and retention, treatment days were selected by participants and did not need to be contiguous. The median number of days between each treatment day was 2 (Q1 = 1.25, Q3 = 2) and the median total days between treatment day 1 and treatment day 3 was 4 days (Q1 = 4, Q3 = 6).

Figure 4. Electric Field Models.

To further investigate the amount of 120% rMT stimulation reaching the l-dlPFC target, we performed electric field modeling in each participant. This figure presents A) the position of the stimulation coil over the F3 target on the standard MNI-152 template head, B) a visualization of electric fields for each of the 22 treatment initiators in which the gray “x” indicates the center of the TMS coil over the F3 location, and C) a boxplot showing the distribution of electric field values per participant. On average, participants received 65.3 V/m (SD = 10.4, range = 48.53–88.54 V/m) from stimulation in the l-dlPFC ROI. This amount of stimulation is safe and within a therapeutic range for effective target engagement.

Participants engaged in computerized cognitive training (BrainHQ) during inter-session breaks. BrainHQ was not dosed for efficacy; this was included to ascertain feasibility, and sensitivity analyses suggest that it did not impact treatment outcomes (see Supplement).

Measures

Sample Characterization.

Baseline cognitive function was assessed using the National Institute of Neurological Disorders and Stroke-Canadian Stroke Network (NINDS-CSN) 30-minute neuropsychological battery [36]. This included the Hopkins Verbal Learning Test-Revised (HVLT-R; age-adjusted scores[37]); the Digit Span Forward and Backward and Digit Symbol Coding subtests of the Wechsler Adult Intelligence Scale-Revised (age, sex, and education adjusted scores [30]), Trail Making Tests A & B, animal fluency, and letter fluency (age, sex, and education adjusted scores [38]). Independence in Daily Functioning was assessed using the Lawton Instrumental Activities of Daily Living scale [33].

Safety.

All safety measures were administered at pre- and post-treatment. Brain MRI was acquired on a 3T Prisma^fit MRI system (Siemens Healthineers, Erlangen, Germany). Imaging included volumetric T1- and varied T2-weighted (T2, FLAIR, diffusion-, and susceptibility-weighted) sequences. Each study was reviewed by a board-certified neuroradiologist (M.A.) to identify any pre- to post-treatment change and to document incidental findings (see Supplement for acquisition parameters). Neuropsychiatric symptoms were monitored using the Columbia Suicide Severity Rating Scale (C-SSRS; [39]) and the Young Mania Rating Scale (YMRS; [40]). Global Cognitive Function was monitored using age- and education-adjusted z-scores [41] on the 30-item Montreal Cognitive Assessment (MoCA; [42]).

Feasibility.

Feasibility was defined as the completion of target enrollment and attainment of a >80% retention rate.

Tolerability.

Review of Systems was taken via self-report at the beginning of each treatment day and at post-treatment assessment to monitor for existing and emergent physical symptoms. Participants rated their experiences of symptoms listed in Figure 5A (from 0 = none/no symptoms to 5 = severe symptoms) during the preceding 24–48 hours. They were also asked to indicate (yes/no) whether they experienced any of the 3 symptoms of mania/hypomania during the preceding 24–48 hours: abnormal elevations in mood, energy, or irritability. Momentary Assessments of Symptoms were administered after each of the 24 iTBS sessions, which asked participants to retrospectively rate their experiences of the symptoms listed in Figure 5B (from 0 = not at all to 10 = extremely) during stimulation and immediately after the stimulation session. Participants could provide narrative descriptions of symptoms.

Figure 5. Tolerability of accelerated iTBS Treatment.

Plots depict ratings over the course of treatment on A) the Review of Systems (ROS) measure given at the beginning of each treatment day and B) the momentary assessments of symptoms in which participants were asked to provide ratings of their experience during active stimulation (left) and immediately following stimulation (right) for each of the 8 stimulation sessions on the three treatment days. Boxplots show medians and quartiles, while points/lines represent individual participant ratings over time.

Acceptability.

A TMS Experience Questionnaire was completed at the post-treatment assessment. Participants were asked to retrospectively rate (from 1 = not at all to 5 = very much so) 15 items querying their perceptions of the treatment, its side effects, and its benefits as well as their preferences regarding treatment format and logistics (Table 2).

Table 2.

Acceptability items and ratings collected immediately post-accelerated iTBS treatment.

Item	Median	[Q1, Q3]
1 = Not at all, 3 = Somewhat, 5 = Very much so
I was motivated to attend the rTMS sessions.	5	[4, 5]
I was committed to completing the treatment course.	5	[5, 5]
I was interested in the rTMS techniques and learning about them.	5	[5, 5]
I understand the purpose of the treatment and how it could help my symptoms.	5	[5, 5]
I would be open to completing another course of rTMS treatment in the future, if needed.	5	[5, 5]
The rTMS treatment helped improve my ability to cope with daily challenges.	3	[2, 4]
Completing multiple rTMS sessions on each day and thus reducing the number of days I had to come to treatment was preferable to attending one rTMS session on each weekday and thus having treatment last for more weeks.	5	[5, 5]
The staff members were attentive and sensitive to my needs during treatment.	5	[5, 5]
The staff members explained all procedures to me and answered my questions.	5	[5, 5]
There was plenty of flexibility in scheduling rTMS sessions around my other obligations.	5	[5, 5]
Reverse Scored Items
1 = Not at all, 3 = Somewhat, 5 = Very much so
At times I wanted to quit treatment.	1	[1, 1]
Treatment sessions were stressful.	1	[1,2]
It was hard to stay awake during the rTMS sessions.	1	[1, 1]
I experienced pain and discomfort during treatment that was difficult to tolerate.	1	[1, 1]
Completing multiple rTMS sessions in each day was at times tiring.	1	[1,2]

Secondary Outcomes.

These measures were administered at pre- and post-treatment. Fluid cognition was measured using the iPad-administered NIH Toolbox Cognition Battery (NIHTB-CB, version 2) which was developed using advanced psychometrics to attenuate floor or ceiling effects [43], thereby making it ideal for assessing change. Fluid Cognition Composite scores [44] were calculated by averaging the demographically adjusted (age, education, sex, race/ethnicity; [45]) T-scores for 4 NIHTB-CB tests: flanker inhibitory control, list sorting working memory, pattern comparison processing speed, and dimensional change card sort. The picture sequence memory test was not administered to mitigate participant burden and to minimize practice effects given that it has been reported to have the largest effect size change over a comparable retest interval [43]. Therefore, the Fluid Cognition Composite scores in this study are abbreviated, akin to previously published “short toolbox” composite scores [46]. Depressive symptoms were measured using total scores on the Hamilton Depression Rating Scale (HAM-D; 17 items [47]) and the Geriatric Depression Scale (GDS; 30 items [48]).

Statistical Analyses

Sample Size Justification.

Since this was a phase I feasibility study, a formal sample size calculation was not performed but was instead based on our primary aims of generating estimates of feasibility, acceptability, and outcome variability to aid in the planning of future sufficiently powered trials [49]. Our final sample size of N = 21 treatment completers will allow us to obtain estimates of the standard deviations of continuous outcome variables with which to conduct sample size calculations for a future definitive trial.

The following statistical analyses were conducted in R version 4.2.3 using the ‘stats’ package (i.e., ‘t.test’, ‘wilcox.test’, and ‘shapiro.test’ functions). Effect sizes for t-tests were generated using the ‘cohens_d’ function of the ‘rstatix’ R package. Graphs were created using the ‘ggplot2’ R package.

Primary Outcomes.

Paired t-tests were conducted to test for pre- to post-treatment differences in MoCA z-scores. The other safety-related clinical variables are described in-text. For Likert-scale tolerability and acceptability ratings, we report descriptive statistics (e.g., median, mode, quartiles) and provide in-text narrative descriptions of rating patterns.

Secondary Outcomes.

Paired t-tests were used to test for changes in depressive symptoms (raw GDS and HAM-D scores) and cognition (NIHTB-CB Fluid Cognition Composite T-scores). As a sensitivity analysis, we also conducted paired t-tests to evaluate change in scores on the standard neuropsychological tests; note that we used nonparametric Wilcoxon signed-rank tests since nearly all scores were not normally distributed (per Shapiro-Wilk tests).

RESULTS

Primary Outcomes

Safety.

There were no serious adverse events during the study. Comparison of pre- and post-treatment brain MRI revealed no significant neuroradiological change. Older adults have a higher likelihood of having brain lesions [50] which may increase risk for TMS-related adverse events [51]; that no adverse events (e.g. seizures) occurred is therefore noteworthy. In addition, there were no adverse neuropsychiatric changes, including no endorsement of current suicidal ideation or behavior or evidence of mania/hypomania at pre- or post-treatment (Table 1). There were also no decrements in global cognitive function as measured by normed MoCA scores from pre- to post-treatment (p = .725; Table 1). E-field modeling (see Supplement for analytic details) suggests that all participants received a safe level of cortical stimulation, with an average E-field magnitude of 65.32 V/m (SD = 10.39; range = 48.53–88.54 V/m; Figure 4B–C). These values are considered to be in the therapeutic range and consistent with expectations given 120% rMT stimulation [52].

Feasibility.

We reached recruitment targets despite conducting the study during the COVID-19 pandemic. We achieved high retention rates (95%; Figure 1), with 21 of the 22 treatment initiators completing all study procedures.

Tolerability.

Throughout treatment, participants reported very few symptoms occurring over the previous 24–48 hours on the Review of Systems (Figure 5A). The modal rating was 0 for all symptoms, and ratings did not exceed moderate levels (i.e., >3/5). Given the broad symptoms queried in this screening, we highlight in Figure 5A that ratings remained in the none to moderate range even for two very common sensations experienced during rTMS (headache and pain).

Momentary assessments (Figure 5B) showed that the modal rating was again 0 for all symptoms experienced during and immediately after each stimulation session. Symptoms largely resolved post-session and over the course of the 3 treatment days. As expected, the most frequently endorsed symptoms were headache and pain. Although one participant rated headache 10/10 both during and after the first stimulation session, this resolved by the end of treatment day 1 and remained low. Five other participants rated headache in the moderate range (≥5/10), but in all cases, this decreased to none-to-mild (≤3/10) after stimulation. A similar pattern was observed for pain, which was most often reported at the stimulation site. Some participants also reported sensations in the upper portion of their face, eyes, or neck and discomfort from the headband or cap, which were adjusted by the treating coordinator as needed.

Acceptability.

At the end of the trial, most participants gave high acceptability ratings (Table 2). They reported very little desire to quit and minimal experiences of pain/discomfort that were difficult to tolerate. However, the treatment was rated as mildly stressful and moderately tiring for some. Participants viewed accelerated iTBS positively, reporting good understanding of the purpose of treatment and how it could address symptoms, and feeling highly motivated, interested, and committed to completing the treatment. There was variability in whether participants felt that the treatment improved their ability to cope with daily challenges, with a median rating of “somewhat.” Nonetheless, the majority were open to completing another course of treatment in the future. They also rated several logistical matters favorably, including the condensed treatment schedule, scheduling flexibility, and interactions with study staff.

Secondary Outcomes

Depression.

This sample was not clinically depressed at baseline, as assessed by the GDS and the HAM-D (Table 1; Figure 6A). Therefore, as expected, there was no change in depressive symptoms as a function of treatment (HAM-D: p = 0.296; GDS: p = 0.897).

Figure 6. Secondary Outcomes at Pre- and Post-Treatment.

Boxplots show medians and quartiles for scores on measures of A) depression (HAM-D and GDS) and B) fluid cognition composite scores (NIHTB-CB) from pre- to post-treatment. Points are individual participants’ scores at pre- and post-treatment, connected by lines. The horizontal lines on the depression plots represent maximum scores on those measures (HAM-D max = 53, GDS max = 30), the horizontal line on the fluid cognition plot shows the normative mean (T = 50). *** p < .001.

Cognition.

We found a significant, large effect-size (Cohen’s d = 0.98) improvement in fluid cognition from pre- to post-treatment (p < .001; Table 1; Figure 6B). At baseline, participants performed about −1 SD below the normative mean (T = 50, SD = 10) on average and this improved to −0.7 SD at post-treatment. From pre- to post-treatment, most participants (17/21) showed improved scores, with over a quarter (6/21) exhibiting an improvement of ≥ 0.5 SD. In contrast, none declined ≥ 0.5 SD. Consistent with our expectation that the Fluid Composite would be more psychometrically sensitive to change, we found no significant changes in performance on the standard neuropsychological battery (Table 1).

DISCUSSION

The results of this open-label phase I trial of accelerated iTBS to l-dlPFC support the safety, feasibility, tolerability, and acceptability of this treatment for individuals with aMCI. Despite neurodegeneration and associated CSF redistribution which could theoretically amplify or attenuate effective voltage, electric field modeling suggests that stimulation was delivered within a safe and therapeutic range. The treatment resulted in no adverse neuroradiological, neuropsychiatric, or neurocognitive effects, and produced minimal and short-lasting side effects. We also observed improved fluid cognition following 3 days of accelerated iTBS (14,400 total pulses), demonstrating target engagement within less than a third of the treatment days required in the prior study of iTBS for AD [27].

One of the greatest benefits of accelerated iTBS is the substantial reduction in the frequency and duration of treatment sessions. This is particularly relevant to older adults with MCI for whom time commitment is a common barrier to engaging in interventions [53]. The treatment course in this study (3 days, 2 hours each) appeared quite manageable as evidenced by the very high retention rate (>95%), and participants reported a strong preference for completing multiple sessions per day on fewer days compared to daily sessions for multiple weeks. Although the optimal inter-session interval for same-day rTMS sessions has not been systematically evaluated, with protocols ranging from 10 minutes to 2 hours [19], our findings suggest that 10–15 minutes is safe and highly acceptable. Feasibility is further supported by most participants reporting high motivation, interest, and commitment to the treatment paired with low desire to quit and an openness to engage in future treatment. In addition to being logistically favorable, treatment was rated as highly tolerable and acceptable.

We took a comprehensive approach to monitoring side effects, incorporating both current and retrospective ratings at a range of timescales (i.e., in the moment, in the previous 24–48 hours, and over the duration of the treatment). Importantly, our findings were not suggestive of side effect sensitization within or between days, which is a potential concern with accelerated protocols. Across all measures, effects were generally rated as minimal/mild when experienced, and most participants reported none. Momentary assessments revealed that although some participants experienced moderate levels of expected symptoms (i.e., headache, scalp pain) during stimulation, these ratings returned to mild/none after stimulation ceased.

Aside from these primary aims, this study also achieved the secondary aim to demonstrate target engagement in the form of a large effect size improvement in fluid cognition—a composite measure of processing speed and executive function from the NIH Toolbox Cognition Battery. Of note, this improvement is greater than the small-to-medium practice effect reported in the normative sample for this battery [44]. As individuals with aMCI typically demonstrate attenuated practice effects [54,55], this larger-than-expected finding suggests that iTBS may exert a measurable effect and that the Fluid Cognition Composite is a viable outcome measure to use in future, more definitive tests of efficacy. Indeed, a notable strength of this study is our use of the NIHTB-CB, which is more psychometrically robust than standard neuropsychological tests and appeared to offer improved sensitivity to intervention-related cognitive changes. Additionally, we employed stringent actuarial neuropsychological criteria for aMCI that minimize false positive diagnoses to enhance diagnostic specificity [56] and stability [57]. Through this careful selection of participants, we reduced potential sources of heterogeneity that could impede our ability to evaluate target engagement. Furthermore, given the well-established efficacy of high frequency rTMS to l-dlPFC for alleviating depression, particularly in older adults [58], the absence of clinical depression in our sample underscores that the observed improvements in cognition were not attributable to the secondary effects of reduced depression.

Nonetheless, our findings are limited by the single-arm, open-label design of this phase I trial; there was no sham or comparator condition, randomization, or blinding of participants or experimenters. Although the target engagement effect is greater than published practice effects (thus making this finding less likely due to systematic measurement error), it remains possible that the improvements in fluid cognition may be attributable, at least in part, to placebo effect. Caution should be exercised when interpreting the observed large effect given the modest sample size in this study, as future randomized controlled trials are needed to control for the influence of expectancy and practice effects on outcomes, providing more robust tests of efficacy than what can be achieved in this phase I study. Corollary measures of target engagement that are commensurate with the treatment (e.g. changes in functional connectivity via resting-state fMRI) can also be related to behavioral outcomes. Such an approach can be utilized in adequately powered studies to mechanistically test the extent to which improvements in executive functions or frontoparietal connectivity due to iTBS can mediate improvements in episodic memory in aMCI. Furthermore, subsequent trials may also go beyond relying on a syndromic characterization of study samples (as was done here) to using biomarkers to enhance diagnostic specificity. While at present a biomarker-based definition of AD is neither required to investigate a well-established syndrome (i.e. aMCI), nor does it supplant current disease definitions for standard medical use [59], future attempts at redefining diagnostic criteria and disease nosology will likely incorporate biomarkers given the rapid evolution of the literature since contemporary criteria were last updated 10+ years ago, i.e. [1,60]. Whether biomarkers will ultimately define the presence of disease [61] or remain a critical supplement to differential diagnosis and treatment planning [62] continues to be up for debate.

These limitations notwithstanding, the promising results from this study point to several future directions. Dosing parameters (e.g., number of sessions, duration of inter-session and inter-treatment day intervals, number of active pulses, stimulation intensity, targeting method) vary widely across existing studies of rTMS [19]. Thus, future trials should comprehensively describe treatment parameters to ensure rigor and reproducibility, to provide a framework for presenting pragmatic and scientific justifications for parameter selection, and to facilitate clinical translation. We have provided transparent reporting of all dosing parameters to facilitate replication or systematic investigations of individual parameters in future studies. For example, studies may use physiological measurements of target engagement to ensure rigorous evaluation of inter-session intervals to determine whether a desired effect is being achieved. Overall, there is a need to systematically determine the optimal dosing parameters for symptom remediation in MCI. Relatedly, the durability of treatment gains in MCI remains an open question that will require investigation with larger trials. Although there is preliminary evidence of sustained improvements in randomized controlled trials of accelerated iTBS (e.g., at 8-weeks for cognition in AD [27] and 5-weeks for treatment-refractory depression [20]), we anticipate that booster sessions may be warranted to facilitate the maintenance of gains in MCI. Furthermore, as MCI is a syndrome encompassing an array of cognitive and behavioral/neuropsychiatric symptoms, and l-dlPFC stimulation remediates several transdiagnostic impairments [63], future work should evaluate whether accelerated iTBS may simultaneously address multiple symptoms such as impairments in discrete cognitive domains and neuropsychiatric symptoms such as depression, which is present in 25–40% of individuals with MCI [64]. We are currently seeking to address these recommended next steps in a randomized, sham-controlled, dose-ranging trial (NCT05992831) which includes study design features to promote inclusivity in recruitment [65,66].

In conclusion, this phase I trial demonstrates that accelerated iTBS, specifically 24 sessions over 3 days, to l-dlPFC is safe, feasible, tolerable, and acceptable, and a relatively low-burden treatment for individuals with aMCI. Further, we provide evidence of target engagement in the form of improved cognition even with this brief course of accelerated iTBS, although future work is needed to replicate this effect in randomized controlled trials and to evaluate their durability. These promising results will directly inform future trials aimed at optimizing treatment parameters, broadening the indication to other MCI subtypes, and testing the augmentation of established cognitive rehabilitation interventions when combined with rTMS.

KEY MESSAGES.

What is already known on this topic:

There is emerging evidence that suggests accelerated iTBS may benefit cognition in prodromal neurodegenerative disease, but the safety, feasibility, tolerability, and acceptability of this treatment in Mild Cognitive Impairment (MCI) have yet to be systematically evaluated.

What this study adds:

This phase I clinical trial demonstrates that three days of accelerated iTBS (14,400 total pulses) is safe, feasible, tolerable, and acceptable in older adults with amnestic MCI due to possible AD. Further, we provide evidence of target engagement in the form of improved fluid cognition following treatment.

How this study might affect research, practice, or policy:

This study provides the critical preliminary data to motivate and inform subsequent clinical trials to determine the efficacy of iTBS to improve cognition in amnestic MCI and related indications.