Quantifying the After Effects of Theta Burst
Stimulation with EEG
Grace Pulsford
Student Number: 11120665 Start date: 1/02/2016 Finish date:18/09/2016
Number of EC: 26 Supervisor: Ilja Sligte Second Assessor: Martijn Wokke
Research Institute: University of Amsterdam, Department of Psychology MSc in Brain and Cognitive Sciences: Track Cognitive Neuroscience
Abstract
Applying repetitive transcranial magnetic stimulation (rTMS) to the human cortex at the theta range has after effects lasting up to an hour. These effects, however, are highly variable between individuals. In order to add to the current understanding as to why this variability occurs, this study aimed to develop a technique to quantify the aftereffects of theta burst stimulation (TBS). Currently, phosphene threshold (PT) is used as an indicator of visual cortex excitability. However, activity levels beyond the visual cortex are missed with this measure. Alternatively, EEG measures whole brain activity and is said to reflect the current state of the stimulated area. Therefore, the present study measured PT before and after TBS to index whether subjects responded in an inhibitory or excitatory way to the stimulation. EEG was also measured, and ERP amplitudes were correlated with PT. A negative correlation was expected, as when ERP amplitude increases indicating excitation, PT decreases, and vice versa for inhibition. The analysis showed a non-significant negative correlation between ERP amplitude and PT for the P1 and N1 components. Conversely, the C1 was non-significantly positively correlated with PT. Taken together, the results indicate with more statistical power, ERP amplitude may be used to quantify the aftereffects of TBS. However, it may not be possible to influence the C1 with TMS, reflected in the positive correlation between this component and PT.
Introduction
When studying the brain, activity can be mapped across various spatial and temporal scales and correlated with behaviour. Common examples include EEG and fMRI, both offering unique insights into the operational organization of the brain. However, correlating activity this way provides no evidence of direct causal relations. An alternative approach is to directly manipulate brain functioning and observe the impact this has on behaviour. One such technique which enables this is transcranial magnetic stimulation (TMS). TMS is a non-invasive brain stimulation technique used to induce electrical fields in the underlying cortex. These fields can cause cell membranes of pyramidal neurons located in the superficial cortical layers to depolarize and lead to the firing of action potentials (Krieg & Mogul, 2013). TMS can be applied as single pulses, online repetitive pulses (during a task) or offline repetitive pulses (before a task), all at varying frequencies. Depending on these stimulation parameters, the cortex is either transiently excited or inhibited. For instance, single pulse TMS has mostly excitatory effects. Low frequency (1hz) repetitive pulse TMS (rTMS) often has an inhibitory effect, while higher frequency rTMS (10hz) can increase excitability (Parkin, Ekhtiari, & Walsh, 2015). As a result, the involvement of multiple brain regions in various cognitive processes and behaviours have been studied with TMS. These include the motor system (Volz, Hamada, Rothwell, Grefkes, 2014), vision (Allen, Sumner, Chambers, 2014), and language (Pobric, Mashal, Faust, & Lavidor, 2008), making it a promising research tool. Furthermore, encouraging results have been found in the treatment of various psychiatric disorders with TMS such as depression (Gershon, Dannon & Grunhaus, 2003), and aiding rehabilitation following neurological damage (Harrington,Chan, Turkeltaub, Dromerick, & Harris-Love, 2015).
The effects of TMS, however, are short lived (approx. 20 mins). This has encouraged the field of neuroscience to look for alternative TMS protocols which have longer lasting effects (Fox, Buckner, White, Greicius, & Pascual-Leone, 2012). One such protocol is theta burst stimulation (TBS) with after-effects lasting up to 1 hour. This is induced by giving three bursts of 50hz stimulation, repeated every 200ms at a 5hz rhythm. TBS presented continuously (cTBS; 40 second train of uninterrupted TBS, 600 pulses) generally has an inhibitory effect, while TBS presented intermittently (iTBS; 2 second train of TBS repeated every 10 seconds for a total of 190 seconds, 600 pulses) tends to have an excitatory effect. These long-lasting after effects have been shown to partly depend upon on pre- and postsynaptic N-Methyl-D-Aspartate (NMDA) receptors. NMDAs facilitate calcium influx, triggering a series of reactions which lead to long-term changes in synaptic strength. Thus, TBS is said to influence neuroplasticity, inducing long-term potentiation (LTP) or depression (LTD) like effects
(Huang, Chen, Rothwell, & Wen, 2007).
However, not all individuals react to TBS in the same way. Intra and inter-individual variability is often found, with some subjects displaying the opposite from the ‘expected' effects (Hinder et al, 2014). A major question is what are the
neurophysiological processes underlying the variability in response to TBS? In order to add to the current understanding, it is essential to accurately quantify the aftereffects of TMS. Currently, there is a significantly higher amount of evidence concerning the effects of TMS to the motor cortex, due to electromyographic responses named motor evoked potentials (MEPs). MEPs are recorded from the muscles contralateral to the site of stimulation, resulting from neural efferent volleys along the corticospinal pathway. The amplitude and latencies of MEPs indicate levels of excitability of this pathway, providing valuable information of the functional state of the motor regions (Bonato, Miniussi, & Rossini, 2006).
Stimulation to most other areas of cortex produces no overt effects, asides from the primary visual cortex (V1). TMS to V1 can induce the perception of light flashes named phosphenes in the absence of visual stimulation. The intensity at which phosphenes are elicited, known as the phosphene threshold (PT), can be used as a reliable index of visual cortex excitability (Franca, Koch, Mochizuki, Huang, & Rothwell, 2006), similar to MEPs and the corticospinal pathways. However, recently, TMS has been used to stimulate areas outside of the motor and visual regions (Sacco, Prior, Poole, & Nurmikko, 2014). Although, due to no direct behavioural (i.e muscle twitch/phosphene), or electrophysiological (MEPs) information being gathered there is limited evidence concerning the effects to such regions. Furthermore, studies show even when motor or visual regions are stimulated, activity of brain areas distant from the site of stimulation can also be influenced (Siebner, Hartwigsen, Kassuba & Rothwell, 2009). Therefore, these corollary discharges of distant neurons connected to those primarily affected by the TMS are also missed.
Therefore, overall, there is a need for a technique to measure facilitation or inhibition of the whole brain, to better understand TMS-induced effects. Previously, the ability to measure the direct effects of TMS has been limited. However, due to recent advances in hardware solutions, amplifier technology, and data analysis techniques, electroencephalogram (EEG) has been shown to record activity both during and after TMS (Rogasch & Fitzgerald, 2013). Due to its high temporal resolution (several milliseconds), synaptic activations induced by TMS are picked up by the EEG. This electrical activity is thought to represent the summed postsynaptic potentials of the underlying pyramidal neurons (Kirschstein & Kohling, 2009), with the amplitude, latency, and scalp topography of such TMS-evoked responses signifying the current state of the stimulated area. Thus, EEG may be used as a quantifiable marker of the cerebral neurophysiological state of the cortex following TMS (Ferreri et al, 2012).
The present study, therefore, combined both EEG and PT to accurately group individuals into those who react to TMS in an excitatory or inhibitory way. Subjects carried out a visual task before and after continuous TBS (cTBS) while EEG was measured as an indicator of cortical excitability. The phosphene threshold was also measured immediately before and after cTBS. If the PT increased after stimulation, this signified inhibition of the cortex as more stimulation was required to reach the
threshold. Conversely, if the threshold decreased following stimulation it suggested excitation, as less stimulation was required to reach the threshold. On the other hand, if the participant responded in an inhibitory way, event related potential (ERP) amplitudes should have decreased. Alternatively, if the stimulation had excitatory effect the ERP amplitude should increase. Thus, it was predicted that PT and ERP amplitude would be negatively correlated, as when one increases, the other decreases (See Figure 1 for example).
Figure 1. Negative correlation expected between the PT and ERP amplitude. An increase of PT
following cTBS suggests inhibition as more stimulation is required to reach the threshold. Conversely, a decrease of ERP amplitude following stimulation indicates inhibition, hence a negative correlation
was predicted.
In addition, a visual task was employed, designed by Kelly, Gomez-Ramirez, and Foxe (2008) to evoke a strong C1. The C1 is the earliest visual component, with an onset latency between 40-70ms and a peak latency between 60-100ms (Di Russo, Martínez, Sereno, Pitzalis, & Hillyard, 2002). Converging evidence suggests it is generated in the striate cortex within the calcarine fissure, which lies in the primary visual cortex (V1). This is a deep, convoluted source, meaning it may not be possible to modulate the C1 with TMS. Indeed, no previous studies have reported this components amplitude altering following stimulation. Thus, a secondary aim of the present study was to determine whether the C1 could be modulated with TMS. The later arising P1 (onset 70-80ms), and N1 (onset 130-150-ms) visual components were also expected to be found. These are said to originate from multiple areas of the extrastriate visual cortex, and represent qualitatively different aspects of visual attention. Namely, inhibitory and facilitatory effects (Slagter et al, 2016). It is much easier for the pulse from the TMS to reach the extrastriate cortex. Therefore, modulation of these components was predicted.
Overall, based on previous findings a large amount of variation in response to the stimulation was expected. If subjects responded in an inhibitory way, the amplitude of
ERPs should decrease, and vice versa if they responded in an excitatory way. Conversely, those who responded inhibitory were expected to have a higher PT, or lower if excitatory. Thus, PT and ERP amplitude were anticipated to be negatively correlated. The ERP components expected to be found were the P1, N1, and C1 due to their involvement in visual attention. The amplitude of the P1 and N1 were predicted to change following stimulation. However, it was unclear whether the C1 would be modulated in the same way by the TMS.
Method and materials Subjects
10 subjects (6 females, mean age 23.3) participated in this study after passing an initial screening. All subjects reported not to have any neurological disorders, a history of seizures, or any other risk factors associated with brain stimulation. Two subjects were excluded from the experiment as they were unable to perceive phosphenes (and replaced with 2 other subjects). All subjects provided written informed consent and participated for financial compensation. The experiment was approved by the local ethics committee of the psychology department of the University of Amsterdam.
Experimental design
Subjects were first screened to ensure they were safe to undergo TBS. During the screening, the subject’s active motor threshold (AMT) was measured to establish the appropriate stimulator output level for cTBS. 4 seconds of cTBS was then applied to the visual cortex so subjects knew what to expect in the experimental sessions. One week after the screening, subjects completed two experimental sessions (cTBS and sham) at least three days apart from each other. In the cTBS condition, subjects first carried out a visual task whilst EEG was measured. Following this, the phosphene threshold was measured followed by receiving 40 seconds of cTBS. Immediately after, the phosphene threshold was measured again. Finally, EEG was measured once more while the visual task was completed. The sham condition was exactly the same apart from receiving sham stimulation rather than cTBS.
Visual task Stimuli
Two stimuli were presented during the task, a standard stimulus consisting of a Gabor patch (Figure 2a), and a target stimulus consisting of the standard Gabor patch with a superimposed ring of reduced luminance (Figure 2b). Both stimuli were presented for 100ms.
Task design
Figure 2 displays the layout of the visual task which was designed to evoke a strong C1. The C1 amplitude is sensitive to the position of the stimulus due to the retinotopic organization of the striate cortex, with its polarity reversing depending on whether the stimulus is presented to the upper or lower visual field. This reversal in polarity is due to the fact the upper visual field is represented on the lower bank of the calcarine fissure, while the lower visual field on the upper bank (Rauss, Schwartz & Pourois, 2011). As a result, a maximal C1 can be identified by taking the amplitude of the upper location from the lower location. Therefore, stimuli were presented to both the upper and lower visual fields, in order to obtain the optimal spatial location for eliciting a C1. Subjects were instructed to fixate on a white central cross on a grey background until they saw the target stimuli which would appear 11% of the time. Gabor stimuli were presented randomly to 1 of 8 locations, with one lying in each visual octant. The locations were numbered as they would on a clock face from 1 – 8. Subjects responded to the target stimuli with a left button press. 10 blocks of 120 trials were run per subject, meaning stimuli were presented 15 times to each visual location. The stimulus onset asynchrony (SOA) was set at 833ms.
Figure 3. Example of visual task. Locations in which Gabor patch stimuli were presented to were numbered as
on a clock face from 1-8. This was designed to identify the location which elicited the strongest C1. Stimuli were presented in a random sequence to 1 of the 8 locations. When subjects saw the target stimulus which appeared
11% of the time, they responded with a left mouse click.
Figure 2. Stimuli for visual task. (a) Standard Gabor stimulus. (b) Target stimulus with
superimposed black ring. Figure taken from Kelly, S. P., Gomez-Ramirez, M., & Foxe, J. J. (2008). Spatial attention modulates initial afferent activity in human primary visual
Electroencephalogram recording, preprocessing and analysis
EEG was recorded using a 64 channel Biosemi ActiveTwo system (Biosemi, Amsterdam, Netherlands) at 512hz. 64 scalp electrodes were measured as well as 4 electrooculographic (EOG) electrodes for vertical and horizontal eye movements. After acquisition, EEG data were high-pass filtered at 1hz and low-pass filtered at 45hz, using the average of two earlobe electrodes as references. Trials were epoched from -0.5 to 0.8 seconds relative to stimulus presentation. All trials were visually inspected, and any trials containing artefacts not related to eye blinks (e.g. muscle tension) were removed. After artefacts were removed, independent component analysis was carried out for each participant. Any component which was clearly related to eye blinks was removed. Finally, the quality of ERPs were for each channel were checked to ensure no noisy channels were still present in the data. All preprocessing steps were carried out using Matlab and EEG lab in Matlab (Mathworks).
Once preprocessed, all data were averaged over all trials to form ERPs. Specifically, the P1, N1, and C1 components were of interest. In order to select the appropriate time latencies and channel locations for each component, ERP waveforms for each participant were derived for each of the 8 locations of all four conditions (pre/post cTBS, pre/post sham). The ERPs were inspected in terms of evolution of topographic maps and morphology (positive or negative amplitude). For the P1 and N1, all data was averaged over all participants, all conditions, and all locations. However, as the C1 topography is shown to be much more sensitive to stimulus location, the data was also averaged over all participants and conditions, but locations were inspected separately. Each pair of diagonally opposite locations were subtracted from each other (1-8, 2-7, 3-6, 4-5), and the pair which elicited the maximal C1 were selected for analysis. Finally, to calculate the difference in amplitude shift from pre to post cTBS/sham stimulation, the amplitude of the pre ERP was subtracted from the post ERP for all three components.
cTBS procedure
In the experimental session, subjects received 40 seconds of cTBS (600 pulses of 50Hz stimulation in a 5Hz rhythm) to their visual cortex (Oz electrode position) whilst wearing the EEG cap. In the sham condition, the coil was tilted away from the visual cortex so that the subject felt and heard the stimulation but the magnetic field did not enter the skull. Stimulation was induced using a Magstim Rapid stimulator connected to a round, hand held coil with a diameter of 90mm. The intensity limit of stimulator output was calculated in the screening session using the active motor threshold (AMT), defined as the minimum amount of stimulation required to elicit a twitch in a contracted muscle in 5 out of 10 trials. To compute this, single pulse TMS was applied to the motor cortex (Cz electrode position). Stimulator output for cTBS was calculated as 80% of the AMT. However, if 80% of the AMT was over 38% of the stimulator output, the output was scaled down to 38%.
Phosphene threshold
First, the stimulation site was determined by applying single pulse TMS to the visual cortex. The position of the coil was moved in steps while subjects focused on a fixation cross displayed in front of them on a computer screen. Stimulation was initially applied
at 20% of the stimulator output then increased in stages of 10% until subject reliably reported a phosphene. This location was then taken as the stimulation site or ‘hotspot’. Following this, the upper and lower boundaries of the threshold were identified, with the lowest being 0 out of 10 trials eliciting a phosphene, and the highest 10 out of 10. Stimulator output was then gradually adjusted in steps of 2% between the two boundaries. At each stage the number of trials out of 10 which produced a phosphene were noted down, resulting in a curve depicting the spread phosphene perception at the different stimulator output levels.
Phosphene threshold analysis
A model was first fitted to the raw data of each subject for both conditions (cTBS, sham) and both blocks (pre, post). Figure 4 shows an example of the raw phosphene curve and the model fitted to this data for one subject.
Figure 4. Example of phosphene curve of subject 01 pre cTBS before and after model fitted to data. The X axis displays the stimulator output level of the TMS machine, the Y axis displays the number of
phosphenes reported out of 10. The number of phosphenes reported increases as the stimulator output increases, until 10 phosphene are perceived on 10 trials.
To test the goodness of fit of the model, the root mean square error was calculated using the raw values and estimated values of the curve. When a good fit was confirmed, the PT was calculated, defined as the stimulator output which induced a phosphene on 5 out of 10 trials. Once the PT was established for each block and each session, the threshold post stimulation was subtracted from the threshold pre stimulation. This was done for both cTBS and sham conditions. The figures were then divided by the pre-condition, and multiplied by 100. This resulted in a percentage shift of PT from pre to post cTBS and pre to post sham stimulation, excluding individual differences between stimulator output levels. For instance, the difference between 42 to 46 and 52 to 56 is the same. However, without standardising between subjects, the shift from 52 to 56 would be calculated as higher. Finally, the variance in phosphene threshold shift between subjects was calculated for both the cTBS and sham conditions.
Correlational analysis
The amplitude of the P1, N1, and C1 components were individually correlated with the shift in PT for the cTBS and sham conditions. For both conditions the data did not meet the assumptions for a Pearson’s correlation. Therefore, a Spearman’s rank-order correlation was carried out.
Results
Phosphene Threshold
To index whether subjects had responded to the stimulation in an inhibitory or excitatory way, PTs were individually compared before and after stimulation. If the PT increased, this indicated they responded inhibitory, if it decreased this suggested an excitatory reaction. Figure 5 displays the raw PT data before and after cTBS and sham stimulation for the two separate groups (inhibitory/excitatory).
Figure 5. Comparison between difference in PT pre to post cTBS and pre to post sham stimulation (a) Slightly larger change in PT in cTBS condition compared to sham for inhibitory group. (b) Similar difference of PT between
Phosphene threshold shift
A model was fitted to the raw data to calculate the shift in PT as a percentage. Figure 6 shows and example of the PT shift pre and post cTBS and sham stimulation with the model fitted for one subject. The model fit well, as all error rates were closed to zero (µ = 0.09, SD = 0.04).
This percentage shift was calculated for all subjects. Table 1 displays the percentage shift for each participant. A larger percentage indicates a larger shift from pre to post cTBS/sham stimulation.
Table1. The phosphene threshold shift for the cTBS and sham conditions for each subject. For both conditions, phosphene threshold post stimulation was taken from pre stimulation then divided by pre
stimulation. This was then multiplied by 100 to create a percentage. subject PT shift cTBS (%) PT shift SHAM (% 01 2.42 3.86 02 1.16 -0.93 03 2.59 2.73 04 -8.09 4.44 05 0.18 -5.00 06 -7.25 6.08 07 1.71 -4.93 08 -4.54 -0.38 09 3.18 4.28 10 -2.59 -10.24
Figure 6. Example of the shift in PT before compared to after stimulation for S01. (a) Shift in PT following cTBS. (b) Shift in PT following sham stimulation.
Phosphene threshold variance
The variance within the cTBS session was 6.53, while the variance within the sham condition was 7.64, indicating overall similar variance in both conditions.
ERP
First, the appropriate channel locations and latencies were selected for analysis of the different components. For the P1 and N1, data were averaged over all subjects, all conditions and all eight locations. Figure 7 displays the average ERP waveform and accompanying topographical maps. The channels selected for the best P1 were PO7 and PO8 with a latency of 110 to 160ms (figure 7b shows peak at 130). For the N1 component, channels O2, PO8, and PO4 were selected between 160 to 210ms (figure 7c shows peak at 190).
Figure 7. Grand average ERP amplitude and topographical maps of electrode PO8 following presentation of visual stimulus. (a) ERP displaying the P1 and N1 components. (b) Topographical map exhibiting activation levels at 130ms when the P1 is at its peak. (c) Topographical map at 190ms
when the N1 is at its peak.
Amp
lit
u
d
For a maximal C1, location 4 was subtracted from location 5. For the analysis, channel POz was selected between 70 to 100 ms. Figure 8 displays average ERP and topographical map with peak at 90ms.
ERP Shift
Once the appropriate channel locations and latencies were identified, ERP amplitudes before and after stimulation were able to be compared. An increase in amplitude would suggest excitation and a decrease indicated inhibition. Figure 9 shows an example of the change in amplitude of the P1 from pre to post cTBS compared to pre to post sham for subjects who responded in an excitatory way. Here it can be seen there is an increase in amplitude of the P1 following cTBS, while there is little change in amplitude following sham stimulation.
Figure 8. Grand average ERP amplitude and topographical map of electrode POz. (a) C1 ERP with a peak latency of 90ms (b) Topographical map exhibiting activation levels at 90ms when the C1 is at its
peak. Amp lit u d e
Shift in P1 amplitude following cTBS and sham stimulation for subjects who responded in an excitatory way
The shift in amplitude of each component from pre to post cTBS and sham stimulation was calculated for every participant individually. Table 2 displays this shift for the P1, N1, and C1 for both conditions (cTBS, sham), for all 10 subjects.
Figure 9. Average P1 amplitude pre to post stimulation in the cTBS condition compared to the sham condition. (a) P1 increases in amplitude post cTBS indicating excitation. (b) Little difference in amplitude of the P1 following sham
Shift in amplitude of ERPs for cTBS and sham conditions for the P1, N1 and C1
Table 2. Shift in ERP amplitude from pre to post stimulation for P1, N1, C1 for both conditions. Pre stimulation amplitudes were subtracted from post stimulation amplitudes to indicate the amount of
shift. Positive numbers imply increase in amplitude and therefore excitation, negative suggest decrease and therefore inhibition.
Correlation analysis
Finally, the PT shift and ERP amplitude shifts of the P1, N1, and C1 were correlated for both conditions. In the cTBS condition, there was a slight, non-significant negative correlation between PT shift and P1 amplitude (rs (8) = -.188, p = .603). There was also a non-significant negative correlation between PT shift and N1 amplitude (rs (8) = -.115, p =.751). The correlation between PT shift and C1 amplitude was positive and non-significant (rs (8) = .188, p = .603. In the sham condition, there was a non-significant positive correlation between PT shift and P1 amplitude (rs (8) = .248, p = .489) as well as PT shift and N1 amplitude (rs (8) = .212, P = .556). And PT shift and C1 amplitude (rs (8) = .442, p = .200).
Discussion
TMS has the capacity to transiently modulate cortical excitability, with some individuals responding in an excitatory way, and others in an inhibitory way. However, the neurophysiological processes underlying such variability are poorly understood. The present study aimed to develop a technique to quantify the aftereffects of TMS using EEG to add towards the understanding of these processes. A visual task was employed, designed to evoke the earliest visual component - the C1. Thus, a secondary aim was to determine whether the C1 can be influenced with TMS. ERP amplitudes were correlated with the shift in phosphene threshold to establish whether the stimulation had excitatory or inhibitory effect. As expected, a large amount of variation was found in response to the stimulation. The results, however, were not significant. Nevertheless, as predicted the correlation between the P1, N1 and PT were negative in the cTBS condition and not in the sham. Conversely, the C1 correlation was positive in both conditions. The implications of these results will be further discussed.
subject P1 cTBS P1 SHAM N1 cTBS N1SHAM C1 cTBS C1 sham
1 0.67 -0.51 1.60 0.48 -0.10 3.12 2 -0.28 -0.51 0.13 0.29 0.14 1.48 3 -0.34 0.86 0.34 0.78 5.02 -0.35 4 0.99 0.14 0.04 0.70 -1.28 -0.77 5 0.50 -0.60 0.60 0.11 4.54 -2.09 6 -0.93 0.36 -0.16 -1.32 -1.47 -2.06 7 -0.12 0.13 0.47 0.42 1.33 -0.32 8 0.54 -0.06 0.09 0.69 2.19 -0.35 9 0.13 0.18 -0.72 0.33 -1.88 1.01 10 0.01 0.17 0.36 -0.05 -2.01 2.10
Large variation in response to continuous theta burst stimulation: The role of interneuron networks
As predicted, there was a large spread of variation in response to the cTBS. Hamada and colleagues showed around 50% of the variation in response to TBS to the motor cortex can be explained by the efficiency of early or late indirect-wave (I-wave) recruitment (Hamada, Murase, Hasan, Balaratnam, & Rothwell, 2012). I-waves are descending volleys from indirect synaptic activation of corticospinal cells following electrical stimulation of the motor cortex (Ziemann & Rothwell, 2000). The short latency of early I-waves indicates they originate from monosynaptic excitatory connections to pyramidal cells, located on proximal parts of the central nervous system (CNS), while late I-waves are produced by oligosynaptic circuits, and target more distal dendrites (Di Lazzaro et al, 2012). Those who suppress early I-waves and more readily recruit late I-waves tend to respond in the expected inhibitory way. Therefore, for these individuals cTBS may mostly suppress the excitatory synapses between early I-wave inputs and the cortico-spinal neurons, producing long-term depression (LTD). On the other hand, those who respond in an excitatory way may do so by facilitating excitatory synapses, leading to long-term potentiation (LTP). The suppression of specific I-waves suggests cTBS does not change the overall excitability of the corticospinal neurons, but rather particular intra-cortical circuits (Di Lazzero et al, 2005). Furthermore, the inputs which are activated (oligosynaptic and monosynaptic) seem to behave differently, resulting in the opposing aftereffects of cTBS (i.e inhibitory or excitatory). Unlike Hamada et al. the present study employed cTBS to the visual cortex, and measured response variability based on the shift in PT before and after stimulation. Nevertheless, Franca et al. (2006) postulated PT and MEP amplitude are comparable measures of different cortical regions. For instance, cTBS generally modulates changes in cortical excitability of visual and motor areas in the same direction (Deblieck, Thompson, Iacoboni & Wu, 2008). If cTBS decreases or increases synaptic connections of circuits involved in MEP generation, then similar mechanisms may account for the visual cortex based on intracortical inhibition and facilitation. Overall, there seems to be an intrinsic difference between individuals in the population of neurons recruited following cTBS, which influences the net direction of excitability. Therefore, the variability in response to the stimulation may be partly a result of recruitment of different interneuron networks between participants.
PT & ERP amplitude pre and post stimulation
The response variability between individuals was inspected and participants were split into two groups (inhibitory or excitatory) based on shift in PT and shift in ERP amplitude pre to post stimulation. Figure 5 displays the average PT pre to post cTBS and sham stimulation for the two groups. Overall, there seems to be a slightly more prominent shift following cTBS compared to sham in the inhibitory group. This suggests the cTBS impacted excitability more than the sham stimulation. There was a less prominent shift in the excitatory group. Nevertheless, after calculating the shift as a percentage there were similar levels of variability in the cTBS and sham conditions. Thus, this figure should be used as a general guide to the level of change following stimulation, rather than a definite indication of larger shift in the cTBS condition compared to sham. The P1 was used as an example to show the impact of cTBS on ERP amplitude. Figure 8
displays a slight increase in amplitude pre to post stimulation. Conversely, minimal change can be seen in the sham condition.
Overall, these figures suggest the cTBS impacted the excitability of the cortex. It also indicates that it is possible to identify the shift in excitability using both PT and ERP amplitude. However, while this is a positive preliminary result, it needs to be stressed this was simply inspected visually rather than statistically due to the limited number of participants.
Negative correlations: P1, N1
As expected, the P1, N1 and C1 components were found in response to the visual task. The amplitude of each component before and after stimulation was correlated with the shift in PT before and after stimulation. An increase in ERP amplitude reflects excitation, while an increase in PT indicates inhibition. Therefore, if the two measures correspond a negative correlation would be expected. While not significant, both the P1 and the N1 were negatively correlated with PT. Due to the limited number of participants a significant result was unlikely. Generally, for a correlational study, 50 or more participants are required to acquire a reliable result (Van Voorhis, & Morgan, 2007). However, this is a preliminary hint that by combining PT and ERP amplitude, it is possible to group individuals as excitatory or inhibitory.
Positive correlation: C1
On the other hand, a non-significant positive correlation was observed between the C1 and PT. This suggests that even with more statistical power, a negative correlation was unlikely to be found. Therefore, it seems the C1 was not influenced like the P1 and N1 were, indicating it may not be possible to modulate this component with TMS. This result is not so surprising, as this has not yet been done before. As previously mentioned, it may be that the calcarine fissure in which the C1 originates is too deep of a structure to reach with TMS. Furthermore, this fissure takes a convoluted path along the medial occipital cortical surface, and has been found to show extreme anatomical variability between individuals (Proverbio et al. 2007). This may also be another reason as to why it is difficult to influence the C1. However, the limited number of participants must once again be stressed. Increasing the statistical power combined with looking into the effects of different stimulation parameters may yield different results. Thus, future researchers should not be discouraged into further investigating whether it is indeed possible to modulate the C1 with TMS.
Limitations and future research
During the PT measurement, subjects were required to verbally indicate whether or not they perceived a phosphene following each TMS pulse. Throughout the experiment, there was high subjectivity and variability of phosphene detection between subjects. Thus, the PT measurement may have not been entirely accurate. Consequently, in future, subjects should complete a practice session to ensure they are definitely capable of discriminating between what is and is not a phosphene. Furthermore, the present study only investigated the impact of cTBS on cortex excitability. In future, it would be interesting to examine the influence of intermittent
TBS in comparison to continuous. The two stimulation parameters are shown to have opposing effects on brain activity (Huang, Edwards, Rounis, Bhatia, & Rothwell, 2005). Thus, quantifying these effects will add to the understanding of how the two parameters differentially influence neuronal networks.
Another factor to consider for future is the method to calculate the stimulator output for cTBS. The present study took 80% of the active motor threshold (AMT) to stimulate the visual cortex. However, studies have either shown motor and visual thresholds not to be correlated with each other (Stewart, Walsh & Rothwell, 2001), or the threshold for eliciting a visual response to be significantly higher than eliciting a motor response (Gerwig, Kastrup, Meyer & Niehaus, 2003). This disparity between the two thresholds implies the AMT may not be the most appropriate guide to visual cortex sensitivity. A more suitable way may be to use a percentage of the PT, rather than AMT.
A final note which has already been mentioned is the small sample size in the current study. Although this could not be avoided due to time constraints, future research should increase the number of participants to determine whether a more robust effect of TBS can be observed.
Implications and conclusion
Overall, the present study provides hints which indicate EEG is a valid technique to quantify the aftereffects of TMS. Once established, this could be used to gather more information of the neurophysiological mechanisms underlying response variability to TBS. Unlike MEPs or PT, EEG measures whole brain activity, thus data from regions outside the motor and visual cortex can be collected. Furthermore, the temporal evolution of the distribution of TMS-induced EEG components can be examined, providing more evidence on the spread of activation from the stimulated site to contra and ipsilateral cortical regions. This may help disentangle focal from distant TMS effects upon different structurers of the central nervous system. This will not only add towards the understanding of the mechanisms behind TMS, but can also provide insights into how neural areas interact during cognition enabling higher cognitive functions to be mapped (Bonato, Miniussi, & Rossini, 2006). Finally, in the future quantifying the effects of TMS may also be useful in clinical settings. For instance, knowing prior to stimulation the direction of which a patient will respond will be useful in determining the stimulation parameters for that individual.
References
Allen, C. P., Sumner, P., & Chambers, C. D. (2014). The timing and neuroanatomy of conscious vision as revealed by TMS-induced blindsight. Journal of cognitive
neuroscience, 26(7), 1507-1518.
Bonato, C., Miniussi, C., & Rossini, P. M. (2006). Transcranial magnetic stimulation and cortical evoked potentials: a TMS/EEG co-registration study.Clinical
neurophysiology, 117(8), 1699-1707.
Di Lazzaro, V., Pilato, F., Saturno, E., Oliviero, A., Dileone, M., Mazzone, P., ... & Rothwell, J. C. (2005). Theta‐burst repetitive transcranial magnetic stimulation suppresses specific excitatory circuits in the human motor cortex. The Journal of
physiology, 565(3), 945-950.
Di Russo, F., Martínez, A., Sereno, M. I., Pitzalis, S., & Hillyard, S. A. (2002).
Cortical sources of the early components of the visual evoked potential.Human brain
mapping, 15(2), 95-111.
Di Russo, F., Martínez, A., & Hillyard, S. A. (2003). Source analysis of event-related cortical activity during visuo-spatial attention. Cerebral cortex,13(5), 486-499.
Deblieck, C., Thompson, B., Iacoboni, M., & Wu, A. D. (2008). Correlation between motor and phosphene thresholds: a transcranial magnetic stimulation study. Human
brain mapping, 29(6), 662-670.
Ferreri, F., Ponzo, D., Hukkanen, T., Mervaala, E., Könönen, M., Pasqualetti, P., ... & Määttä, S. (2012). Human brain cortical correlates of short-latency afferent inhibition: a combined EEG–TMS study. Journal of neurophysiology, 108(1), 314-323.
Fox, M. D., Buckner, R. L., White, M. P., Greicius, M. D., & Pascual-Leone, A. (2012). Efficacy of transcranial magnetic stimulation targets for depression is related to intrinsic functional connectivity with the subgenual cingulate. Biological
psychiatry, 72(7), 595-603.
Franca, M., Koch, G., Mochizuki, H., Huang, Y. Z., & Rothwell, J. C. (2006). Effects of theta burst stimulation protocols on phosphene threshold. Clinical
neurophysiology, 117(8), 1808-1813.
Gershon, A. A., Dannon, P. N., & Grunhaus, L. (2003). Transcranial magnetic stimulation in the treatment of depression. American Journal of Psychiatry.
Gerwig, M., Kastrup, O., Meyer, B. U., & Niehaus, L. (2003). Evaluation of cortical excitability by motor and phosphene thresholds in transcranial magnetic stimulation. Journal of the neurological sciences, 215(1), 75-78.
Hamada, M., Murase, N., Hasan, A., Balaratnam, M., & Rothwell, J. C. (2012). The role of interneuron networks in driving human motor cortical plasticity. Cerebral cortex, bhs147.
Harrington, R. M., Chan, E., Turkeltaub, P. E., Dromerick, A. W., & Harris-Love, M. L. (2015). Simple Partial Status Epilepticus One-day Post Single-pulse TMS to the Affected Hemisphere in a Participant With Chronic Stroke.Brain stimulation, 8(3), 682-683.
Hinder, M. R., Goss, E. L., Fujiyama, H., Canty, A. J., Garry, M. I., Rodger, J., & Summers, J. J. (2014). Inter-and intra-individual variability following intermittent theta burst stimulation: implications for rehabilitation and recovery. Brain stimulation, 7(3), 365-371.
Huang, Y. Z., Edwards, M. J., Rounis, E., Bhatia, K. P., & Rothwell, J. C. (2005). Theta burst stimulation of the human motor cortex. Neuron, 45(2), 201-206.
Kelly, S. P., Gomez-Ramirez, M., & Foxe, J. J. (2008). Spatial attention modulates initial afferent activity in human primary visual cortex. Cerebral cortex, 18(11), 2629-2636.
Krieg, T., & Mogul, D. J. (2013). Transcranial magnetic stimulation. In Neural
Engineering (pp. 405-453). Springer US.
Kirschstein, T., & Köhling, R. (2009). What is the Source of the EEG?. Clinical EEG
and neuroscience, 40(3), 146-149.
Parkin, B. L., Ekhtiari, H., & Walsh, V. F. (2015). Non-invasive human brain stimulation in cognitive neuroscience: a primer. Neuron, 87(5), 932-945.
Pobric, G., Mashal, N., Faust, M., & Lavidor, M. (2008). The role of the right cerebral hemisphere in processing novel metaphoric expressions: a transcranial magnetic stimulation study. Journal of Cognitive Neuroscience,20(1), 170-181.
Proverbio, A. M., Del Zotto, M., & Zani, A. (2007). Inter-individual differences in the polarity of early visual responses and attention effects. Neuroscience letters, 419(2), 131-136.
Rauss, K., Schwartz, S., & Pourtois, G. (2011). Top-down effects on early visual processing in humans: A predictive coding framework. Neuroscience & Biobehavioral
Reviews, 35(5), 1237-1253.
Rogasch, N. C., & Fitzgerald, P. B. (2013). Assessing cortical network properties using TMS–EEG. Human brain mapping, 34(7), 1652-1669.
Sacco, P., Prior, M., Poole, H., & Nurmikko, T. (2014). Repetitive transcranial magnetic stimulation over primary motor vs non-motor cortical targets; effects on experimental hyperalgesia in healthy subjects. BMC neurology, 14(1), 1.
Siebner, H. R., Hartwigsen, G., Kassuba, T., & Rothwell, J. C. (2009). How does transcranial magnetic stimulation modify neuronal activity in the brain? Implications for studies of cognition. Cortex, 45(9), 1035-1042.
Slagter, H. A., Prinssen, S., Reteig, L. C., & Mazaheri, A. (2016). Facilitation and inhibition in attention: Functional dissociation of pre-stimulus alpha activity, P1, and N1 components. NeuroImage, 125, 25-35.
Stewart, L. M., Walsh, V., & Rothwell, J. C. (2001). Motor and phosphene thresholds: a transcranial magnetic stimulation correlation study.Neuropsychologia, 39(4), 415-419.
Van Voorhis, C. R. W., & Morgan, B. L. (2007). Understanding power and rules of thumb for determining sample sizes. Tutorials in Quantitative Methods for
Psychology, 3(2), 43-50.
Volz, L. J., Hamada, M., Rothwell, J. C., & Grefkes, C. (2014). What makes the muscle twitch: motor system connectivity and TMS-induced activity.Cerebral cortex, bhu032. Ziemann, U., & Rothwell, J. C. (2000). I-waves in motor cortex. Journal of Clinical
