Participants will be recruited from Chicago metropolitan area US Department of Veterans Affairs (VA) hospitals: Edward Hines Jr. Veterans Administration Hospital (hereafter referred to as “Hines VA”) and Jesse Brown VA Medical Center in the state of Illinois, United States. The intervention will be conducted at the Hines VA.

Veteran participants will be included if they have a diagnosis of PD (idiopathic or atypical) as determined by a VA neurologist and are aged 50 years or older. They must be on stable medication (ie, no changes in medication and/or dose in the previous month) and are expected to remain on stable medication for the duration of the study. Participants must demonstrate decision-making capacity prior to enrolling in the study, that is, verbally demonstrate (1) understanding of the purpose and activities required in the study, (2) appreciation for the potential risks and benefits, (3) appreciation of alternative choices other than participating in the study, and (4) expression of choice about whether they want to participate. PD-MCI will be determined using the International Parkinson and Movement Disorder Society level 2 diagnostic criteria [24], as detailed in Table 1. Patients will be excluded if they exhibit dementia (score of <21 on the Montreal Cognitive Assessment [MoCA]) or severe depression (score of >29 on the Beck Depression Inventory–II). Patients will be excluded if they have a history of deep brain stimulation surgery. Other reasons for exclusion are based on contraindication for magnetic resonance imaging (MRI), including implanted cardiac pacemaker or defibrillator, implanted medical pump, increased intracranial pressure, metal in the eyes or face, or shrapnel or bullet remnants in the brain. Patients who have contraindications for rTMS, including history of psychotic spectrum disorders such as bipolar and schizophrenia, history of suicide attempts, metal or medical implants in the head or neck, increased intracranial pressure, or resting head tremor, or are taking medications that lower seizure threshold will also be excluded. Medication lists and the current evidence in the literature for safety profiles of medications will be reviewed by the principal investigator (PI) and study team clinicians to determine eligibility

Participants will receive a total of 10 active or sham rTMS sessions. They will be scheduled to receive 2 sessions per day. Treatment days do not need to be consecutive; rather, they can be spread across approximately 10 days (eg, Monday, Wednesday, Thursday, Tuesday, and Wednesday). Participants randomized to active rTMS will receive stimulation applied to the left DLPFC at 110% the RMT and a 15-Hz rate, 5 seconds per train, 10-second intertrial interval, and a total of 40 trains per session. Each patient will receive approximately 3000 pulses per session. Sham stimulation will be administered at the same parameters as the real rTMS; however, no magnetic field will be produced on the placebo side of the active or placebo coil. Patients will be blinded as to which group they are in (treatment or sham). Participants will be unblinded to their treatment group after the last participant completes the study via a letter that will be mailed to the veterans (see details in the rTMS Sessions section).

Each participant will be instructed to take their PD medication within an hour before their first scheduled session of the day. rTMS will be administered at approximately the same time each day, with an hour break in between sessions.

The primary study outcome is the change in score from baseline to end point on the NIH-EXAMINER Executive Composite score. The composite score is advantageous because it combines performance across various aspects of executive function consistently reported to be impacted by PD [36]. It is derived using item response theory, which can improve sensitivity and increase statistical power over standard scoring methods [37]. Seven tests in the NIH-EXAMINER will be used to compute the composite score, as detailed in Table 2. This battery takes approximately 30 minutes to administer, and each test has 3 alternate forms to help avoid practice or learning effects when administered multiple times during the study.

The NIH-EXAMINER establishes a 3-factor model defined by cognitive control, working memory, and fluency. A confirmatory factor analysis indicates that these 3 factors load on to 1 factor: the Executive Composite score [38]. The ecological validity of the composite score was assessed in 225 patients with various neurological conditions, including PD [45]. Exhibiting concurrent validity, the composite score is a robust predictor of real-world executive behavior, as measured using the Frontal Systems Behavior Scale [36]. Furthermore, poor composite scores correlate with atrophy in frontal cortical regions that mediate executive function [36]. In patients with PD, the sensitivity of the NIH-EXAMINER was compared to that of other commonly used tests of executive function recommended by the International Parkinson and Movement Disorder Society task force on PD-MCI [24]; the NIH-EXAMINER Executive Composite score was the most sensitive and specific in detecting cognitive impairment, with an effect size of 0.94 [36].

Secondary outcomes include change from baseline to end point in the cognitive control composite score from the NIH-EXAMINER, working memory composite score from the NIH-EXAMINER, delayed verbal memory score for the Hopkins Verbal Learning Test–Revised [30], total score on the oral version of the Symbol Digit Modalities Test [46], total score on the Hooper Visual Organization Test [33], total score on the Beck Depression Inventory–II [47], total score on the Beck Anxiety Inventory [48], total score on the Apathy Evaluation Scale [49], total score on the MoCA [50], total score on the PD Cognitive Functional Rating Scale [40], and total score on the Activity Measure for Post-Acute Care applied cognitive domain questionnaire [41].

Time schedule of enrollment, interventions (including any run-ins and washouts), assessments, and visits for participants. A schematic diagram of the study timeline is shown in Figure 1.

Effect sizes for rTMS on cognition in the PD literature are variable, with a range of 0.04 [51] to 0.36 [14]. However, there are several limitations to these studies that lead us to believe that the effect size will be larger in the population with PD-MCI. First, given that the existing literature has studied patients with PD-N, it is reasonable to expect that cognitive effects in these studies may have been limited by a ceiling effect such that cognitive function could not have improved any further [8]. In addition, not all the existing PD studies have used neuronavigation, so variability was likely increased in terms of the actual site being stimulated [52]. Such treatment variability can increase variability in cognitive outcome measures and, thus, lower the effect size [53]. Finally, all existing PD literature examining cognitive effects has used 100% the RMT or less. As people age, cortical atrophy occurs, leading to increased scalp-to-cortex distance. This atrophy occurs more in the prefrontal regions compared to the motor cortex [54]. If this is not compensated for, it can weaken rTMS stimulation intensity. Thus, effects observed using relatively lower stimulation intensity (80%-100%) likely underestimate the benefits of rTMS for executive function in PD studies. One way to compensate for this is to increase the RMT to above 100%.

The limitations of existing data indicate that rTMS effect sizes on cognition in Alzheimer disease are likely more accurate for sample size determination in this study. Similar to PD, patients with Alzheimer disease experience cognitive deficits and have overlapping brain regions undergoing neurodegeneration. A meta-analysis [55] estimated a moderate effect size of 0.48 in terms of the benefits of rTMS (with many of the included studies targeting the DLPFC) for cognition among older patients with cognitive impairment related to Alzheimer disease.

Taking all this into consideration, as well as the fact that this is a small-scale trial, the effect size of 0.48 found in the aforementioned meta-analysis [55] was estimated for this RCT. The sample size was determined using a 2-level mixed-effects longitudinal model with random participant intercept and random participant slope. The end point effect size was fixed at 0.48, which is considered moderate. The participant intercept and slope variances were fixed at 2.3 and 0.3, respectively. In addition, the error variance was fixed at 2.0, which should be considered moderate. Using this combination of parameters, we determined that a sample size of 71 per group (active and sham) considering an attrition rate of 10% will be needed to achieve 80% power while maintaining the type 1 error rate at 5%. The RMASS software was used for this power analysis [56]. Hence, a total of 156 participants will be recruited for this proposed study.

Military veteran participants will be recruited through VA national databases, clinician referrals, and flyers placed in the 2 participating VA hospitals.

Figure 1 [25,50] illustrates plans for assessment and collection of outcomes during screening, baseline, end point, and follow-up visits.

For safety outcomes, the Poisson distribution for the number of research-related adverse events will be fitted for each treatment group. Negative binomial distributions will be used in case overdispersion is present. The mean parameters (λ) of these 2 distributions will be tested to see whether any difference in the average number of adverse events in each treatment group exists. Lack of significance will suggest that active rTMS is as safe as sham rTMS in terms of adverse events. To further compare the trend in adverse events between the 2 groups, we will use a 3-level mixed-effects Poisson regression model, and the group-by-time interaction parameter will be tested to examine whether active rTMS has more adverse events than sham rTMS over time.

For neuropsychological outcomes, all data will be analyzed using a 2-level longitudinal mixed-effects model with random participant effects to incorporate correlation of measurements nested within the same participant. The model will also include covariates such as medication dose and anxiety and depression levels. Parameters will be estimated using measurements from all 3 time points (ie, baseline, end point, and follow-up). The group-by-time interaction parameter will be used to examine the difference in the trends of the active rTMS group compared to the sham rTMS group. The significance of this parameter will indicate whether treatment is efficacious in terms of executive function over time when adjusted for the identified covariates. To assess the immediate effects of rTMS, one contrast using means of test scores at baseline and end point from the experimental group will be obtained, and the other contrast will be obtained from those same types of means for the control group. These 2 contrasts will then be tested using a 2-tailed t test to compare the change scores of the 2 groups. To assess sustainability of the rTMS effects, contrasts using means of test scores at end point and follow-up will be obtained for each group, and the contrast will then be compared using a t test. Estimates of variance components from mixed-model analysis will be used to calculate test statistics. To account for the many-to-many comparisons of secondary outcomes, we will use a multivariate mixed-effects model with the aforementioned covariates. For every secondary outcome component, there will be fixed as well as random components. The random component will address the within-subject correlation and the heterogeneity between multiple outcomes. Every component will have its own trend, which means that the recovery curve may be outcome specific. The group-by-time interaction parameter of each secondary outcome will be tested to examine whether cognitive function in the active rTMS group is different from that in the sham rTMS group. The false discovery rate will be addressed using Benjamini and Hochberg [57] procedures, with a false discovery rate level of 0.05. Data analysis will be performed using the SAS PROC MIXED statement (version 9.3; SAS Institute). This modeling process will be reported for each outcome of interest for the safety and neuropsychological assessments. Every effort will be made to avoid missing data; we will try to understand the missingness pattern and why and how the missingness occurred. No missing value imputation technique will be used if Little test supports that data are missing completely at random. Otherwise, proper missing data imputation techniques (eg, imputation with mean, nearest neighbor, or last observation carried forward) will be used.

Potential moderators of the effects of rTMS will be explored through Pearson correlations and regression analyses. Factors include, for example, age at study enrollment, age at PD onset, disease duration, MoCA score, levodopa equivalent dose, and change in mood (eg, depression, anxiety, and apathy). Blinding success will be reported alongside the primary results.

There is no data monitoring committee for this pilot study. There is a medical monitor with whom the study team consults for safety.

All study procedures and parameters are in accordance with the current safety and application guidelines for rTMS [19,21]. Treatment will be delivered by personnel trained in the administration of rTMS. An rTMS data safety monitoring sheet (DSMS), adverse event tracking log, medication log, and sleep log for each participant will be kept to monitor session attendance and any side effects or adverse events according to prespecified criteria. The Columbia-Suicide Severity Rating Scale will also be administered at the baseline, end point, and follow-up assessments.

The DSMS contains a customized severity indicator scale with before-and-after variables, including rates of change from baseline in vital signs (temperature, blood pressure, heart rate, and oxygen saturation levels), fatigue, tinnitus, sleep, dizziness, nausea, vomiting, confusion, seizure, syncope (fainting), headache, neck pain, skin integrity of the scalp, and movement. PD-related motor assessments, including speech, facial expression, finger taps, rapid alternating movement of the hands, tremor at rest, leg agility, ability to rise from a chair, gait, and body bradykinesia and hypokinesia, will also be conducted before and after each rTMS session and compared to baseline.

An adverse event tracking log will also be developed for this project for further safety monitoring. All serious and nonserious adverse events will be tracked on this log.

Sleep disruption and deprivation is associated with a lower seizure threshold, which may put the participant at a heightened seizure risk when receiving rTMS. Therefore, the sleep log asks the participants to record total hours of sleep at night and total minutes of nap time during the day. After the baseline assessment, a study team member will provide a sleep packet (collection of blank sleep logs) for the participants. The participants will then receive instructions on how to use the log. Data collected before the first rTMS session will be averaged, and variance will be documented and used as the baseline for the DSMS. At least 3 days of “total hours of sleep” will be required for the participant to complete the first session of rTMS. During the study period, self-reported sleep based on the sleep log will be checked before the intervention is administered for a given day.

Change in medication can also affect the seizure threshold. Therefore, a personalized medication log is generated for each participant. A thorough review of active medications (including prescriptions, over-the-counter medications, and supplements) will be conducted by a study team member, checked against the medical record, and reconciled with the participant to generate an accurate log of medications the participant is taking. After the baseline assessment visit, a study team member will provide a packet (collection of blank medication logs) for the participants. The participants will then receive instructions on how to use the log. During the study period, self-reported medication intake based on the medication log will be checked before the intervention is administered for a given day.

The PI has the right to terminate participation if a participant becomes uncooperative or unwilling to complete study tests, if the participant is experiencing undue stress from study procedures, or if the participant has medical problems that interfere with completion of study tests. As the long-term effects of the experimental treatment are unknown, if a participant withdraws before study completion, we recommend that they complete follow-up assessments.

This study was approved by both the Hines VA (1337782) and Jesse Brown VA Medical Center (1395002) institutional review boards (IRBs) and is registered on ClinicalTrials.gov (NCT03836950). This study has also received an FDA investigational device exemption (G190076). Important modifications to the protocol will be communicated to the sponsor, IRB, and FDA for approval before implementation. The registration on ClinicalTrials.gov will also be updated.

Consent will be obtained by trained study team members. All study participants will provide informed consent before completing research procedures. Participants will be given the option to opt out of consenting or research procedures at any time without loss or benefits. Research participants will have access to research staff to assist with any questions or concerns until understanding is achieved to the judgment of the individual asking the questions. A copy of the signed consent form will be provided to the participants. The consent form will indicate that data will be placed in a Hines VA IRB-approved data repository for potential future use. Only IRB-approved studies with approved data use agreements can request and use the research data for ancillary studies.

All hard copies of assessments will include a unique participant identification number and the date the assessment was completed. Thus, data will be deidentified. The only identifiable information will be the demographic questionnaire and contact information. The demographic questionnaire will be kept in a locked filing cabinet in the study PI’s locked office. The participants’ names and contact information will be linked to their unique study identification number and saved electronically on a secure VA crosswalk server at the Hines VA. This secure crosswalk file is the only place where the participants’ names will be linked with their unique participant identification number. The study PI will have control over the research team members who are allowed access to the crosswalk file.

Participants will receive up to US $230 for completing all study procedures. They will receive US $25 after completing the screening visit, US $25 after completing baseline assessments, US $30 for completing the MRI scan, US $50 after the 1st rTMS session, US $50 after the 10th rTMS session, US $25 after completing end-point assessments, and US $25 after completing follow-up assessments. This compensation is intended to facilitate participation without adding undue influence.