The relationship of neuroanatomy on resting motor threshold and induced electric field strength on treatment outcomes in mild to moderate traumatic brain injury patients during transcranial magnetic stimulation Open Access

Transcranial magnetic stimulation (TMS) is a non-invasive neuromodulation technique approved by the Food and Drug Administration to treat depression, migraine, smoking cessation, and obsessive-compulsive disorder.^1–3 Repetitive transcranial magnetic stimulation (rTMS) has previously shown effectiveness in treating neurological conditions by changing site-specific cortical excitability.^4,5 Stimulation to the dorsolateral prefrontal cortex (DLPFC) has shown improvements in depression symptoms, post-concussive symptoms, and sleep disorders.^6–10 However, treatment outcomes of TMS are known to vary, and the mechanisms behind this variance is poorly understood.^5,11,12

Resting motor threshold (RMT) as a percentage of maximum stimulator output (%MSO) is defined as the minimum stimulation required to elicit a motor evoked potential (MEP) of 50 microvolts in 5 out of ten attempts.^13–15 RMT is used as the primary dosage parameter for rTMS treatment protocols as stimulation is set between 80-130% of patients’ RMT.^5,15,16 However, variance in the inter- and intra-participant variability of RMT values is not well characterized.^17,18 Brain scalp distance (BSD) or coil to cortex distance has been previously shown to account for some of this variability but is not completely predictive.^5,19–22 Other participant-specific attributes, including age, gray matter volume, functional connectivity, and tractography-derived structural connectivity, have all shown mixed results in providing explanations for variations in RMT.^{5,19,21,23,24}

In our previous study, rTMS treatment showed improvements in 26 mild to moderate brain injury (mTBI) patients in depression, post-concussive symptoms, and sleep dysfunction.²⁵ In the present study, we developed anatomically accurate head models from the MRIs and used finite element analysis to investigate the correlation between empirically collected TMS treatment outcomes in depression, post-concussive symptoms, and sleep dysfunction alongside RMT. The goal was to increase understanding of the relationship between RMT with MRI-based anatomical features, including brain scalp distance (BSD), gray matter volume (GMV), age, and depolarized gray matter volume (DGMV). These relationships in mild to moderate traumatic brain injury patients (mTBI) could provide an increased understanding of both positive and negative treatment outcomes in TMS patients.

Veterans who met inclusion criteria resulted in 26 mTBI participants who fully completed rTMS sessions and were MRI scanned as described in Franke et al.²⁵ RMTs were collected as a dosage parameter for rTMS sessions using the NextStim NBS4 figure-of-eight double coil and navigation system.^25,26 All participants in the active group received a full week of either sham or active rTMS, followed by a washout period before returning for a full week of the remaining form of stimulation.²⁵ Participants and outcome assessors were double blinded to the type of stimulation received.²⁵ Active sessions with the coil positioned over the right DLPFC began with 80% of the participant’s RMT for the first session and increased to 100% of RMT for the duration of the six-week study. The sham stimulation consisted of intensity set at 25% of RMT and the coil tilted 90 degrees from the scalp. Participants and outcome assessors were blinded to treatment group.²⁵ T1-weighted MRI images were taken with a Philips Ingenia 3.0 Tesla Scanner for all 26 participants with 1 mm slices (Fig. 1).²⁵ Outcome assessments were completed at five different time points as described in Franke et al.²⁵ Among the significant improvements reported, three were included for this project: depression using the patient health questionnaire (PHQ-9), sleep dysfunction using Pittsburgh sleep quality index (PSQI), and post-concussive symptoms using the Neurobehavioral symptom inventory (NSI).²⁵ 

T1-weighted magnetic resonance imaging (MRI) was processed through SimNIBS’ mri2mesh pipeline (SimNIBS Developers 2019, v2.0.1). The mri2mesh pipeline reconstructs and centers a tetrahedral head mesh from T1-weighted images and segments head models into seven distinct segments. Gray matter, cerebellum, cerebrospinal fluid, skin, skull, ventricles, and white matter segments were all imported into Autodesk Meshmixer (AutoDesk, Inc., v11.2.37) where artifacts and abnormalities were smoothed (Fig. 2). Total gray matter volume (GMV) was taken in Meshmixer (AutoDesk, Inc., v11.2.37). The region of interest (ROI) was identified by the ‘inverted omega’ landmark characterizing the precentral gyrus as described in Yoursey et al. corresponding to the site of empirical RMT collection (Fig. 3(c)).²⁷ Brain scalp distance (BSD) was taken between the skin and gray matter at the same ROI in Meshmixer.^20,22

Segmented head models were imported into Sim4Life (Zurich Med Tech, v6.2.2) and assigned individual material properties based on the IT’IS LF database (IT’IS Foundation, v4.0) also found in Table I.²⁸ The NextStim NBS4 figure-of-8 double coil was designed to match the specifications given by NextStim.²⁶ To match the empirical RMT data collection, the coil was centered over the ROI at a 45° orientation to the coronal plane tangential to the skin as shown in Fig. 3(b).^5,27,29 Finite element analysis RMT simulations were run at 3.6 kHz and 4750 Amps to correspond to 100% of the stimulator output (%MSO).²⁶ For DLPFC simulations, the coil was centered over the region between the inferior and superior frontal sulcus corresponding to the connection between Brodman area 9 and Brodman area 49 as described in Mylius et al. (Fig. 4(b)).²⁹ DLPFC simulations were run at 3.6 kHz, and the current was set at each participant’s RMT to match the empirical setup (Fig. 4(c)).²⁶ Both sets of simulations had the coil set 5 mm from the skin, and voxels were created to ensure size uniformity throughout all head model segments.²² Sim4life surface viewer was used to locate the hotspot EFS and verify maximum intensity matched the desired target (Figs. 3(d) and 4(d)). Peak EFS was extracted as a magnitude, which has shown greater influence than directional EFS at inducing a response.³⁰ Depolarized gray matter volume (DGMV) was calculated as a percentage of voxels with magnitude of electric field strength above 100 V/m out of all gray matter voxels in MATLAB (Mathworks, R2022a) for PreCG simulations.²³ 100 V/m is the threshold induced electric field previously found to initiate depolarization in cortical pyramidal neurons.³¹ 

Linear regression was performed in MATLAB (MathWorks, vR2022a) using the Statistics and Machine Learning toolbox and the film function. Linear regressions between the empirically collected RMT and MRI-derived parameters (BSD, DGMV, and EFS of the motor cortex) were evaluated. Change scores for PHQ-9, NSI, and PSQI were calculated for pre-post treatment differences for both the sham and active groups.²⁵ The difference between active and sham change scores were used for linear regression with participant-specific measures.²⁵ 

Linear mixed effects models were used to identify differences in these relationships between RMT and MRI-derived parameters when accounting for one another as each is a different representation of neuroanatomical structure. Age was used as a random effect to account for cortical changes with aging.¹⁹ Participant ID was a random effect to account for relationships within data from repeated sampling.¹⁹ Linear mixed effects models were completed in R (The R Foundation, v 4.2.1) using the statistical package LME4.³² 

Linear mixed effects models were used to test the relationships between MRI-derived and baseline parameters (EFS at DLPFC and RMT) on outcome measures (PHQ, NSI, and PSQI). Sham data were excluded as the efficacy of rTMS had been previously shown and to prevent blunting these relationships because model derived parameters were individualized but not condition specific.²⁵ Time identified if the outcome measurement was pre-intervention, post-intervention, or follow-up. Age was used as a random effect to account for cortical changes with aging.¹⁹ Participant ID was a random effect to account for relationships within data from repeated sampling.¹⁹ RMT represented empirical responsiveness to TMS and was used as a random effect to address differences in responsiveness to individualized dosing with EFS derived from MRI using a simulated maximal stimulation to the motor cortex.⁵ Interactions between model-derived parameters and time were used to investigate if these parameters had predictive value of the effects of rTMS on the outcome measures.

Previously, we have reported rTMS treatment showed statistically significant improvements in depression (p = 0.037), post concussive symptoms (p = 0.030), and sleep dysfunction (p = 0.025).²⁵ In the present study, BSD, DGMV, GMV, and maximum EFS strength at both the PreCG and DLPFC locations were calculated for all 26 participants (Table II).

Linear regression showed associations between measured BSD and the following: (1) empirically collected RMT (R² = 0.290, p = 0.005) shown in Fig. 5; (2) simulated EFS (R² = 0.311, p = 0.003) shown in Fig. 6; (3) and calculated DGMV (R² = 0.256, p = 0.008) shown in Fig. 7. There was insufficient evidence of relationships between the outcome measures, i.e., PHQ-9, PSQI, or NSI, and any of the MRI-derived measures and RMT (Table III).

BSD had no effect on RMT when accounting for the effect of other MRI-derived metrics of neuroanatomical features (p = 0.069). DGMV had a positive effect on RMT (slope = 46.8, p < 0.001), and EFS had a weakly positive effect on RMT (slope = 0.316, p = 0.002). The effect of time validated the previous findings that rTMS did affect all three outcome measures, which were improvements in depression, post concussive symptoms, and sleep dysfunction.²⁵ There were no significant effects of EFS at DLPFC or RMT on the outcome measures (Table IV). Further, there were no effects of the MRI-derived metrics with an interaction with Time on any outcome measure (Table IV).

The goal of this study was to investigate the role of neuroanatomy on rTMS outcomes, specifically using RMT and MRI-derived parameters such as simulated induced electric field on mTBI participants. Relationships between BSD, RMT, EFS, and DGMV have been previously observed and shown to account for some of the variances in inter-patient RMT measurements.^5,19,21–23 However, these studies were conducted on healthy participants where neuroanatomy may influence TMS response (MEP and RMT) more than the resting functional status of the brain.

Linear regression modeling showed expected associations for BSD with RMT, EFS, and DGMV. This relationship was expected due to the exponential decay of the induced electric field with distance²⁸ where participants with a higher BSD also had a higher RMT but lower EFS and DGMV. The relationship between BSD and RMT has been previously observed where BSD is a weak indicator of RMT, and was confirmed in mTBI participants.^19,33

None of the MRI-derived parameters or RMT exhibited direct influence on rTMS outcomes or an assessment measure of mTBI. These findings serve to further validate the methodology for derivation of neuroanatomically affected parameters from MRI in mTBI. However, the results also show there is no obvious predictive value through the evaluated MRI-derived metrics or RMT, of rTMS effect on PHQ-9, NSI, or PSQI, in the context of mTBI. The lack of a direct relationship between MRI-derived parameters or RMT on rTMS outcomes and previous studies indicating multiple factors influencing TMS response led to the use of linear mixed effects modeling to account for multiple parameters’ influence on rTMS response.^5,19,23

The lack of a correlation between RMT and BSD was unexpected in linear mixed effects modeling. As the BSD increases, more magnetic field strength is needed to initiate neuronal firing and depolarization, hence higher current is needed. This suggests that BSD alone does not capture the effects of stimulation on the cortex, as previously discussed by Mittal et al.⁵ The correlation between DGMV and RMT was similarly unexpected, as larger stimulation needed to achieve RMT was expected to correlate with less cortical response to a simulated maximal stimulation. A possible explanation is that model overstimulation could have obscured the nuance of this relationship. Finally, the lack of correlation with EFS was unexpected, as we had previously determined that EFS was a more comprehensive parameter than BSD.⁵ The linear mixed effect model, however, showed that relating RMT to multiple parameters lessened the influence of BSD to insignificance, while EFS became more notable. This supports the conjecture that neurological architecture plays a role, not just distance as measured in BSD, in stimulation effects.

While the efficacy of rTMS on outcomes after TBI was reaffirmed in this study, the MRI-derived parameter of dose-scaled, rTMS-site-specific EFS did not relate to either immediate outcome measure values nor to changes in outcome measures after rTMS. RMT also had no notable relationship to outcome measures. This suggests that neither the metric of neuroanatomical architecture (EFS) nor the representation of baseline TMS responsiveness (RMT) serve as appropriate biomarkers for any of the outcome measures evaluated. EFS and RMT also did not serve to predict the effect of rTMS on changes in outcome measures in this population of individuals with TBI.

This study was limited by the number of participants and the lack of a negative control or healthy population group. Follow up should be completed with larger cohorts and negative controls. This study was also limited by the lack of T2-weighted imaging, which would improve model quality, and diffusion tensor imaging (DTI). DTI would allow for fiber tractography; cortical fiber tract architecture has been previously shown to connect MRI-derived parameters with rTMS outcomes.^5,23 The role of medication, alcohol and the presence of a neurological or psychiatric condition have shown significant variations in RMT and motor cortical excitability which were not addressed and could have influenced results.^34–36 

Future studies investigating the relationship between RMT, maximum EFS, and treatment outcome in participants with the above conditions should consider functional connectivity as an independent variable. This Variations in Z-scores or independent component analysis in functional MRI (fMRI) or electroencephalography (EEG) delta power spectra coefficient have been reported in the literature for pre- and post-TMS treatments.^25,37 These functional connectivity parameters may have a higher influence on RMT and motor cortical excitability than the neuroanatomy parameters such as BSD, DGMV, or volumetric BSD.

The present study investigated how participant-specific features including age, GMV, BSD, DGMV, and maximum EFS at the PreCG influenced the primary dosing parameter (RMT) used in TMS treatment protocols in 26 mTBI patients. We also investigated how maximum EFS at the DLPFC and RMT influenced rTMS treatment outcomes, including depression, sleep dysfunction, and post-concussive symptoms in these same participants. We found BSD to be a potential indicator of RMT, simulated maximum EFS at the PreCG and calculated DGMV as a single metric. Future investigations regarding inter-participant variability of RMT should include neuroanatomical features, functional connectivity, white matter tractography-based structural connectivity, and EEG data.

Research operations were supported by the Commonwealth Cyber Initiative (Proposal No. FP00010500). This work was also supported by the U.S. Department of Veterans Affairs Award No. I01 CX002097. The U.S. Army Medical Research Acquisition Activity, 839 Chandler Street, Fort Detrick, MD 21702, is the awarding and administering acquisition office. Opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the Department of Defense.

Dr. Hadimani has two granted patents on TMS coils (US10792508B2 and US11547867B2), one granted patent on anatomically accurate brain phantom (US11373552B2), and one patent published and pending on TMS coil, US Patent Application (US20220241605A1).

All procedures were approved by the Institutional Review Boards of the McGuire VA Medical Center and Virginia Commonwealth University, and the trial was registered at clinical trials.gov with identifier NCT03642158. All participants gave informed consent for the present study.

Connor J. Lewis: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal); Writing – review & editing (equal). Laura M. Franke: Conceptualization (equal); Data curation (lead); Investigation (equal); Methodology (equal); Writing – review & editing (equal). Joseph V. Lee: Investigation (equal); Methodology (equal); Writing – review & editing (equal). Neil Mittal: Formal analysis (equal); Writing – original draft (equal); Writing – review & editing (equal). George T. Gitchel: Investigation (equal); Methodology (equal); Writing – review & editing (equal). Robert A. Perera: Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – review & editing (equal). Kathryn L. Holloway: Investigation (equal); Methodology (equal); Writing – review & editing (equal). William C. Walker: Investigation (equal); Methodology (equal); Writing – review & editing (equal). Carrie L. Peterson: Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – review & editing (equal). Ravi L. Hadimani: Conceptualization (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Resources (equal); Software (equal); Supervision (equal); Validation (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

BSD

brain scalp distance

DGMV

depolarized gray matter volume

DLPFC

dorsolateral prefrontal cortex

EFS

electric field strength

GMV

gray matter volume

LMEM

linear mixed effects modeling

MRI

magnetic resonance imaging

MSO

maximum stimulator output

mTBI

mild to moderate traumatic brain injury

NSI

neurobehavioral symptom inventory

PHQ-9

patient health questionnaire

PreCG

precentral gyrus

PSQI

Pittsburgh sleep quality index

RMT

resting motor threshold

ROI

region of interest

rTMS

repetitive transcranial magnetic stimulation

TMS

transcranial magnetic stimulation