Advertisement

Transcranial Magnetic Stimulation-induced Heart-Brain-Coupling: Implications for site selection and frontal thresholding – preliminary findings

Open AccessPublished:January 24, 2023DOI:https://doi.org/10.1016/j.bpsgos.2023.01.003

      ABSTRACT

      Background

      Neuro-cardiac-guided transcranial magnetic stimulation (NCG-TMS) employs repetitive TMS (rTMS)-induced heart rate deceleration to confirm activation of the frontal-vagal pathway. Here we test a novel NCG-TMS method that utilizes Heart-Brain-Coupling (HBC) in order to quantify rTMS-induced entrainment of the inter-beat-interval as a function of TMS cycle-time. Since prior NCG-TMS studies indicated no association between motor and frontal excitability threshold, we also introduce the approach of using HBC to establish individualized frontal excitability thresholds for optimally dosing frontal TMS.

      Methods

      In study 1A and 1B we validated intermittent theta-burst stimulation (iTBS)-induced HBC (2s iTBS-on; 8s off: HBC=0.1Hz) in fifteen (1A) and twenty-two (1B) patients with major depressive disorder from two double-blind placebo-controlled studies. In study 2, HBC was measured in ten healthy subjects during the 10Hz "dash”-protocol (5s 10Hz-on; 11s off: HBC=0.0625Hz) applied with fifteen increasing intensities to four evidence-based TMS locations.

      Results

      Using blinded electrocardiogram-based HBC-analysis, we successfully identified sham from real-iTBS sessions (accuracy Study 1A=83%, Study 1B=89.5%) and found a significantly stronger HBC at 0.1Hz in active compared to sham iTBS (d=1.37) (Study 1A). In study 2, clear dose-dependent entrainment (p=.002) was observed at 0.0625Hz in a site-specific manner.

      Conclusions

      We demonstrated rTMS-induced HBC as a function of TMS cycle-time for two commonly used clinical protocols (iTBS and 10Hz-dash). These preliminary results supported individual site-specificity and dose-response effects, indicating that this is a potentially valuable method for clinical rTMS site-stratification and frontal-thresholding. Further research should control for TMS side-effects, such as pain of stimulation, to confirm these findings.

      Keywords

      INTRODUCTION

      Repetitive transcranial magnetic stimulation (rTMS) is a non-invasive neuromodulation technique, which is increasingly used as intervention for the treatment of refractory major depressive disorder (MDD) (
      • Donse L.
      • Padberg F.
      • Sack A.T.
      • Rush A.J.
      • Arns M.
      Simultaneous rTMS and psychotherapy in major depressive disorder: Clinical outcomes and predictors from a large naturalistic study.
      ). rTMS for MDD is often targeted to the dorsolateral prefrontal cortex (DLPFC), and clinical response is thought to be mediated by network connectivity between the DLPFC and the subgenual anterior cingulate cortex (sgACC)(
      • Fox M.D.
      • Buckner R.L.
      • White M.P.
      • Greicius M.D.
      • Pascual-Leone A.
      Efficacy of Transcranial Magnetic Stimulation Targets for Depression Is Related to Intrinsic Functional Connectivity with the Subgenual Cingulate.
      ). There is great potential to optimize rTMS parameters and move towards an individualized approach of rTMS therapy, including optimizing target engagement and parameters such as rTMS pattern, intensity, and frequency of stimulation (
      • Caulfield K.A.
      • Brown J.C.
      The Problem and Potential of TMS’ Infinite Parameter Space: A Targeted Review and Road Map Forward.
      ,
      • Fitzgerald P.B.
      • Hoy K.
      • Gunewardene R.
      • Slack C.
      • Ibrahim S.
      • Bailey M.
      • et al.
      A randomized trial of unilateral and bilateral prefrontal cortex transcranial magnetic stimulation in treatment-resistant major depression.
      ).
      It is well known that MDD is associated with an increased risk of cardiovascular disease, and heart rate (HR) is often dysregulated in MDD, quantified by a higher HR and lower HR variability (HRV) (
      • Iseger T.A.
      • Bueren NER van
      • Kenemans J.L.
      • Gevirtz R.
      • Arns M.
      A frontal-vagal network theory for Major Depressive Disorder: Implications for optimizing neuromodulation techniques.
      ). Relatedly, we recently proposed the frontal-vagal network theory for MDD, stating that major hubs such as the DLPFC, sgACC and vagus nerve (VN) share overlap with the networks involved in autonomic control as well as in MDD(5). The neuro-anatomical framework of the heart-brain connection in MDD is provided in (
      • Iseger T.A.
      • Bueren NER van
      • Kenemans J.L.
      • Gevirtz R.
      • Arns M.
      A frontal-vagal network theory for Major Depressive Disorder: Implications for optimizing neuromodulation techniques.
      ). Based on this theory, a new target-engagement method for treatment with rTMS was recently proposed by Iseger and colleagues(
      • Iseger T.A.
      • Padberg F.
      • Kenemans J.L.
      • Gevirtz R.
      • Arns M.
      Neuro-Cardiac-Guided TMS (NCG-TMS): Probing DLPFC-sgACC-vagus nerve connectivity using heart rate - First results.
      ), called neuro-cardiac guided TMS (NCG-TMS). In NCG-TMS, HR-deceleration in response to 10Hz-rTMS or intermittent theta burst stimulation (iTBS) is used to confirm activation of the frontal-vagal network in a site-specific manner, with HR-deceleration at sites often used in TMS (e.g. F3, FC3) and HR-acceleration in control sites such as the motor (C3, C4) or parietal (Pz) cortex (
      • Iseger T.A.
      • Padberg F.
      • Kenemans J.L.
      • Gevirtz R.
      • Arns M.
      Neuro-Cardiac-Guided TMS (NCG-TMS): Probing DLPFC-sgACC-vagus nerve connectivity using heart rate - First results.
      ) .
      This finding has now been replicated in healthy controls (
      • Iseger T.A.
      • Bueren NER van
      • Kenemans J.L.
      • Gevirtz R.
      • Arns M.
      A frontal-vagal network theory for Major Depressive Disorder: Implications for optimizing neuromodulation techniques.
      ,
      • Iseger T.A.
      • Padberg F.
      • Kenemans J.L.
      • Gevirtz R.
      • Arns M.
      Neuro-Cardiac-Guided TMS (NCG-TMS): Probing DLPFC-sgACC-vagus nerve connectivity using heart rate - First results.
      ,
      • Kaur M.
      • Michael J.A.
      • Hoy K.E.
      • Fitzgibbon B.M.
      • Ross M.S.
      • Iseger T.A.
      • et al.
      Investigating high- and low-frequency neuro-cardiac-guided TMS for probing the frontal vagal pathway.
      ,
      • Iseger T.A.
      • Padberg F.
      • Kenemans J.L.
      • Dijk H van
      • Arns M.
      Neuro-Cardiac-Guided TMS (NCG TMS): A replication and extension study.
      ) and MDD patients(
      • Zwienenberg L.
      • Iseger T.A.
      • Dijkstra E.
      • Rouwhorst R.
      • Dijk H van
      • Sack A.T.
      • et al.
      Neuro-cardiac guided rTMS as a stratifying method between the ‘5cm’ and ‘BeamF3’ stimulation clusters.
      ), and therefore holds promise as a possible biomarker of response and target-engagement method allowing for determination of the best prefrontal rTMS target in MDD treatment(
      • Zwienenberg L.
      • Iseger T.A.
      • Dijkstra E.
      • Rouwhorst R.
      • Dijk H van
      • Sack A.T.
      • et al.
      Neuro-cardiac guided rTMS as a stratifying method between the ‘5cm’ and ‘BeamF3’ stimulation clusters.
      ). This is supported by preliminary results where HR-deceleration at a first rTMS session was associated with clinical response post-treatment (
      • Iseger T.A.
      • Arns M.
      • Downar J.
      • Blumberger D.M.
      • Daskalakis Z.J.
      • Vila-Rodriguez F.
      Cardiovascular differences between sham and active iTBS related to treatment response in MDD.
      ). As an intermediate step, we recently proposed that NCG-TMS could be employed as an enhanced rTMS stratification technique to select between one of two evidence-based DLPFC-sites: the Beam-cluster (Beam F3 and F4) and 5cm-rule (5CM) method (
      • Zwienenberg L.
      • Iseger T.A.
      • Dijkstra E.
      • Rouwhorst R.
      • Dijk H van
      • Sack A.T.
      • et al.
      Neuro-cardiac guided rTMS as a stratifying method between the ‘5cm’ and ‘BeamF3’ stimulation clusters.
      ).
      Importantly, iTBS showed stronger effects on HR relative to standard 10Hz-rTMS (iTBS: 8.4 vs 10Hz: 1.9 BPM deceleration)(
      • Iseger T.A.
      • Padberg F.
      • Kenemans J.L.
      • Dijk H van
      • Arns M.
      Neuro-Cardiac-Guided TMS (NCG TMS): A replication and extension study.
      ,
      • Iseger T.A.
      • Arns M.
      • Downar J.
      • Blumberger D.M.
      • Daskalakis Z.J.
      • Vila-Rodriguez F.
      Cardiovascular differences between sham and active iTBS related to treatment response in MDD.
      ), but it also resulted in more side effects, such as light-headedness, emotional reactions, and painfulness. For this reason, we set-out to improve the current 10Hz NCG-TMS method. Recently, the FDA approved the 10Hz “dash”-protocol(
      • Carpenter LindaL.
      • ScottT Aaronson
      • Hutton T.M.
      • Mina M.
      • Pages K.
      • Verdoliva S.
      • et al.
      Comparison of clinical outcomes with two Transcranial Magnetic Stimulation treatment protocols for major depressive disorder.
      ), where the inter-train interval (ITI) was shortened to 11s, allowing for more rapid delivery of the stimulation trains during the session.
      Given that effects of TMS-induced HR-deceleration are nearly immediate, since the vagus nerve is myelinated (whereas the sympathetic inputs to the heart are unmyelinated; for review see (
      • Iseger T.A.
      • Bueren NER van
      • Kenemans J.L.
      • Gevirtz R.
      • Arns M.
      A frontal-vagal network theory for Major Depressive Disorder: Implications for optimizing neuromodulation techniques.
      )), we hypothesized that specific TMS-protocol parameters could induce entrainment of the cardiac rhythm as a function of TMS cycle-time (Figure 1), as described in (
      • Thut G.
      • Veniero D.
      • Romei V.
      • Miniussi C.
      • Schyns P.
      • Gross J.
      Rhythmic TMS Causes Local Entrainment of Natural Oscillatory Signatures.
      ) for rTMS and EEG alpha oscillations. If entrained, the 2s (iTBS) or 5s (10Hz) stimulation would decelerate HR during and immediately after each train, followed by a recovery of the HR during the ITI. Over the session with repeated trains, an entrainment of the inter-beat-interval at a specific frequency could be observed. The heart-brain directionality of TMS would thus be top-down from the brain on the HR(13). This hypothesis is visualized in Figure 1 for the 10Hz dash-protocol (top) and iTBS (bottom). This novel ‘Heart-Brain-Coupling’ (HBC) marker is proposed, since HBC is independent of ‘baseline’ HR state and other influences on HR, such as respiratory sinus arrhythmia (RSA), making it a more reliable and stable biomarker. Furthermore, RSA is generally faster (0.15–0.4 Hz) than these TMS cycle-times (
      • Candia-Rivera D.
      • Catrambone V.
      • Barbieri R.
      • Valenza G.
      Functional assessment of bidirectional cortical and peripheral neural control on heartbeat dynamics: A brain-heart study on thermal stress.
      ).
      Figure thumbnail gr1
      Figure 1rTMS-induced Heart-Brain-Coupling. This figure visualizes rTMS-induced entrainment of the inter-beat-interval as a function of rTMS cycle-time. The cycle-time of the 10Hz dash-protocol comprises one rTMS train every 16 seconds (5s stimulation on, 11s stimulation off). This results in a specific entrainment frequency of 1/16 = 0.0625Hz. For iTBS, the cycle-time is 10 seconds (2s on and 8s off), resulting in an entrainment frequency of 1/10 = 0.1Hz. The neuro-anatomical framework of the heart-brain connection illustrated on the left is provided in (
      • Iseger T.A.
      • Bueren NER van
      • Kenemans J.L.
      • Gevirtz R.
      • Arns M.
      A frontal-vagal network theory for Major Depressive Disorder: Implications for optimizing neuromodulation techniques.
      ). rTMS=Repetitve transcranial magnetic stimulation, iTBS=intermittent theta-burst stimulation.
      We thus set-out to test the above predictions for iTBS and the 10Hz dash-protocol in three independent studies. Finally, as prior NCG-TMS studies indicate that the frontal excitability threshold (FT) was not associated with the motor threshold (MT, which is used to ‘dose’ prefrontal rTMS treatment) (
      • Iseger T.A.
      • Padberg F.
      • Kenemans J.L.
      • Dijk H van
      • Arns M.
      Neuro-Cardiac-Guided TMS (NCG TMS): A replication and extension study.
      ), we also tested if this method could be utilized to establish an FT. Two previous studies investigated the dose-response relationship of stimulation intensity of DLPFC-iTBS and clinical response. In healthy individuals, the largest neurophysiological changes were found at 75%MT, compared to 50% and 100%MT (
      • Chung S.W.
      • Rogasch N.C.
      • Hoy K.E.
      • Sullivan C.M.
      • Cash R.F.H.
      • Fitzgerald P.B.
      Impact of different intensities of intermittent theta burst stimulation on the cortical properties during TMS‐EEG and working memory performance.
      ) and subthreshold iTBS resulted in a larger decrease of depressive symptoms than suprathreshold iTBS (
      • Lee J.C.
      • Corlier J.
      • Wilson A.C.
      • Tadayonnejad R.
      • Marder K.G.
      • Ngo D.
      • et al.
      Subthreshold stimulation intensity is associated with greater clinical efficacy of intermittent theta-burst stimulation priming for Major Depressive Disorder.
      ). These studies imply that higher TMS intensities could lead to attenuated effects and research is needed to define the individual sweet spot of stimulation intensity.

      METHODS AND MATERIALS

      Study 1: iTBS induced Heart-Brain-Coupling

      To test the specific entrainment of iTBS on the cardiac rhythm, we used data from two previous iTBS studies: Baseline ECG data collected at a Canadian health science center (author FVR) from a double-blind placebo controlled iTBS trial (CARTBIND; ClinicalTrials.gov Identifier: NCT02729792), reported in more detail in Blumberger et al. (
      • Blumberger D.M.
      • Vila-Rodriguez F.
      • Wang W.
      • Knyahnytska Y.
      • Butterfield M.
      • Noda Y.
      • et al.
      A randomized sham controlled comparison of once vs twice-daily intermittent theta burst stimulation in depression: A canadian rTMS treatment and biomarker network in depression (CARTBIND) study.
      ) and Iseger et al.(
      • Iseger T.A.
      • Arns M.
      • Downar J.
      • Blumberger D.M.
      • Daskalakis Z.J.
      • Vila-Rodriguez F.
      Cardiovascular differences between sham and active iTBS related to treatment response in MDD.
      ) (Study 1A), and ECG data from a double-blind randomized controlled trial performed at the Department of Psychiatry at Stanford University (SNT: ClinicalTrials.gov Identifier: NCT03068715), reported in more detail in Cole et al.(
      • Cole E.J.
      • Phillips A.L.
      • Bentzley B.S.
      • Stimpson K.H.
      • Nejad R.
      • Barmak F.
      • et al.
      Stanford Neuromodulation Therapy (SNT): A Double-Blind Randomized Controlled Trial.
      ) (Study 1B). In summary, the methods comprised the following:

      Participants

      Fifteen (Study 1A) and twenty-two (Study 1B) MDD patients were included, and all participants provided written informed consent.

      iTBS device and protocol

      iTBS-treatment localization was MRI-based, and rTMS was applied with a MagPro X100 system (Magventure, Denmark) equipped with a B70 figure-of-eight coil (Study 1A) and a double-sided Cool-B65 A/P coil (Study 1B).
      Study 1A: All subjects received two sessions of both sham and active iTBS. The active condition comprised stimulation at the DLPFC location with triplet 50Hz bursts, repeated at 5Hz; 2s on and 8s off; 600 pulses per session. A sham (internally shielded) coil without active electrical stimulation was positioned over the vertex during the sham condition. Subjects were randomized to one of two treatment arms (arm A: sham-active, 54 min. pause, active-sham, or arm B: sham-sham, 54 min. pause, active-active). ECG was acquired simultaneous with iTBS stimulation using a Biopac MP 150 system (Biopac Systems Inc., Goleta, CA, USA) comprising modular hardware and “AcqKnowledge” software.
      Study 1B: Subjects received either active or sham “Stanford Neuromodulation Therapy” (SNT): A high-dose iTBS protocol over the DLPFC, which consisted of ten sessions (18,000 pulses) per day, on 5 consecutive days. Both sham and active iTBS were targeted at the left DLPFC. Participants and study staff were blinded to treatment assignments. All iTBS sessions utilized the same stimulation coil, with no indication of active or sham orientation. During sham, participants wore noise-cancelling earphones connected to a sham noise generator to simulate the stimulation noise pattern. Additionally, lidocaine was applied to the stimulation site to reduce sensation. Spontaneous side effects were recorded daily. ECG was acquired simultaneous with iTBS using NCG-engage (neuroCare), only the first recording for each participant was used for analysis.

      Coil positioning and MT

      Study 1A

      The target location was specified by reverse co-registration from a stereotaxic coordinate on the standard Montreal neurological Institute (MNI-152) template brain, onto each individual anatomical MRI. MNI coordinates for left DLPFC were [x-38 yþ44 zþ26], drawn from a study identifying this site as optimal based on clinical outcomes and resting-state functional connectivity (
      • Fox M.D.
      • Buckner R.L.
      • White M.P.
      • Greicius M.D.
      • Pascual-Leone A.
      Efficacy of Transcranial Magnetic Stimulation Targets for Depression Is Related to Intrinsic Functional Connectivity with the Subgenual Cingulate.
      ). Stimulation was delivered at 120% MT.

      Study 1B

      During active and sham stimulation, the coil was positioned over the DLPFC using MRI-guided neuronavigation. The stimulation intensity at 90% MT was adjusted for depth of the identified MRI-target. For safety, stimulation intensity never exceeded 120% MT.

      Analysis

      From the ECG data of the first session, RR intervals (the time intervals between consecutive heartbeats) were determined using Kubios Premium (© 2022 Kubios OY, version 3.0.2(18)), after which the data was analyzed using a custom-built analysis package (https://github.com/brainclinics/NCGTMS-2.0) in Python (©2001-2022 Python Software Foundation). In short, HR was computed for each RR interval and HR was interpolated for each ECG timepoint by a moving average of 5 consecutive HRs. Resulting HRs were convolved with a Hann window of 1.5 seconds (Figure 2A). Time-frequency representations (TFR) were computed using the tfr_array_morlet function from the MNE-Python package (
      • Gramfort A.
      • Luessi M.
      • Larson E.
      • Engemann D.A.
      • Strohmeier D.
      • Brodbeck C.
      • et al.
      MEG and EEG data analysis with MNE-Python.
      ) using a frequency range of 0.02 to 0.18 Hz in steps of 5e-4Hz, with three cycles yielding results in high time resolution (Figure 2B) and ten cycles for high frequency resolution (Figure 2C). Before TFR-analysis, data was padded with the first block of rTMS and rest period of data at the beginning, and the last block of data at the end to allow for analysis at these low frequencies. To compute HBC, the mean power (μV2) at 0.1Hz was computed for each block and averaged.
      Figure thumbnail gr2
      Figure 2HBC report of active and sham iTBS condition. This overview visualizes the effect of one active and one sham iTBS session on HR (panel A) over 3-min time (x-axis), the 0.1Hz high time resolution (panel B) and the 0.1Hz high frequency resolution (panel C). The vertical lines represent the start of a stimulation train. Panel A visualizes that during stimulation, the HR consistently decreases, and normalizes during the ITI, in line with the hypothesized effects in . This rTMS-induced cardiac rhythm of 0.1Hz (green line) is also clearly visible as increased power (red) in the high time and high frequency plots. HBC=Heart-brain-coupling, iTBS=intermittent theta-burst stimulation, HR=heart rate.

      Study 1A

      For three subjects, insufficient ECG data was available, resulting in n=12 included in the analysis. Researcher HvD - blinded to group assignment and stimulation details - processed the ECG data based on the above hypothesis of 0.1Hz entrainment in RR-signal and made predictions if ECG data belonged to sham or real stimulation within each subject. These predictions were examined after data-lock by an unblinded researcher (MA). Next, a repeated measures ANOVA was conducted to test the effect of active and sham iTBS on the HBC marker (0.1Hz).

      Study 1B

      Three participants were excluded for low quality ECG, resulting in n=19. The ECG was analysed by blinded researcher HvD, who used the same method as in Study 1A to classify their active/sham status.

      Study 2: DASH (10Hz) NCG-TMS

      Participants

      Ten healthy participants (seven female, aged 25-46 [mean 34.0] years) were included in this study after providing written informed consent and meeting safety criteria for rTMS. Exclusion criteria for both studies were (
      • Donse L.
      • Padberg F.
      • Sack A.T.
      • Rush A.J.
      • Arns M.
      Simultaneous rTMS and psychotherapy in major depressive disorder: Clinical outcomes and predictors from a large naturalistic study.
      ) neurological/psychiatric disease, (
      • Fox M.D.
      • Buckner R.L.
      • White M.P.
      • Greicius M.D.
      • Pascual-Leone A.
      Efficacy of Transcranial Magnetic Stimulation Targets for Depression Is Related to Intrinsic Functional Connectivity with the Subgenual Cingulate.
      ) age under 18 years and (
      • Caulfield K.A.
      • Brown J.C.
      The Problem and Potential of TMS’ Infinite Parameter Space: A Targeted Review and Road Map Forward.
      ) standard exclusion criteria for rTMS, such as epilepsy. This study was approved by the local Ethics Committee of Maastricht University.

      rTMS device and protocol

      rTMS was applied with either a DuoMag XT-100 system (Deymed Diagnostic, Czech Republic) or a MagPro R20 system (Magventure, Denmark), both equipped with a focal figure-of-eight coil. Stimulation was applied in trains of 10Hz for 5s with an ITI of 11s (“dash”-protocol). Subjects were presented with an intensity-sweep of fifteen stimulation trains from low to high intensities defined in 2% machine output (%MSO) steps, with one stimulation train (10Hz for 5s + 11s ITI) per intensity, and with the 15th intensity matching 120%MT. This stimulation train was preceded by 16s of no-stimulation (total 256s) and the starting intensity of machine output was thus defined as 28%MSO below 120%MT (Figure 3). This stimulation sequence was applied to four different locations (Beam and 5CM, left and right hemisphere; the starting location was randomized between subjects). During rTMS sessions, HR was simultaneously measured using an H10 Polar band (© Polar Electro 2022) connected through Bluetooth with the ECG recorder app (© 2021 Philipp Pöml, version 1.4).
      Figure thumbnail gr3
      Figure 3Visualization of the intensity-sweep in study 2. This figure shows the intensity-sweep that was applied in this pilot study 2, using fifteen increasing intensities with 2%MSO-steps. Step 15 corresponds to 120%MT. This rTMS-protocol was applied on four sites: the Beam and 5CM location on both hemispheres. %MSO=% of machine outpout, MT=motor threshold. rTMS=repetitive transcranial magnetic stimulation, 5CM location=the location 5cm anterior to the scalp position for optimal activation of the first dorsal interosseus muscle in a para-sagittal line.

      Coil positioning & MT

      The rTMS-coil was positioned over the Beam and 5CM location at a 45° angle relative to the parasagittal plane (the coil handle pointing posteriorly), which seems to be the optimal angle to stimulate frontal areas (
      • Thomson R.H.
      • Cleve T.J.
      • Bailey N.W.
      • Rogasch N.C.
      • Maller J.J.
      • Daskalakis Z.J.
      • et al.
      Blood Oxygenation Changes Modulated by Coil Orientation During Prefrontal Transcranial Magnetic Stimulation.
      ) and which is used in most depression trials (
      • Opitz A.
      • Fox M.D.
      • Craddock R.C.
      • Colcombe S.
      • Milham M.P.
      An integrated framework for targeting functional networks via transcranial magnetic stimulation.
      ). If the distance between these two locations was smaller than one cm, the Beam and 5CM location on that hemisphere were taken as one location (n=1). The 5CM location is defined as the location 5cm anterior to the scalp position for optimal activation of the first dorsal interosseus muscle in a para-sagittal line. Beam locations were defined using the Beam-F3 algorithm and software (
      • Beam W.
      • Borckardt J.J.
      • Reeves S.T.
      • George M.S.
      An efficient and accurate new method for locating the F3 position for prefrontal TMS applications.
      ). MT intensity was defined as the lowest stimulation intensity that in four trials, induced at least two visible twitches in the contralateral hand.

      Analysis

      Data were analyzed using the custom-built analysis package described above, now including the determination of the RR intervals. For this, ECG was bandpass-filtered with a bidirectional 4th order Butterworth filter between 5 and 49Hz, after which signal.find_peaks (Scipy (
      • Virtanen P.
      • Gommers R.
      • Oliphant T.E.
      • Haberland M.
      • Reddy T.
      • Cournapeau D.
      • et al.
      SciPy 1.0: Fundamental Algorithms for Scientific Computing in Python.
      )) was used to detect R-peaks. RR intervals were then corrected for ectopic beats and subsequent analysis was performed as described above, in this case for 0.0625Hz. HBC was quantified across the intensity-sweep and four sites and visualized in a custom HBC-marker report (Figure 4 and supplement). Primary outcome measure was the HBC-marker for the 10Hz dash-protocol (0.0625Hz). Using the HBC reports, the ‘best’ location was determined for every subject, based on the location with the highest average oscillatory power (μV2) at 0.0625Hz (i.e. over all intensities) (supplement). Data for all target locations and for the best location was subsequently analyzed in SPSS with a repeated measures ANOVA, using within subject factors Intensity (0-15) and Location (BF3, BF4 and 5CM left and right).
      Figure thumbnail gr4
      Figure 4HBC report example. This figure shows HBC-power of 0.0625Hz over time during the intensity-sweep for one location in two different individuals (Panel A1 and A2). Step 0-15 indicates the intensity-sweep of increasing stimulator output, where step 0 is no stimulation. HBC is seen later in time for individual A1 compared to individual A2, suggesting individual differences in low and high threshold response. Panel B shows the summary of HBC-power of 0.0625Hz during the intensity-sweep for all stimulation locations for individual A1. Here, BF4 would be the selected best location given the highest average oscillatory power at 0.0625 Hz (“pow”) at that location. HBC=Heart-brain-coupling, BF4=Beam F4 location.

      RESULTS

      Study 1A

      The blinded classifications on sham and active iTBS conditions from the ECG of each subject were highly accurate with 10/12 predicted correctly by the blinded researcher (accuracy=83%), suggesting successful unblinding of stimulation only by inspection of the ECG-based HBC-marker.
      Two examples of HBC-output are seen in Figure 2 for a representative example of sham and real iTBS in the same subject. This overview visualizes the effect of one 3-min iTBS session on mean-HR (panel A) and 0.1Hz HBC, visualized using a high time resolution (panel B) and high frequency resolution (panel C). Vertical lines represent the start of each stimulation train. During active iTBS, HR decelerates during every stimulation train (2s) and normalizes afterwards (8s), until the next stimulation train, resulting in an rTMS-entrained HR oscillation at 0.1Hz (horizontal green line) also visible as increased power (red) in the high time and high frequency plots. This is not seen during sham stimulation.
      Repeated measures ANOVA yielded a main effect of Stimulation (F(1,10)=16.318; p=.002, d=1.37) and a main effect of Time (F(1,10)=11.767, p=.006; d=0.54), but no Stimulation X Time interaction (p=.318). Median and mean self-rated pain scores during active stimulation, on a scale from 1 [no pain] to 10 [intolerable pain], were higher (4.0 and 4.3, IQR=2.7-6.0, SD=2.0) than during sham stimulation (1.0 and 1.1, IQR=1.0-1.1, SD=0.1)

      Study 1B

      The generated HBC values predicted active/sham conditions with 89.5% accuracy. These results show that iTBS stimulation can be clearly distinguished from sham iTBS, based on the ECG-based HBC-marker only, also when sham and active stimulation are targeted at the exact same location. Of measured spontaneous side effects, there was only a higher incidence of headache in the active iTBS group compared to sham (Fisher’s exact test, p<0.06). Participants did not guess their treatment allocation beyond chance in both groups.

      Study 2

      The repeated measures ANOVA on data of all locations showed a significant effect of Intensity (F(15,135)=3.540, p<.001) on the HBC-marker, and no effect of Location (p=0.997) or Location x Intensity interaction (p=0.955). The repeated measures ANOVA conducted on data for the best location per individual, demonstrated a significant effect of Intensity (F(15,135)=3.718, p<.001). The HBC-peak for the best location, defined as the stimulation intensity at which the HBC reached the highest oscillatory power at 0.0625 Hz, reaches its maximum at intensity 8, with a large effect size (d = 2.40) (Figure 5). The data in Figure 5 is derived from the individual HBC reports (supplement), which were normalized to a 0-1 scale. The Beam F3 and F4 were the individual best location for two subjects each, the 5CM left and right for three subjects each. We investigated whether order of stimulation had an impact on the HBC effect at the different rTMS sites. No order-effect was found, implying that HBC effects were independent of which site was stimulated first.
      Figure thumbnail gr5
      Figure 5Dose-response effect of increasing stimulation intensities on HBC. Dose-response effect of increasing intensities of machine output on the HBC-marker in all locations (
      • Donse L.
      • Padberg F.
      • Sack A.T.
      • Rush A.J.
      • Arns M.
      Simultaneous rTMS and psychotherapy in major depressive disorder: Clinical outcomes and predictors from a large naturalistic study.
      ,
      • Fox M.D.
      • Buckner R.L.
      • White M.P.
      • Greicius M.D.
      • Pascual-Leone A.
      Efficacy of Transcranial Magnetic Stimulation Targets for Depression Is Related to Intrinsic Functional Connectivity with the Subgenual Cingulate.
      ,
      • Caulfield K.A.
      • Brown J.C.
      The Problem and Potential of TMS’ Infinite Parameter Space: A Targeted Review and Road Map Forward.
      ,
      • Fitzgerald P.B.
      • Hoy K.
      • Gunewardene R.
      • Slack C.
      • Ibrahim S.
      • Bailey M.
      • et al.
      A randomized trial of unilateral and bilateral prefrontal cortex transcranial magnetic stimulation in treatment-resistant major depression.
      ) and the individual best location (
      • Iseger T.A.
      • Bueren NER van
      • Kenemans J.L.
      • Gevirtz R.
      • Arns M.
      A frontal-vagal network theory for Major Depressive Disorder: Implications for optimizing neuromodulation techniques.
      ). The HBC-effect for the best location has a higher maximum than the separate locations. The data in is derived from the individual HBC reports (supplement A), which were normalized to a 0-1 scale. HBC=Heart-brain-coupling.
      The results indicate a clear dose-response effect of rTMS-induced HBC, with no difference between sites at the group level. This is as expected, as we are investigating a method to individualize rTMS targeting, based on the hypothesis that large inter-individual differences exist. When the best site was chosen for each individual participant, a strong dose-response effect was observed. The dose-response relationship follows an inverted U-curve. More importantly, the data visualized in Figure 5 suggest large inter-individual differences in FT, also confirmed by individual trajectories as visualized in Figure 4 and 6. The peak effect of stimulation on the HBC-marker arises at different stimulation strengths for individual subjects (Figure 6), with the greatest difference between subject 10 (intensity 7) and 1 (intensity 15). These HBC-peaks correspond with a relative %MT of 81% for subject 10 and 120% for subject 1. Importantly, for most of the individual subjects, the greatest effect arises sooner than step 15, which is currently the standard stimulation strength in clinical practice applied to frontal rTMS.
      Figure thumbnail gr6
      Figure 6Individual trajectories of the HBC-marker over time. Dose-response effect of increasing intensities of machine output on the HBC-marker in all subjects. HBC=Heart-brain-coupling.

      DISCUSSION

      This study is the first to test and validate a novel NCG-TMS method utilizing Heart-Brain-Coupling, that quantifies rTMS-induced entrainment of the heart inter-beat-interval as a function of TMS cycle-time (Figure 1). The proposed model makes clear predictions about two commonly used clinical rTMS protocols used for the treatment of depression (
      • Blumberger D.M.
      • Vila-Rodriguez F.
      • Thorpe K.E.
      • Feffer K.
      • Noda Y.
      • Giacobbe P.
      • et al.
      Effectiveness of theta burst versus high-frequency repetitive transcranial magnetic stimulation in patients with depression (THREE-D): a randomised non-inferiority trial.
      ): iTBS with a cycle-time of 10s and 10Hz-dash with a cycle-time of 16s. First, we validated this model for iTBS, in which the cycle-time resulted in a 0.1Hz entrainment, where we could successfully differentiate between active and sham iTBS with a large effect size and unblind two datasets with high accuracy. The effect of iTBS on HBC did not attenuate over time in study 1A (opposed to the effects of HR-deceleration in the older NCG-TMS 1.0 (
      • Iseger T.A.
      • Arns M.
      • Downar J.
      • Blumberger D.M.
      • Daskalakis Z.J.
      • Vila-Rodriguez F.
      Cardiovascular differences between sham and active iTBS related to treatment response in MDD.
      )), which could be caused by neuroplastic effects of the first iTBS exposure. We further validated this finding for a second rTMS protocol (10Hz-dash), where stimulation resulted in specific 0.0625Hz HR-entrainment.
      In addition to target engagement results described above, study 2 also demonstrated clear dose-response effects on the HBC-marker with large inter-individual variability in FT. Prior studies already demonstrated that NCG-TMS strength was not correlated to %MT but to %MSO, suggesting %MT is a poor proxy for the FT (
      • Iseger T.A.
      • Padberg F.
      • Kenemans J.L.
      • Dijk H van
      • Arns M.
      Neuro-Cardiac-Guided TMS (NCG TMS): A replication and extension study.
      ). In our limited sample, the greatest difference in HBC-peak was found between two subjects with a relative %MT of 81%MT and 120%MT (Figure 6). These different TMS-HBC distributions could reflect a combination of TMS intensity and possibly another factor, such as pain, target engagement, or individual differences in susceptibility to TMS. In larger samples it is likely that larger differences will be found, necessitating studies to investigate this method further as an FT-technique. The non-linear and non-sigmoidal dose-response effects suggest possible ‘over-stimulation’, which is in line with a recent case of TMS-induced syncope (
      • Rouwhorst R.
      • Oostrom I van
      • Dijkstra E.
      • Zwienenberg L.
      • Dijk H van
      • Arns M.
      Vasovagal syncope as a specific side effect of DLPFC-rTMS: A frontal-vagal dose-finding study.
      ) and previous studies that imply that the stimulation intensity sweet spot neither under- nor over-doses (
      • Chung S.W.
      • Rogasch N.C.
      • Hoy K.E.
      • Sullivan C.M.
      • Cash R.F.H.
      • Fitzgerald P.B.
      Impact of different intensities of intermittent theta burst stimulation on the cortical properties during TMS‐EEG and working memory performance.
      ,
      • Lee J.C.
      • Corlier J.
      • Wilson A.C.
      • Tadayonnejad R.
      • Marder K.G.
      • Ngo D.
      • et al.
      Subthreshold stimulation intensity is associated with greater clinical efficacy of intermittent theta-burst stimulation priming for Major Depressive Disorder.
      ). Compared to those studies, we used more and smaller incremental steps in the current study and investigated HBC on the individual level, enabling the determination of an individual fine-grained dose-response curve.
      Some potential limitations should be acknowledged. In all three studies, small sample sizes were included, decreasing the power of the studies. Study 1A and 1B are hard to compare, as they differ in the number of TMS pulses (1A: 1200 pulses in 2 sessions; 1B: 90000 pulses in 50 sessions) and the design (1A: within-subject; 1B: between-subject design). Also, in study 1B and study 2 subjective unpleasantness of stimulation was not measured, study 1A lacked a good control condition (vertex vs. DLPFC) and active condition was more painful than sham, and study 1B was not a within-subject design (participants could not compare sham and active condition directly). Therefore, distinctive scalp sensations might have contributed to differences between active and sham condition. The sensory effects of TMS, such as the pain and discomfort associated with stimulation, and its influence on heart rate and HBC should be controlled for in future TMS studies, by measuring pain and discomfort during stimulation and use that as a covariate, or by conducting stimulation inside an MRI to demonstrate correlation between BOLD changes and HR changes. This is important, as it is known that frontal TMS is especially uncomfortable (
      • Han S.
      • Ogawa A.
      • Osada T.
      • Suda A.
      • Tanaka M.
      • Nanjo H.
      • et al.
      More subjects are required for ventrolateral than dorsolateral prefrontal TMS because of intolerability and potential drop-out.
      ) and can influence results such as reaction times in cognitive tasks (
      • Holmes N.P.
      • Meteyard L.
      Subjective Discomfort of TMS Predicts Reaction Times Differences in Published Studies.
      ). This was not sufficiently controlled for in the current study, necessitating future research to investigate the potential relevance of these results. Finally, it should be acknowledged that much is still unknown about the FT, hence it is unclear whether the individualized FT has any associations or implications for clinical outcomes of rTMS treatment.
      Despite of the limitations, there are at least three good reasons to believe that the TMS protocol has direct effects on the frontal-vagal system and that non-specific effects of rTMS, such as somatosensory stimulation, are unlikely to have affected the results: (
      • Donse L.
      • Padberg F.
      • Sack A.T.
      • Rush A.J.
      • Arns M.
      Simultaneous rTMS and psychotherapy in major depressive disorder: Clinical outcomes and predictors from a large naturalistic study.
      ) in study 2, the HBC-effect demonstrated clear site-specificity. For non-specific effects, a similar effect on the HR would have been expected for all four stimulated locations. In addition, (
      • Fox M.D.
      • Buckner R.L.
      • White M.P.
      • Greicius M.D.
      • Pascual-Leone A.
      Efficacy of Transcranial Magnetic Stimulation Targets for Depression Is Related to Intrinsic Functional Connectivity with the Subgenual Cingulate.
      ) discomfort/pain at the stimulation site could affect HR, but mostly by sympathetic activation and thus HR increases (
      • Chen J.
      • Abbod M.
      • Shieh J.S.
      Pain and Stress Detection Using Wearable Sensors and Devices—A Review.
      ). Therefore, HR-decelerations during rTMS stimulation can be interpreted as specific effect on the frontal-vagal system. Finally, (
      • Caulfield K.A.
      • Brown J.C.
      The Problem and Potential of TMS’ Infinite Parameter Space: A Targeted Review and Road Map Forward.
      ), in earlier work (
      • Hong B.
      • Kuwaki T.
      • Ju K.
      • Kumada M.
      • Akai M.
      • Ueno S.
      Changes in blood pressure and heart rate by repetitive transcranial magnetic stimulation in rats.
      ), rTMS was applied in anesthetized rats and they found HR-decelerations after active stimulation but not sham, emphasizing active signaling, rather than just the effect of pain.
      Our findings suggest that the use of rTMS-induced HBC might be a valuable method for rTMS target engagement of the frontal-vagal pathway. Given this novel approach deviates substantially from the earlier published NCG-TMS studies, we propose to refer to this method as NCG-TMS 2.0 or Heart-Brain-Coupling. In previous studies, it was shown that the effects of rTMS on HR are similar between healthy subjects and MDD patients (
      • Iseger T.A.
      • Bueren NER van
      • Kenemans J.L.
      • Gevirtz R.
      • Arns M.
      A frontal-vagal network theory for Major Depressive Disorder: Implications for optimizing neuromodulation techniques.
      ,
      • Iseger T.A.
      • Padberg F.
      • Kenemans J.L.
      • Gevirtz R.
      • Arns M.
      Neuro-Cardiac-Guided TMS (NCG-TMS): Probing DLPFC-sgACC-vagus nerve connectivity using heart rate - First results.
      ,
      • Kaur M.
      • Michael J.A.
      • Hoy K.E.
      • Fitzgibbon B.M.
      • Ross M.S.
      • Iseger T.A.
      • et al.
      Investigating high- and low-frequency neuro-cardiac-guided TMS for probing the frontal vagal pathway.
      ,
      • Iseger T.A.
      • Padberg F.
      • Kenemans J.L.
      • Dijk H van
      • Arns M.
      Neuro-Cardiac-Guided TMS (NCG TMS): A replication and extension study.
      ,
      • Zwienenberg L.
      • Iseger T.A.
      • Dijkstra E.
      • Rouwhorst R.
      • Dijk H van
      • Sack A.T.
      • et al.
      Neuro-cardiac guided rTMS as a stratifying method between the ‘5cm’ and ‘BeamF3’ stimulation clusters.
      ), which demonstrates the possible clinical implications of this novel method for target stratification in MDD patients. In line with the suggestions by (
      • Zwienenberg L.
      • Iseger T.A.
      • Dijkstra E.
      • Rouwhorst R.
      • Dijk H van
      • Sack A.T.
      • et al.
      Neuro-cardiac guided rTMS as a stratifying method between the ‘5cm’ and ‘BeamF3’ stimulation clusters.
      ), this method could be specifically used to stratify MDD patients between one of two evidence-based stimulation clusters: The Beam and 5CM cluster. Large effectiveness studies have demonstrated comparable response (47-58%) and remission (29-37%) rates for these clusters (
      • Blumberger D.M.
      • Vila-Rodriguez F.
      • Thorpe K.E.
      • Feffer K.
      • Noda Y.
      • Giacobbe P.
      • et al.
      Effectiveness of theta burst versus high-frequency repetitive transcranial magnetic stimulation in patients with depression (THREE-D): a randomised non-inferiority trial.
      ,
      • Carpenter L.L.
      • Janicak P.G.
      • Aaronson S.T.
      • Boyadjis T.
      • Brock D.G.
      • Cook I.A.
      • et al.
      Transcranial magnetic stimulation (TMS) for major depression: A multisite, naturalistic, observational study of acute treatment outcomes in clinical practice.
      ), thus NCG TMS 2.0 could be employed to effectively address inter-individual differences between MDD patients and stratify patients to their optimal rTMS target. This could enhance response and remission rates on the individual level, while adhering to validated rTMS-protocols, as both targets are widely implemented in clinical practice.
      Future studies should focus on replicating these findings in larger samples and investigate whether NCG TMS 2.0 should be used for rTMS target stratification and probing the FT in MDD patients to improve treatment outcome. Further research should focus on optimization of other rTMS parameters as well, such as frequency of stimulation. Also, the potential of NCG-TMS 2.0 could be investigated further for TMS in other (cognitive) applications. The current study will hopefully aid to individualize rTMS treatment and might have important implications for the field of stratified psychiatry. NCG TMS may become for the prefrontal cortex, what the thumb-twitch is for the motor system.

      Uncited reference

      • Tarvainen M.P.
      • Niskanen J.P.
      • Lipponen J.A.
      • Ranta-aho P.O.
      • Karjalainen P.A.
      Kubios HRV – Heart rate variability analysis software.
      .

      Acknowledgements

      No acknowledgements.

      Supplementary Material

      REFERENCES

        • Donse L.
        • Padberg F.
        • Sack A.T.
        • Rush A.J.
        • Arns M.
        Simultaneous rTMS and psychotherapy in major depressive disorder: Clinical outcomes and predictors from a large naturalistic study.
        Brain Stimul [Internet]. 2017; 11 (Available from:): 337-345https://doi.org/10.1016/j.brs.2017.11.004
        • Fox M.D.
        • Buckner R.L.
        • White M.P.
        • Greicius M.D.
        • Pascual-Leone A.
        Efficacy of Transcranial Magnetic Stimulation Targets for Depression Is Related to Intrinsic Functional Connectivity with the Subgenual Cingulate.
        Biol Psychiat. 2012; 72: 595-603
        • Caulfield K.A.
        • Brown J.C.
        The Problem and Potential of TMS’ Infinite Parameter Space: A Targeted Review and Road Map Forward.
        Frontiers Psychiatry. 2022; 13867091
        • Fitzgerald P.B.
        • Hoy K.
        • Gunewardene R.
        • Slack C.
        • Ibrahim S.
        • Bailey M.
        • et al.
        A randomized trial of unilateral and bilateral prefrontal cortex transcranial magnetic stimulation in treatment-resistant major depression.
        Psychol Med. 2011; 41: 1187-1196
        • Iseger T.A.
        • Bueren NER van
        • Kenemans J.L.
        • Gevirtz R.
        • Arns M.
        A frontal-vagal network theory for Major Depressive Disorder: Implications for optimizing neuromodulation techniques.
        Brain Stimul. 2020; 13: 1-9
        • Iseger T.A.
        • Padberg F.
        • Kenemans J.L.
        • Gevirtz R.
        • Arns M.
        Neuro-Cardiac-Guided TMS (NCG-TMS): Probing DLPFC-sgACC-vagus nerve connectivity using heart rate - First results.
        Brain Stimul. 2017; 10: 1006-1008
        • Kaur M.
        • Michael J.A.
        • Hoy K.E.
        • Fitzgibbon B.M.
        • Ross M.S.
        • Iseger T.A.
        • et al.
        Investigating high- and low-frequency neuro-cardiac-guided TMS for probing the frontal vagal pathway.
        Brain Stimul. 2020; 13: 931-938
        • Iseger T.A.
        • Padberg F.
        • Kenemans J.L.
        • Dijk H van
        • Arns M.
        Neuro-Cardiac-Guided TMS (NCG TMS): A replication and extension study.
        Biol Psychol. 2021; 108097
        • Zwienenberg L.
        • Iseger T.A.
        • Dijkstra E.
        • Rouwhorst R.
        • Dijk H van
        • Sack A.T.
        • et al.
        Neuro-cardiac guided rTMS as a stratifying method between the ‘5cm’ and ‘BeamF3’ stimulation clusters.
        Brain Stimul. 2021; 14: 1070-1072
        • Iseger T.A.
        • Arns M.
        • Downar J.
        • Blumberger D.M.
        • Daskalakis Z.J.
        • Vila-Rodriguez F.
        Cardiovascular differences between sham and active iTBS related to treatment response in MDD.
        Brain Stimul. 2020; 13: 167-174
        • Carpenter LindaL.
        • ScottT Aaronson
        • Hutton T.M.
        • Mina M.
        • Pages K.
        • Verdoliva S.
        • et al.
        Comparison of clinical outcomes with two Transcranial Magnetic Stimulation treatment protocols for major depressive disorder.
        Brain Stimul. 2021; 14: 173-180
        • Thut G.
        • Veniero D.
        • Romei V.
        • Miniussi C.
        • Schyns P.
        • Gross J.
        Rhythmic TMS Causes Local Entrainment of Natural Oscillatory Signatures.
        Curr Biol. 2011; 21: 1176-1185
        • Candia-Rivera D.
        • Catrambone V.
        • Barbieri R.
        • Valenza G.
        Functional assessment of bidirectional cortical and peripheral neural control on heartbeat dynamics: A brain-heart study on thermal stress.
        Neuroimage. 2022; 251119023
        • Chung S.W.
        • Rogasch N.C.
        • Hoy K.E.
        • Sullivan C.M.
        • Cash R.F.H.
        • Fitzgerald P.B.
        Impact of different intensities of intermittent theta burst stimulation on the cortical properties during TMS‐EEG and working memory performance.
        Hum Brain Mapp. 2018; 39: 783-802
        • Lee J.C.
        • Corlier J.
        • Wilson A.C.
        • Tadayonnejad R.
        • Marder K.G.
        • Ngo D.
        • et al.
        Subthreshold stimulation intensity is associated with greater clinical efficacy of intermittent theta-burst stimulation priming for Major Depressive Disorder.
        Brain Stimul. 2021; 14: 1015-1021
        • Blumberger D.M.
        • Vila-Rodriguez F.
        • Wang W.
        • Knyahnytska Y.
        • Butterfield M.
        • Noda Y.
        • et al.
        A randomized sham controlled comparison of once vs twice-daily intermittent theta burst stimulation in depression: A canadian rTMS treatment and biomarker network in depression (CARTBIND) study.
        Brain Stimul. 2021;
        • Cole E.J.
        • Phillips A.L.
        • Bentzley B.S.
        • Stimpson K.H.
        • Nejad R.
        • Barmak F.
        • et al.
        Stanford Neuromodulation Therapy (SNT): A Double-Blind Randomized Controlled Trial.
        Am J Psychiat. 2021; (appiajp202120101429)
        • Tarvainen M.P.
        • Niskanen J.P.
        • Lipponen J.A.
        • Ranta-aho P.O.
        • Karjalainen P.A.
        Kubios HRV – Heart rate variability analysis software.
        Comput Meth Prog Bio. 2014; 113: 210-220
        • Gramfort A.
        • Luessi M.
        • Larson E.
        • Engemann D.A.
        • Strohmeier D.
        • Brodbeck C.
        • et al.
        MEG and EEG data analysis with MNE-Python.
        Front Neurosci-switz. 2013; 7: 267
        • Thomson R.H.
        • Cleve T.J.
        • Bailey N.W.
        • Rogasch N.C.
        • Maller J.J.
        • Daskalakis Z.J.
        • et al.
        Blood Oxygenation Changes Modulated by Coil Orientation During Prefrontal Transcranial Magnetic Stimulation.
        Brain Stimul. 2013; 6: 576-581
        • Opitz A.
        • Fox M.D.
        • Craddock R.C.
        • Colcombe S.
        • Milham M.P.
        An integrated framework for targeting functional networks via transcranial magnetic stimulation.
        Neuroimage. 2016; 127: 86-96
        • Beam W.
        • Borckardt J.J.
        • Reeves S.T.
        • George M.S.
        An efficient and accurate new method for locating the F3 position for prefrontal TMS applications.
        Brain Stimul. 2009; 2: 50-54
        • Virtanen P.
        • Gommers R.
        • Oliphant T.E.
        • Haberland M.
        • Reddy T.
        • Cournapeau D.
        • et al.
        SciPy 1.0: Fundamental Algorithms for Scientific Computing in Python.
        Nature Methods. 2020 Jan 1; 17: 261-272
        • Blumberger D.M.
        • Vila-Rodriguez F.
        • Thorpe K.E.
        • Feffer K.
        • Noda Y.
        • Giacobbe P.
        • et al.
        Effectiveness of theta burst versus high-frequency repetitive transcranial magnetic stimulation in patients with depression (THREE-D): a randomised non-inferiority trial.
        Lancet. 2018; 391: 1683-1692
        • Rouwhorst R.
        • Oostrom I van
        • Dijkstra E.
        • Zwienenberg L.
        • Dijk H van
        • Arns M.
        Vasovagal syncope as a specific side effect of DLPFC-rTMS: A frontal-vagal dose-finding study.
        Brain Stimul. 2022; 15: 1233-1235
        • Han S.
        • Ogawa A.
        • Osada T.
        • Suda A.
        • Tanaka M.
        • Nanjo H.
        • et al.
        More subjects are required for ventrolateral than dorsolateral prefrontal TMS because of intolerability and potential drop-out.
        Plos One. 2019; 14e0217826
        • Holmes N.P.
        • Meteyard L.
        Subjective Discomfort of TMS Predicts Reaction Times Differences in Published Studies.
        Front Psychol. 2018; 9: 1989
        • Chen J.
        • Abbod M.
        • Shieh J.S.
        Pain and Stress Detection Using Wearable Sensors and Devices—A Review.
        Sensors. 2021; 21: 1030
        • Hong B.
        • Kuwaki T.
        • Ju K.
        • Kumada M.
        • Akai M.
        • Ueno S.
        Changes in blood pressure and heart rate by repetitive transcranial magnetic stimulation in rats.
        Neurosci Lett. 2002; 329: 57-60
        • Carpenter L.L.
        • Janicak P.G.
        • Aaronson S.T.
        • Boyadjis T.
        • Brock D.G.
        • Cook I.A.
        • et al.
        Transcranial magnetic stimulation (TMS) for major depression: A multisite, naturalistic, observational study of acute treatment outcomes in clinical practice.
        Depress Anxiety. 2012; 29: 587-596