The non-transcranial TMS-evoked potential is an inherent source of ambiguity in TMS-EEG studies

Transcranial Magnetic Stimulation (TMS) excites populations of neurons in the stimulated cortex, and the resulting activation may spread to connected brain regions. The distributed cortical response can be recorded with electroencephalography (EEG). Since TMS also stimulates peripheral sensory and motor axons and generates a loud “click” sound, the TMS-evoked EEG potential (TEP) not only reflects neural activity induced by transcranial neuronal excitation but also neural activity reflecting somatosensory and auditory processing. In 17 healthy young individuals, we systematically assessed the contribution of multisensory peripheral stimulation to TEPs using a TMS-compatible EEG system. Real TMS was delivered with a figure-of-eight coil over the left para-median posterior parietal cortex or superior frontal gyrus with the coil being oriented perpendicularly or in parallel to the target gyrus. We also recorded the EEG responses evoked by sham stimulation over the posterior parietal and superior frontal cortex, mimicking the auditory and somatosensory sensations evoked by real TMS. We applied state-of-the-art procedures to attenuate somatosensory and auditory confounds during real TMS, including the placement of a foam layer underneath the coil and auditory noise masking. Despite these precautions, the temporal and spatial features of the cortical potentials evoked by real TMS at the prefrontal and parietal site closely resembled the cortical potentials evoked by realistic sham TMS, both for early and late TEP components. Our findings stress the need to include a peripheral multisensory control stimulation in the study design to enable a dissociation between truly transcranial and non-transcranial components of TEPs.


Introduction (906 words)
4 8 activity evoked by focal TMS targeting non-motor prefrontal and posterior parietal cortex. Although we 4 0 4 implemented state-of-the art measures to attenuate multisensory co-stimulation, the cortical potentials 4 0 5 evoked by real and sham TMS at the prefrontal and parietal site closely resembled each other, both in 4 0 6 temporal shape and spatial distribution. This similarity might be even greater than the one shown in the 4 0 7 present study, because our realistic sham condition did not perfectly match the multisensory input evoked 4 0 8 by TMS in the somatosensory domain. The close resemblance of EEG responses evoked by real TMS and 4 0 9 realistic sham stimulation shows that the non-transcranial TEP is an inherent source of ambiguity in 4 1 0 TMS-EEG studies. Therefore, future TMS-EEG studies need to actively show that multisensory co-4 1 1 stimulation was suppressed completely. This could be achieved by showing that participants perform at 4 1 2 chance level in a two-alternative forced choice test in which they indicate whether they have received 4 1 3 TMS or not. If participants still can dissociate between TMS and no-TMS trials after all measures are 4 1 4 taken to suppress multisensory co-stimulation, the experimental design needs to include a realistic sham 4 1 5 control condition which mimics multi-sensory co-stimulation as closely as possible. 4 1 6 4 1 7 Peripherally evoked potentials evoked by multisensory stimulation 4 1 8 Although our realistic sham stimulation did not perfectly match the multisensory input associated with 4 1 9 real TMS, the temporal and spatial patterns of the peripherally-evoked cortical responses closely 4 2 0 resembled the spatiotemporal patterns of TEPs evoked in the real TMS conditions. In the temporal 4 2 1 domain, evoked peak latencies closely matched the TEP latencies evoked by real TMS at early, middle, 4 2 2 and late post-stimulation intervals. Peak correspondence was found 40-400 ms post stimulation for the 4 2 3 frontal target site and 70-400 ms for the parietal target side, including the classic N100 central negativity 4 2 4 often reported in TMS-EEG studies (Du et al., 2017). Likewise, the topographical distribution of the 4 2 5 evoked responses showed a significant correlation between sham and real TMS conditions for almost the 4 2 6 entire 20-410 ms post-stimulation time window. Using a sham condition that consisted of real TMS 4 2 7 delivered to the shoulder, Herring et al. (Herring et al., 2015) showed that sham stimulation induced a 4 2 8 cortical response pattern that was similar to the one evoked by real TMS over the scalp, primarily at late 4 2 9 peak latencies (> 80 ms post stimulation). Extending these findings, we show that concurrent cranial 4 3 0 somatosensory and auditory stimulation mimicking TMS contributes substantially to the TEP also at early 4 3 1 latencies. 4 3 2 The similarity between realistic sham and real TMS between 20 and 80 ms after TMS can be attributed to We also found a close resemblance of the EEG response between sham and real TMS stimulation 4 5 0 conditions for the later components evoked by both realistic sham and real TMS, including the N100 and 4 5 1 P180 components, commonly described as the N1-P2 complex for both auditory and somatosensory 4 5 2 stimulation (Goff et al., 1977;Hyde, 1997). The auditory N1-P2 peaks at frontocentral scalp electrodes as 4 5 3 a result of respectively oriented dipoles in bilateral temporal cortices (Zouridakis et al., 1998), and 4 5 4 somatosensory components at > 100 ms originate from bilateral secondary somatosensory cortices 4 5 5 (Allison et al., 1992). The N100 is of particular interest as has been associated with GABA-B-ergic 4 5 6 inhibition based on pharmacological interventions (Premoli et al., 2014a) and paired-pulse TMS (Opie et 4 5 7 al., 2017;Premoli et al., 2014b;Rogasch et al., 2012), as well as by its amplitude correlation with the 4 5 8 silent period duration (Farzan et al., 2013). Notably, Du et al. (Du et al., 2017) observed a vertex N100 of 4 5 9 similar amplitude after TMS of prefrontal, motor, primary auditory cortices, vertex, and cerebellum, and 4 6 0 concluded that the N100 is a ubiquitous TEP reflecting a general property of the cerebral cortex. Our 4 6 1 findings point rather to the conclusion that the N100 observed over the vertex is at least to a great extent a 4 6 2 non-transcranial sensory evoked potential. The close resemblance of TMS and sham-evoked potentials does by no means imply that specific TEP 4 6 6 components can be always and fully explained by multisensory-evoked potentials. On the contrary, TEP 4 6 7 recordings hold great potential for probing the local and distributed brain response to focal TMS. Since 4 6 8 the multisensory components overlap substantially with the truly transcranial components, it is necessary 4 6 9 to disentangle the multisensory temporal and spatial response patterns from the truly transcranially-4 7 0 evoked brain response. The true TEP components may become only evident after subtraction of the 4 7 1 multisensory components or in experimental designs that effectively account for multisensory stimulation 4 7 2 as a confound. In the study of Herring et al. (Herring et al., 2015), for instance, the authors found a left 4 7 3 occipital N40 component following left visual cortex TMS but not multisensory sham that can hardly be 4 7 4 explained by somatosensory or auditory co-stimulation. If the topography of a TEP component is clearly 4 7 5 lateralized and confined to the stimulation site, such component is less likely to be the mere result of 4 7 6 multisensory stimulation which often shows a different voltage distribution. Also, the GABA-B-receptor-4 7 7 mediated amplitude modulation of an N100 component lateralized to the stimulated left sensorimotor 4 7 8 1 cortex most likely reflects a local cortical effect at the target site (Premoli et al., 2014a). In contrast, 4 7 9 GABA-A receptor-mediated amplitude modulations of the TEP have been reported to only be significant 4 8 0 in the hemisphere contralateral to stimulation (Premoli et al., 2014a), and future work has to clarify the 4 8 1 degree to which remote effects like this are due to distant scalp projections of a local dipole, a network 4 8 2 spread of transcranially-induced activity, or pharmacological effects on multisensory cortical processing. 4 8 3 Studies using similar GABA-mediating drugs such as benzodiazepines have consistently reported effects 4 8 4 on AEPs and SEPs also at 100 ms, reinforcing the need to further investigate the purely transcranial 4 8 5 effects of drugs on the TEP (Abduljawad et al., 2001;Scaife et al., 2006). Our findings are compatible 4 8 6 with the notion that local activations at the target site may predominantly arise from transcranial 4 8 7 stimulation particularly in the early post-stimulation period. For electrodes close to the stimulated region, 4 8 8 the similarity between sham and real TMS was less consistent. The stronger dissimilarity of evoked 4 8 9 responses 24-70 ms after stimulation may thus be due to the local activations after real TMS as compared 4 9 0 to sham. Alternatively, this dissimilarity may have resulted from methodological issues since the decay 4 9 1 artefacts resulting from transcutaneous electric stimulation were also strongest at the stimulation site, and 4 9 2 the early post-stimulation interval included less time points than the middle or late post-stimulation 4 9 3 intervals potentially decreasing similarity between stimulation conditions. 4 9 4 In a recent study aiming to disentangle the cortical origin of TEPs, Gosseries et al. targeted both lesioned 4 9 5 and preserved cortical tissue in two patients with unresponsive wakefulness syndrome and multi-focal 4 9 6 brain injury (Gosseries et al., 2015). In these patients, TEPs were completely absent when TMS directly 4 9 7 targeted the lesioned cortex, whereas TEPs were preserved when targeting non-lesioned cortex, keeping 4 9 8 multisensory co-stimulation comparable (Gosseries et al., 2015). These results show that a local cortical 4 9 9 response can be evoked by TMS, but does not rule out a substantial multisensory contribution to TEPs 5 0 0 recorded in healthy conscious individuals. It should also be noted that the first patient had additional brain 5 0 1 stem lesions in the pons, medulla and cerebellar peduncles. These additional lesions might have blocked 5 0 2 peripheral somatosensory input from the lesioned but not from the non-lesioned hemisphere. The second 5 0 3 patient had massive bilateral hemispheric lesions, involving auditory and somatosensory cortex 5 0 4 bilaterally. Again, this might have prevented the occurrence of cortical responses caused by multisensory 5 0 5 co-stimulation. It also seems that substantially higher TMS intensities were applied by Gosseries et al. 5 0 6 and the local responses had much larger amplitudes than those normally obtained in healthy conscious 5 0 7 individuals. Finally, in patients with disorders of consciousness it is not possible to individually adjust the 5 0 8 sound pressure of the noise masking, potentially resulting in higher sound pressures than those tolerated 5 0 9 by healthy individuals. 5 1 0

1 1
Can auditory and somatosensory stimulation be completely suppressed in awake individuals without 5 1 2 brain lesions? 5 1 3 The evidence obtained in unresponsive patients with massive multi-focal brain damage (Gosseries et al., 5 1 4 2015) cannot be generalized to other studies and does not imply that those components are principally of 5 1 5 transcranial origin when observed under different conditions. Special care needs to be taken when 5 1 6 contrasting different physiological states (e.g., drug challenges, vigilance or attentional states, etc.) or 5 1 7 groups (e.g., psychiatric or neurological patients) for which also a modulatory effect on auditory or 5 1 8 somatosensory evoked potentials is conceivable or in some cases known. It has been proposed that 5 1 9 multisensory co-stimulation does not account for any TEP components as long as both auditory and 5 2 0 somatosensory perception are suppressed by noise masking and foam padding (Gosseries et al., 2015).

2 1
Unfortunately, a complete suppression is often not achievable when studying fully awake individuals, 5 2 2 even when following best practice procedures as reported in the present study. We implemented all 5 2 3 measures currently advised to attenuate multisensory co-stimulation (i.e., individualized noise masking, 5 2 4 foam padding, and stimulation sites close to the midline) and still observed multisensory evoked 5 2 5 potentials, while almost all participants reported residual auditory and tactile perception of the TMS 5 2 6 pulses. Unlike in other studies for which complete suppression of TMS "click" sound perception has been 5 2 7 reported (Casula et al., 2017;Gosseries et al., 2015;Massimini et al., 2005), we systematically asked 5 2 8 participants to rate perceptual intensity after each stimulation condition. Only one participant reported 5 2 9 complete suppression, whereas all others reported perceptual intensities between 1 and 8 (out of max 10 5 3 0 points on the VAS) despite the maximal tolerable noise volume being used.

3 1
While it may be feasible to completely suppress concurrent auditory stimulation by applying noise 5 3 2 masking at very high sound pressures, we doubt that TMS-related inductive electric stimulation of 5 3 3 peripheral sensory and motor axons can be effectively suppressed given the biophysics of TMS. The fast-5 3 4 conducting myelinated peripheral axons passing through the tissue in close proximity to the induced 5 3 5 electric filed are readily excitable by TMS (Siebner et al., 1999), and these nerves are exposed to a much 5 3 6 larger electric field than the cortex because they are located much closer to the coil. Since myelinated 5 3 7 fast-conducting sensory trigeminal fibers are present in parasagittal parts of the dura mater (Lv et al., 5 3 8 2014), concurrent stimulation of dural trigeminal nerve fibers may also contribute significantly to the 5 3 9 TEPs. Notably, these nerve fibers are not effectively stimulated by bipolar electric cutaneous stimulation 5 4 0 due to the poor electric conductivity of the skull, so that not even our realistic sham condition would be 5 4 1 able to control for those responses.

4 2
One pioneering TMS-EEG study used electric stimulation of the scalp and did not observe any 5 4 3 somatosensory evoked cortical potentials (Paus et al., 2001), yet did neither report the precise stimulation 5 4 4 area nor any electric artifact removal procedures. Moreover, it has been argued that SEPs should be 5 4 5 "click" sound (middle panel) and overall discomfort (lower panel). 6 3 0 The columns represent the mean VAS scores (range: 0 to 10) and the error bars equal onefold standard 6 3 1 deviation for each stimulation condition. The bold horizontal lines with an asterisk on top represent 6 3 2 significant differences between two conditions for the same stimulation site (continuous lines) or between 6 3 3 the frontal and parietal conditions (stippled line). Statistical comparisons used a Wilcoxon Signed-Ranked 6 3 4 test with an alpha of 0.05/n (Bonferroni-Holm corrected for multiple comparisons). each correlation analysis as a bold timeline. The interruptions indicate periods during which correlation 6 6 5 did not reach significance. With a few exceptions, spatial correlations were significant between conditions 6 6 6 across the entire post-stimulation interval. 6 6 7 6 6 8 n  e  a  n  d  d  i  a  z  e  p  a  m  o  n  6  7  0  p  r  e  p  u  l  s  e  i  n  h  i  b  i  t  i  o  n  o  f  t  h  e  a  c  o  u  s  t  i  c  s  t  a  r  t  l  e  r  e  s  p  o  n  s  e  a  n  d  t  h  e  N  1  /  P  2  a  u  d  i  t  o  r  y  e  v  o  k  e  d  p  o  t  e  n  t  i  a  l  i  n  6  7  1  m  a  n  .  J  P  s  y  c  h  o  p  h  a  r  m  a  c  o  l  1  5  ,  2  3  7  -2  4 a  r  i  n  a  z  z  o  ,  D  .  ,  G  o  s  s  e  r  i  e  s  ,  O  .  ,  B  o  l  y  ,  M  .  ,  L  e  d  o  u  x  ,  D  .  ,  R  o  s  a  n  o  v  a  ,  M  .  ,  M  a  s  s  i  m  i  n  i  ,  M  .  ,  N  o  i  r  h  o  m  m  e  ,  Q  .  ,  7  6  1  L  a  u  r  e  y  s  ,  S  .  ,  2  0  1  4  .  D  i  r  e  c  t  e  d  i  n  f  o  r  m  a  t  i  o  n  t  r  a  n  s  f  e  r  i  n  s  c  a  l  p  e  l  e  c  t  r  o  e  n  c  e  p  h  a  l  o  g  r  a  p  h  i  c  r  e  c  o  r  d  i  n  g  s  :  7  6  2  i  n  s  i  g  h  t  s  o  n  d  i  s  o  r  d  e  r  s  o  f  c  o  n  s  c  i  o  u  s  n  e  s  s  .  C  l  i  n  E  E  G  N  e  u  r  o  s  c  i  4  5  ,  3  3  -3  9  .  7  6  3  M  a  s  s  i  m  i  n  i  ,  M  .  ,  F  e  r  r  a  r  e  l  l  i  ,  F  .  ,  H  u  b  e  r  ,  R  .  ,  E  s  s  e  r  ,  S  .  K  .  ,  S  i  n  g  h  ,  H  .  ,  T  o  n  o  n  i  ,  G  .  ,  2  0  0  5  .  B  r  e  a  k  d  o  w  n  o  f  c  o  r  t  i  c  a  l  7  6 R  o  s  s  i  ,  S  .  ,  F  e  r  r  o  ,  M  .  ,  C  i  n  c  o  t  t  a  ,  M  .  ,  U  l  i  v  e  l  l  i  ,  M  .  ,  B  a  r  t  a  l  i  n  i  ,  S  .  ,  M  i  n  i  u  s  s  i  ,  C  .  ,  G  i  o  v  a  n  n  e  l  l  i  ,  F  .  ,  P  a  s  s  e  r  o  ,  S  .  ,  2  0  0  7  .  A  8  0  9  r  e  a  l  e  l  e  c  t  r  o  -m  a  g  n  e  t  i  c  p  l  a  c  e  b  o  (  R  E  M  P  )  d  e  v  i  c  e  f  o  r  s  h  a  m  t  r  a  n  s  c  r  a  n  i  a  l  m  a  g  n  e  t  i  c  s  t  i  m  u  l  a  t  i  o  n  (  T  M  S  )  .  8  1  0  C  l  i  n  N  e  u  r  o  p  h  y  s  i  o  l  1  1  8  ,  7  0  9  -7  1  6  .  8  1  1  S  a  r  a  s  s  o  ,  S  .  ,  B  o  l  y  ,  M  .  ,  N  a  p  o  l  i  t  a  n  i  ,  M  .  ,  G  o  s  s  e  r  i  e  s  ,  O  .  ,  C  h  a  r  l  a  n  d  -V  e  r  v  i  l  l  e  ,  V  .  ,  C  a  s  a  r  o  t  t  o  ,  S  .  ,  R  o  s  a  n  o  v  a  ,  M  .  ,  8  1