CLINICAL TRIAL published: 08 July 2020 doi: 10.3389/fnagi.2020.00200 Corticomotor Plasticity Predicts Clinical Efficacy of Combined Neuromodulation and Cognitive Training in Alzheimer’s Disease Anna-Katharine Brem 1,2† , Riccardo Di Iorio 3 , Peter J. Fried 1 , Albino J. Oliveira-Maia 4,5,6 , Camillo Marra 3 , Paolo Profice 3 , Davide Quaranta 3 , Lukas Schilberg 1,7 , Natasha J. Atkinson 1 , Erica E. Seligson 1 , Paolo Maria Rossini 8 and Alvaro Pascual-Leone 9,10,11* 1 Berenson-Allen Center for Noninvasive Brain Stimulation, Division of Cognitive Neurology, Department of Neurology, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, United States, 2 Department of Experimental Psychology, University of Oxford, Oxford, United Kingdom, 3 Department of Geriatrics, Neurosciences and Orthopaedics, Edited by: Polyclinic A. Gemelli Foundation-IRCCS, Rome, Italy, 4 Champalimaud Research and Clinical Centre, Champalimaud Centre Panteleimon Giannakopoulos, for the Unknown, Lisbon, Portugal, 5 Department of Psychiatry and Mental Health, Centro Hospitalar de Lisboa Ocidental, Université de Genève, Switzerland Lisbon, Portugal, 6 NOVA Medical School – Faculdade de Ciências Médicas de Lisboa, Universidade Nova de Lisboa, Reviewed by: Lisbon, Portugal, 7 Department of Cognitive Neuroscience, Faculty of Psychology and Neuroscience, Maastricht University, Rosa Manenti, Maastricht, Netherlands, 8 Department of Neuroscience and Neurorehabilitation, IRCCS San Raffaele, Rome, Italy, 9 Hinda Centro San Giovanni di Dio and Arthur Marcus Institute for Aging Research and Center for Memory Health, Hebrew SeniorLife, Boston, MA, Fatebenefratelli (IRCCS), Italy United States, 10 Department of Neurology, Harvard Medical School, Boston, MA, United States, 11 Guttmann Brain Health Markus Kosel, Institut, Institut Guttmann, Universitat Autonoma Barcelona, Barcelona, Spain Geneva University Hospitals (HUG), Switzerland Objective: To investigate the efficacy of repetitive transcranial magnetic stimulation *Correspondence: Alvaro Pascual-Leone (rTMS) combined with cognitive training for treatment of cognitive symptoms in patients apleone@hsl.harvard.edu with Alzheimer’s disease (AD). A secondary objective was to analyze associations † Present address: between brain plasticity and cognitive effects of treatment. Anna-Katharine Brem, University Hospital of Old Age Methods: In this randomized, sham-controlled, multicenter clinical trial, 34 patients Psychiatry, University of Bern, Bern, with AD were assigned to three experimental groups receiving 30 daily sessions of Switzerland; Department of Neuropsychology, combinatory intervention. Participants in the real/real group (n = 16) received 10 Hz Lucerne Psychiatry, Lucerne, repetitive transcranial magnetic stimulation (rTMS) delivered separately to each of six Switzerland cortical regions, interleaved with computerized cognitive training. Participants in the Received: 19 August 2019 sham rTMS group (n = 18) received sham rTMS combined with either real (sham/real Accepted: 09 June 2020 group, n = 10) or sham (sham/sham group, n = 8) cognitive training. Effects of Published: 08 July 2020 treatment on neuropsychological (primary outcome) and neurophysiological function Citation: Brem A-K, Di Iorio R, Fried PJ, were compared between the 3 treatment groups. These, as well as imaging measures Oliveira-Maia AJ, Marra C, Profice P, of brain atrophy, were compared at baseline to 14 healthy controls (HC). Quaranta D, Schilberg L, Atkinson NJ, Seligson EE, Rossini PM and Results: At baseline, patients with AD had worse cognition, cerebral atrophy, and TMS Pascual-Leone A (2020) Corticomotor measures of cortico-motor reactivity, excitability, and plasticity than HC. The real/real Plasticity Predicts Clinical Efficacy of Combined Neuromodulation group showed significant cognitive improvement compared to the sham/sham, but and Cognitive Training in Alzheimer’s not the real/sham group. TMS-induced plasticity at baseline was predictive of post- Disease. Front. Aging Neurosci. 12:200. intervention changes in cognition, and was modified across treatment, in association doi: 10.3389/fnagi.2020.00200 with changes of cognition. Frontiers in Aging Neuroscience | www.frontiersin.org 1 July 2020 | Volume 12 | Article 200 Brem et al. Combinatory Treatment in AD Interpretation: Combined rTMS and cognitive training may improve the cognitive status of AD patients, with TMS-induced cortical plasticity at baseline serving as predictor of therapeutic outcome for this intervention, and potential mechanism of action. Clinical Trial Registration: www.ClinicalTrials.gov, identifier NCT01504958. Keywords: clinical trial, randomized controlled, Alzheimer’s disease, transcranial magnetic stimulation, cognitive training, plasticity, combinatory intervention INTRODUCTION rTMS alone yielded positive results in AD, in a few studies with relatively small samples (Cotelli et al., 2011; Ahmed et al., 2012; Alzheimer’s disease (AD) is the most common neurodegenerative Haffen et al., 2012). The concomitant application of these two disease (Alzheimer’s Association, 2012) with cognitive decline approaches has shown therapeutic promise in a small industry- significantly affecting quality of life (Alzheimer’s Association, sponsored open trial (Bentwich et al., 2011), and three follow-up 2012). Given that pharmacological agents (Ballard et al., 2011) trials comparing the combined intervention with a control group have limited efficacy and entail unfavorable side effects (Ryu receiving sham rTMS and sham cognitive training (Rabey et al., et al., 2005) there is a pressing need for non-pharmacological 2013; Lee et al., 2015; Rabey and Dobronevsky, 2016). A recent interventions to complement current options. Importantly, AD is review emphasizes the need for larger studies, and further associated with alterations in synaptic function and mechanisms suggests investigation of treatment-related neurophysiological of neuroplasticity (Koch et al., 2012; Rossini et al., 2015), markers (Buss et al., 2019). Our study builds on and extends which can be assessed indirectly using transcranial magnetic these preliminary studies in two important ways. First, we stimulation (TMS). TMS is a non-invasive technique to generate included a control group of AD patients who received sham brief and relatively focal electric currents in the brain via rTMS with real cognitive training, in order to elucidate effects electromagnetic induction. When applied as single or paired of cognitive training alone. Furthermore, we aimed to provide pulses to the motor cortex, it provides a diagnostic tool to mechanistic insight into the intervention by characterizing interrogate cortico-motor reactivity and inhibitory/excitatory the neurophysiology of participants using TMS assessments of intracortical circuits, respectively. As a therapeutic tool, TMS is cortical reactivity, inhibitory/excitatory intracortical circuits, applied in trains of pulses, termed repetitive TMS (rTMS) to and plasticity, that were assessed longitudinally and also induce changes in excitability of the activated neural circuits compared at baseline to a group of healthy matched controls that outlast the period of stimulation. TMS can thus be used (HC). Specifically, we hypothesized that such rTMS-cognitive to modulate activity in brain areas implicated in the cognitive training-dependent modulation of plasticity is associated with and behavioral dysfunctions of AD. Furthermore, over the lasting changes in cognition. Moreover, we hypothesized that last 15 years, a low-intensity form of patterned rTMS, termed cortical plasticity is reduced in patients with AD relative to intermittent theta-burst stimulation (iTBS), has emerged as a HC, as has been shown in previous studies (Koch et al., 2012; means to induce NMDA receptor (NMDAR)-dependent long- Di Lorenzo et al., 2016) and is modulated by neuronavigated term potentiation (LTP)-like plasticity (Huang et al., 2005, 2007). rTMS to multiple brain regions interleaved with region-specific When applied over the motor cortex, the after-effects of iTBS cognitive training. are measured as the change in the amplitude of motor-evoked potentials (MEPs) elicited by suprathreshold single-pulse TMS, i.e., the change in cortico-motor reactivity, which is thought to MATERIALS AND METHODS reflect cortical plasticity (Pascual-Leone et al., 1994). The after- effects of rTMS and TBS protocols are state-dependent (Silvanto Subjects and Pascual-Leone, 2008) and mechanisms underlying rTMS- In this proof-of-principle, randomized, double-blind, sham- mediated neuromodulation are similar to neuroplastic effects controlled study we examined 35 patients diagnosed with involved in learning (Cooke and Bliss, 2006). Therefore, targeting mild-moderate AD [DSM 5, NIA-AA (McKhann et al., neural networks with rTMS while engaging them, for example 1984) Figure 1A and Table 1). Patients were randomized in cognitive exercises, has been suggested to lead to functional using concealed randomization practices (study-independent improvements (Freitas et al., 2011) resulting from synergistic researcher) and a 2:1:1 group distribution to one of three groups: effects if repeated in time. (1) real cognitive training combined with real rTMS (real/real: Cognitive training interventions designed to attenuate n = 16), (2) sham cognitive training and sham stimulation cognitive decline in dementia typically train basic functions with over the same brain regions (sham/sham: n = 8), or (3) real simple tasks, tapping, for example, into non-verbal and verbal cognitive training and sham rTMS (real/sham: n = 11) (see memory, language, and visuo-spatial skills (Hill et al., 2017). Figure 1B for an overview of the study design). Except for Cognitive interventions alone have yielded inconsistent results to the technicians applying the intervention, all experimenters date (Clare et al., 2003; Hill et al., 2017; Kallio et al., 2017), while and patients were blinded. Findings from two trials, one in Frontiers in Aging Neuroscience | www.frontiersin.org 2 July 2020 | Volume 12 | Article 200 Brem et al. Combinatory Treatment in AD FIGURE 1 | CONSORT flowchart: Enrollment and study design. (A) Flow diagram of the enrollment process and final study participants analyzed. (B) Schematic representation of study design. Boston (n = 21) and one in Rome (n = 13), were combined to unstable/chronic medical conditions; major structural/vascular increase the sample size and reach more reliable conclusions. abnormalities, agitation, psychiatric disorders, substance abuse, The eligibility criteria were the following: subjects diagnosed other progressive neurological disorders different from diseases with mild to moderate Alzheimer’s disease according to DSM- causing cognitive impairment, or conditions considered a 5 criteria and criteria established by the NIA-AA (McKhann potential hazard for the application of rTMS (Rossi et al., 2009). et al., 1984) for AD, at medium level of certainty according to Patients treated with cholinesterase inhibitors or ginkgo-biloba PET and/or MRI examination; age between 55 and 90 years; were allowed to participate if the treatment had started at written informed consent; MMSE between 18 and 24; normal least 3 months prior to screening and remain stable for the or corrected ability to see and hear; English (Boston) or duration of the study. One participant from the real/sham group Italian (Rome) as primary language. The exclusion criteria were: was excluded due to the incidental finding of a brain lesion. Frontiers in Aging Neuroscience | www.frontiersin.org 3 July 2020 | Volume 12 | Article 200 Frontiers in Aging Neuroscience | www.frontiersin.org Brem et al. TABLE 1 | Demographic, neuropsychological, morphometric, and neurophysiologic features of study participants (before intervention). AD group Healthy controls p-value Real/Real Sham/Sham Real/Sham p-value (n = 21, B) (n = 13, B) intervention (n = 16) intervention (n = 8) intervention (n = 10) (n = 13, R) Age (years)* 68.79 ± 7.17 66.00 ± 7.19 0.317 69.25 ± 6.80 67.50 ± 10.27 69.10 ± 5.24 0.710 Gender* 20 Female, 14 Male 7 Female, 6 Male 0.520 12 Female, 4 Male 3 Female, 5 Male 5 Female, 5 Male 0.169 Education (years)* 14.91 ± 4.72 15.62 ± 2.22 0.745 14.25 ± 4.64 17.50 ± 4.00 13.90 ± 5.07 0.064 Medication* 3 MEM 0 MEM 2 MEM 0.053 7 COM, 2 COM 3 COM 6 AChEI 5 AChEI 1 AChEI 0 None 1 None 4 None SBD left IPL (mm) 24.60 ± 9.78 17.15 ± 2.87 0.008 23.50 ± 6.66 26.13 ± 13.49 24.98 ± 12.02 0.961 SBD mean (mm) 18.96 ± 3.57 16.27 ± 2.29 0.005 18.36 ± 3.01 20.38 ± 4.17 18.46 ± 4.17 0.555 MMSE* 21.71 ± 2.47 29.46 ± 0.88 <0.000 21.19 ± 2.69 20.88 ± 2.95 22.00 ± 1.83 0.662 ADAS-Cog* 23.76 ± 9.90 4.25 ± 1.97 <0.001 23.00 ± 9.92 23.61 ± 11.29 25.10 ± 9.67 0.670 4 ADCS-ADL* 56.77 ± 19.69 74.31 ± 3.45 0.002 60.93 ± 15.96 54.67 ± 17.99 51.80 ± 25.64 0.473 GDS 2.43 ± 2.34 0.69 ± 1.11 0.020 2.00 ± 2.00 2.67 ± 3.08 3.00 ± 2.35 0.576 rMT 42.33 ± 10.57 46.82 ± 12.33 0.326 44.12 ± 7.02 42.44 ± 14.64 38.60 ± 12.40 0.381 aMT 42.54 ± 9.99 46.42 ± 8.74 0.309 45.53 ± 5.70 43.00 ± 12.74 36.00 ± 12.10 0.179 SICI 0.66 ± 0.39 0.34 ± 0.23 0.025 0.63 ± 0.52 0.69 ± 0.20 0.70 ± 0.29 0.784 ICF 1.43 ± 0.85 1.43 ± 0.56 0.611 1.23 ± 0.73 2.10 ± 0.97 0.94 ± 0.26 0.081 LICI 0.59 ± 0.79 0.04 ± 0.06 0.004 0.41 ± 0.58 1.06 ± 1.13 0.34 ± 0.43 0.289 MC reactivity 1246 ± 1010 1184 ± 620 0.727 1222 ± 1182 814 ± 473 1815 ± 995 0.203 MC max. plasticity 1.50 ± 0.92 1.87 ± 0.65 0.020 1.53 ± 1.34 1.33 ± 0.15 1.66 ± 0.34 0.147 MC mean plasticity 1.09 ± 0.49 1.45 ± 0.60 0.040 1.04 ± 0.63 0.99 ± 0.12 1.32 ± 0.44 0.412 *These parameters include the Rome sample (n = 34). AChEI, acetylcholinesterase inhibitors; ADAS-Cog, Alzheimer’s Disease Assessment Scale-Cognitive Subscale; ADCS-ADL, Alzheimer’s Disease Cooperative Study–Activities of Daily Living Inventory; B, Boston sample (n = 21); COM, combination therapy with AChEI and memantine; GDS, Geriatric Depression Scale; ICF, intracortical facilitation; IPL, inferior parietal lobule; LICI, long-interval intracortical inhibition; MC, motor cortex; MEM, memantine; MMSE, Mini-Mental State Examination; R, Rome sample (n = 13); rMT/aMT, resting/active motor threshold; SBD, Scalp-brain distance; July 2020 | Volume 12 | Article 200 SICI, short-interval intracortical inhibition; T5/T10, Time-points at 5 and 10 min post intermittent Theta Burst Stimulation. Group values are presented as mean ± standard deviation (SD), p-values are two-tailed. Bold font indicates significant p-values. Combinatory Treatment in AD Brem et al. Combinatory Treatment in AD AD patients had a score of 1 (n = 30) or 2 (n = 4 participants figure-of-eight focal biphasic and monophasic coil, respectively. studied in Rome) on the Clinical Dementia Rating scale (CDR), The coils were held over the left motor cortex to elicit motor and a Mini-Mental State Examination (MMSE) score between 18 responses in the contralateral first dorsal interosseous (FDI) and 24, and 29 of them were medicated (Table 1). We also studied muscle. Intensities during assessment of the rMT and aMT 14 age-matched healthy controls (HC), who underwent the same were expressed as a percentage of the maximum output of the baseline evaluation as the AD participants in Boston. One HC stimulator. The stimulation was performed positioning the was excluded as a statistical outlier in several measures. All HC virtual cathode of the coil centered over the site of the scalp to had normal physical, neurological and cognitive exams (CDR = 0 be stimulated and the holder oriented at a 45◦ angle with respect and MMSE score ≥ 28). to the approximate direction of the central sulcus (current Neurophysiological assessment of cortical function was flow in postero-anterior direction). The rMT was defined as completed in the participants studied in Boston (real/real: n = 10; the minimum intensity of the magnetic field able to generate combined sham groups: n = 11). Two participants in the sham a motor evoked potential (MEP) of 50 µV in approximately group did not complete post-intervention measures (due to 50% of 10 consecutive stimuli, i.e., generating 5 MEPs out of their distance from the testing site). Interventions took place at 10 trials (Rossini et al., 1994, 2015). The aMT was defined as the Berenson-Allen Center for Noninvasive Brain Stimulation the minimum stimulus intensity that produced a liminal MEP and the Harvard-Thorndike Clinical Research Center, Beth (about 200 µV in 50% of 10 trials) during isometric contraction Israel Deaconess Medical Center, Boston, United States or at of the tested muscle. the Polyclinic A. Gemelli, Rome, Italy. The local Institutional After assessing rMT and aMT, MEP amplitudes were Review Boards at both institutions (Beth Israel Deaconess measured from the right FDI muscle using 30 × 22 mm gel Medical Center and Polyclinic A. Gemelli Foundation-IRCCS) surface electrodes (Ambu, United Kingdom; active over muscle approved the study. Participants and their legally authorized belly, reference over proximal interphalangeal joint of the index representatives (if appropriate) gave written informed consent finger, ground on the ulnar styloid process). Stimulation intensity prior to study onset. AD data from the Boston study originate for all single-pulse TMS trials was set at 120% of rMT. For from the trial registered at ClinicalTrials.gov NCT01504958. paired-pulse TMS, test pulse intensity was set at 120% rMT, while the conditioning pulse intensity and inter-pulse interval Cognitive and Behavioral Measures (IPI) determined the type of measurement: 80% rMT with a Cognitive and behavioral functions were assessed within 2 weeks 3-ms IPI to assess short-interval intracortical inhibition (SICI), before and 1 week after the therapeutic intervention using 80% rMT with a 12-ms IPI to evaluate intracortical facilitation tests and inventories drawn from the National Alzheimer’s (ICF), and 120% rMT with an IPI of 100 ms to measure Coordinating Center – Uniform Data Set (v1) (Beekly et al., long-interval intracortical inhibition (LICI). Intermittent TBS 2007). The cognitive subscale of the Alzheimer’s Disease (iTBS) was applied at 80% of aMT. Paired pulse measures were Assessment Scale (ADAS-Cog) was defined as the primary assessed in separate sets (50 pulse pairs with a jittered inter-trial outcome and re-assessed at 1-month follow-up in the Boston interval of 5–6 s) and expressed as a ratio of the test-MEP to and 6 weeks in the Rome sample. The Geriatric Depression a block of 50 unconditioned pulses. Baseline cortical reactivity Scale (GDS) and the CDR were evaluated at baseline, and (three sets of 30 single TMS pulses) was followed by a 600- the Clinical Global Impression of Change (ADCS-CGIC) pulse regimen of iTBS (20 two-second trains of 50 Hz burst- was administered after the intervention. To minimize triplets repeated every 200 ms; MagPro X100, MagVenture A/S, practice/learning effects, alternate forms were used and Denmark). Cortical reactivity was reassessed at 5, 10, 20, and counterbalanced whenever possible. The ADAS-Cog and 30 min post-iTBS in sets of 30 single TMS pulses. Cortico-motor ADCS-CGIC and CDR were administered by board-certified plasticity was expressed as the percent change from baseline in neuropsychologists/neurologists, while all other measures, cortical reactivity. Maximum brain plasticity was defined as the including the Activities of Daily Living (ADCS-ADL) inventory largest of these measures taken at any of the time-points assessed. were administered by trained psychometrists. The same experimenter administered the neuropsychological tests at all Intervention: Cognitive Training and time points. Participants studied in Rome did not undergo assessment with ADCS-CGIC or GDS. rTMS Intervention consisted of daily (monday-to-friday) 1-h sessions of combined cognitive training (real or sham) synchronized with Motor Thresholds, Brain Reactivity, and rTMS (real or sham) for 6 weeks (total of 30 sessions) using Plasticity NeuroAD (Neuronix Medical, Yoqneam, Israel). The NeuroAD Resting and active motor thresholds (rMT, aMT), motor system combines brief trains of 10 Hz rTMS targeting brain cortical reactivity, inhibitory/excitatory intracortical circuits, regions known to be affected in AD (i.e., frontal, temporal, and and plasticity were evaluated with TMS combined with parietal regions) (Mattson, 2004) interleaved with an adaptive electromyography (EMG), at baseline and after the therapeutic computerized cognitive training program designed to engage intervention (Figure 1B). The eXimia Navigated Brain the regions targeted by rTMS (Table 2A). Trains of rTMS (20 Stimulation (NBS) system 4 (Nexstim, Finland) was used trains of 2 s, 10 Hz per region per day at 120% rMT) were for single- and paired-pulse stimulation using a handheld applied using a handheld figure-of-eight focal coil (SuperRapid Frontiers in Aging Neuroscience | www.frontiersin.org 5 July 2020 | Volume 12 | Article 200 Brem et al. Combinatory Treatment in AD TABLE 2A | Cognitive training and morphometric and technical parameters of interest. Brain Region L IFG L STG R DLPFC L DLPFC R IPL L IPL Cognitive tasks Sentence Differentiate Action naming, Remember Identify red/blue Identify letters similarities, words/pseudo word recall color/location of rectangles B/T/M in a cluster differentiate words, assign rectangles, word of letters right/wrong pictures to recall sentences categories Task examples (A) Cognitive training tasks were designed to train the following functions: syntax, grammar, lexical meaning, categorization, action naming, object naming, spatial memory, and spatial attention. Cognitive tasks were associated with the stimulation of specific brain areas: L IFG (Broca’s area): determine whether sentences are grammatically correct; L STG (Wernicke’s area): differentiate between words and pseudo words, or to assign pictures to one of two categories; R DLPFC: match written descriptions with pictures; L DLPFC: remember locations and colors of rectangles; R IPL: identify red or blue vertical rectangles within an array of stimuli; L IPL: identify a particular letter in a cluster of random letters. Notably, stimulating several brain areas per session might in turn activate remotely associated hubs of the same networks, possibly leading to a more extensive modulation. For each cognitive task, the difficulty level was adjusted individually and increased following successful completion of a previous level (≥80% correct answers). An increase in difficulty level was either achieved by increasing stimulus-response time, or by increasing the number of displayed target and distracter stimuli. (B) Difficulty levels (12 per task) achieved by the real/real and the real/sham group, scalp-brain distances, and stimulation intensity applied per TMS-targeted region. Values are presented as mean standard deviation (SD). (C) Correlations between maximum difficulty level achieved during cognitive training tasks per TMS-targeted region with stimulation intensity (real/real group) and with scalp-to-brain distance (real/real and real/sham group). Neurophysiological measures include the Boston sample only (n = 21). L IFG, left inferior frontal gyrus; L STG, left superior temporal gyrus; R and L DLPFC, right and left dorsolateral prefrontal cortex; R and L IPL, right and left inferior parietal lobule; SDB, scalp-to-brain distance; t, Kendall’s tau. P-values are one-tailed. TABLE 2B | Brain Region L IFG L STG R DLPFC L DLPFC R IPL L IPL Average Difficulty level real/real 6.90 ± 2.74 7.40 ± 2.16 6.70 ± 3.06 5.35 ± 3.38 7.90 ± 0.99 6.10 ± 2.81 6.73 ± 1.90 Difficulty level real/sham 5.25 ± 1.56 5.30 ± 2.49 6.40 ± 3.21 5.00 ± 3.45 7.00 ± 3.46 6.20 ± 3.21 5.86 ± 2.62 Scalp-to-brain distance 18.21 ± 5.64 13.29 ± 2.18 21.04 ± 9.50 14.43 ± 3.28 21.78 ± 6.82 23.50 ± 6.66 18.36 ± 3.01 Stimulation intensity (training) 63.92 ± 12.97 63.92 ± 12.97 64.73 ± 14.46 63.92 ± 12.97 64.73 ± 14.46 63.92 ± 12.97 64.19 ± 13.42 Values are presented as mean ± standard deviation (SD). TABLE 2C | Brain Region L IFG L STG R DLPFC L DLPFC R IPL L IPL Average τ (difficulty level, stimulation intensity) real/real 0.25 −0.75 0.454 0.66 0.48 0.57 0.54 p-value 0.486 0.836 0.188 0.040 0.161 0.085 0.111 τ (difficulty level, SBD) real/real and real/sham 0.13 −0.46 −0.06 −0.18 −0.53 −0.71 −0.40 p value 0.658 0.087 0.828 0.515 0.040 0.003 0.136 L IFG, left inferior frontal gyrus; L STG, left superior temporal gyrus; R and L DLPFC, right and left dorsolateral prefrontal cortex; R and L IPL, right and left inferior parietal lobule; SDB, scalp-to-brain distance; τ, Kendall’s tau. P-values are one-tailed. Bold font indicates significant p-values and correlation coefficients. stimulator, Magstim Co. Ltd., Whitland, United Kingdom) stereotaxic neuronavigation system Brainsight (Rogue Research, and were each followed by a cognitive task presented on a Inc., Canada) was used to individually select the targets, guide touch-screen monitor (20–40 s). During each session, focal coil placement, and measure scalp-brain distance, as a surrogate stimulation was applied to 3 of 6 predefined brain regions measure of cortical atrophy. The following predefined cortical in a pseudo-randomized sequence, without counterbalancing regions were individually targeted by transforming published the stimulation sequence between patients. This strategy was group average coordinates into individual MRI space: right/left devised to ensure the same amount of stimulation per target- DLPFC (R/L DLPFC) (Rusjan et al., 2010), right/left IPL (R/L region, with each region stimulated 15 times over the course IPL) (Sajonz et al., 2010), Wernicke area in the left superior of the intervention. Though rTMS was performed using a focal temporal gyrus (L STG) (Price et al., 2003) and Broca’s area in coil to only one region at a time, the combined effects of the left inferior frontal gyrus (L IFG) (Rogalsky et al., 2008). stimulating multiple regions over the course of each session could Sham rTMS was delivered with a sham TMS coil (Magstim) lead to similar effects to those of multifocal stimulation (i.e., that imitated auditory and somatosensory sensations. Regarding concurrent stimulation to several brain regions). The frameless cognitive training, for each task, difficulty levels were adapted for Frontiers in Aging Neuroscience | www.frontiersin.org 6 July 2020 | Volume 12 | Article 200 Brem et al. Combinatory Treatment in AD each patient, according to performance (Table 2B). The sham predictors, and relevant interaction terms, namely interactions cognitive training consisted of a picture preference task using with active rTMS intervention, were tested. Model assumptions the same stimuli as in the real cognitive training. Participants were tested by analyses of residuals, and influence diagnostics had to indicate for each picture whether they find it “nice” or were conducted using Cook’s distance. “not nice.” The presentation of cognitive stimuli and rTMS were synchronized via the NeuroAD system. RESULTS Scalp-Brain Distance Measurements Scalp-brain distance was defined as the distance between the Baseline Comparisons outer edge of the cortical surface and the outer edge of the scalp See Table 1 for an overview of baseline comparisons between AD and was measured for each target in the plane perpendicular and HC and across the AD intervention groups. The AD groups to the scalp tangent (i.e., orthogonal to the plane of the TMS did not differ in any of the baseline measures. AD and HC did coil) on each individual’s brain MRI for the six stimulated not differ in age, education, gender, and TMS-measures of rMT, regions and the left motor cortex (site of TMS measures) aMT and motor-cortical reactivity. However, depression scores as using Brainsight (Rogue Research Inc., Canada). These measures assessed with the GDS were significantly higher in AD (U = 71.5, served as a surrogate indicator of regional brain atrophy z = −2.41, p = 0.020, r = −0.41), as was brain atrophy (scalp- (Rusjan et al., 2010) (Table 2B). brain distance) for the left IPL (U = 63.0, z = −2.61, p = 0.008, r = −0.45) and averaged across seven brain regions (U = 58.5, Data Analysis z = −2.77, p = 0.005, r = −0.47) (Tables 1, 2B). Data were analyzed using SPSS Statistics 21.0 (SPSS Inc., Chicago, Plasticity indices measured up to 30 min after iTBS were IL, United States) and SAS (version 9.3, SAS Institute, Cary, significantly different between AD patients and HC at T5 NC, United States). Significance was set using a 95% confidence (U = 183, z = 2.13, p = 0.033, r = 0.37), T10 (U = 179, z = 1.98, interval (α = 0.05). p = 0.048, r = 0.35), and averaged over T5 to T30 (U = 181, Individual MEP amplitudes that were greater than 1.5 z = 2.06, p = 0.040, r = 0.36) (Figure 2A). Furthermore, the standard deviations from the mean were excluded. Baseline maximum plasticity change (U = 187.5, z = 2.302, p = 0.020, motor cortical reactivity was defined as the mean MEP amplitude r = 0.40) was significantly smaller in AD. ICF was similar in AD across all 90 trials. To assess plastic changes after iTBS, MEPs and HC, while SICI and LICI were significantly reduced in AD from each post-iTBS time point were averaged and divided by (SICI: U = 56, z = −2.23, p = 0.025, r = −0.40; LICI: U = 42, baseline. Additionally, the achieved degree of difficulty during z = −2.81, p = 0.005, r = −0.50) (Figure 2B). cognitive training was assessed for its relationship with atrophy After adjusting for a 5% False Discovery Rate (FDR), non- and stimulation intensity. significant trends were observed for plasticity indices at T5 Group and post hoc analyses were calculated with non- and T10, as well as overall and maximum plasticity change parametric Kruskal–Wallis and Mann–Whitney U tests, (p < 0.1), while differences in SICI and LICI remained significant respectively. Correlations [bootstrapped with 10,000 resamples (p < 0.05). Furthermore, atrophy of the left IPL and average with bias-corrected confidence intervals] were calculated using atrophy remained significantly larger in AD (p < 0.05). Pearson’s r tests for whole group analysis and Kendall’s tau tests for subgroup analyses. After finding no difference between the Intervention Effects in Alzheimer Disease real/sham and sham/sham groups (see section “Results”) both Patients were combined into a single sham rTMS group to increase power. The largest score change in ADAS-Cog within 4–6 weeks after Change in ADAS-Cog (ratio post/pre) was defined as primary the end of intervention was significantly different between groups outcome and the effect of depression on ADAS-Cog outcome was (H[2] = 10.16, p = 0.006) (Figure 2C). Post hoc tests revealed assessed with a mediation analysis. The difference in ADAS-Cog a significantly higher improvement in the real/real group as in the real/real group was tested with a two-sided paired t-test compared to the sham/sham (U = 115, z = 3.13, p = 0.001, (pre vs. post-treatment and pre-treatment vs. follow-up). r = 0.64) but not the real/sham group (U = 110, z = 1.56, p = 0.121, Multivariable linear models were built to test either r = 0.31). Immediately after the intervention, the real/real associations between ADAS-Cog and other clinical, group improved their cognitive score on average by 2.18(±3.70) demographic, and physiological variables at baseline, or the points, while the real/sham group improved by 0.4(±4.14) points prediction of the clinical response by said variables, and/or and the sham/sham group showed a decline of 0.49(±2.48) intervention parameters (real vs. sham rTMS and real vs. sham points. During the following 4–6 weeks post-intervention, the cognitive stimulation). Longitudinal mixed effects models were real/real group continued to improve on average by an additional built to test associations between ADAS-Cog and physiological 1.69(±4.07) points, that is, at the end of the follow-up period variables across repeated assessment moments, when adjusting the real/real group improved on average by 3.87(±4.54) points for baseline clinical and demographic variables. Initial models from baseline (Figure 2D). Notably, the immediate change from were built according to prior knowledge, and sequential pre to post-treatment in the real/real group reached statistically regression models used to test the relevance of additional significant levels (t(15) = 2.408, p = 0.029, CI 95% [0.253, 4.16]) as variables of interest. Data transformations and polynomial well as the change from pre to follow-up (t(15) = 3.413, p = 0.004 models were used to test the better alternative to fit continuous CI 95% [1.45, 6.29]). Neither of the sham groups (real/sham and Frontiers in Aging Neuroscience | www.frontiersin.org 7 July 2020 | Volume 12 | Article 200 Brem et al. Combinatory Treatment in AD FIGURE 2 | (A) Plasticity change at T5, T10, and on average at T5–30 (T mean) expressed as the mean ratios of single-pulse TMS-measured MEP amplitudes pre-post iTBS before the intervention. (B) Paired-pulse TMS-measures of short-interval intracortical inhibition (SICI), intracortical facilitation (ICF), and long-interval intracortical inhibition (LICI) in AD (before intervention) and HC. (C) (including Italian sample): Best individual score change (ratio post/pre) in ADAS-Cog after real/real (light grey), sham/sham (medium grey), and after combined real cognitive training with sham rTMS (black). (D) Average ADAS-Cog score at pre, post and follow-up in the three treatment groups. (E) Average ADCS-CGIC scores. A score of 4 (dotted line) is equivalent to no change from pre- to post-intervention, scores < 4 indicate improvement, scores > 4 indicate decline. Indicated values correspond to mean ± standard error (SE). sham/sham) showed a significant cognitive change from pre to In these analyses the difference for ADAS-Cog score change post (Supplementary Table S2). Importantly, two-thirds of the remained significant (U = 225, z = 2.78, p = 0.004, r = 0.48) patients in the real group continued to improve in ADAS-Cog and the change in ADCS-CGIC (sham n = 11, real/real n = 10) even after the end of intervention. Comparing across groups, a became significant (U = 87, z = 2.35, p = 0.024, r = 0.51). marginally significant difference was observed in ADCS-CGIC Moreover, the change in ADAS-Cog was correlated with clinical (real/real: 3.25 ± 0.72, sham/sham: 4.08 ± 1.02, real/sham: changes as indicated by CGIC (r = 0.473, p = 0.031, 95% CI 0.08 4.40 ± 0.89; H[2] = 5.950, p = 0.051). We thus decided to explore to 0.75). Furthermore, a mediation analysis with bootstrapping post-hoc results and found a significantly larger improvement (10,000 samples, Preacher and Hayes, 2008), to examine the effect for the real/real as compared to the real/sham group (U = 42, of depression on intervention outcome, showed that depression z = 2.21, p = 0.04, r = 0.57) (Figure 2E), while the other group scores did not account for the variance in ADAS-Cog (R2 = 0.15, comparisons did not reach significance. The score change in F (2,18) = 1.625, p = 0.225; 95% bias corrected confidence interval ADCS-ADL immediately after the end of the intervention was for the mediating variable: [−0.007 to 0.037]). No major side significantly different between groups (H[2] = 6.95, p = 0.031). effects were reported in any participants. Post hoc tests showed that the real/real group (U = 18, z = −2.12, p = 0.036, r = −0.46) as well as the real/sham group (U = 53, z = 2.5, p = 0.011, r = 0.63) showed a significantly better outcome Association of Cognitive With than the sham/sham group. The real/real and real/sham groups Physiological and Other Measures did not differ significantly in ADL outcome (U = 89, z = 0.79, At baseline, ADAS-Cog (lower values indicate better cognitive p = 0.461, r = 0.16). function) was significantly correlated with rMT (r = −0.44, As hypothesized, the two sham groups did not differ p = 0.009, 95% CI −0.64 to −0.20), SICI (r = 0.564, p = 0.001, 95% significantly with regard to their largest ADAS-Cog score change CI 0.21 to 0.77) and marginally with LICI (r = 0.304, p = 0.096, (U = 23, z = −1.56, p = 0.122, r = 0.34) and their outcomes 95% CI 0.03 to 0.71), as well as atrophy in the left IPL (r = 0.42, in ADCS-CGIC (U = 19, z = 0.76, p = 0.537, r = 0.23). We p = 0.013, 95% CI 0.09 to 0.70), and mean atrophy (r = 0.37, therefore followed our strategy in accordance with the 2:1:1 group p = 0.033, 95% CI 0.00 to 0.68). Atrophy in the motor cortex was randomization and further compared the real/real group (n = 16) positively correlated with rMT (r = 0.389, p = 0.023) across all with the combined sham groups (n = 18) in order to assess participants. Cortical reactivity and plasticity were not associated rTMS-specific intervention effects and increase statistical power. with cortico-motor atrophy (ps > 0.05). The maximum difficulty Frontiers in Aging Neuroscience | www.frontiersin.org 8 July 2020 | Volume 12 | Article 200 Brem et al. Combinatory Treatment in AD level reached in region-specific tasks was significantly correlated adjusting for diagnosis. In additional linear regression analyses with absolute TMS intensity for the left DLPFC (r = 0.66, p = 0.04) of ADAS-Cog ratio (post-intervention/baseline) as a measure in the real/real group and with atrophy in the left (r = −0.707, of response to treatment in patients with AD, when adjusting p = 0.003) and the right IPL (r = −0.534, p = 0.04) (Table 2C). for rTMS intervention, a significant predictor of response in In multivariable linear regression analyses with data from all models, age, education, gender, and cognitive training were all participants at baseline, the association of ADAS-Cog with not significant predictors of response. However, mean atrophy rMT and SICI was confirmed, and found also for aMT, and atrophy over the left DLPFC, but not other cortical areas, mean plasticity indices averaged over T5–T30, maximum were found to be significant predictors of response (Table 3). plasticity, and LICI (Table 3 and Figure 3). None of these Importantly, in similar rTMS adjusted analyses, MEP amplitude, associations were confounded by mean atrophy. However, as well as plasticity at T5 and T10, mean plasticity from T5 to the associations between baseline ADAS-Cog and plasticity T30 and maximum plasticity, were all found to be significant measures, as well as LICI, were no longer significant when predictors of response (Table 3 and Figure 3). Interaction TABLE 3 | Parameter estimates and model fit statistics for multivariable linear models testing associations between TMS-driven measures of cortico-motor reactivity evaluated at baseline, and cognitive function (ADAS-Cog) also measured at baseline, or clinical response (ADAS-Cog ratio) in the sample from Boston (n = 21). Linear Models Dependent variable Baseline ADAS-Cog (AD and HC) ADAS-Cog ratio (post-intervention/baseline; AD only) Base model Independent variables Beta ± SE p R2 Beta ± SE p R2 Age (years) 0.3 ± 0.3 0.3 0.09 Education (years) 0.3 ± 0.7 0.6 Mean SBD 1.3 ± 0.6 0.04 rTMS −0.25 ± 0.06 <0.001 0.42 Single clinical or demographic Diagnosis −18 ± 3.6 <0.0001 0.5 variables added to base model Cognitive stimulation −0.1 ± 0.08 0.1 0.47 Age (years) −2.4*10−4 ± 4.6*10−3 0.96 0.39 Gender (male) −0.02 ± 0.07 0.8 0.39 Education (years) 6.9*10−3 ± 8.8*10−3 0.4 0.41 Mean SBD −0.02 ± 8.7*10−3a 0.02 0.54 Single structural variables Left DLPFC −0.02 ± 0.01a 0.03 0.51 (scalp-brain distance) added to base model Right DLPFC −5.2*10−3 ± 4.3*10−3 0.2 0.44 Left IFG −3.2*10−3 ± 5.5*10−3 0.6 0.4 Left IPL −3.6*10−4 ± 3.4*10−3 0.9 0.39 Right IPL −2,6*10−3 ± 4*10−3 0.5 0.41 MC −6.1*10−3 ± 0.01 0.6 0.4 Left STG −4.6*10−3 ± 5.3*10−3 0.4 0.42 Mean distance (7 areas) −0.02 ± 8.7*10−3a 0.02 0.54 Single physiological variables rMT −0.7 ± 0.2 <0.001 0.42 1.8*10−3 ± 3.2*10−3 0.6 0.4 added to base model aMT −0.7 ± 0.2 0.001 0.35 2.4*10−3 ± 3.4*10−3 0.5 0.41 MEP amplitude 1.9*10−3 ± 2.5*10−3 0.4 0.08 −1*10−4 ± 4*10−5a 0.02 0.56 Plasticity T5 −4.3 ± 3.9 0.3 0.08 −0.2 ± 0.1a,b <0.05 0.5 Plasticity T10 −8.7 ± 4.3a 0.06 0.25 −0.2 ± 0.08a 0.02 0.56 Mean Plasticity T5-T30 −8.9 ± 4.3a <0.05 0.26 −0.3 ± 0.1a 0.02 0.54 Maximum plasticity T5-T30 −9.9 ± 4.1a 0.02 0.31 −0.2 ± 0.09a,b <0.05 0.52 SICI 19.5 ± 5.1 <0.001 0.39 0.08 ± 0.09 0.4 0.42 ICF −1.9 ± 3.2 0.6 0.06 0.02 ± 0.04 0.7 0.39 LICI 12.2 ± 4.4a 0.01 0.34 0.05 ± 0.04 0.2 0.43 a One or more subjects were not included in this model due to being influential observations according to Cook’s distance analysis. b Becomes only marginally significant (0.05 < p < 0.08) when adjusted for mean scalp-brain distance. ADAS-Cog, cognitive subscale of the Alzheimer’s Disease Assessment Scale; ICF, intracortical facilitation; LICI, long-interval intracortical inhibition; MEP, motor evoked potential; rMT/aMT, resting/active motor threshold; rTMS, repetitive transcranial magnetic stimulation; SBD, Scalp-brain distance; SICI, short-interval intracortical inhibition; T5/T10/T30, Time-points at 5 10 and 30 min post intermittent Theta Burst Stimulation. Bold font indicates significant values. Frontiers in Aging Neuroscience | www.frontiersin.org 9 July 2020 | Volume 12 | Article 200 Brem et al. Combinatory Treatment in AD FIGURE 3 | Relationship between TMS-driven measures and cognitive function (ADAS-Cog). Relationship at baseline (A–D) in AD and HC, or clinical response over time (ADAS-Cog ratio) in AD patients (E–H) treated with real or sham rTMS. Cortico-motor plasticity was expressed as the change of cortical reactivity from baseline to post-iTBS. Circles reflect data for individuals and lines are unadjusted regression lines for the specified groups. When adjusting for age, education and mean atrophy, significant associations were found between baseline ADAS-Cog and resting motor threshold [rMT – (A)], short-interval intracortical inhibition [SICI – (B)], long-interval intracortical inhibition [LICI – (C)] and mean plasticity from 5 to 30 min post-iTBS (D). However, in models adjusting for rTMS intervention (real vs. sham), mean plasticity (H), but not rMT (E), SICI (F), and LICI (G) was found to be a significant predictor of response. terms with rTMS were not significant and none of these cognitive symptoms in AD patients. Furthermore, we aimed to associations were confounded by cognitive stimulation. However, analyze if cortical plasticity was different in patients with AD plasticity at T5 and maximum plasticity became only marginally versus HC as shown in previous studies, whether its modulation significant predictors of clinical response when adjusting for by the combined rTMS-training intervention was possible, and mean atrophy. The effects for the remaining predictors were not whether it might be associated with changes in cognitive function. significantly affected. For the primary aim, in order to reach a suitable sample size, Even though change in neurophysiological parameters findings from two parallel trials utilizing the same recruitment exploring plasticity from pre to post-treatment did not criteria, interventions, and cognitive assessments (Boston and differ significantly between groups (H[2] = 0.464, p = 0.793) Rome) were combined. (Supplementary Table S1), longitudinal regression models were built to test physiological and other factors associated to ADAS-Cog across intervention. These multivariable analyses Baseline Measures were adjusted for rTMS, that was again a significant parameter At baseline, we found no significant difference in rMT and in the models. In separate rTMS-adjusted models, age, gender, aMT between patients with AD and HC. However, both were education, cognitive stimulation and mean atrophy were not robust indicators of severity of cognitive dysfunction, even when associated to ADAS-Cog, and the latter two variables were controlling for brain atrophy. Cortico-motor plasticity revealed also not confounders in any of the subsequent models. As significant differences between AD patients and HC, as could found at baseline, rMT and aMT were significantly associated be expected from findings of a positive correlation between to ADAS-Cog, but the associations with SICI, LICI, and T5 amyloid-β cerebrospinal fluid levels and long-term potentiation and T10 plasticity measures were no longer significant in (LTP)-like effects induced by iTBS (Mori et al., 2011). Another these longitudinal analyses. However, the association between study (Koch et al., 2012) showed significant differences in brain ADAS-Cog and both mean and maximum plasticity from T5 to plasticity between AD and HC at several time-points between 11 T30 remained significant. Importantly, these were also the only and 25 min after iTBS. We found significant differences for T5, instances where an interaction term between the physiological T10, average plasticity (T5–30), and for maximum iTBS-induced measure and rTMS was also significant, showing not only that plasticity, as well as significant linear associations between all these associations between cognition and physiology persist of these plasticity measures and ADAS-Cog at baseline, when across intervention, but also that they were modulated by active adjusting for several demographic and clinical factors, as well rTMS treatment (Table 4). as atrophy. Change in neurophysiological parameters exploring plasticity from pre to post-treatment did not differ significantly between AD groups. The differential time-course of post-iTBS DISCUSSION plasticity could be related to the fact that our patients were mostly medicated, while Koch et al. (2012) investigated medication- The primary aim of the present study was to assess the efficacy naïve patients. Huang et al. (2007) found that memantine of rTMS combined with cognitive training for the treatment of significantly impacts measures of TBS-induced plasticity in Frontiers in Aging Neuroscience | www.frontiersin.org 10 July 2020 | Volume 12 | Article 200 Brem et al. Combinatory Treatment in AD TABLE 4 | Parameter estimates and model fit statistics for multivariable mixed effects longitudinal models testing associations between TMS-driven measures of corticomotor reactivity and cognitive function (ADAS-Cog), measured across time (pre and post-treatment), in the sample from Boston (n = 21). Longitudinal Mixed Effects Models Dependent variable: ADAS-Cog (AD only) Independent variables Beta ± SE p Base model rTMS −10.5 ± 3.8 0.006 Time (pre- vs. post treatment) −2.8 ± 0.9 0.002 Single clinical or demographic variables added to base model Cognitive stimulation −2.5 ± 5.4 0.6 Age (years) −0.07 ± 0.3 0.8 Gender (male) 4.8 ± 4 0.2 Education (years) 0.1 ± 0.5 0.9 Mean scalp-brain distance −0.3 ± 0.6 0.5 Single physiological variables added to base model rMT −0.5 ± 0.2 0.004 aMT aMT 1.9 ± 0.9 < 0.05 aMT2 −0.02 ± 0.01 0.03 MEP amplitude 4.7*10−4 ± 6.4*10−4 0.5 Plasticity T5 1.5 ± 1.2 0.2 Plasticity T10 −0.4 ± 1.3 0.7 Mean Plasticity T5–T30 MeanT5-T30 5.2 ± 2.2 0.02 MeanT5-T30*rTMS −6 ± 2.6 0.02 Maximum Plasticity T5–T30 MaxT5-T30 4.9 ± 1.7 0.005 MaxT5-T30*rTMS −5.2 ± 1.9 0.005 SICIa 3.4 ± 2.8 0.2 ICF −0.4 ± 0.8 0.6 LICI −0.5 ± 1.5 0.7 a One or more subjects were not included in this model due to being influential observations according to Cook’s distance analysis. ADAS-Cog, cognitive subscale of the Alzheimer’s Disease Assessment Scale; ICF, intracortical facilitation; LICI, long-interval intracortical inhibition; MEP, motor evoked potential; rMT/aMT, resting/active motor threshold; rTMS, repetitive transcranial magnetic stimulation; SICI, short-interval intracortical inhibition; T5/T10/T30, Time-points at 5 10 and 30 min post intermittent Theta Burst Stimulation. normal subjects, which might have played a role for some cognitive improvements and not progression in the training. patients. And in an own study, we found that pharmacological High-frequency rTMS protocols (such as the 10 Hz stimulation treatment can also have a specific impact on plasticity measures protocol used in the treatment phase of the present study) in AD (Brem et al., 2013). are established as a safe and effective method to increase We also found reduced SICI and LICI in AD patients, which brain excitability and have been applied in a wide range of were associated to baseline ADAS-Cog. LICI has been shown to patient populations in a similar way as in the present study interact with short-latency afferent inhibition (SAI) in healthy (Bentwich et al., 2011; Cotelli et al., 2011; Chang et al., 2012; subjects (Udupa et al., 2009), which again has been shown to Kakuda et al., 2013; Khedr et al., 2014; Yang et al., 2013; be reduced in AD (Koch et al., 2012), and may be a valuable Kim et al., 2014; Wobrock et al., 2015; McGirr et al., 2015; biomarker that could be assessed in future studies using TMS- Quan et al., 2015). The degree of improvement in the real/real EEG across regions of the cortical convexity (Farzan et al., 2010). group is clinically meaningful (Schrag et al., 2012) and was not SICI has been previously reported to be reduced in AD patients mediated by potential effects of rTMS on depression. Previous and to be related to disease severity (Liepert et al., 2001), however, studies using the same approach (Bentwich et al., 2011; Rabey findings are divergent (Freitas et al., 2011). Atrophy over the et al., 2013; Lee et al., 2015; Rabey and Dobronevsky, 2016) primary motor cortex was not associated with cortico-motor also reported similar changes in ADAS-Cog, which further reactivity and plasticity, and mean atrophy did not confound the strengthens our findings. Interestingly, substantial cognitive associations between cognitive and motor measures. Atrophy in improvement occurred after the end of intervention, possibly AD was significantly larger, specifically for the left IPL, which is reflecting time-prolonged modulation of neurophysiological in accordance with previous findings stating that IPL is among processes underlying cognitive maintenance. After combining the first brain regions to show atrophy (Jacobs et al., 2011). the two sham groups, an improvement in the ADCS-CGIC in the real intervention group that was associated with cognitive Clinical Effect improvement became significant, stressing the importance of the Overall training progression did not differ between the two addition of rTMS. Both the real/real and the real/sham group groups who received active cognitive training, indicating that improved significantly more in ADL than the sham/sham group the addition of rTMS to cognitive training is crucial to achieve after the intervention. Frontiers in Aging Neuroscience | www.frontiersin.org 11 July 2020 | Volume 12 | Article 200 Brem et al. Combinatory Treatment in AD In contrast to previous studies (Bentwich et al., 2011; Rabey not only at baseline but also after intervention, in an rTMS- et al., 2013; Lee et al., 2015; Rabey and Dobronevsky, 2016), we dependent fashion. used neuronavigation to ensure precise targeting of the desired During the intervention, rTMS was not applied to the motor cortical brain regions. Furthermore, we demonstrated that sham cortex, and yet this was the cortical area from which plasticity rTMS combined with real cognitive training (real/sham group) measures were collected. The observed correlations between showed similar effects as the sham/sham group, emphasizing cortico-motor plasticity and cognitive measures strongly suggest the importance of rTMS. Though, we cannot establish whether that alterations in plasticity may occur globally. Further support rTMS in combination with cognitive training produces the for this assumption is suggested by the positive correlation same effect as rTMS on its own, we strongly believe that found between maximum level achieved in training tasks and the combination is crucial (Brem and Sensi, 2018). Added stimulation intensity. A neuroimaging study has shown that efficacy might arise from Hebbian mechanisms of synaptic memory training can drive non-motor brain activation patterns reinforcement induced by the combined impact of cognitive of patients with mild cognitive impairment toward normalization training and rTMS on the same neural networks engaged in (Belleville et al., 2011). Changes in motor cortical measures after the different tasks. This is supported by the demonstrated stimulation of non-motor areas have been reported previously cortico-motor reactivity and plasticity effects in the participants (Pallanti et al., 2012), as has prediction of non-motor effects studied in Boston. Blinding is difficult to assess given the of prefrontal stimulation by motor cortical measures (Oliveira- nature of this cohort. Nonetheless, when asked to indicate Maia et al., 2017). These effects could arise through changes group assignment among the participants studied in Boston, in overlapping subcortical glutamatergic and dopaminergic 6/10 patients (60%) in the real/real group and 6/11 patients pathways (Wang et al., 2012), which in turn could be driven via in the sham groups (55%) thought they had received real cortico-subcortical projections from stimulated non-motor areas. intervention. However, as no electrical surface electrodes were used in the sham protocol, we cannot entirely exclude non- Study Limitations neural effects. This study did not assess the effects of rTMS alone, which would Our patients progressed less in training tasks related to the left have necessitated adding a fourth study arm. Though magnetic and right IPL the greater the cortical atrophy in these regions. stimulation alone could affect cognition without additionally It is important to consider that greater atrophy would have inducing brain activity (Wang et al., 2019), the combination resulted in relatively lower rTMS intensity at cortical level in of rTMS with cognitive interventions is thought to be more IPL. Furthermore, training gains were significantly correlated effective (Brem and Sensi, 2018). Given our main interest in the with absolute TMS intensity for the left DLPFC. Therefore, combined method, we actively decided against using resources considering that atrophy was a negative predictor of clinical for an additional study arm and to favor the inclusion of response, future studies may base stimulation intensity on the two rTMS sham groups (2:1:1 randomization). The challenges modeling of current distribution (Wagner et al., 2008). Notably, in enrolling AD patients in such an intensive protocol (daily the activation of a broad range of brain areas known to be treatment for 6 weeks) account for the modest sample size. associated with AD pathology was thought to maximize cognitive Though the two study sites used slightly different criteria, we effects. Though we did not find clear benefits of cognitive training believe that the effects of a center bias on the primary outcome are alone, we nevertheless observed numerical improvements across minimal as results remained largely unchanged after adding the a range of parameters. Optimizing and individualizing the results from Rome to the results from Boston (Supplementary cognitive training could possibly improve interaction effects with Results). However, the patients in Rome did not undergo rTMS and enhance clinical outcomes. neurophysiological measures, which might have had an impact on compliance, tiredness at pre- and posttest and likelihood to Correlation of Physiological Measures adhere to the study plan. Finally, although we chose a control cognitive task with minimal cognitive engagement in order to With Cognitive Performance Across avoid having the control condition represent an “uncontrolled Treatment brain state,” it is possible that changes in brain state could Transcranial magnetic stimulation has been previously found to have lead to unspecific improvements in the sense of an active be precise enough to track disease-related neuroplastic changes control training. of motor cortex output in early AD (Ferreri et al., 2003). Notably, in our study, MEP amplitude, as well as plasticity at T5 and T10, mean and maximum plasticity were all significant predictors CONCLUSION of clinical response, even when adjusting for mean atrophy and other intervention parameters. Furthermore, the association We found that the combinatory treatment of rTMS and between ADAS-Cog and several plasticity measures, as well as cognitive training resulted in significant cognitive improvement motor thresholds, was preserved across the intervention resulting as assessed with the ADAS-Cog, while cognitive training alone in significant cognitive improvement. Importantly, in the case of did not lead to significant improvement compared to placebo plasticity measures, active rTMS intervention was a significant treatment. Confirming prior reports, we show abnormalities in modulator of this association. Thus, we found that measures the mechanisms of plasticity and cortical reactivity in patients of iTBS-induced plasticity were associated to cognitive function with AD, and show for the first time that these can be modulated Frontiers in Aging Neuroscience | www.frontiersin.org 12 July 2020 | Volume 12 | Article 200 Brem et al. Combinatory Treatment in AD by a combined non-invasive stimulation and cognitive training and the NCATS NIH, UL1 RR025758), and an investigator- paradigm resulting in behavioral, clinical and cognitive benefits. initiated grant from Neuronix Ltd. A-KB was supported by the This study supports the relevance of non-pharmacological Young Academics Support and the Stiefel-Zangger Foundation combinatory interventions in individuals with AD and provides of the University of Zurich, Switzerland, and the Swiss National important groundwork for future studies to build upon. Many Foundation (PBZHP1_147196). AO-M was supported by a questions remain to be answered, including how cognitive Junior Research and Career Development Award (HMSP- improvements translate into patients’ daily lives and whether ICJ/0020/2011) from the Harvard Medical School and Portugal neurophysiological measures could act as potential biomarkers Program, Fundação para a Ciência e Tecnologia (Portugal). for interventions in AD. Future research should assess a broader The content is solely the responsibility of the authors and range of patients and aim to improve applicability in the does not necessarily represent the official views of Harvard clinical environment. Catalyst, Harvard University and its affiliated academic health care centers, the National Institutes of Health or the Sidney R. Baer, Jr. Foundation. DATA AVAILABILITY STATEMENT The datasets generated for this study are available on request to ACKNOWLEDGMENTS the corresponding author. We are grateful to all patients and caregivers for their participation in the study. We thank Léonie Asboth, MSc, ETHICS STATEMENT Edward Gold, Christina Carbone, BA, Ilya Vidrin, BS, Cory Kosche, Jessica Olson, BS, Fritz Ifert-Miller, BA, and Zachary The studies involving human participants were reviewed and Gray, BS for their support in data collection; Lisa Iguchi, Ph.D. approved by the local Institutional Review Boards at both Beth for the neuropsychological exams; Ann Connor, RN, MS for Israel Deaconess Medical Center and Polyclinic A. Gemelli help with patient assessments and study oversight, and Drs. Foundation-IRCCS. The patients/participants provided their Daniel Press, MD and Mark Eldaief, MD for assistance with the written informed consent to participate in this study. neurological exams. AUTHOR CONTRIBUTIONS SUPPLEMENTARY MATERIAL A-KB and AP-L designed the study in Boston. CM, PP, and PR The Supplementary Material for this article can be found designed the study in Rome. A-KB, LS, NA, and ES collected the online at: https://www.frontiersin.org/articles/10.3389/fnagi. data in Boston. RD and DQ collected the data in Rome. A-KB, PF, 2020.00200/full#supplementary-material AO-M, LS, NA, and ES analyzed the Boston data. RD, CM, PP, and DQ analyzed the Rome data. A-KB prepared the initial draft TABLE S1 | Plasticity measures in the AD sample from Boston (n = 21) at baseline report. AP-L edited the report, which was critically reviewed by and after the end of the intervention (T5, T10, T20, T30). all authors. A-KB, PR, and AP-L had full access to all the data in TABLE S2 | ADAS-Cog results from each site (Boston and Rome) and the study and take responsibility for the integrity of the data and combined results. the accuracy of the data analysis. RESULTS | Before adding the results from Rome to the results from Boston, we find very similar results. In Boston, the largest score change in ADAS-Cog within one month after the end of intervention was significantly different between groups FUNDING (Boston: H[2] = 11.626, p = 0.003; Boston combined with Rome: (H[2] = 10.16, p = 0.006)). Post hoc tests also revealed a significantly higher improvement in the Work on this study was supported in part by grants from real/real group as compared to the sham/sham group [Boston: U = 58, z = 3.04, the Sidney R. Baer, Jr. Foundation, the National Institutes of p = 0.001, r = 0.76; Boston combined with Rome: (U = 115, z = 3.13, p = 0.001, r = 0.64)]. However, while the real/real group was also significantly different form Health (R01HD069776, R01NS073601, R21 MH099196, R21 the real/sham group in Boston (U = 43, z = 2.21, p = 0.028, r = 0.57), this NS082870, R21 NS085491, and R21 HD07616), Harvard Catalyst difference became non-significant after combining the results from Boston with | The Harvard Clinical and Translational Science Center (NCRR Rome (U = 110, z = 1.56, p = 0.121, r = 0.31). REFERENCES Ballard, C., Gauthier, S., Corbett, A., Brayne, C., Aarsland, D., and Jones, E. (2011). Alzheimer’s disease. Lancet 377, 1019–1031. doi: 10.1016/S0140- Ahmed, M. A., Darwish, E. S., Khedr, E. M., El Serogy, Y. M., and Ali, A. M. 6736(10)61349-9 (2012). Effects of low versus high frequencies of repetitive transcranial magnetic Beekly, D. L., Ramos, E. M., Lee, W. W., Deitrich, W. D., Jacka, stimulation on cognitive function and cortical excitability in Alzheimer’s M. E., Wu, J., et al. (2007). The National Alzheimer’s coordinating dementia. J. Neurol. 259, 83–92. doi: 10.1007/s00415-011-6128-4 center (NACC) database: the uniform data set. Alzheimer Dis. Alzheimer’s Association (2012). 2012 Alzheimer’s disease facts and figures. Assoc. Disord. 21, 249–258. doi: 10.1097/WAD.0b013e31814 Alzheimers Dement. 8, 131–168. doi: 10.1016/j.jalz.2012.02.001 2774e Frontiers in Aging Neuroscience | www.frontiersin.org 13 July 2020 | Volume 12 | Article 200 Brem et al. Combinatory Treatment in AD Belleville, S., Clément, F., Mellah, S., Gilbert, B., Fontaine, F., and Gauthier, combined with mobility training for gait disturbance after stroke: a preliminary S. (2011). Training-related brain plasticity in subjects at risk of developing study. Brain Inj. 27, 1080–1086. doi: 10.3109/02699052.2013.794973 Alzheimer’s disease. Brain 134, 1623–1634. doi: 10.1093/brain/awr037 Kallio, E.-L., Öhman, H., Kautiainen, H., Hietanen, M., and Pitkälä, K. (2017). Bentwich, J., Dobronevsky, E., Aichenbaum, S., Shorer, R., Peretz, R., Khaigrekht, Cognitive Training Interventions for patients with Alzheimer’s disease: a M., et al. (2011). Beneficial effect of repetitive transcranial magnetic stimulation systematic review. J. Alzheimers Dis. 56, 1349–1372. doi: 10.3233/JAD-16 combined with cognitive training for the treatment of Alzheimer’s disease: a 0810 proof of concept study. J. Neural. Transm. 118, 463–471. doi: 10.1007/s00702- Khedr, E. M., Abo El-Fetoh, N., Ali, A. M., El-Hammady, D. H., Khalifa, H., Atta, 010-0578-1 H., et al. (2014). Dual-hemisphere repetitive transcranial magnetic stimulation Brem, A.-K., Atkinson, N. J., Seligson, E. E., and Pascual-Leone, A. (2013). for rehabilitation of poststroke aphasia: a randomized, double-blind clinical Differential pharmacological effects on brain reactivity and plasticity in trial. Neurorehab. Neural Re. 28, 740–750. doi: 10.1177/1545968314521009 Alzheimer’s disease. Front. Psychiatry 4:124. doi: 10.3389/fpsyt.2013.00124 Kim, C., Choi, H. E., Jung, H., Lee, B.-J., Lee, K. H., and Lim, Y.-J. (2014). Brem, A.-K., and Sensi, S. L. (2018). Towards combinatorial approaches for Comparison of the Effects of 1 Hz and 20 Hz rTMS on Motor Recovery in preserving cognitive fitness in aging. Trends Neurosci. 41, 885–897. doi: 10. Subacute Stroke Patients. Ann. Rehabil. Med. 38, 585–591. doi: 10.5535/arm. 1016/j.tins.2018.09.009 2014.38.5.585 Buss, S. S., Fried, P. J., and Pascual-Leone, A. (2019). Therapeutic noninvasive brain Koch, G., Di Lorenzo, F., Bonnì, S., Ponzo, V., Caltagirone, C., and Martorana, A. stimulation in Alzheimer’s disease and related dementias. Curr. Opin. Neurol. (2012). Impaired LTP- but not LTD-like cortical plasticity in Alzheimer’s disease 32, 292–304. doi: 10.1097/WCO.0000000000000669 patients. J. Alzheimers Dis. 31, 593–599. doi: 10.3233/JAD-2012-120532 Chang, W. H., Kim, Y.-H., Yoo, W.-K., Goo, K.-H., Park, C.-H., Kim, S. T., Lee, J., Choi, B. H., Oh, E., Sohn, E. H., and Lee, A. Y. (2015). Treatment of et al. (2012). rTMS with motor training modulates cortico-basal ganglia- Alzheimer’s disease with repetitive transcranial magnetic stimulation combined thalamocortical circuits in stroke patients. Restor. Neurol. Neurosci. 30, 179– with cognitive training: a prospective, randomized, double-blind, placebo- 189. doi: 10.3233/RNN-2012-110162 controlled study. J. Clin. Neurol. 12, 57–64. Clare, L., Woods, R. T., Moniz Cook, E. D., Orrell, M., and Spector, A. (2003). Liepert, J., Bär, K. J., Meske, U., and Weiller, C. (2001). Motor cortex disinhibition Cognitive rehabilitation and cognitive training for early-stage Alzheimer’s in Alzheimer’s disease. Clin. Neurophysiol. 112, 1436–1441. doi: 10.1016/s1388- disease and vascular dementia. Cochrane Database Syst. Rev. 2003:CD003260. 2457(01)00554-5 doi: 10.1002/14651858.CD003260 Mattson, M. P. (2004). Pathways towards and away from Alzheimer’s disease. Cooke, S. F., and Bliss, T. V. P. (2006). Plasticity in the human central nervous Nature 430, 631–639. doi: 10.1038/nature02621 system. Brain 129, 1659–1673. doi: 10.1093/brain/awl082 McGirr, A., Van den Eynde, F., Tovar-Perdomo, S., Fleck, M. P. A., and Berlim, M. Cotelli, M., Calabria, M., Manenti, R., Rosini, S., Zanetti, O., Cappa, S. F., et al. T. (2015). Effectiveness and acceptability of accelerated repetitive transcranial (2011). Improved language performance in Alzheimer disease following brain magnetic stimulation (rTMS) for treatment-resistant major depressive disorder: stimulation. J. Neurol. Neurosurg. Psychiatry 82, 794–797. doi: 10.1136/jnnp. an open label trial. J. Affect. Disord. 173, 216–220. doi: 10.1016/j.jad.2014. 2009.197848 10.068 Di Lorenzo, F., Ponzo, V., Bonnì, S., Motta, C., Negrão Serra, P. C., Bozzali, M., McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., and Stadlan, et al. (2016). LTP-like cortical plasticity is disrupted in Alzheimer’s disease E. M. (1984). Clinical diagnosis of Alzheimer’s disease: report of the NINCDS- patients independently from age of onset. Ann. Neurol. 80, 202–210. doi: 10. ADRDA work group under the auspices of department of health and human 1002/ana.24695 services task force on Alzheimer’s disease. Neurology 34, 939–944. doi: 10.1212/ Farzan, F., Barr, M. S., Levinson, A. J., Chen, R., Wong, W., Fitzgerald, P. B., et al. wnl.34.7.939 (2010). Reliability of long-interval cortical inhibition in healthy human subjects: Mori, F., Rossi, S., Sancesario, G., Codecà, C., Mataluni, G., Monteleone, F., et al. a TMS-EEG study. J. Neurophysiol. 104, 1339–1346. doi: 10.1152/jn.00279. (2011). Cognitive and cortical plasticity deficits correlate with altered amyloid- 2010 β CSF levels in multiple sclerosis. Neuropsychopharmacology 36, 559–568. doi: Ferreri, F., Pauri, F., Pasqualetti, P., Fini, R., Dal Forno, G., and Rossini, 10.1038/npp.2010.187 P. M. (2003). Motor cortex excitability in Alzheimer’s disease: a transcranial Oliveira-Maia, A. J., Press, D., and Pascual-Leone, A. (2017). Modulation of magnetic stimulation study. Ann. Neurol. 53, 102–108. doi: 10.1002/ana. motor cortex excitability predicts antidepressant response to prefrontal cortex 10416 repetitive transcranial magnetic stimulation. Brain Stimul. 10, 787–794. doi: Freitas, C., Mondragón-Llorca, H., and Pascual-Leone, A. (2011). Noninvasive 10.1016/j.brs.2017.03.013 brain stimulation in Alzheimer’s disease: systematic review and perspectives for Pallanti, S., Di Rollo, A., Antonini, S., Cauli, G., Hollander, E., and Quercioli, the future. Exp. Gerontol. 46, 611–627. doi: 10.1016/j.exger.2011.04.001 L. (2012). Low-frequency rTMS over right dorsolateral prefrontal cortex Haffen, E., Chopard, G., Pretalli, J.-B., Magnin, E., Nicolier, M., Monnin, J., et al. in the treatment of resistant depression: cognitive improvement is (2012). A case report of daily left prefrontal repetitive transcranial magnetic independent from clinical response, resting motor threshold is related stimulation (rTMS) as an adjunctive treatment for Alzheimer disease. Brain to clinical response. Neuropsychobiology 65, 227–235. doi: 10.1159/00033 Stimul. 5, 264–266. doi: 10.1016/j.brs.2011.03.003 6999 Hill, N. T. M., Mowszowski, L., Naismith, S. L., Chadwick, V. L., Valenzuela, Pascual-Leone, A., Valls-Solé, J., Wassermann, E. M., and Hallett, M. M., and Lampit, A. (2017). Computerized cognitive training in older adults (1994). Responses to rapid-rate transcranial magnetic stimulation of the with mild cognitive impairment or dementia: a systematic review and meta- human motor cortex. Brain 117(Pt 4), 847–858. doi: 10.1093/brain/117. analysis. Am. J. Psychiatry 174, 329–340. doi: 10.1176/appi.ajp.2016.1603 4.847 0360 Preacher, K. J., and Hayes, A. F. (2008). Asymptotic and resampling strategies for Huang, Y.-Z., Chen, R.-S., Rothwell, J. C., and Wen, H.-Y. (2007). The assessing and comparing indirect effects in multiple mediator models. Behav. after-effect of human theta burst stimulation is NMDA receptor Res. Methods 40, 879–891. doi: 10.3758/brm.40.3.879 dependent. Clin. Neurophysiol. 118, 1028–1032. doi: 10.1016/j.clinph.2007. Price, C. J., Winterburn, D., Giraud, A. L., Moore, C. J., and Noppeney, U. 01.021 (2003). Cortical localisation of the visual and auditory word form areas: a Huang, Y.-Z., Edwards, M. J., Rounis, E., Bhatia, K. P., and Rothwell, J. C. (2005). reconsideration of the evidence. Brain Lang. 86, 272–286. doi: 10.1016/s0093- Theta burst stimulation of the human motor cortex. Neuron 45, 201–206. doi: 934x(02)00544-8 10.1016/j.neuron.2004.12.033 Quan, W. X., Zhu, X. L., Qiao, H., Zhang, W. F., Tan, S. P., Zhou, D. F., Jacobs, H. I. L., Van Boxtel, M. P. J., Uylings, H. B. M., Gronenschild, et al. (2015). The effects of high-frequency repetitive transcranial magnetic E. H. B. M., Verhey, F. R., and Jolles, J. (2011). Atrophy of the parietal lobe stimulation (rTMS) on negative symptoms of schizophrenia and the follow-up in preclinical dementia. Brain Cogn. 75, 154–163. doi: 10.1016/j.bandc.2010. study. Neurosci. Lett. 584, 197–201. doi: 10.1016/j.neulet.2014.10.029 11.003 Rabey, J. M., and Dobronevsky, E. (2016). Repetitive transcranial magnetic Kakuda, W., Abo, M., Watanabe, S., Momosaki, R., Hashimoto, G., Nakayama, stimulation (rTMS) combined with cognitive training is a safe and Y., et al. (2013). High-frequency rTMS applied over bilateral leg motor areas effective modality for the treatment of Alzheimer’s disease: clinical Frontiers in Aging Neuroscience | www.frontiersin.org 14 July 2020 | Volume 12 | Article 200 Brem et al. Combinatory Treatment in AD experience. J. Neural Transm. 123, 1449–1455. doi: 10.1007/s00702-016- Wagner, T., Eden, U., Fregni, F., Valero-Cabre, A., Ramos-Estebanez, C., Pronio- 1606-6 Stelluto, V., et al. (2008). Transcranial magnetic stimulation and brain atrophy: Rabey, J. M., Dobronevsky, E., Aichenbaum, S., Gonen, O., Marton, R. G., and a computer-based human brain model study. Exp. Brain Res. 186, 539–550. Khaigrekht, M. (2013). Repetitive transcranial magnetic stimulation combined doi: 10.1007/s00221-007-1258-8 with cognitive training is a safe and effective modality for the treatment of Wang, M., Wong, A. H., and Liu, F. (2012). Interactions between NMDA and Alzheimer’s disease: a randomized, double-blind study. J. Neural. Transm. 120, dopamine receptors: a potential therapeutic target. Brain Res. 1476, 154–163. 813–819. doi: 10.1007/s00702-012-0902-z doi: 10.1016/j.brainres.2012.03.029 Rogalsky, C., Matchin, W., and Hickok, G. (2008). Broca’s area, sentence Wang, X., Mao, Z., Ling, Z., and Yu, X. (2019). Repetitive transcranial magnetic comprehension, and working memory: an fMRI study. Front. Hum. Neurosci. stimulation for cognitive impairment in Alzheimer’s disease: a meta-analysis of 2:14. doi: 10.3389/neuro.09.014.2008 randomized controlled trials. J. Neurol. 267, 791–801. doi: 10.1007/s00415-019- Rossi, S., Hallett, M., Rossini, P. M., and Pascual-Leone, A. (2009). Safety, ethical 09644-y considerations, and application guidelines for the use of transcranial magnetic Wobrock, T., Guse, B., Cordes, J., Wölwer, W., Winterer, G., Gaebel, W., stimulation in clinical practice and research. Clin. Neurophysiol. 120, 2008– et al. (2015). Left prefrontal high-frequency repetitive transcranial magnetic 2039. doi: 10.1016/j.clinph.2009.08.016 stimulation for the treatment of schizophrenia with predominant negative Rossini, P. M., Barker, A. T., Berardelli, A., Caramia, M. D., Caruso, G., Cracco, symptoms: a sham-controlled, randomized multicenter trial. Biol. Psychiatry 77, R. Q., et al. (1994). Non-invasive electrical and magnetic stimulation of the 979–988. doi: 10.1016/j.biopsych.2014.10.009 brain, spinal cord and roots: basic principles and procedures for routine Yang, Y.-R., Tseng, C.-Y., Chiou, S.-Y., Liao, K.-K., Cheng, S.-J., Lai, K.-L., et al. clinical application. Report of an IFCN committee. Electroencephalogr. Clin. (2013). Combination of rTMS and treadmill training modulates corticomotor Neurophysiol. 91, 79–92. doi: 10.1016/0013-4694(94)90029-9 inhibition and improves walking in Parkinson disease: a randomized Rossini, P. M., Burke, D., Chen, R., Cohen, L. G., Daskalakis, Z., Di trial. Neurorehabil. Neural. Repair. 27, 79–86. doi: 10.1177/15459683124 Iorio, R., et al. (2015). Non-invasive electrical and magnetic stimulation 51915 of the brain, spinal cord, roots and peripheral nerves: Basic principles and procedures for routine clinical and research application. An updated Conflict of Interest: AO-M is recipient of a grant from Schuhfried GmBH for report from an I.F.C.N. Committee. Clin. Neurophysiol. 126, 1071–1107. doi: norming and validation of cognitive tests, and national coordinator for Portugal 10.1016/j.clinph.2015.02.001 of a non-interventional study (EDMS-ERI-143085581, 4.0) to characterize a Rusjan, P. M., Barr, M. S., Farzan, F., Arenovich, T., Maller, J. J., Fitzgerald, P. B., treatment-resistant depression cohort in Europe, sponsored by Janssen-Cilag Ltd, et al. (2010). Optimal transcranial magnetic stimulation coil placement for and trial of a psilocybin therapy for treatment-resistant depression, sponsored by targeting the dorsolateral prefrontal cortex using novel magnetic resonance Compass Pathways, Ltd. (EudraCT NUMBER: 2017-003288-36). AP-L serves on image-guided neuronavigation. Hum. Brain Mapp. 31, 1643–1652. doi: 10.1002/ the scientific advisory boards for Starlab Neuroscience, Neuroelectrics, Constant hbm.20964 Therapy, Cognito, and Neosync; and is listed as an inventor on several issued and Ryu, S.-H., Katona, C., Rive, B., and Livingston, G. (2005). Persistence of and pending patents on the real-time integration of transcranial magnetic stimulation changes in neuropsychiatric symptoms in Alzheimer disease over 6 months: (TMS) with electroencephalography (EEG) and magnetic resonance imaging the LASER-AD study. Am. J. Geriatr. Psychiatry 13, 976–983. doi: 10.1176/appi. (MRI). ajgp.13.11.976 Sajonz, B., Kahnt, T., Margulies, D. S., Park, S. Q., Wittmann, A., Stoy, M., et al. The remaining authors declare that the research was conducted in the absence of (2010). Delineating self-referential processing from episodic memory retrieval: any commercial or financial relationships that could be construed as a potential common and dissociable networks. Neuroimage 50, 1606–1617. doi: 10.1016/j. conflict of interest. neuroimage.2010.01.087 Schrag, A., Schott, J. M., and Alzheimer’s Disease Neuroimaging Initiative (2012). Copyright © 2020 Brem, Di Iorio, Fried, Oliveira-Maia, Marra, Profice, Quaranta, What is the clinically relevant change on the ADAS-Cog? J. Neurol. Neurosurg. Schilberg, Atkinson, Seligson, Rossini and Pascual-Leone. This is an open-access Psychiatr. 83, 171–173. doi: 10.1136/jnnp-2011-300881 article distributed under the terms of the Creative Commons Attribution License Silvanto, J., and Pascual-Leone, A. (2008). State-dependency of transcranial (CC BY). The use, distribution or reproduction in other forums is permitted, provided magnetic stimulation. Brain Topogr. 21, 1–10. doi: 10.1007/s10548-008-0067-0 the original author(s) and the copyright owner(s) are credited and that the original Udupa, K., Ni, Z., Gunraj, C., and Chen, R. (2009). Interactions between short publication in this journal is cited, in accordance with accepted academic practice. latency afferent inhibition and long interval intracortical inhibition. Exp. Brain No use, distribution or reproduction is permitted which does not comply with Res. 199, 177–183. doi: 10.1007/s00221-009-1997-9 these terms. Frontiers in Aging Neuroscience | www.frontiersin.org 15 July 2020 | Volume 12 | Article 200