Noninvasive theta-burst stimulation of the human striatum enhances striatal activity and motor skill learning

Abstract

The stimulation of deep brain structures has thus far only been possible with invasive methods. Transcranial electrical temporal interference stimulation (tTIS) is a novel, noninvasive technology that might overcome this limitation. The initial proof-of-concept was obtained through modeling, physics experiments and rodent models. Here we show successful noninvasive neuromodulation of the striatum via tTIS in humans using computational modeling, functional magnetic resonance imaging studies and behavioral evaluations. Theta-burst patterned striatal tTIS increased activity in the striatum and associated motor network. Furthermore, striatal tTIS enhanced motor performance, especially in healthy older participants as they have lower natural learning skills than younger subjects. These findings place tTIS as an exciting new method to target deep brain structures in humans noninvasively, thus enhancing our understanding of their functional role. Moreover, our results lay the groundwork for innovative, noninvasive treatment strategies for brain disorders in which deep striatal structures play key pathophysiological roles.

Fig. 1. Results of the task-based fMRI experiment—local activity.

a, Average BOLD activity in the putamen (top left) and caudate (top right) during tTIS and HF control stimulation (n = 13, one influential point removed based on Cook’s distance). tTIS led to significantly higher activity in the putamen (two-sided pairwise comparisons via estimated marginal means: t(276) = −2.55, P = 0.01, d = −0.41, Tukey adjustment) but not in the caudate. The average BOLD activity in the posterior (bottom left) and anterior (bottom right) putamen during tTIS and HF control stimulation was also studied (n = 14). tTIS led to significantly higher activity in both areas (one-sided ANOVA with Satterthwaite’s approximations: F(1, 299) = 13.47, P = 0.0003, pη² = 0.04 (small)). The lines indicate the measure of center (mean value across the stimulation condition) and the shaded areas represent standard errors (SEs). b, Voxels showing a trend of linear changes (a one-sided t contrast, uncorrected P = 0.01 at the voxel level, and uncorrected at the cluster level) over time in the striatum during tTIS are shown on the left and during HF control stimulation on the right, on a group level. The sections are ordered from caudal to rostral. Hot colors represent increased activity over time, while cold colors represent decreased activity. The green area indicates the striatum. c, Qualitative characterization of the location of the activity during each of the six blocks during tTIS (top) and HF control stimulation (bottom). Data are shown on a group level, highlighting voxels involved in the task when compared with baseline (a one-sided t contrast, uncorrected P = 0.001 at the voxel level, and uncorrected at the cluster level). The left side corresponds to the early training phase, and the right side corresponds to the later training phase. The shift from the superior to the inferior striatum is consistent with previous observations^,. The green area indicates the striatum. The color bar indicates the t statistic.

Fig. 2. Results of the task-based fMRI experiment—network activity.

a, BOLD activity during the motor task with concomitant HF control stimulation. The regions in the motor network involved in the SFTT are shown. Significant clusters are shown for an one-sided t contrast, uncorrected P = 0.001 at the voxel level, and FDR-corrected P = 0.05 at the cluster level. b, Comparison of BOLD activity between tTIS and HF control stimulation. Hot colors represent higher activity during tTIS. Significant clusters are shown for an one-sided t contrast, uncorrected P = 0.001 at the voxel level, and FDR-corrected P = 0.05 at the cluster level. c, Behavioral results of experiment 1 (n = 13, one influential point removed based on Cook’s distance). Performance is shown as the correct number of key presses normalized to the baseline. A significant effect of the stimulation was present, with tTIS leading to overall higher performance (one-sided ANOVA with Satterthwaite’s approximations: F(1, 1,560) = 6.35, P = 0.01, pη² = 0.004 (micro)). The lines indicate the measure of center (mean value across the stimulation condition), and the shaded areas represent standard errors (SEs). d, Areas in the right striatum where activity was significantly modulated by the behavioral score (correct key presses) during tTIS. Significant clusters are shown for an one-sided t contrast, uncorrected P = 0.001 at the voxel level, and FDR-corrected P = 0.05 at the cluster level. No significant clusters were observed during HF control stimulation.

Fig. 3. tTIS exposure strength in control regions with respect to the targeted striatum.

Histogram depicting the tTIS exposure distribution within specific ROIs computed for a 2 mA current intensity per channel (peak to baseline). a, tTIS exposure distribution of voxels in 10-mm-radius spheres underneath the four stimulating electrodes, averaged for the frontal and posterior electrodes, compared with that in the bilateral striatum (putamen, caudate and nucleus accumbens). The horizontal axis scale was limited to the range [0, 1] for visualization purposes. As a result, nine values greater than 1 V m⁻¹ were omitted, which most likely represented noise values at the edges of the brain mask. b, tTIS exposure distribution of voxels in subparts of the target region, namely the right putamen and right caudate. c–e, tTIS exposure distribution of voxels in supratentorial hubs showing stronger BOLD activation during the task-based fMRI experiment with concurrent tTIS than during HF control stimulation compared with that in the right striatum (putamen, caudate and nucleus accumbens). c, tTIS exposure distribution of voxels in the right striatum compared with voxels in the specific BNA regions of the thalamus, which contained voxels showing higher BOLD activity during the task-based fMRI experiment with concurrent tTIS than during HF control stimulation. d, tTIS exposure distribution of voxels in the right striatum compared with voxels in the specific BNA regions of the amygdala, which contained voxels showing higher BOLD activity during the task-based fMRI experiment with concurrent tTIS than during HF control stimulation. e, tTIS exposure distribution of voxels in the right striatum compared with voxels in the specific BNA regions of the SMA, which contained voxels showing higher BOLD activity during the task-based fMRI experiment with concurrent tTIS than during HF control stimulation.

Fig. 4. Behavioral experiment results in experiment 2.

a, Motor task performance for the younger cohort (n = 14, one influential point removed based on Cook’s distance). Performance is shown as the correct number of sequences normalized to the baseline. No differences across stimulation conditions were observed. b, Motor task performance for the older cohort (n = 15). Performance is shown as the correct number of sequences normalized to the baseline. The post hoc analysis showed that this cohort performed significantly better during tTIS than HF control stimulation (two-sided pairwise comparisons via estimated marginal means: t(351) = 3.26, P = 0.001, d = 0.45, Tukey adjustment). c, Motor task performance during the follow-up (FU) sessions for the younger cohort (n = 15). Performance is shown as the correct number of sequences normalized to the last block of training. No differences across stimulation conditions were observed. d, Motor task performance during the follow-up sessions of the older cohort (n = 14, one influential point removed based on Cook’s distance). Performance is shown as the correct number of sequences normalized to the last block of training. No differences across stimulation conditions were observed. In a–d, the lines indicate the measure of center (mean value across the stimulation condition) and the shaded areas represent standard errors (SEs).

Fig. 5. Experimental setup.

a, Illustration of the SFTT^,: participants were asked to reproduce a sequence displayed on a screen with a four-button box. b, Protocol of the task sessions in experiment 1: the baseline measurement was followed by training with concomitant stimulation, consisting of six blocks containing ten 30-s task repetitions alternated with 30-s rest periods. c, Protocol of experiment 2: the baseline measurement was followed by training with concomitant stimulation, consisting of seven repetitions of 1 min 30 s task blocks. A post assessment (Post) with three block repetitions was performed immediately after and in a follow-up assessment ~90 min (FU1) after and ~24 h (FU2) after the end of the stimulation. d, TI stimulation concept. On the left, two pairs of electrodes are shown on a head model, and currents are applied with frequencies of f₁ and f₁ + Δf. On the right, the interference between the two electric fields within the brain is plotted at two different locations with high and low envelope modulation. e, In the pulsed stimulation mode, the frequency shift is only introduced during an exposure phase, for example, a burst. This allows the delivery of patterned intermittent theta-burst protocols in which three pulses of amplitude modulation at 100 Hz (burst phase) are repeated every 200 ms (theta frequency) for a 2-s train. The trains are repeated every 10 s. f, Electric field modeling with the striatal montage. Top: the electric tTIS exposure distribution in three chosen slices passing through the target region. Bottom: a 3D reconstruction of the structural MRI data, highlighting the electrode positioning.