This study represents the first endeavour to combine LFP recordings from the PerceptTM PC with high-density EEG, aiming to investigate thalamo-cortical oscillatory connectivity patterns via phase synchronisation in TS patients with an implanted DBS system in the medial thalamus. This innovative approach underscores the practicality of using sensing-enabled neurostimulators, exemplified by the Percept PC from Medtronic, for research purposes, enabling the acquisition of unique data otherwise unattainable, thereby offering invaluable contributions to our understanding of TS. Our findings revealed a spatially and spectrally distinct oscillatory network, connecting medial thalamus and frontal regions in the alpha (8–12 Hz) band, with functional connectivity strength negatively correlated with TS symptom severity. Additionally, we demonstrated a reduction in thalamo-frontal alpha connectivity immediately preceding tic onset, suggesting its involvement in tic generation. Further analysis refining the spatiality of this finding revealed that this modulation extended to sensorimotor regions, including the primary motor cortex, primary somatosensory cortex, premotor cortex, as well as the inferior parietal cortex. Notably, pre-tic-related temporal dynamics are specific to phase synchronisation and not evident in the pure power spectra for both LFP and cortical sources.
Before proceeding to the discussion of our main findings, it is essential to address the absence of a tic suppression effect in the present patient group in this study. While our findings indicate no significant impact on tic frequency from voluntary suppression, we lack additional data related to tic suppression efforts for further insights. Therefore, only speculations can be made about possible explanations. One possibility is that patients did not voluntarily suppress their tics when requested to do so, for various reasons. Some patients may find tic suppression too challenging because they cannot resist the urge [41]. In other patients, symptom severity may have been sufficiently low to render tic suppression superfluous or impractical, given the difficulty in detecting impending tics due to low urge intensity [42]. However, another possibility is that tics were inadvertently suppressed even in the tic-freely condition. Individuals with TS often develop habitual tic control mechanisms, particularly in social settings, implying that tic suppression may occur automatically over time [41, 43, 44]. Consequently, tic control may persist even when patients are not actively attempting to suppress their tics, potentially explaining the lack of differences between the tic-freely and tic-suppression conditions. Therefore, we cannot rule out the potential influence of habitual/automatic tic control in the supposed tic-freely condition. Importantly, even if both habitual/automatic and voluntary tic control resulted in similar tic frequencies, the underlying mechanisms may differ. Our findings show that voluntary tic suppression did not change thalamo-frontal alpha connectivity or thalamic/frontal alpha power, which may indicate that voluntary suppression did not occur, or that both habitual/automatic and voluntary control processes lead to the same effects on these neural patterns. This remains an open question, and we can only conclude that our task manipulation did not have the intended effect. This topic warrants further investigation, as understanding the differences between habitual/automatic and voluntary tic control and their underlying neural mechanisms is crucial for advancing our knowledge of tic control.
Beyond that, it should be noted that the small number of patients (n = 4) included in the suppression analysis likely challenged the detection of a tic suppression effect. Furthermore, the heterogeneity of our patient group, which included individuals with very severe symptoms as well as those with minor symptoms due to several months of successful DBS treatment, likely impacts both habitual/automatic and voluntary tic control expression.
Unique to the present study is the comprehensive characterisation of thalamo-cortical functional connectivity patterns across different frequency ranges covering the entire cortex. We discovered a spatially and spectrally distinct thalamo-cortical network in patients with TS at rest, restricted to the alpha frequency band (8–12 Hz) in frontal regions. Notably, simultaneous resting cortical alpha power peaking in posterior, rather than frontal, regions suggests that the observed thalamo-frontal alpha connectivity pattern is independent of overall power activity. Interestingly, we observed a negative correlation between thalamo-frontal alpha connectivity and tic/urge severity. At the same time, no similar correlation pattern could be observed for thalamic and frontal alpha power, emphasising the distinctiveness of the relation between the identified functional connectivity pattern and TS symptomatology. This finding underscores the importance of considering TS as a network disorder characterised by pathophysiological functional connections within CBGTC circuits. It aligns with prior neuroimaging findings of abnormal connections between the thalamus and various frontal regions, encompassing motor and sensory cortices, the cingulate cortex, and the supplementary motor area [22, 45].
The specific mechanism underlying the observed association between increased thalamo-frontal alpha connectivity and reduced symptom severity remains speculative. Building on the earlier notion that habitual/automatic tic control may have been engaged during the tic-freely condition, one plausible hypothesis is that increased thalamo-frontal connectivity could potentially enhance (habitual/automatic) tic control. This is supported by previous research linking fronto-striatal hyperconnectivity as well as general cortical alpha network connectivity to chronic tic control [43, 46]. Also, it has been postulated that tic control involves top-down control mechanisms originating from frontal to subcortical regions, potentially normalising abnormal activity within CBGTC circuits responsible for tics [41, 47, 48]. However, considering that thalamo-frontal alpha connectivity also negatively correlated with urge severity, an alternative or complementary hypothesis could be that increased connectivity may be associated with a reduced PMU. This would also be in line with the observed dynamical decrease of thalamo-frontal alpha connectivity preceding tic execution, as discussed later. Given the thalamus’ role in sensorimotor function as a central mediator of sensory input and perception it is reasonable to posit that thalamo-frontal connections may influence the PMU [49, 50]. Moreover, previous research has highlighted the critical role of frontal regions in the PMU [2, 51, 52].
Although the precise mechanisms are yet to be fully understood, the negative correlation between thalamo-frontal alpha connectivity and symptom severity suggests its potential as a target for stimulation-based treatments in patients with TS. Consistent with this, neuroimaging studies have shown that DBS is most effective when structural or functional connectivity networks linking the thalamus to the frontal cortex, particularly sensorimotor regions such as the (pre-)SMA, cingulate cortex, primary motor cortex, and primary sensory cortex, are stimulated [53,54,55,56]. Furthermore, research utilising median nerve stimulation (MNS) highlights the importance of targeting the alpha frequency range, as rhythmic 10-Hz pulse trains have shown significant tic improvement [57, 58]. Rhythmic 10-Hz median nerve stimulation may increase the thalamo-frontal alpha connectivity, which may reduce the occurrence of tics.
Interestingly, we observed dynamic changes in thalamo-frontal alpha connectivity in relation to the tic. These were characterised by distinct functional connectivity decreases between the thalamus and frontal regions, particularly in sensorimotor areas and the inferior parietal cortex, at different timings before the tic. A notable reduction in connectivity around one second before the tic involved various brain regions, including the SMA, cingulate motor cortex, primary motor cortex, primary somatosensory cortex, premotor cortex, inferior parietal cortex as well as the insular and frontal opercular cortex. This indicates that neural processes underlying tic occurrence start well before tic onset, which is in line with the typically observed pre-tic symptomology in TS patients, i.e. the PMU [2]. Furthermore, immediate connectivity decreases around tic onset involved the primary motor cortex, primary somatosensory cortex, premotor cortex, and inferior parietal cortex. This finding is particularly interesting as it implies a direct link to tic generation. Notably, these immediate pre-tic changes were specific to functional connectivity patterns, with no similar direct tic-related dynamic changes detected for mere thalamic or frontal alpha power at either sensor or source level.
Importantly, it should be noted that the tic analysis included tics from both the tic-freely and tic-suppression conditions. Our decision to pool these epochs was driven by the goal of identifying a general tic marker applicable to both conditions. In addition, given the absence of voluntary tic suppression effects, as discussed above, we have no strong reason to believe that the neural dynamics of tics differ between these conditions in the present patient group. However, we must acknowledge that the neural patterns immediately preceding tic onset may differ between the tic-freely condition, which represents usual tics, and the tic-suppression condition, which reflect tics that failed to be suppressed — or potentially not, as it remains unclear whether the tics were actively suppressed, given the absence of significant voluntary suppression effects. Nevertheless, previous studies suggest the potential of a single tic-generation process that is unaffected by voluntary suppression, indicating that all tics, whether attempted to be suppressed or not, might arise from the same fundamental tic-generating mechanism [46, 59]. For this reason, and due to the limited number of tics available, we did not conduct further subgroup analyses comparing these two conditions, and we preliminarily interpret our tic-related functional connectivity pattern as reflecting a general tic-generation process. However, we cannot rule out potential differences between usual tics and those that failed to be suppressed, and we suggest that future studies with a larger dataset explore whether distinct functional connectivity patterns emerge between these states or whether the observed patterns generalize across different tic control contexts.
Our understanding of the precise mechanisms underlying the observed pre-tic disconnections remains speculative. Building on our earlier hypothesis regarding the nature of the resting thalamo-frontal connectivity pattern, the observed disconnection immediately preceding tics might indicate a transient lapse in (habitual/automatic) tic control, potentially facilitating tic execution. However, the temporal pattern of gradually decreasing connectivity over time leading up to the tic suggests more a progressive development of underlying processes, culminating in the manifestation of the apparent tic. Such a process could be more likely related to the PMU, which typically increases before the tic until reaching its peak just before tic onset [2]. It is also plausible that the decreases observed around one second before the tic and immediately before tic onset represent different underlying processes. In fact, the present functional connectivity patterns may stem from a complex interplay of processes involving both tic control and PMU, engaging different brain regions at different timings.
The observed tic-related dynamic functional connectivity changes, encompassing different sensorimotor, frontal, and parietal brain areas are in line with various observations from imaging studies on tic-preceding neural activity [23, 38, 39]. Previous LFP studies have primarily focused on tic-related thalamic power changes, consistently reporting a distinct unrhythmic low-frequency (2–10 Hz) increase following tic onset [10, 13, 14, 16, 18]. Based on this feature, closed-loop DBS approaches in TS have already demonstrated feasibility, safety, and efficacy comparable to continuous DBS [19, 20]. However, these studies did not identify any pre-tic activity changes. Similarly, a recent EEG study found no pre-tic alterations in alpha or beta power in sensorimotor cortices, contrasting with the well-known movement-related beta suppression observed before voluntary movements [60]. This highlights the absence of a distinct electrophysiological power marker preceding tic onset, suggesting the involvement of a complex neural network in tic generation. Prior electrophysiological research on tic-related thalamo-cortical functional connectivity patterns in TS is very limited [10, 18]. One study combining chronic LFP recordings with surface electrocorticogram (ECoG) recordings over the motor cortex detected no thalamo-motor cortex coherence during rest, movement, or tics, which could be related to the limited coverage provided by subdural strips [18]. In another study, intraoperative combined LFP and EEG recordings in three patients revealed repetitive increases in thalamo-cortical coherence preceding tics across broad frequency ranges, including alpha and beta [10]. Discrepancies between these findings and ours may be attributed to factors such as the timing of the recordings and potential cross-subject variability.
In light of this, our results add valuable insights to the existing literature by demonstrating a consistent pattern of pre-tic-related functional connectivity changes across patients, extending beyond the intraoperative time window. These findings may pave the way for future research aimed at identifying electrophysiological pre-tic markers, particularly for closed-loop DBS in TS.
Various limitations of the present study need to be acknowledged. First, it was limited by a small sample size, which restricts the generalisability of our findings to a broader population. Additionally, while the patients exhibited very heterogeneous symptoms, the sample was homogeneous in terms of gender, with all participants being male. We also acknowledge that the small sample size increases the likelihood of spurious correlations, as even strong-looking associations may not replicate or could show different directions in larger samples. Therefore, future studies with larger samples are essential to confirm the reliability and generalizability of these correlations. In addition, while we applied a data-driven approach to focus on Fz for the correlational analyses, the channel showing the strongest connectivity during the rest analysis, we acknowledge that this selection process could introduce bias and inflate the observed correlations. A broader investigation of multiple channels in future studies with larger samples and increased statistical power would help confirm whether these correlations are specific to Fz or generalize across the frontal region. Next, correlation results may be influenced by DBS effects, as clinical parameters reflect symptom severity over the past week when DBS was active. To accurately assess the relationship between thalamic activity and symptom severity in the DBS-Off state, it would be necessary to collect clinical parameters after turning off DBS for at least a week. However, this is unfeasible due to clinical and ethical constraints. A further limitation may arise from potential synaptic plasticity changes following long-term stimulation, especially given the broad range of DBS durations (3–164 months) across patients. The effects of prolonged stimulation might not fully reverse within a wash-out period of 2 min, potentially influencing our results and increasing the heterogeneity in our small sample. In our tic-related analysis, a major limitation is the lack of a control condition for comparison, such as voluntary movements. Additionally, we cannot rule out the potential influence of other movements during the pre-tic state, as patients performed mouse movement as part of the task. The considerable heterogeneity in the phenomenological appearance of tics introduces another limitation, as we were not able to investigate the distinction between vocal and motor tics, which may exhibit different connectivity patterns. However, the current study could not differentiate between these tic types due to the limited number of tics recorded and the presence of combined motor and vocal tics in three out of five patients, as well as the fact that one patient exhibited only motor tics with no vocal tics (see Table 2). It should be emphasised that our primary aim was to identify a common neural substrate underlying tics, irrespective of their specific characteristics. Future research with larger datasets may explore the differences in connectivity between vocal and motor tics, as well as simple and complex tics, more thoroughly.
In conclusion, the present study, combining LFP recordings using the PerceptTM PC with high-density EEG in TS patients with thalamic DBS, extends beyond previous intraoperative LFP studies, providing valuable new insights. Our findings implicate the role of a distinct thalamo-frontal network within the alpha frequency band (8–12 Hz) in the TS pathophysiology. Thereby, they underscore the importance of investigating electrophysiological oscillatory synchronisation between subcortical and cortical regions to characterise pathological functional connections within CBGTC circuits. These identified functional connectivity patterns may serve as targets for stimulation-based interventions in TS, informing future research on closed-loop DBS for TS.