Continuous and intermittent theta burst stimulation to the visual cortex do not alter GABA and glutamate concentrations measured by magnetic resonance spectroscopy

Abstract Background Theta burst stimulation (TBS), a form of repetitive transcranial magnetic stimulation (rTMS), uses repeated high‐frequency bursts to non‐invasively modulate neural processes in the brain. An intermittent TBS (iTBS) protocol is generally considered “excitatory,” while continuous TBS (cTBS) is considered “inhibitory.” However, the majority of work that has led to these effects being associated with the respective protocols has been done in the motor cortex, and it is well established that TMS can have variable effects across the brain. Objectives and method We investigated the effects of iTBS and cTBS to the primary visual cortex (V1) on composite levels of gamma‐aminobutyric acid + co‐edited macromolecules (GABA+) and glutamate + glutamine (Glx) since these are key inhibitory and excitatory neurotransmitters, respectively. Participants received a single session of cTBS, iTBS, or sham TBS to V1. GABA+ and Glx were quantified in vivo at the stimulation site using spectral‐edited proton magnetic resonance spectroscopy (1H‐MRS) at 3T. Baseline pre‐TBS GABA+ and Glx levels were compared to immediate post‐TBS and 1 h post‐TBS levels. Results There were no significant changes in GABA+ or Glx following either of the TBS conditions. Visual cortical excitability, measured using phosphene thresholds, remained unchanged following both cTBS and iTBS conditions. There was no relationship between excitability thresholds and GABA+ or Glx levels. However, TBS did alter the relationship between GABA+ and Glx for up to 1 h following stimulation. Conclusions These findings demonstrate that a single session of TBS to the visual cortex can be used without significant effects on the tonic levels of these key neurotransmitters; and add to our understanding that TBS has differential effects at visual, motor, and frontal cortices.


INTRODUCTION
Non-invasive brain stimulation (NIBS) techniques such as repetitive transcranial magnetic stimulation (rTMS) are an invaluable tool to induce neural changes safely and effectively across brain networks in healthy and patient populations. TMS uses strong focused magnetic field pulses to induce lasting neural changes at the stimulation site and remote regions (Hallet, 2007;. Theta burst stimulation (TBS), a type of high-frequency rTMS, is increasingly being used as an alternative to conventional rTMS protocols since it can be delivered more quickly while achieving similar outcomes as conventional protocols (Blumberger et al., 2018;Suppa et al., 2016). Intermittent TBS (iTBS) typically generates an "excitatory" response, due to its interval pattern, in a similar fashion to longterm potentiation (LTP) of synaptic transmission in the brain. Continuous TBS (cTBS) typically produces an "inhibitory" response, due to its continuous pattern, mimicking a reaction like long-term depression (LTD) (Hess et al., 1996;Huang et al., 2005). However, this dichotomy that iTBS is excitatory and cTBS is inhibitory is inadequate since it has been shown that both high-and low-frequency rTMS can have mixed excitatory and inhibitory effects (Houdayer et al., 2008). Even when the rTMS effect appears specific, doubling the duration of stimulation can reverse the outcome from inhibition to excitation and vice versa (Gamboa et al., 2010(Gamboa et al., , 2011Goldsworthy et al., 2015).
Similar adverse effects are reported between rTMS and TBS protocols (Blumberger et al., 2018). If TBS performs comparably to rTMS, then employing TBS over rTMS would considerably improve capacity, compliance, and cost by enabling stimulation over shorter sessions.
This would have significant implications for both experimental and clinical use.
The majority of work investigating the underlying effects of rTMS protocols have studied motor and frontal cortices (for reviews, see Hoffman & Cavus, 2002;Ridding & Ziemann, 2010;Sandrini et al., 2011;Suppa et al., 2016). Accordingly, the use of rTMS to modulate disorders has focused on these cortical regions, for example, stroke (for reviews, see Dionisio et al., 2018;Smith & Stinear, 2016;Webster et al., 2006), and psychiatric disorders (for reviews, see Ferrarelli & Phillips, 2021;Guo & Wang et al., 2017), respectively. Despite extensive research on motor and frontal cortices, the effects of TMS at the visual cortex are relatively under-investigated and, therefore, poorly understood. Although rTMS has been used frequently in exploratory vision research (e.g., Bona et al., 2014;Chiou & Ralph, 2016;Groen et al., 2021;Julian et al., 2016;Rafique et al., 2015;Solomon-Harris et al., 2016), very few studies have investigated the underlying mechanisms and aftereffects of rTMS, particularly TBS protocols, at the visual cortex. The use of TMS to the occipital cortex, whether clinical or investigative therefore relies on the assumption of underlying neurophysiological effects determined mainly from regions outside of the occipital cortex (Cárdenas-Morales et al., 2010;Hallet, 2007;Hoogendam et al., 2010;Thut & Pascual-Leone, 2010). However, underlying neurophysiological effects of NIBS differ fundamentally across cortical and subcortical regions (Castrillon et al., 2020). As a result, rTMS applications are less widely used in the clinical application of visual-related dis-orders compared to non-visual-related conditions. Clinically, TMS has great potential as a valuable therapeutic tool in several visual and ophthalmological disorders (Mahayana et al., 2017). Previously, we have successfully reduced visual hallucinations that occurred as a consequence of occipital stroke using rTMS to the visual cortex . Others have used rTMS to manage visual hallucinations from Charles Bonnet syndrome (Merabet et al., 2003). TBS to the posterior parietal cortex improves visual-spatial neglect in patients with right-hemispheric stroke (Cazzoli et al., 2012). Both rTMS and TBS protocols to the occipital cortex have significantly improved visual acuity, stereoacuity, and contrast sensitivity measures in amblyopia (Clavagnier et al., 2013;Thompson et al., 2008;Tuna et al., 2020).
Neurotransmitters such as γ-aminobutyric acid (GABA) and glutamate are key actors in inhibitory and excitatory neural processes corresponding to LTD-and LTP-like changes (for a review, see Lüscher & Malenka, 2012) and provide insight into TMS mechanisms. Levels of neurotransmitters and other metabolites can be measured non-invasively in vivo using magnetic resonance spectroscopy (MRS).
Although MRS is not able to distinguish between vesicular and synaptic or intra-and extracellular pools, instead, it quantifies "bulk" tonic levels averaged over a macroscopic region of interest. MRS-measured GABA contains co-detected GABA-like metabolites like homocarnosine and macromolecules with similar spectra. The composite measure is consequently referred to as GABA+. Similarly, MRS-measured glutamate contains the co-detected glutamine signal, and the composite signal is referred to as Glx (Ramadan et al., 2013;Schmidt-Wilcke et al., 2018). Table 1 provides an overview of studies investigating the effects of rTMS on MRS-measured GABA and glutamate in healthy participants. Although there is a lack of data investigating changes in metabolites following rTMS, Table 1 highlights variability in findings across studies owing to differences in stimulation parameters and the cortical region being stimulated. We previously investigated how a single-session of low-frequency rTMS and multiple sessions within a day (termed accelerated/within-session) influence GABA and glutamate at the visual cortex. We found that accelerated sessions significantly reduced GABA+ at the stimulation site for up to 24 h, whereas a single rTMS session had no effect, and Glx remained unchanged with both protocols (Rafique & Steeves, 2020).
If TBS offers similar effects to rTMS but with shorter stimulation, then based on our previous work, we would expect a single session of cTBS to have no effect on GABA and glutamate at the visual cortex. However, Allen et al. (2014)  Note. cTBS consists of bursts containing three pulses at 50 Hz (20 ms between each stimulus), repeated at 5 Hz intervals (i.e., 200 ms ISI), applied continuously/uninterrupted for 40 s, providing a total of 600 pulses; and iTBS consists of the same bursts containing three pulses at 50 Hz, repeated at 5 Hz intervals but applied in 2 s trains repeated every 10 s for a total of 190 s, also providing a total of 600 pulses (Huang et al., 2005). MRS = magnetic resonance spectroscopy; TBS = primary visual cortex.
addition, it will help establish safety profiles for various protocols and populations.
Participants had no known contraindications to TMS and magnetic resonance imaging (MRI), no underlying medical conditions, and no history of neurological or psychological disorders (Kim et al., 2019;Levinson et al., 2010;Schür et al., 2016). We employed further strict exclusion criteria to control for confounding factors and to minimize potential extraneous interactions associated with metabolite receptors and/or TMS mechanisms. Participants were not taking any medications at the time of participation (Stell et al., 2003) including hormonal contraceptives (Kaore et al., 2012;Smith et al., 2002), had no history of frequent or chronic migraines (Bohotin et al., 2002;Russo et al., 2005), no history of alcohol/substance dependence (Brust, 2004;Ke et al., 2004;Lobo & Harris, 2008;Malcolm, 2003), and were non-smokers (Epperson et al., 2005). Additionally, participants were asked to attempt a good night's sleep (Clow et al., 2014), and not to consume alcohol 48 h (Lobo & Harris, 2008) before each visit. Participants were assigned to one of three experimental groups (cTBS, iTBS, sham) in a pseudo-random fashion and were naive to the stimulation condition. Ten participants were discarded due to high MRS data fit errors, motion artifacts, or failure to meet exclusion criteria following the initial visit/screening. The remaining participants (mean age ± SD = 22.17 ± 3.52 years; N = 29; 12 males/17 females) had normal or corrected-to-normal vision. All participants gave informed consent, and the protocol was approved by the Office of Research Ethics at York University in accordance with the Declaration of Helsinki. Participants received monetary compensation.

Vision and cognitive assessments
All participants were required to complete and pass three basic visual assessments to ensure eligibility for normal or corrected-tonormal vision (>0.04 logMAR; stereoacuity ≥ 50″, normal color vision).

Magnetic resonance imaging
Anatomical and MRS data were acquired with a 3 Tesla Siemens Mag-netom® Prisma magnetic resonance scanner with a 32-channel highresolution array head coil (Siemens, Erlangen, Germany). Anatomical high-resolution T 1 -weighted images were acquired first followed by MRS. Participants were instructed to remain still, keep their eyes closed throughout, and refrain from falling asleep. Imaging was performed with the room lights turned off.

Anatomical T 1 -weighted
The T 1 -weighted magnetisation-prepared rapid gradient echo times for a total of 128 "edit-on" averages and 128 "edit-off" averages, number of samples = 2048, acquisition time = 15 min. A separate unsuppressed water reference scan was also acquired to allow for a tissue concentration reference: averages = 10, total acquisition time = 52 min. Two automated shimming procedures were performed using Siemens B1 Shim mode and TrueForm. Figure 1a shows an example of the standard voxel placement in the occipital cortex. Figure 1b shows an example of the difference-edited spectra for GABA+ and Glx peaks, as well as the creatine (Cr) signal in the "edit-off" spectrum, and

TMS
A Magstim Rapid 2 and Plus 1 Stimulator 70-mm diameter Double Air Film figure-of-eight coil and its sham coil counterpart (Magstim, Whiteland, Wales, UK) were used to deliver stimulation pulses or mimic stimulation pulses, respectively, to the defined target site.

Phosphene threshold
Phosphene threshold (PT) is a method of measuring visual cortex excitability through the perception of phosphenes. A phosphene is a phenomenon of light that can be produced from direct stimulation of the occipital cortex in the absence of visual stimuli. PTs can be used to determine the individual intensity for TMS administration at the visual cortex in the same way that the motor threshold is used to determine TMS intensity when applied to the motor cortex. PT, therefore, pro-vides an individual excitability threshold for TMS administration since thresholds vary greatly across individuals (Stewart et al., 2001).
In a dimly lit room, wearing a blindfold with eyes closed, participants were instructed to lean forward with their forehead resting on a table while placing no pressure on their eyes. Phosphenes are elicited when stimulation is applied from 1 to 5 cm above the inion and 0 to 3 cm laterally, depending on the hemisphere being tested (Elkin-Frankston et al., 2010). We marked four locations to form a square area to be tested: at the inion, 2 cm above the inion, 2 cm to the left of the inion, and 2 cm above the 2 cm to the left of the inion marker.
PTs were measured for each participant using single-pulse stimulation with the coil center held tangential to the scalp and handle orientated at 90 0 to the midline. The minimum output began at 50% intensity, and 10 pulses were administered to the marker 2 cm above the inion, with each pulse 6 s apart. Following a single TMS pulse, participants were instructed to respond "yes/no/maybe" corresponding to whether a phosphene was perceived that could vary in shape, color, motion, and size. At each location, the stimulator output was increased by 5% until phosphenes were evoked with the maximum output setting restricted to 90% intensity according to safety regulations (Wassermann, 1998).
If no phosphenes were evoked after 10 pulses, the coil was moved to a new position in the marked region. The coil was placed in a new location until the individual responded "yes," which was then designated as the hotspot. Subsequently, at the hotspot, the threshold was modified in 1% increments to refine the PT. A threshold was defined as the intensity at which 50% of pulses (5/10 pulses) resulted in a "yes" response.
The blindfold was removed every 10−15 min, when necessary, for a minimum of 2 min, to prevent dark adaption (Boroojerdi et al., 2000).

TBS
Participants underwent one of three TBS stimulation conditions per-

Experimental design
The study was divided into two visits. Day 1) participants initially underwent pre-TBS (baseline) screening consisting of eligibility ques- Metabolites are, however, reliably stable for at least several weeks (Henry et al., 2011;Near, Ho, et al., 2014). MRI acquisition began within 5 min of TBS cessation (immediate post-TBS) and was repeated 1 h later (1 h post-TBS) at visit 2. Vision assessments and the MoCA screening were repeated after the immediate post-TBS scan and before the 1 h post-TBS scan at the second visit. Participants were also asked to report any adverse effects of TBS. Finally, the PT was measured after the 1 h post-TBS scan. Figure 2 shows an overview of the experimental procedure. and subtraction of the "on" and "off" spectra to generate the edited difference spectrum Near, Edden, et al., 2014). Modeling of the different signals of interest was performed using the Gannet- The final output value of GannetQuantify was used providing tissuecorrected (relaxation-and alpha-corrected, voxel-average-normalized) GABA+ and Glx concentration estimates relative to water. These corrections account for the effects of tissue composition, tissue water content, and water and metabolite relaxation, as well as the fact that GABA is present in higher concentrations in GM compared to WM at approximately a 2:1 ratio Harris et al., 2015). We also include results for GABA+ and Glx concentrations as ratios relative to Cr, as is commonly reported in the literature, to allow the comparison of our data with a greater number of studies. Using Cr as an internal reference signal reduces the risk of error in the propagation of waterbased scaling, although care needs to be taken regardless of the choice of the reference signal as it is difficult to determine whether group differences arise from changes to the nominator or the denominator (Alger, 2010). For MEGA-PRESS of GABA+, water-referenced quantification including tissue-correction shows similar levels of performance to Cr referencing, and the reliability of the two referencing strategies is comparable (Mikkelsen et al., 2017(Mikkelsen et al., , 2019. For data quality control, datasets with a fit error over 10% (defined as the standard deviation of the fit residual divided by the model amplitude) or visible subtraction artifacts in the difference spectrum were removed (Mikkelsen et al., 2019). This resulted in the removal of all time points for one participant as mentioned in Section 2.1.

Statistical analyses
Statistical analyses were conducted in R statistical software (v1.

Tissue fractions
For GM, there was no significant interaction between TBS condition  Figure 3c shows GABA+/Cr concentrations across all time points.

Correlation between GABA+, Glx, and phosphene threshold
Since there were no significant changes following either of the

Adverse effects
Following TBS, one participant in the cTBS condition and one participant in the sham condition reported a minor headache, and one participant in the sham group reported craniofacial discomfort near the left eye. These effects are consistent with common and temporary reports following TMS (Oberman et al., 2011;Rossi et al., 2009).

DISCUSSION
This study is the first to provide data on the immediate and short-term effects of cTBS and iTBS to the visual cortex on V1 GABA and glutamate concentrations. We found that a single session of cTBS or iTBS had no significant effects on MRS measures of GABA+ and Glx levels and visual cortical excitability at a cohort level. However, TBS did alter the relationship between GABA+ and Glx for up to 1 h following stimulation. These preliminary findings suggest that unlike TBS to the motor or frontal cortices, TBS to the visual cortex can be used in investigative or clinical settings without significant implications or alterations to these neurotransmitter levels at the stimulation site.
These effects following cTBS to the visual cortex are consistent with our previous findings where a single session of conventional "inhibitory" 1 Hz rTMS to the visual cortex also did not affect GABA+ and Glx concentrations (Rafique & Steeves, 2020  ship between the neurotransmitters may be driven by glutamate since glutamate is a precursor to GABA. While MRS-measured glutamate is considered to reflect synaptic glutamatergic activity (Stagg & Nitsche, 2011), MRS measures of GABA represent the total sum of GABA A and GABA B receptor activity (inhibitory and excitatory activity, respectively) (Luo et al., 2011;Rae, 2014). GABA B receptor mechanisms are facilitated to an extent by glutamatergic activity (Chalifoux & Carter, 2011;Prout & Eisen, 1994). The effects of TBS are also in part mediated by glutamatergic mechanisms (Huang et al., 2011). Since it is still unclear how TBS protocols precisely interact with LTP and LTD mechanisms, if TBS acts on glutamatergic mechanisms, then it may affect the overall homeostatic balance between GABA and glutamate concentrations rather than simply impacting one metabolite over the other. Donahue et al. (2010) have used a number of haemodynamic approaches to determine the relationship between GABA and cerebral blood measures. In the presence of higher GABA concentrations, they suggest that the greater associated vascular and metabolic response occurs presumably to promote an increase in excitatory activity required to overcome the inhibition that may result from higher GABA availability. The lack of causal effects in our study may also be masked by other factors. We observed high interindividual variability across participants in the present study. Offline NIBS protocols tend to demonstrate larger intra-and interindividual variability in their aftereffects, which are dependent on numerous factors. The current brain state at the time of stimulation, history of synaptic activity, structural asymmetry, neurochemistry, the specific interneuron networks recruited, hormonal levels, circadian rhythms, sex, age, genetics (e.g., polymorphisms) are but some of the factors influencing interindividual variability (Hamada et al., 2013;Miniussi et al., 2013;Polania et al., 2018;Ridding & Ziemann, 2010;Silvanto et al., 2007). Although we attempted to constrain these variables as much as possible in our participant recruitment, there are factors that we could not control for that may have led to null findings, for example, structural differences and genetics. Intraindividual variability is, however, considerably consistent over weeks (Hinder et al., 2014;Vernet et al., 2014). The null finding in the present study may, therefore, be partially owed to the fact that NIBS triggers a complex chain of effects confounded by several variables, some of which are not directly observable (Bergmann & Hartwigsen, 2021).
Our findings stand in contrast to Allen and colleagues (2014) who found that cTBS significantly increased GABA+ at V1 following stimulation. One aspect of the contrasting finding could be owed to differences in MRI scanner and acquisition parameters. The most notable disparities include differences in stimulation coil, inclusion criteria, threshold determination, and active versus passive viewing during MRS. We used a figure-of-eight coil to deliver stimulation, which has been shown to produce more focal effects than the circular coil used by Allen et al. (2014); however, some circular coils may be capable of producing deeper effects depending on the geometry and design (Deng et al., 2013). We employed much stricter exclusion criteria to control for confounding variables associated with TMS and metabolites to minimize interindividual variability mentioned above that substantially impacts the response to TMS. Full details on our exclusion/inclusion criteria are provided in Section 2.1. Allen et al. (2014) used the MT rather than the PT to determine stimulation intensity. We used the PT as it offers a more accurate and relevant measure of visual cortical excitability, and quantifies suprathreshold stimulation of target neurons at the visual cortex. The MT is shown to not accurately reflect visual cortical excitability (Boroojerdi et al., 2002;Gerwig et al., 2003;Stewart et al., 2001). Further, MTs are markedly lower than PTs, which would result in lower stimulation intensity in the study by Allen and colleagues. When the intensity of TMS is lowered, a reversal of TMS effects has been demonstrated (Abrahamyan et al., 2011). It is recommended that at least allowances in intensity should be made to account for the cortical distance from the motor cortex and associated changes in neural tissue when using the MT for non-motor regions to minimize the risk of substantial under-or overstimulation (Stokes et al., 2005(Stokes et al., , 2007. Lastly, Allen et al. (2014) employed an active viewing task where participants watched a film during MRS acquisition, whereas our acquisition was performed at rest with the eyes closed in a dark room. Task viewing stimulates cortical activity (Vanderwal et al., 2017) and activates large-scale brain networks (van der Meer et al., 2020) that can impact metabolite levels as has been shown previously for MRS measures of GABA and Glx (Duncan et al., 2014;Kurcyus et al., 2018). Thus, task viewing in itself may have impacted metabolite levels and/or interacted with stimulation effects.
We also found that TBS had no significant impact on visual cortical excitability using 80% PT. Two studies using a circular coil have found that cTBS increased PTs when applied at 80% PT (Allen et al., 2014;Franca et al., 2006), and iTBS had no effect (Franca et al., 2006).
Conversely, Brückner and Kammer (2016) have found that a figure-ofeight coil significantly decreased PTs following cTBS applied at 80% PT, whereas a circular coil had no significant effect. At least with corticospinal excitability, thresholds are higher with a circular rather than figure-of-eight coil; however, reliability is better for the figure-of-eight than circular coil (Fleming et al., 2012). No significant change in PT is observed following cTBS or iTBS at 100% PT using a figure-of-eight coil, but a change in PT is seen following cTBS that is related to whether a visual acuity task is presented (Brückner & Kammer, 2015). A significant change in PT occurs when presented with high visual demands following cTBS, but no change in PT is observed in a low visual demand condition (i.e., rest), consistent with our protocol and our data. Stochastic resonance, a phenomenon that exists in systems with measurement thresholds, may account for these observed differences (Schwarzkopf et al., 2011). The phenomenon suggests that information is enhanced by the injection of low levels of noise that in turn lower the response threshold, whereas higher noise levels disrupt performance. Brückner and Kammer (2016) propose that cTBS to the visual cortex at 80% of individual PT using a figure-of-eight-coil may add low levels of noise to the visual system, thereby lowering the PT compared to the circular coil that has more diffuse effects and can depolarize a greater number of neurons. Using a higher intensity of 100% PT may exceed the amount of noise that would improve signal detection in the stochastic resonance framework. Reducing the stimulation intensity to 80% of PT may correspondingly reduce the volume of depolarized neurons differently and the resultant network modulation would be different (Brückner & Kammer, 2016). Subthreshold TMS is associated with opposing effects compared with suprathreshold TMS (Nahas et al., 2001;Nakamura et al., 1997). Consequently, the choice of threshold, as well as the coil, likely affects any impact on metabolite levels. However, the inhibitory effect following cTBS at 80% PT is only apparent in participants with higher PTs, with the slope of baseline PT predicting the direction of modulation irrespective of coil type (Brückner & Kammer, 2016). For participants with lower baseline PTs, there may be low levels of noise to the system, thus, increasing excitability in an already excitable system and lowering the threshold for both circular and figure-of-eight coils. In an individual with a higher PT and reduced visual cortical excitability (Áfra et al., 1998;Aurora & Welch, 1998;Terhune et al., 2015), there may be larger background neuronal activity.
Such elevated baseline noise in participants with higher PTs could be increased further by TBS as per the stochastic resonance phenomenon and would lead to increased PTs. Particularly with the round coil, the induced noise would be expected to be greater as well as stimulating a greater number of neurons. Additionally, a visual acuity task (i.e., high visual demand) following TBS modifies the state-dependent neuronal effects (Pasley et al., 2009;Perini et al., 2012), including raising noise levels that may have led to increased PTs (Brückner & Kammer, 2016).
Optimal levels of noise are necessary to push weak subthreshold signals over the threshold, thereby improving information transfer.
Continued investigation into the neurophysiological effects of NIBS will allow us to refine the poorly generalized assumption that stimulation interaction with underlying brain activity, structure, and its ability to target specific neuronal pathways is homogenous across the brain. Cortical excitability and neural noise depend not only on the stimulation parameters but also the morphophysiological properties of the stimulated area, spontaneous and task-induced brain states, and intrinsic connectivity patterns (Fertonani & Miniussi, 2017;Nettekoven et al., 2015;Silvanto & Muggleton, 2008). Stimulating cortical and subcortical surfaces with TMS induces changes in a mixture of neuronal populations that utilize distinct neurotransmitters and perform specific actions/functions. These physiological, functional, and anatomical properties as well as connectivity profiles greatly influence the diverse responses to stimulation depending on the site of stimulation.
The concept of "excitatory" effects of iTBS and "inhibitory" effects of cTBS are shown to be highly variable even at the motor cortex depending on differences in the interneuronal cortical networks that are preferentially recruited by the TMS pulse (Hamada et al., 2013) and the subregion stimulated (Martin et al., 2006). LTP and LTD in the visual cortex vary depending on the layer in the visual cortex that is stimulated since each layer depends on specific receptor and neurotransmitter activation (Daw et al., 2004). Identical stimulation protocols induce a differential cascade of effects since not all brain regions respond equally (Castrillon et al., 2020;Funke & Benali, 2010). It has been suggested that the effects of NIBS are determined by the extent of functional integration of a target region rather than the frequency range of the stimulation protocol (Castrillon et al., 2020). Numerous studies choose only to employ cTBS and not iTBS based on the findings of Huang et al. (2005) that demonstrate the effect of iTBS was not as strong nor as long-lasting as cTBS. Franca et al. (2006) also report that iTBS does not have effects at the visual cortex; however, we have described earlier why findings differ between TBS studies to the visual cortex and, therefore, why conclusions regarding iTBS null effects are not so straightforward. iTBS protocols may simply need modification to produce maximal effects. For example, iTBS effects are strongly related to baseline network connectivity (Nettekoven et al., 2014).
Priming and conditioning with opposing cTBS and iTBS protocols have been shown to magnify the conditioned aftereffects (Murakami et al., 2012

Limitations and future directions
The immediate and shorter-lasting effects of TBS may be lost during the acquisition of anatomical images and time taken for VOI positioning that cannot be overcome (∼10 min in total and was consistent across participants). Additionally, the limited spectral resolution at clinical field strengths (∼3T) hampers the reliable separation of glutamate resonances from glutamine, and the composite measure of Glx may dilute the effects of glutamate. However, Glx is considered primarily driven by the glutamate signal as glutamate is found at much greater concentrations than glutamine in the brain (Stagg, 2014). While the edited GABA signal is also contaminated with co-detected spectra, variations in macromolecules between individuals are estimated to be small (Hofmann et al., 2001;Kreis et al., 2005). As such, it is unlikely that macromolecule contamination affects interpretability. Importantly, MRS GABA estimates are reliable across weeks at the occipital cortex (Near, Ho, et al., 2014), and show low interindividual variability, but are not observed to correlate across cortical regions (Greenhouse et al., 2016).
Glutamate concentrations are considered stable for at least 1 month (Henry et al., 2011). Overall, GABA and glutamate concentrations show excellent reproducibility in the visual cortex using the MEGA-PRESS sequence (Evans et al., 2010;Henry et al., 2011;O'Gorman et al., 2011).
Since TMS focality is not confined to the stimulation site but also regions nearby as well as inducing direct and indirect distal effects on neurophysiological and cognitive behavior, it is essential to measure stimulation effects in distal brain regions. Given the multifactorial nature of variables contributing to TMS effects, this will help to establish how TMS affects metabolites across the brain, provide more insight into effects on biochemical pathways, and enable fine-tuning of parameters appropriately. Vidal-Piñeiro et al. (2015) have previously found no change in GABA at the stimulation site but found significant changes distally. Due to the single voxel limit of MRS, this requires further separate sessions of the highly controlled experimental design with identical stimulation protocols.
We used 80% threshold intensity in line with the original TBS protocol (Huang et al., 2005) and due to uncertainty over safety at higher intensities especially at the visual cortex since PTs are generally higher than MTs (Boroojerdi et al., 2002;Gerwig et al., 2003;Stewart et al., 2001). In pathophysiology, greater unpredictability to NIBS is expected (Maeda et al., 2000;Wassermann, 2002). Studies have safely employed TBS up to 120% MT to the frontal cortex in disease states without reducing tolerability (Bakker et al., 2015;Fregni et al., 2006;McClintock et al., 2017). In addition, intensities of 100% and 120% PT have been used safely at the visual cortex (Brückner & Kammer, 2015. Since the response to TMS varies depending on whether subthreshold, threshold, or suprathreshold intensities are employed, it is important to establish whether cTBS and iTBS have differing effects on V1 GABA+ and Glx at 100% and even 120% PT. The present study and our previous rTMS study (Rafique et al., 2020) are not directly comparable due to differences in the total number of pulses (600 versus 1200 pulses, respectively). Offline protocols of rTMS tend to be longer in order to produce longer-lasting changes (Thut & Pascual-Leone, 2010). The general analogy that TBS produces similar effects to rTMS does not account for the discrepancy in that the number of pulses is usually unequal. Previous studies have demonstrated that doubling the duration of stimulation can produce a contrasting effect (Gamboa et al., 2010(Gamboa et al., , 2011Goldsworthy et al., 2015). Future work is required to compare equivalent protocols in terms of pulses between conventional rTMS and TBS at the visual cortex since analogous effects may not be seen when the number of pulses is equal. The effects of accelerated/within-session TBS also need to be studied at the visual cortex. Accelerated rTMS produces accumulative, stable, and longer-lasting effects than a single session (Rafique & Steeves, 2020) and single daily sessions over consecutive days/weeks Holtzheimer et al., 2010).
Accelerated TBS is considered safe and well tolerated in clinical populations (Desmyter et al., 2014;Duprat et al., 2016), but investigations at the visual cortex and in visual-related disorders are crucially lacking.

CONCLUSION
We demonstrate that cTBS and iTBS protocols at the visual cortex have different effects than those seen at the motor and frontal cortices.
With the protocol employed in this study, neither paradigm caused significant shifts in visual cortex GABA+ or Glx. TBS did, however, change the relationship between visual cortex GABA+ and Glx up to 1 h post-TBS suggesting that effects may be subtly sufficient to alter the relationship between metabolites but not induce significant changes to the metabolite concentrations. Modified protocols of TBS may be needed at the visual cortex to produce substantial changes in LTD and LTP mechanisms, particularly if the tool is to hold value for translation to therapeutic use in patient populations.