Cortico-thalamic tremor circuits and their associations with deep brain stimulation effects in essential tremor

Abstract

Essential tremor is one of the most common movement disorders in adults. Deep brain stimulation (DBS) of the ventralis intermediate nucleus of the thalamus and/or the posterior subthalamic area has been shown to provide significant tremor suppression in patients with essential tremor, but with significant inter-patient variability and habituation to the stimulation. Several non-invasive neuromodulation techniques targeting other parts of the CNS, including cerebellar, motor cortex or peripheral nerves, have also been developed for treating essential tremor, but the clinical outcomes remain inconsistent. Existing studies suggest that pathology in essential tremor might emerge from multiple cortical and subcortical areas, but its exact mechanisms remain unclear. By simultaneously capturing neural activities from motor cortices and thalami and recording hand tremor signals via accelerometers in 15 human subjects who had undergone lead implantations for DBS, we systematically characterized the efferent and afferent cortico-thalamic tremor networks. Through the comparisons of these network characteristics and tremor amplitude between DBS off and on conditions, we also investigated the associations between different tremor network characteristics and the magnitude of the DBS effect. Our findings implicate the thalamus, specifically the contralateral hemisphere, as the primary generator of tremor in essential tremor, also with a significant contribution of the ipsilateral hemisphere. Although there is no direct correlation between the cortico-tremor connectivity and tremor power or reduced tremor by DBS, the strength of connectivity from the motor cortex to the thalamus and vice versa at tremor frequency predicts baseline tremor power and effect of DBS. Interestingly, there is no correlation between these two connectivity pathways themselves, suggesting that, independent of the subcortical pathway, the motor cortex appears to play a relatively distinct role, possibly mediated through an afferent/feedback loop in the propagation of tremor. DBS has a greater clinical effect in those with stronger cortico-thalamo-tremor connectivity involving the contralateral thalamus, which is also associated with bigger and more stable tremor measured with an accelerometer. Interestingly, stronger cross-hemisphere coupling between left and right thalami is associated with more unstable tremor. This study provides important insights for a better understanding of the cortico-thalamic tremor-generating network and its implication for the development of patient-specific therapeutic approaches for essential tremor.

Keywords: deep brain stimulation; directed connectivity; efferent and afferent; essential tremor; local field potential.

Figure 1

Experimental protocol. (A) Schematic diagram of the posture-holding task performed when the DBS is switched off (left) and on (right). (B) Time line for the experimental protocol, which consists of 10 posture-holding blocks (∼20 s per block) when both arms are raised up and 10 resting blocks when both arms are put down. (C and D) 3D reconstruction in coronal (C) and coronal–axial (D) views of all analysed DBS leads localized in standard Montreal Neurological Institute (MNI)-152_2009b space using Lead-DBS.^, Electrodes in the left hemisphere were mirrored to the right hemisphere. ACC = accelerometer; Ch = channel; DBS = deep brain stimulation; GND = ground; LFP = local field potential; OUH = Oxford University Hospital; SGH = St George’s Hospital; UHC = University Hospital Cologne; VIM = ventral intermediate thalamus; ZI = zona incerta.

Figure 2

Comparisons of tremor characteristics between DBS off and DBS on conditions. (A) An example of 30 s postural tremor (P1L), showing the instability of tremor in essential tremor. (B) Demonstration of the quantifications of tremor amplitude and frequency instability from a segment of 2 s measurement from an accelerometer. (C) Normalized PSD of accelerometer measurements showed peaks at tremor frequency band in both DBS off (black) and DBS on (red) conditions (top), with a significant reduction of the normalized power (as a percentage) in the individualized tremor frequency band during DBS on (bottom). (D–F) Comparisons of tremor power (D), amplitude instability (E) and frequency instability (F) between DBS off (black) and DBS on (red) conditions using raincloud plots. Here, the shaded areas indicate distributions (probability density) of the data. (G–I) Tremor power during DBS off (baseline) is positively correlated (G) whereas tremor amplitude (H) and frequency (I) instability are negatively correlated with the reduction in tremor power during DBS (Pearson correlation). Solid lines in C and bars in C–F indicate the mean, while shaded areas in C and error bars in C–F indicate the standard error of the mean. Statistics were applied between DBS off and DBS on conditions using a non-parametric cluster-based permutation procedure in C (PSD) on a hand-by-hand basis, or using generalized linear mixed-effect modelling in all bar plots (C–F) on a trial-by-trial basis. Multiple comparisons were corrected by controlling the false discovery rate. ***P < 0.001 after false discovery rate correction. CDBS = continuous deep brain stimulation; DBS = deep brain stimulation; P1L = participant1, left hand; PCA = principal component analysis; PSD = power spectral density; std = standard deviation.

Figure 3

Characteristics of thalamic-tremor and cortico-tremor networks when DBS was switched off. (A) A demonstration of left-hand postural tremor and thalamic LFP recordings from Participant 1, left hand (P1L) during DBS off condition. (B) Directed connectivity between VIM thalamus and hand tremor quantified using gOPDC. Solid lines indicate efferent connectivity from thalamus to hand tremor, while dashed lines indicate afferent connectivity from hand tremor to thalamus. Orange and purple represent the connectivity with ipsilateral and contralateral VIM thalami, respectively. The top and bottom panels indicate gOPDC involving only one thalamus (unconditioned) and both thalami (HC), respectively. (C) Efferent connectivity from the contralateral thalamus was significantly stronger than that from the ipsilateral hemisphere in both unconditioned (left) and HC (right) models. When conditioning the impact from the other hemisphere, the efferent connectivities from the contralateral (purple) and ipsilateral (orange) thalami to hand tremor were both significantly reduced. (D) Afferent connectivity from hand tremor to the contralateral thalamus was significantly weaker than that to the ipsilateral hemisphere in both unconditioned (left) and hemisphere-conditioned (right) models. When conditioning the impact from the other hemisphere, the afferent connectivities from hand tremor to the contralateral (purple) and ipsilateral (orange) thalami were both significantly reduced. (E–H) The same as A–D but for cortico-tremor network. Bars and error bars indicate mean and standard error of the mean, respectively. Statistics were applied on each comparison using generalized linear mixed-effect modelling on a trial-by-trial basis. Multiple comparisons were corrected by controlling the false discovery rate. *P < 0.05, **P < 0.01 and ***P < 0.001, after false discovery rate correction. Acc = accelerometer; cCort = contralateral motor cortex; cThal = contralateral thalamus; DBS = deep brain stimulation; gOPDC = generalized orthogonalized partial directed coherence; HCgOPDC = hemisphere-conditioned generalized orthogonalized partial directed coherence; iCort = ipsilateral motor cortex; iThal = ipsilateral thalamus; LFP = local field potential; VIM = ventral intermediate thalamus.

Figure 4

Characteristics of cortico-thalamo-tremor network. (A) Directed efferent connectivity from sensorimotor cortex and VIM thalamus to hand tremor quantified using gOPDC. (B) Comparing with the model involving only bilateral thalami in Fig. 3, conditioning cortical input significantly reduced the efferent connectivity from thalamus to hand tremor in both DBS off and DBS on conditions. (C) Comparing with the model involving only bilateral sensorimotor cortices in Fig. 3, conditioning thalamic input significantly reduced the efferent connectivity from cortex to hand tremor in both DBS off and DBS on conditions. (D) Directed afferent connectivity from hand tremor to sensorimotor cortex and VIM thalamus quantified using gOPDC. (E) Comparing with the model involving only bilateral thalami in Fig. 3, conditioning cortical input significantly reduced the afferent connectivity from hand tremor to thalamus in both DBS off and DBS on conditions. (F) Comparing with the model involving only bilateral sensorimotor cortices in Fig. 3, conditioning thalamic input significantly reduced the afferent connectivity from hand tremor to cortex in both DBS off and DBS on conditions. Here, the connectivity in A–F was quantified in tremor frequency band. (G) Directed connectivity between sensorimotor cortices and the contralateral VIM thalamus relative to the focused hand tremor quantified using gOPDC. (H) The directed top-down connectivity from cortex to thalamus (black) was significantly and consistently stronger than bottom-up connectivity from thalamus to cortex (red) in tremor (left), alpha (middle) and beta (right) frequency bands. Bars and error bars indicate mean and standard error of the mean, respectively. Statistics were applied on each comparison using generalized linear mixed-effect modelling on a trial-by-trial basis. Multiple comparisons were corrected by controlling the false discovery rate. ***P < 0.001 after false discovery rate correction. Acc = accelerometer; cCort = contralateral motor cortex; CDBS = continuous deep brain stimulation; cThal = contralateral thalamus; DBS = deep brain stimulation; gOPDC = generalized orthogonalized partial directed coherence; iCort = ipsilateral motor cortex; iThal = ipsilateral thalamus; NC = network-conditioned; VIM = ventral intermediate thalamus.

Figure 5

Correlations between cortico-thalamo-tremor network characteristics and the reduced tremor power with DBS. (A and B) Correlations between the efferent connectivity from the contralateral (A) or ipsilateral (B) thalami with hand tremor and the reduced tremor power with DBS. (C) Correlation between the sum of thalamus to cortex and cortex to thalamus connectivity at tremor frequency band and the reduced tremor power with DBS. (D) Correlation between the sum of all connectivity at tremor frequency involving the contralateral thalamus and the reduced tremor power with DBS. (E and F) There was no correlation between the efferent connectivity from the contralateral (E) or ipsilateral (F) sensorimotor cortices to hand tremor and the reduced tremor power with DBS. (G and H) There was no correlation between the sum of thalamus to cortex and cortex to thalamus connectivity at alpha (G) or beta (H) frequency band and the reduced tremor power with DBS. P-values were corrected for multiple comparisons by controlling the false discovery rate. CDBS = continuous deep brain stimulation; DBS = deep brain stimulation; gOPDC = generalized orthogonalized partial directed coherence; HC = hemisphere-conditioned.

Figure 6

Comparisons between thalamo-cortical and cortico-thalamic connectivity. (A) Directed connectivity at tremor frequency band (gOPDC) from thalamus to cortex (x-axis) was not correlated with that from cortex to thalamus (y-axis). (B and C) The strongest thalamo-cortical (B) and cortico-thalamic (C) gOPDC clustered at different areas in the standard MNI-152_2009b space. (D) A demonstration of the VTA with DBS at 1 mA applied to the selected bipolar local field potential channels (P13). (E) Results from Spearman rank correlation between the intersection of the VTA in VIM thalamus and directed connectivity from thalamus to cortex. (F) Results from Spearman rank correlation between the intersection of the VTA in ZI and directed connectivity from cortex to thalamus. gOPDC = generalized orthogonalized partial directed coherence; VIM = ventral intermediate thalamus; VTA = volume of tissue activated; ZI = zona incerta.

Figure 7

A summary of the present study. (A) Our study suggests that tremor in essential tremor originates from the contralateral thalamus (path 1). The motor cortex is involved through an indirect pathway, probably via a feedback loop, by receiving afferent input from the tremulous hand through ascending pathways (paths 2 and 3) and sending it back to the thalamus (path 4). There is also significant cross-hemisphere coupling at both subcortical (path 5) and cortical (path 6) levels. (B) Potential clinical implications of this study. cCort = contralateral motor cortex; iCort = ipsilateral motor cortex; cThal = contralateral thalamus; DBS = deep brain stimulation; iHand = ipsilateral hand; iThal = ipsilateral thalamus.