Single session of intermittent theta burst stimulation alters brain activity of patients in vegetative state

Participants

We enrolled medically stable patients with prolonged VS (after brain injury, the patient remains in a vegetative state lasting for at least 28 days) in this study. The exclusion criteria were: contraindications for MRI scans and iTBS applications (e.g. pacemakers, aneurysm clips, nerve stimulators, or brain/subdural electrodes), severe cerebral atrophy or injury in the left frontal area upon MRI scans, a history of epilepsy or EEG epileptiform activity. Finally, 18 patients (8 females and 10 males, average age: 42.38 ± 13.57, average time since injury: 5.75 ± 2.38 months) completed the experiment. Their clinical and demographic characteristics of the patients are shown in Table 1. All the patients were assessed by one week’s repeated Coma Recovery Scale-Revised (CRS-R) process [25]. The CRS-R were conducted by trained neurologists at least three times during the afternoon. All the patients received routine medication and rehabilitation courses but did not obtain consciousness improvement for at least two weeks. They had not taken any drugs such as zolpidem, modafinil, midazolam or baclofen, and were free of any acute medical complications (e.g. acute pneumonia) for at least two weeks. Written informed consent to participate in the study was obtained from the patients’ caregivers. This study was approved by the ethics committee of the Seventh Medical Center of PLA General Hospital. Clinical register: ChiCTR1800017623.

Experimental protocol

A randomized double-blind, crossover experimental was set up (Figure 1A). Each patient received both a real and sham iTBS in a randomized order. A computer-generated randomization sequence was used to assign, the first session as either a real or sham stimulation. For each patient, the experimenter received two blinded codes from a third person, one for the real stimulation and one for the sham stimulation. Before each stimulation, patients received CRS-R and fifteen minutes of resting-state EEG recording. In order to capture the modulation effects of iTBS on patients’ brain activities, 70 minutes of resting-state EEG followed by CRS-R was conducted immediately after the stimulation. The real and sham stimulations had at least a three-day washout period in between. To evaluate the long-term functional outcomes, we followed up with the Glasgow Outcome Scale Extension (GOS-E) scores at 6 months post-experiment via a phone call.

Stimulation

As shown in Figure 1B, iTBS with an intensity of 80% resting motor threshold (RMT) was conducted at the left dorsolateral prefrontal cortex (DLPFC) of patients. The iTBS included 20 trains of 50Hz bursts (3 pulses) repeated at 5Hz. They were applied at every 10 seconds (total 600 stimuli). The magnetic stimulation was administered in accordance with safety guidelines [26]. TMS pulses were delivered using a Magstim R² stimulator with a 70 mm figure-of-eight coil (Magstim Company Limited, Whitland, UK). Stimulation intensities varied across this experiment and were determined relative to the RMT, defined as the lowest TMS intensity that can be evoked in at least five out of 10 trials of an electromyogram with an amplitude >50μV peak-to-peak in the relaxed first dorsal interosseous muscle of the right hand. For the real iTBS session, the coil was positioned tangentially (perpendicularly in sham session) to the scalp pointing in an anteromedial direction, 45° from the midsagittal axis of patient’s head.

EEG recordings and pre-processing

In this study, the EEG recordings were acquired from patients by 62 channels (BrainAmp 64 MRplus, BrainProducts) with positions of the international 10-20 system. The equipment used sintered Ag/AgCl-pin electrodes. We set a band-pass filtered at DC to 1000Hz in the recorder. The EEG signals were digitized at a sampling rate of 2.5 kHz. During the recording, the skin/electrode impedance was maintained below 5kΩ. We monitored patients for possible EEG signs of drowsiness and sleep onset (an increase of tonic theta rhythms, sleep spindles). An arousal procedure of CRS-R would be performed in the patients who showed above EEG signs.

An off-line analysis was performed with the EEGLAB 12.0.2.5b, running in a MATLAB environment (Version 2013b, MathWorks Inc., Natick, USA). An independent component analysis (ICA) function was used to identify and remove the artefact’s relevant components such as eye movement and muscle activities [27]. The EEG data from two patients were excluded because of serious artefacts (over one of three ICA components were identified as artefacts). The EEG data were down-sampled to 500Hz, bandpass filtered (1-45Hz) and average referenced. Then, the EEG data were divided into epochs of 10 seconds. Artefact-free epochs were preserved and recorded for EEG analysis.

EEG analysis

Spectrum power, nonlinear neural dynamics of permutation entropy (PE) and the functional connectivity of the weighted phase lag index (wPLI) were measured for the EEG at different time points (baseline, T1, T2 and T3). For patients, the PE and wPLI at each time point were calculated by averaging the indexes within the epochs.

Permutation entropy

PE is a quantitative complexity measure that explores the local order structure of a dynamical time series. It transforms given time series into series of ordinal patterns, each describing the order relation between the present and a fixed number of equidistant past values at a given time [28].

The scalar time series are given as {x(i) : 1 ≤ i ≤ N}. Firstly, the reconstruction time series is given as the following;

$$X_i=\{x(i),x(i+\tau ),\mathrm{...},x(i+(m-1)\tau )\},i=1,2,\mathrm{...}N-(m-1)\tau$$ (1)

where τ is time delay and m is the embedding dimension.

Then, rearrange X_i in an increasing order:

$$\{x(i+(j_1-1)\tau )\le x(i+(j_2-1)\tau )]\le \dots \le x(i+(j_m-1)\tau )\}$$ (2)

There are m! permutations for m dimensions. Each vector X_i can be mapped to one of the m! permutations.

Next the probability of the jth permutation occurring p_j can be defined as:

$$p_j=\frac{n_j}{{\displaystyle \sum _{j=1}^{m!}{n}_j}}$$ (3)

where n_j is the number of times the jth permutation occurs.

The permutation entropy of the time series {x(i) : 1 ≤ i ≤ N} is defined by

$$H_x(m)=-{\displaystyle \sum _{j=1}^{m!}{p}_j\ln p_j}$$ (4)

when the time series is random, the H_x (m) approaches its maximum value of ln(m!); when the time series is regular, the H_x (m) approaches to zero.

Finally, normalizing H_x (m) by diving ln(m!):

$$PE=\frac{H_x(m)}{\ln (m!)}$$ (5)

Weighted phase lag index

The wPLI is a conservative measure of phase synchronisation between electrodes [29]. It enables the analysis of the properties of phase synchronization without the deleterious impact of volume conduction. Here, we assumed that wPLI could be used to quantify the modulation effects of iTBS on neural dynamics reflected in the phase oscillatory activity of the scalp EEGs. The wPLI were computed for each EEG channel across the other channels based on the following equation:

$$\begin{array}{lll}{\text{wPLI}}_{i,j}\hfill & =\hfill & \frac{\left|E\left\{\Im \left\{X_i{X}_j^{*}\right\}\right\}\right|}{E\left\{\left|\Im \right\{X_i{X}_j^{*}\left\}\right|\right\}}\hfill \\ \hfill & =\hfill & \frac{\left|E\right\{\left|\Im \right\{X_i{X}_j^{*}\left\}\right|sgn[\Im \{X_i{X}_j^{*}\}]\left\}\right|}{E\left\{\right|\Im \left\{X_i{X}_j^{*}\right\}\left|\right\}}\hfill \end{array}$$ (6)

where i and j are channel indices, X_i is the time-frequency spectrum of channel i, $X_j^{*}$is the complex conjugate of X_j, $\Im \left\{X_i{X}_j^{*}\right\}$indicates an imaginary section of cross frequency spectra $\left\{X_i{X}_j^{*}\right\}$, E{.} is the expected value operator and sgn{.} is the sign function operator.

Statistic

A statistical analysis was performed using SPSS (version 17.0). The effect of the stimulation on consciousness state was analyzed based on the modification of the CRS-R total score, using the paired t-test. The modification of the brain activity was assessed by comparing EEG indexes between different time points in each stimulation session. The differences were checked by one-way repeated ANOVA. Post-hoc paired t-tests were used in the comparison of different time points: T1 vs. baseline, T2 vs. baseline, and T3 vs. baseline. Bonferroni correction was performed after multiple comparisons. The comparisons of PE values and functional connectivity at sensor level were conducted by the paired t-tests with FDR correction.

CRS-R

No significant difference (p > 0.05, paired t-test) of total CRS-R scores between before and after stimulation (Table 2), either in the real or sham sessions. Some patients showed fluctuation of total CRS-R scores, but none of the patients gained an improvement of consciousness state. Patient 6 and patient 16 showed a change in the arousal sub-scale, showing stable sign of slight spontaneous opening eyes without external stimulation (2 scores in arousal sub-scale) after real stimulation. None of the other patients displayed any valuable changes of CRS-R scores exceeding those in the previous one week’s diagnosis.

Spectrum

There was no significant difference of global average spectrum between T1, T2, T3 and baseline in the real stimulation. Topography indicated a decreasing of delta power and an increasing of theta power (Figure 2A). There was no significance of the frontal average delta power among the different time points (F_3,71 = 3.45, p>0.05). No significance in the pairwise comparisons of the frontal average delta power. However, the frontal average delta power of T1 (p = 0.03) and T2 (p = 0.04) were significantly lower than that of baseline before multiple comparison correction (Figure 2B). Sham stimulation did not induce any significant changes of spectrum power of EEG.

Permutation entropy

PE values were measured at the baseline and the post-stimulation time points. Then, PE values at the T1, T2 and T3 were compared with that of the baseline for the electrodes. The comparisons indicated that electrodes of bilateral frontal and right parietal regions showed significantly different PE values (p<0.05, paired t-tests with FDR correction) between T1 and baseline (Figure 3A). PE values of electrodes at left hemisphere were significantly different between T2 and baseline. None of electrodes showed marked different PE values at T3 in comparison with baseline. Electrodes (FP1, AF3, AF7, F1, F3, F5, FC1, FC5 and C3) have significantly higher PE values (p<0.05, paired t-tests with FDR correction) between the T2 and the baseline (Figure 3B).

Figure 3C shows the boxplots of average PE values of the left frontal region (including FP1, AF3, AF7, F1, F3, F5, FC1, FC5 and C3) of the patients at different time points. There was significant difference between time points in the real stimulation (F_3,71 = 5.12, p = 0.008), but not in the sham stimulation (F_3,71 = 1.69, p > 0.05). Post-hoc t-tests showed a significant increase of the PE values at T2 than the baseline (p=0.003, paired t-tests with Bonferroni correction) after the real stimulation. The sham stimulation did not significantly alter the average PE values of the left frontal region.

Weighted phase lag index

The functional connectivity was indexed by wPLI and measured at the baseline and post-stimulation time points for the pairwise electrodes. Figure 4 shows the significantly strengthened connectivity by the real stimulation (p<0.05, paired t-tests with FDR correction). The significantly strengthened connectivity were mostly relevant with electrodes at frontal region. Consistently, the electrodes, which showed significant increasing of connectivity degree, were mainly located at bilateral frontal regions (lower panel of Figure 4) at T1 and T2.

Inter-hemispheric connectivity of frontal regions showed marked increasing at T1 and T2 in comparison with baseline (Figure 5A) after the real stimulation. The average strength of frontal inter-hemispheric connectivity showed significant difference in the time points of the real stimulation (F_3,71 = 17.33, p < 0.001), but not in the sham stimulation (F_3,71 = 0.89, p > 0.05). Post-hoc t-tests showed a significant increase at T1 (p=0.03, paired t-tests with Bonferroni correction) and T2 (p=0.008) after real stimulation than that at the baseline (Figure 5B). There was no significant difference between T3 and baseline in the real stimulation. The sham stimulation did not induce any significant change of the inter-hemispheric connectivity at the T1, T2 and T3 between that of the baseline (Figure 5C).

GOS-E

The functional outcomes of vegetative state patients at 6 months post-experience, as reflected in the GOS-E scores (Table 2), show that out of the total, 13 patients (72.2%) passed away, 4 patients (22.2%) remained in a vegetative state, and only 1 patient (5.6%) (P16) experienced severe disability, regained consciousness, yet lost the capacity for autonomous movement.