Short pauses in thalamic deep brain stimulation promote tremor and neuronal bursting

Abstract

Objective: We conducted intraoperative measurements of tremor during DBS containing short pauses (⩽50 ms) to determine if there is a minimum pause duration that preserves tremor suppression.

Methods: Nine subjects with ET and thalamic DBS participated during IPG replacement surgery. Patterns of DBS included regular 130 Hz stimulation interrupted by 0, 15, 25 or 50 ms pauses. The same patterns were applied to a model of the thalamic network to quantify effects of pauses on activity of model neurons.

Results: All patterns of DBS decreased tremor relative to 'off'. Patterns with pauses generated less tremor reduction than regular high frequency DBS. The model revealed that rhythmic burst-driver inputs to thalamus were masked during DBS, but pauses in stimulation allowed propagation of bursting activity. The mean firing rate of bursting-type model neurons as well as the firing pattern entropy of model neurons were both strongly correlated with tremor power across stimulation conditions.

Conclusions: The temporal pattern of stimulation influences the efficacy of thalamic DBS. Pauses in stimulation resulted in decreased tremor suppression indicating that masking of pathological bursting is a mechanism of thalamic DBS for tremor.

Significance: Pauses in stimulation decreased the efficacy of open-loop DBS for suppression of tremor.

Keywords: Computational model; Deep brain stimulation; Essential tremor; Movement disorders; Thalamus.

Figure 1

Tremor was measured during temporal patterns of thalamic deep brain stimulation (DBS) containing short pauses. (A) Patterns of DBS, labeled with pause duration, frequency of pauses, and instantaneous pulse frequency (IPF) between pauses. All patterns had a geometric mean frequency of 130 Hz. (B) Timeline of intraoperative trials to measure tremor. Twenty-second measurements of tremor were made following thirty seconds without stimulation (Pre), and again after thirty seconds with one of the stimulation patterns in (A) or DBS OFF.

Figure 2

Computational model of thalamic DBS. (A) Representation of model thalamocortical (TC) neuron and four terminating axons providing input to model TC neuron. (B) Oblique rectangular prism representation of the ventral intermediate (Vim) thalamus with coordinates of 50 TC model neurons randomly positioned within the nucleus. Abbreviations: CTX, cortex; RN, reticular nucleus; TIN, thalamic interneuron; CER, cerebellum; NMDA, N-methyl-D-aspartate; AMPA, DL-α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid; GABA, γ-Aminobutyric acid Reprinted from Birdno et al., 2012, with permission.

Figure 3

Tremor was quantified using an accelerometer on the back of the hand. Triaxial accelerometer signals collected during a postural tremor task (left) and power spectra for each axis (middle). Spectra were summed and integrated across tremor frequencies to yield tremor power (right). The base 10 logarithmic transform of tremor power (LTP) is correlated with clinical ratings of tremor severity.

Figure 4

Effects of DBS with temporal patterns of stimulation containing short pauses on postural tremor. Mean ± standard error is plotted as a function of stimulation pattern. Log-transformed tremor power (LTP) varied across stimulation patterns (F = 18.8, p<0.0001, linear mixed-effects model). Levels not sharing the same letter are significantly different (p<0.05, Tukey’s HSD).

Figure 5

Responses of model thalamocortical (TC) neurons to simulated DBS. Cerebellar (CER) burst-driver input and somatic firing activity for 10 of 50 model TC neurons during a one s period (A) with DBS off and (B) during 130 Hz constant frequency DBS. Rastergrams for three regular-firing, two random-firing, and five bursting-type TC cells are shown in each plot.

Figure 6

Responses of model thalamocortical (TC) neurons to simulated pattern of DBS containing short pauses. Somatic and axonal firing pattern activity (left and right columns, respectively) for 10 representative model thalamocortical (TC) neurons during a one s of DBS with 50 or 15 ms pauses at a rate of 4.4 Hz (top and bottom rows, respectively). Rastergrams for three regular-firing, two random-firing, and five bursting-type TC cells are shown in each plot.

Figure 7

Correlation between tremor measured in subjects with ET and thalamic DBS and mean firing rate in bursting-type model neurons across different temporal patterns of DBS (R²=0.733 across stimulation conditions plotted above; R²=0.866 when including stimulation OFF condition).

Figure 8

Correlation between tremor measured in subjects with ET and thalamic DBS and firing pattern entropy of model neurons across different temporal patterns of DBS. (A) First order estimate of entropy of log-transformed interspike intervals (ISIs) for each TC neuron sub-type (Random, n=15; Regular, n=10; Bursting, n=25) and averaged across sub-types (All, n=50), reported as mean ± standard error for each stimulation pattern tested. (B) Correlation between tremor and model neuron entropy (R²=0.927 across stimulation conditions plotted above; R²=0.964 when including stimulation OFF condition).

Figure 9

Responses of model thalamocortical (TC) neurons to simulated patterns of cycling DBS. Cerebellar (CER) burst-driver input and somatic firing activity for 10 representative model TC neurons during a one s period with trains of 130Hz DBS triggered by the (A) beginning or (B) end of cerebellar bursts. Rastergrams for three regular-firing, two random-firing, and five bursting-type TC cells are shown in each plot.