Baseline clinical assessment.

At the outset of the study, we will perform a series of baseline assessments. All participants who meet the inclusion/exclusion criteria will be administered shortened second edition of the Wechsler Abbreviate Scale for intelligence (WASI-II), Edinberg Handedness Inventory (EHI), Adult Repetitive Behavior Questionnaire-2 (ARBQ-2), the Social Responsiveness Scale-2 (SRS-2), and SPQ [24,34–37]. To screen for the presence of general psychopathology in control adults, neurotypical participants will also complete the Personality Assessment Inventory (PAI) [38]. Furthermore, other than participating in ADOS-2, all autistic participants will complete the Ritvo Autism Diagnostic Scale-Revised (RAADS-R) [39].

Phase I. Assessment of plasticity Transcranial Magnetic Stimulation. TMS procedures will be performed using a Magstim 200 stimulator (Magstim Co., Whitland, UK) equipped with a 70-mm figure-of-eight coil. To enable neuronavigated TMS, structural T1-weighted magnetic resonance (MR) images will be obtained prior to the TMS session. These images will be acquired with a 3T MRI scanner (Discovery MR750, GE Healthcare, Chicago, IL, USA) for each participant at the Research Imaging Centre, CAMH. Subsequently, these MR images will be imported into BrainSight software (Rogue, Research Inc, Canada) for individual MRI-guided navigated TMS (nTMS).

The TMS stimulation session will begin with the identification of the cortical representation of the abductor pollicis brevis (APB) muscle, a process known as ‘hotspotting’. To locate the APB hotspot, we will position the intersection of the coil tangentially to the participants’ skull, optimizing it around the C3 electrode (the approximate location of the primary motor cortex for the hand). The coil’s handle will be oriented backward, forming a 45° angle with the midline. Once the hotspot will be located, we will determine the resting motor threshold (RMT). The RMT will be identified as the lowest stimulus intensity required to induce MEPs with a peak-to-peak amplitude of at least 50 microvolts (μV) in the relaxed right APB muscle, in at least half of 10 consecutive trials. Subsequently, the stimulation intensity is set to 1 mV, i.e. the intensity of TMS required to produce MEPs with a mean peak-to-peak amplitude of 1 mV across 20 TMS trials. 1 mV MEP is typically obtained with a stimulator intensity of 120% of the RMT, and stimulation intensities below this level are generally inadequate for reaching depths greater than 2 cm beneath the scalp. Therefore, we initially set the stimulation intensity to 120% of RMT and then adjust it to achieve the target 1 mV amplitude (SI1_mV). TMS stimulation will be then applied to the left S1 site (MNI coordinates: X = -48 mm, Y = -21 mm, Z = 50 mm), with the intensity set at SI1_mV throughout the session. The orientation of the coil is adjusted to minimize artifacts, specifically aiming to keep observed decay artifacts within 10 ms following TMS pulses to ensure that early evoked responses remain unaffected. This adjustment may cause a shift of the stimulation site within a range of ±1 mm. In total, 100 single pulses will be delivered, with an inter-stimulus interval (ISI) of 5 s. This paradigm will be administered both before and after the PAS intervention at different time points (0, 15, 30, and 60 minutes) (Fig 2A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Schematic figure of plasticity assessment and intervention phases of the current study.

(A). The plasticity assessment procedure consists of delivering 100 single TMS pulses over the left S1 using SI1 mV which will be applied before and after (0, 15, 30, and 60 min) PAS intervention. Throughout the PAS intervention, electrical stimulation of the right median nerve will precede TMS stimulation of the left S1 at an ISI of 20 ms. The plasticity will be then assessed using the ratio of the maximum CEA following PAS to the baseline CEA. CEA is quantified as the area under the curve of rectified TMS-EEG response within a time window of 25–275 ms from TMS stimulus onset. (B). Autistic participants will be further randomized to receive either a single session sham or active 20 Hz rTMS. The rTMS intervention will be delivered bilaterally over the S1 consisting of 120 trains of 50 pulses with an ITI of 30s (in total, 6000 pulses). The sequence of delivery will be randomized for each participant, that is left S1 followed by right S1 and vice versa, and the pulses will be applied at an intensity of 90% RMT. Following rTMS, on the same day, TMS-EEG plasticity assessment paradigm will be applied.

https://doi.org/10.1371/journal.pone.0305013.g002

Participants will remain seated in a comfortable chair for the entire session, and to mask the auditory stimulation from TMS, white noise will be played through earplugs. The volume of the white noise will be carefully calibrated to effectively mask TMS-related sounds while staying within the participant’s comfort limits. The volume, will be gradually increased, monitoring the participant’s comfort level to find the maximum volume that is both comfortable and effective in masking the TMS clicks. Additionally, a layer of foam will be placed on the coil. This combination of foam and adapted white noise has been shown to achieve the greatest reduction in auditory evoked responses associated with the clicking sound of TMS [40].

Electroencephalographic Recording and Preprocessing. EEG recordings will be continuously acquired concurrently with TMS using a TMS-compatible 64-channel amplifier (Compumedic Inc) while the electrode impedances will be kept below 5 kΩ. In our EEG recording setup, the reference and ground electrodes are integrated into the EEG cap, with the reference electrode positioned posterior to the CZ electrode and the ground electrode located posterior to the FPZ electrode. An electrooculogram (EOG) electrode is placed under the left eye to record the vertical eye movements. The sampling rate of 20 kHz and a low-pass filter of 100 Hz will be selected, to minimize the artifacts introduced by TMS and avoid amplifier saturation [41].

Thereafter, the acquired EEG data will be preprocessed offline using customized scripts in MATLAB (MathWorks Inc., Natick, MA, USA) and EEGLAB open source MATLAB toolbox, in accordance with a defined cleaning pipeline. According to this pipeline, the EEG data will be (1) trimmed from -2 ms to 10 ms after TMS pulse; (2) decay artifact will be removed with ICA and then data will be down-sampled to 1 kHz; (3) segmented into epochs (-1000 ms to 1000 ms with respect to the TMS pulses); (4) baseline-corrected using the artifact-free pre-TMS interval (from -500 ms to -5 ms); (5) another round of ICA will be applied to remove residual artifacts such as EOG and muscle artifacts. In the last step, data will be filtered using a zero-phase shift band-pass filter with a frequency range of 1 to 100 Hz and 60 re-referenced using an average reference for further analysis.

Paired Associative Stimulation Intervention. To induce plasticity in the S1, a PAS paradigm will be adopted. During this PAS paradigm, one hundred eighty electrical stimuli will be delivered peripherally at 0.1 Hz over the right median nerve. The intensity of these stimuli will be set to three times the participant’s perceptual threshold, which is the minimum intensity at which they can perceive the stimulation. To generate an LTP-like plasticity effect, the peripheral stimulation of the median nerve will be preceded the TMS stimulation of the contralateral S1 by 20 ms (Fig 2A). The 20 ms interval will be employed to ensure that the timing of the peripheral nerve stimulation and TMS is aligned with the arrival of the peripheral sensory input in the S1 [42]. Participants will be instructed to maintain their attention on their stimulated wrist and count the number of pulses they will receive throughout the protocol. This instruction is based on the recognition that attention can modulate the effects of PAS, as suggested by the study conducted by Stefan et al [43].

Phase II. Repetitive transcranial magnetic stimulation procedure in autistic participants Stimulation will be conducted with an X100 stimulator with a double-sided B65 A/P type fluid-cooled coil (MagVenture, Farum, Denmark Inc.) which is a combined active/sham coil with identical sides. Single-session 20 Hz rTMS will be delivered bilaterally over the S1 at 90% of RMT, with 120 trains of 50 pulses and an inter-train interval (ITI) of 30 s (Fig 2B) (15). In total, 6000 pulses will be administered to each autistic participant. The sequence of delivery will be randomized for each participant, i.e., left S1 followed by the right S1 (MNI coordinates: X = 48 mm, Y = -21 mm, Z = 50 mm) or vice versa. Sham stimulation will be delivered using the sham side of the coil, which does not generate a magnetic field but mimics the sensation of magnetic stimulation and produces the same clicking sounds. We opted not to use electrical scalp stimulation for the sham condition due to the anatomical location of S1, where effective attachment of electrodes cannot be achieved, preventing the production of discernible sensations.

Outcomes Primary Outcome. The primary outcome measure of this study is the PAS-induced LTP-like plasticity in the S1. PAS-induced LTP-like plasticity will be assessed using CEA. CEA is quantified as the area under the curve of rectified TMS-EEG response within a time window of 25–275 ms from TMS stimulus onset [44]. This chosen time window is designed to exclude any early artifacts and account for the subsiding effects of TMS. CEA will be averaged across 100 epochs from the electrodes of interest corresponding to the S1 region. The mean post-PAS CEA will be then divided by the pre-PAS CEA, reflecting the potentiation of CEA at each of post-PAS time points (0, 15, 30, and 60 minutes). Given that the timing of the post-PAS maximum potentiation could differ among participants, we will use the maximum CEA ratio for each participant for the statistical analysis [21].

To asses TR, the total score on the “touch” category of SPQ questionnaire will be used.

A Priori power calculation A sample size of N = 64 participants, consisting of 32 autistic participants and 32 neurotypical controls, has been determined as adequate for hypothesis testing. This sample size provides 90% power. This calculation is based on the preliminary data, the observed LTP-like plasticity difference of 20.6 points [15], as evidenced in our PAS-EEG pilot study involving both autistic adults and control subjects. Moreover, the same sample size is deemed suitable for evaluating hypotheses 2, offering an 80% statistical power to detect a moderate effect size (Cohen’s f2 = 0.15). These power calculations were executed using G*Power [45].

Drawing upon our previously published findings [15], we envision a "substantial" impact of active rTMS in a group of 16 autistic participants when compared to a sham group also consisting of 16 autistic participants. It is important to note that, given the relatively small sample size, the application of a single session, and the exploratory nature of this outcome, our approach to presenting results and their interpretation will extend beyond statistical significance values. Instead, we will place significant emphasis on estimated effects and their precision, including the calculation of 95% confidence intervals. This comprehensive set of information, combined with data related to tolerability, will play a pivotal role in shaping the design of future clinical trials.

Statistical analysis.

Hypothesis 1: To test this hypothesis, we will employ a linear regression model with CEA ratio (LTP-like plasticity) as dependent variable and study groups (autistic and neurotypical controls) as independent variable. Age, sex, hemisphere of PAS-EEG assessment, IQ and pre-PAS CEA (an index of individual baseline cortical excitability) will enter the model as ‘control’ independent variables. To ensure robustness, a model diagnostic will be conducted through visual inspection of model residuals for lack of normality, heteroscedasticity, linearity and outliers [46] and Cook’s distance [47] will be used to inspect for influential data points. Potential confounding effect of medication use among autistic adults on plasticity will be investigated in a post-hoc sensitivity analysis that will include mediation as an additional control.

Hypothesis 2: We will use a multivariate analysis of variance to test this hypothesis where TR, as dependent variable, is associated with plasticity in the autism group; S1 plasticity, study groups and their interaction will be the main predictors; and sex, age, hemisphere of PAS-EEG assessment and IQ will enter the model as controls. If the interaction between S1 plasticity and study group proves to be significant at confidence level 0.05, the model estimated slopes of S1 within each group will be investigate using graphs.

Hypotheses 3 & 4: In order to study the effect of active vs. sham rTMS on LTP-like plasticity, we will use an analysis of covariance with CEA ratio (LTP-like plasticity) as dependent variable, groups (active and sham rTMS) as fixed factor, and pre-rTMS CEA ratio as covariate. We will count discontinuation of rTMS session and adverse events to assess tolerability.