Thirty stroke survivors participated in this study. Stroke survivors were included into this study if they had a single stroke less than 18 months prior to enrollment. All stroke survivors were screened for eligibility to receive iTBS and excluded if they were using medications that reduce seizure threshold or had history of seizure disorder; pregnant; or any implanted devices or metal that might be affected by iTBS (Rossi et al., 2009). Stroke survivors were also excluded if there was a presence of cognitive impairment as defined by inability to comprehend and follow three step commands (Ding et al., 2018). Upper-extremity component of the Fugl-Meyer motor function assessment (FMA) and action research arm test (ARAT) were used to assess motor impairment and upper extremity motor performance, respectively. Demographic characteristics are reported in Tables 1, 2.

Subjects gave their written informed consent for the experimental procedures that were approved by the Guangzhou First People’s Hospital Human Research Ethics Committee. The study was performed in accordance with the Declaration of Helsinki.

We used a sham-controlled, randomized single-blinded design. Stroke survivors were randomly assigned to the experimental (Active iTBS) and control (Sham iTBS) groups, with fifteen subjects in each group. Stroke survivors were blinded with respect to the group they were assigned to, that is, whether the subject received Active or Sham iTBS.

Intermittent theta burst stimulation was applied over the M1 in the ipsilesional hemisphere (IH) using a NS5000 Magnetic Stimulator (YIRUIDE Medical Co., Wuhan, China). The iTBS pattern consists of bursts containing three pulses at 50 Hz repeated at 5 Hz. A 2 s train of TBS was repeated every 10 s for a total of 192 s (600 pulses in total) (Huang et al., 2005). Of note, the stimulation intensity was set at 70% resting motor threshold (RMT) instead of 80% active motor threshold (AMT) in the original iTBS protocol (Volz et al., 2016). The reason for setting stimulation intensity based on RMT rather than AMT is that the latter would require stroke survivors to perform constant submaximal contractions of the target muscle which is often impossible for the paretic hand, especially in low-functioning stroke survivors (Volz et al., 2016). Furthermore, previous studies have shown similar aftereffects of iTBS with a stimulation intensity of 70% RMT or 80% AMT (Gentner et al., 2008; Cardenas-Morales et al., 2014). Therefore, a stimulation intensity of 70% RMT can be considered as an effective variant for increasing cortical excitability after iTBS (Volz et al., 2016; Yu et al., 2021).

Resting motor threshold determination was performed prior to the application of iTBS. Surface electromyography (EMG) was recorded from the first dorsal interosseous (FDI) in the paretic hand. Stroke survivors were seated in a comfortable chair with back support (Ding et al., 2018). TMS was applied over M1 using a figure-of-eight-shaped coil (70 mm diameter) positioned tangentially 45° from midline. Stroke survivors were asked to remain static while determining the optimal scalp position (i.e., “hotspot”) for eliciting maximal responses in the FDI (Ding et al., 2018). RMT was determined experimentally as the lowest stimulation intensity that produced MEP greater than 50 μV in at least 50% of consecutive stimulations at rest (Chen et al., 1998). A neuronavigation system (Visor2, ANT Neuro, Hengelo, Netherlands) was used to ensure reliable and consistent coil positioning over the “hotspot” throughout the experiment (Ding et al., 2021a).

Intermittent theta burst stimulation was applied over the “hotspot.” During the application of iTBS, stroke survivors were asked to remain static. As 40% maximum stimulator output (MSO) is the upper limit for iTBS with the NS5000 Magnetic Stimulator, stimulation intensity was set at 40% MSO for iTBS if the calculated stimulation intensity was greater than 40% MSO (i.e., for those whose RMT was greater than 57% MSO). For sham stimulation, the same stimulation intensity was used as for iTBS, and the TMS coil was held perpendicular to the skull, touching the skull with the rim opposite the handle (Nettekoven et al., 2014).

All statistical analyses were performed in JMP Pro Version 13.2 (SAS Institute Inc., Cary, NC, United States). Linear mixed effects (LME) modeling was performed to test differential changes in EEG power, coherence and global efficiency after iTBS between groups. Group, Timepoint, and Group×Timepoint interaction were included as fixed effects, and subject was included as a random effect. Timepoint was set as repeated covariance structure. Normality of the residuals was visually assessed for each model with conditional residual quantile-quantile plots, and all were found to reasonably conform to the assumption of normality. Post hoc tests were performed when F-tests were significant. Multiple comparisons between Timepoints or Groups were performed with Tukey-Kramer adjustment.

Data were found to meet the normality assumption using the Kolmogorov-Smirnov test. Pearson correlations were performed to investigate the relationship between baseline and changes in neurophysiological measures (e.g., EEG power, coherence, and global efficiency) and patients characteristics (e.g., age, chronicity, FMA, and ARAT). For all analyses, the statistical significance was set at P < 0.05.

The LME modeling did not reveal any significant Group×Timepoint interaction in EEG power in the ipsilesional [F_(1,28) = 0.02, P = 0.893; F_(1,28) = 1.59, P = 0.218; F_(1,28) = 0.64, P = 0.429; F_(1,28) = 0.70, P = 0.409] or contralesional [F_(1,28) = 0.40, P = 0.534; F_(1,28) = 1.79, P = 0.192; F_(1,28) = 0.10, P = 0.753; F_(1,28) = 0.14, P = 0.707] hemisphere in the delta, theta, alpha or beta band, respectively.

In the delta band, results of LME modeling revealed significant Group×Timepoint interaction [F_(1,28) = 5.03, P = 0.033]. Post hoc revealed that in the Active iTBS group, coherence was significantly increased after iTBS compared with baseline (P = 0.038), while there was no significant change in coherence over time in the Sham iTBS group (P = 0.327) (Figure 1).

In the theta band, results of LME modeling revealed significant Group×Timepoint interaction [F_(1,28) = 6.75, P = 0.015]. Post hoc revealed that in the Active iTBS group, coherence was significantly increased after iTBS compared with baseline (P = 0.011), while there was no significant change in coherence over time in the Sham iTBS group (P = 0.342) (Figure 2).

In the beta band, results of LME modeling revealed significant main effect of Timepoint [F_(1,28) = 6.38, P = 0.018], but there was no significant Group×Timepoint interaction [F_(1,28) = 3.25, P = 0.082], suggesting coherence was increased after iTBS in both groups without group differences (Figure 3).

In the alpha band, the LME modeling did not reveal significant Group×Timepoint interaction [F_(1,28) = 0.38, P = 0.544].

In the delta band, results of LME modeling revealed significant Group×Timepoint interaction [F_(1,28) = 5.11, P = 0.032]. Post hoc revealed that in the Active iTBS group, global efficiency was significantly increased after iTBS compared with baseline (P = 0.013), while there was no significant change in global efficiency over time in the Sham iTBS group (P = 0.586) (Figures 4A–C).

In the beta band, results of LME modeling revealed significant main effect of Timepoint [F_(1,28) = 8.54, P = 0.007] and Group×Timepoint interaction [F_(1,28) = 8.06, P = 0.008]. Post hoc revealed that global efficiency was significantly increased after iTBS compared with baseline in the Active iTBS group (P < 0.001), while there was no significant change in global efficiency over time in the Sham iTBS group (P = 0.954) (Figures 4D–F).

In the theta and alpha band, the LME modeling did not reveal any significant Group×Timepoint interaction [F_(1,28) = 3.08, P = 0.090; F_(1,28) = 2.19, P = 0.150, respectively].

No significant correlation was observed between neurophysiological measures (i.e., EEG power, coherence and global efficiency) and subject characteristics (i.e., age, chronicity, FMA, and ARAT).

We observed an increase in delta and theta bands coherence between bilateral sensorimotor cortices after iTBS, indicating increased interhemispheric functional connectivity after iTBS. The acute effects of iTBS on functional connectivity have not been investigated in stroke survivors, but it has been investigated in healthy adults (Nettekoven et al., 2014; Hoy et al., 2016). Our results are in line with Hoy et al.’s (2016) study that reported increased interhemispheric functional connectivity in theta band after iTBS in healthy adults. Similarly, an fMRI study (Nettekoven et al., 2014) also reported an increase in functional connectivity between bilateral sensorimotor areas after the application of iTBS on M1 in healthy adults. Despite different methodology among studies, our current study for the first time extends these findings from healthy adults to stroke population, suggesting that iTBS produces similar effects on interhemispheric functional connectivity in stroke survivors and healthy adults.

Neural mechanisms underlying the increase in interhemispheric functional connectivity after iTBS in stroke survivors remain unclear, which possibly relates to the simultaneous induction of neural activity in the whole motor network during the application of iTBS (Nettekoven et al., 2014). It has been reported that rTMS-induced changes in cortical activity are not exclusively local, but also extending to remote, interconnected regions (Bestmann et al., 2004; Suppa et al., 2008; Cardenas-Morales et al., 2014). As bilateral sensorimotor cortices are interconnected by transcallosal fibers, iTBS applied on ipsilesional M1 would induce simultaneous activation in the contralesional sensorimotor cortex (Nettekoven et al., 2014). The simultaneous activation of bilateral sensorimotor cortices would contribute to an increase in the coherence of brain activity that represents an important neurophysiological mechanism enforcing communication between the interconnected brain regions via transcallosal connections, and thus increases interhemispheric functional connectivity (Fries, 2005; Di Lazzaro et al., 2008).

Interestingly, beta band coherence was increased in both Active and Sham iTBS groups without significant group difference, suggesting changes in beta band coherence might not relate to neural effects of iTBS but result from other confounding factors such as the noise of iTBS click (Fuggetta et al., 2008). An increase in beta band coherence was also reported after sham rTMS in healthy adults (Fuggetta et al., 2008). The increased beta band coherence after sham rTMS may be caused by a cumulative effect of the rapid sequency of auditory TMS-click sounds produced during the application of rTMS (Fuggetta et al., 2008). Some neuroimaging studies (Bestmann et al., 2004; Takano et al., 2004) suggested that TMS clicks induce activations of the auditory systems and influence cerebral blood flow and synaptic activity in the brain regions interconnected with the auditory systems, which possibly influences beta band coherence between bilateral sensorimotor cortices. Collectively, external influences on cortical oscillations due to concomitant auditory stimulation need to be carefully controlled in TMS studies (Fuggetta et al., 2008).

We observed an increase in delta and beta bands global efficiency after iTBS. Acute changes in global efficiency induced by iTBS or high frequency rTMS have not been investigated in stroke survivors. Park et al. (2014) investigated acute changes in global efficiency after high frequency rTMS in healthy adults. No significant change in global efficiency was observed in the whole sample, but the authors reported an increase in global efficiency in individuals with behavioral facilitation after rTMS (Park et al., 2014). In our current study, we observed an increase in global efficiency after iTBS in the whole sample. The inconsistent results between Park et al.’s (2014) and our study may result from differences in subjects’ characteristics (i.e., healthy adults in Park et al.’s (2014) study vs. stroke survivors in ours) and experimental methodology. For example, Park et al. (2014) used 10 Hz rTMS, while we used iTBS in the present study. As the neural effects produced by iTBS have been suggested to be stronger than traditional rTMS (Fuggetta et al., 2008), it is reasonable to speculate that increase in global efficiency is possibly more robust after iTBS compared with 10 Hz rTMS.

The mechanisms for the increase in global efficiency after iTBS has not been fully elucidated. It has been suggested that iTBS induces long-term potentiation (LTP)-like changes at synaptic connections (Huang et al., 2007), and would increase efficiency of synaptic transmission in both local and remote brain regions from the stimulation site (Philips et al., 2017). Furthermore, iTBS causes simultaneous activation in the interconnected brain regions which increases neural synchrony in the global brain network (Di Lazzaro et al., 2008). Therefore, iTBS possibly facilitates global information exchange and thus increases global efficiency.

We did not observe any change in EEG power after iTBS. Although there was no stroke study investigating aftereffects of rTMS (including iTBS) on EEG power, our results are in line with studies conducted in healthy adults, which reported no change in EEG power after iTBS (Hoy et al., 2016) or high frequency rTMS (Oliviero et al., 2003; Fuggetta et al., 2008); however, increased EEG power after high frequency rTMS has also been reported (Azila Noh and Fuggetta, 2012). These conflicting results may be due to the different methodological details among studies. For example, 10 Hz rTMS was applied in Azila Noh and Fuggetta (2012) study, while 5 Hz rTMS was applied in Oliviero et al.’s (2003) and Fuggetta et al.’s (2008) studies. These results suggest that different types of rTMS might influence its effect on EEG power.

This study for the first time used EEG to investigate the aftereffects of iTBS following stroke. Our results revealed increased interhemispheric functional connectivity and global efficiency after iTBS in stroke survivors. Dynamics of functional brain network has been suggested to be altered following stroke due to focal brain lesions (Vecchio et al., 2019). Normal functioning of brain network (i.e., interhemispheric functional connectivity and global efficiency) plays an important role in recovery of motor performance following stroke.

Interhemispheric functional connectivity has been suggested to play a supportive role in motor recovery following stroke (Rehme et al., 2011). van Meer et al. (2010) reported that impaired motor function acutely after experimental stroke in rats was related to partial loss of interhemispheric functional connectivity, and interhemispheric functional connectivity was increased subsequently concomitant to motor recovery. In humans, reduced interhemispheric functional connectivity was also observed acutely after stroke (Philips et al., 2017; Desowska and Turner, 2019; Hoshino et al., 2020; Li et al., 2020). It has been reported that the increase in interhemispheric functional connectivity was associated with motor improvements in stroke survivors, and restoration of interhemispheric functional connectivity was noted only in well recovered individuals, but not in the poorly recovered stroke survivors (Golestani et al., 2013; Desowska and Turner, 2019; Hoshino et al., 2020; Li et al., 2020), suggesting that interhemispheric functional connectivity is a potential biomarker indicating stroke recovery (Caliandro et al., 2017; Philips et al., 2017; Vecchio et al., 2019). Increase in interhemispheric functional connectivity after iTBS observed in our current study provides evidence that iTBS could normalize brain network functioning in stroke survivors.

Global efficiency exhibits the efficiency in transporting information at a global scale between genetic brain areas (Vecchio et al., 2019). Increased global efficiency after iTBS suggests alterations in how efficiently information is transferred over the brain, reflecting an acute shift of the brain state induced by iTBS (Park et al., 2014). Reduced global efficiency indicates lower efficiency in global information flow, which has been suggested to be related to motor deficits associated with aging (Park et al., 2012) or neurological conditions such as stroke (Philips et al., 2017). Contrary to the brain state of motor deficits, increased efficiency in global information flow could reflect the brain state of intact or enhanced motor function (Park et al., 2014). Therefore, the shift of brain state toward an emphasis on global information exchange after iTBS suggests that iTBS has the potential to be utilized in stroke rehabilitation.

The influence of stroke characteristics (e.g., chronicity, motor impairment, age, etc.) on the effects of iTBS is less clear. In present study, no significant correlation observed between clinical characteristics and neurophysiological measures was observed. Our results suggest that the effects of iTBS on functional brain network were not influenced by stroke characteristics. As our sample size is small (N = 30), cautions are needed when interpreting these results. Further studies with larger sample sizes are still needed to investigate how the heterogeneity of stroke survivors influences the effects of iTBS.

As a pilot study, the sample size of current study is small (N = 30). Chronicity of stroke survivors in current study ranged from 1 to 18 months, so cautions are needed when generalizing our findings to more chronic stroke survivors. Chronicity of stroke survivors was not evenly distributed in our sample, and most subjects were within 3 months post-stroke. Therefore, we did not perform subgroup analysis for chronicity. Further research is required to test our results in stroke survivors with a wider range of chronicity with larger sample sizes and to perform subgroup analysis for individuals in acute, subacute and chronic phases of stroke.

Our current study only measured resting-state EEG for 6 min immediately after iTBS without a follow-up. We understand that it would be more meaningful to measure EEG at multiple time points after iTBS. However, it has already been a long experiment for stroke survivors, and many subjects could not tolerate for a longer time of data collection. Further studies are needed to monitor changes in EEG at multiple time points after iTBS.

Another limitation is that sex of stroke survivors was not very balanced between groups with 12 males in the Active iTBS group vs. 9 males in the Sham iTBS group. To the best of our knowledge, no previous study has reported sex difference in the aftereffects of TBS. There was a tDCS study (Kuo et al., 2006) reporting sex differences in the aftereffects of cathodal (i.e., inhibitory) but not anodal (i.e., excitatory) tDCS due to changes in ovarian hormones over the menstrual cycle. In current study, most female subjects (8 out of 9) were postmenopausal women. Those postmenopausal women did not have a menstrual cycle, so they were less likely to be influenced by changes in ovarian hormones. Although it is still unclear whether sex influences aftereffects of iTBS, our results are unlikely to be attributed to sex difference. Further studies are still needed to investigate sex difference in the aftereffects of iTBS.

In this study, 40% MSO was the upper limit for iTBS with the TMS machine. We set the stimulation intensity of iTBS at 40% MSO if the calculated stimulation intensity (i.e., 70% RMT) was greater than 40% MSO. Therefore, the actual stimulation intensity of iTBS was lower than the calculated stimulation intensity for those whose RMT was greater than 57% MSO. In 11 out of 15 subjects in the Active iTBS group, RMT was greater than 57% MSO. We acknowledge that the relatively lower stimulation intensity of iTBS for subjects with high RMT is a limitation of current study. However, as higher stimulation intensity may produce stronger neurophysiological effects, neuroplastic changes observed in current study were induced by relatively lower stimulation intensity, suggesting that the neuroplastic changes observed in current study were robust.

Ours is the first study that used EEG to investigate the aftereffects of iTBS on functional brain network in stroke survivors. This study for the first time provides evidence that iTBS modulates functional brain network in stroke survivors. Our results revealed an increase in interhemispheric functional connectivity and global efficiency after iTBS, suggesting that iTBS has the potential to normalize brain network functioning following stroke, which can be utilized in stroke rehabilitation.