Motor and non-motor circuitry activation induced by subthalamic nucleus deep brain stimulation (STN DBS) in Parkinson’s disease patients: Intraoperative fMRI for DBS

Abstract

Objective

To test the hypothesis suggested by previous studies that subthalamic nucleus (STN) deep brain stimulation (DBS) in patients with PD would affect the activity of both motor and non-motor networks, we applied intraoperative fMRI to patients receiving DBS.

Patients and Methods

Ten patients receiving STN DBS for PD underwent intraoperative 1.5T fMRI during high frequency stimulation delivered via an external pulse generator. The study was conducted between the dates of January 1, 2013 and September 30, 2014.

Results

We observed blood oxygen level dependent (BOLD) signal changes (FDR<.001) in the motor circuitry, including primary motor, premotor, and supplementary motor cortices, thalamus, pedunculopontine nucleus (PPN), and cerebellum, as well as in the limbic circuitry, including cingulate and insular cortices. Activation of the motor network was observed also after applying a Bonferroni correction (p<.001) to our dataset, suggesting that, across subjects, BOLD changes in the motor circuitry are more consistent compared to those occurring in the non-motor network.

Conclusions

These findings support the modulatory role of STN DBS on the activity of motor and non-motor networks, and suggest complex mechanisms at the basis of the efficacy of this treatment modality. Furthermore, these results suggest that, across subjects, BOLD changes in the motor circuitry are more consistent compared to those occurring in the non-motor network. With further studies combining the use of real time intraoperative fMRI with clinical outcomes in patients treated with DBS, functional imaging techniques have the potential not only to elucidate the mechanisms of DBS functioning, but also to guide and assist in the surgical treatment of patients affected by movement and neuropsychiatric disorders.

Introduction

Deep brain stimulation (DBS) is an effective treatment option for movement disorders, including Parkinson disease (PD), essential tremor, and dystonia,^1,2 and its applications are expanding to other neurological and neuropsychiatric disorders such as epilepsy,³ chronic pain,⁴ obsessive-compulsive disorder,⁵ Tourette’s syndrome,⁶ and major depression.⁷

The main brain targets for DBS treatment of PD are the subthalamic nucleus (STN) and the globus pallidus internus.⁸ Although they share commonalities in efficacy, STN stimulation seems to be more commonly correlated with neuropsychological and behavioral alterations, even though this association is still subject to investigation.^8,9 These effects would be concordant with the studied basal ganglia connections, whereby STN has been shown to connect to motor, limbic, and associative networks.^10–13

Although a large number of patients affected by movement disorders have been effectively treated with DBS, the exact mechanisms that lead to this success are still elusive.^13,14 Because of the similar efficacy of DBS to lesion surgeries, the first hypothesis concerning the mechanism of DBS advocated the inhibition of the targeted area and consequent facilitation of the basal ganglia direct pathway.¹⁰ However, it is now established that DBS can function through more complex mechanisms such as antidromic and orthodromic activation of STN input and output regions, modulation of complex circuits, and normalization of STN neuronal activity.^13–17

In order to characterize DBS mechanisms, the functional connectivity of the STN as well as the brain regions that may mediate the clinical efficacy of this treatment are still subject to investigation. The cortical activity during STN DBS has been studied with electrophysiology and cortex excitability studies in rodent, non-human primate (NHP), and human subjects,^16,18–24 and with optogenetic techniques in rodents.²⁵

These studies are of fundamental importance in mapping the relationships and effects of STN activity on other areas related to motor, limbic, and associative functions, and target each component of these networks with high specificity and selectivity. However, because of the complex role and interactions of the basal ganglia, techniques that are able to globally assess brain function and activity, such as functional magnetic resonance imaging (fMRI) and positron emission tomography (PET), must also be implemented to elucidate the effects of DBS.²⁶ fMRI in particular is able to evaluate global brain activation with high temporal and spatial resolution.

To date, single-case or small group studies have utilized fMRI or PET imaging to investigate brain activity in PD patients undergoing STN DBS, reporting changes in the activity of sensorimotor cortex, premotor cortex, supplementary motor area, basal ganglia, thalamus, cerebellum, prefrontal cortex, cingulate gyrus, insular cortex, and brainstem.^27–32 To investigate the neural circuitry associated with basal ganglia stimulation, we previously developed a method for fMRI in swine³³ and NHP,³⁴ reporting increase activation in the sensorimotor associative and limbic networks by STN DBS.

In this study, we adopted an intraoperative 1.5T fMRI design and studied the blood oxygen level dependent (BOLD) response induced by STN DBS in ten patients affected by PD. We aimed to expand upon previous studies by translating the use of intraoperative fMRI during DBS to the clinical environment. With this development we tested the hypothesis supported by previous findings that STN DBS would lead to both motor and non-motor functional neural network activation.

Materials and Methods

Safety Testing

A phantom study was performed to assess radiofrequency (RF) induced heating at the DBS lead tips using the methods described by the American Society for Testing and Materials in ASTM F2182-11a.³⁵ For this study, a DBS electrode (Model 3387, Medtronic Inc, Minneapolis, MN) was placed in a polyacrylic gelled saline head-and-trunk phantom, designed to mimic the tissue heat transfer properties,³⁵ such that the orientation of the lead and extension wiring would replicate that of the electrode in a patient (Figure 1B). The electrode was connected to a percutaneous lead extension (Medtronic Model 3550, Minneapolis, MN) and a custom extension wire which extended outside the MR scan room to allow stimulation using an external device (DualScreen 3628 Medtronic. Medtronic, Inc., Minneapolis, MN).³⁶ During MRI scanning, temperatures at the DBS electrodes and inside the phantom were monitored using fluoroptic temperature probes placed on each of the most distal and proximal contacts of the DBS electrode as shown in Figure 1B (.8-mm tip, STF-2, Luxtron 750 system, Lumasens, Santa Clara, CA). Identical pulse sequences used in the patient studies were tested, including the gradient echo-echo planar imaging (GE-EPI) sequence and MPRAGE. The scanner specific absorption rate (SAR) predictions for the phantom study were calculated based on an assumed patient weight of 50kg, however, in the patient study patient-specific SAR values were based on individual patient weight.

Figure 1. Intraoperative fMRI setup and Phantom Testing.

A) Photograph of the intraoperative MRI suite. B) Schematic drawing of the anthropomorphic phantom in the MR bore illustrating placement extension wiring (purple), positioning of DBS electrode, and placement of temperature probes (red) on the proximal and distal DBS electrode contacts. C) Plot of change in temperature (°C) vs. time (sec) during a series of pulse sequences: A- MP-RAGE, stimulation on, wires along iso-line (SAR=.064W/kg; ΔT_max=.27±.01°C) C-GE-EPI, stimulation on, wires along iso-line (SAR=.016W/kg; ΔT_max=.12±.01°C); Temperature data shown were sampled at 1sec intervals smoothed using a 20-point running average.

Subjects

Ten subjects scheduled for STN DBS surgery for treatment of PD were recruited. Informed consent was obtained from all subjects prior to participation. All study procedures were approved by Mayo Clinic Institutional Review Board and were performed in compliance with Health Insurance Portability and Accountability Act guidelines. The study was registered at ClinicalTrials.gov (I.D. # NCT01809613). The study was completed between January 1, 2013 and September 30, 2014.

The group of patients included 5 males and 5 females, of a mean (±standard deviation, SD) age at surgery of 61.8 (±8.6) years. Mean (±SD) duration of disease before surgery was 10.2 (±4.4) years. UPDRS-III scores (range 0–108) were recorded: preoperatively, off and on levodopa; postoperatively off levodopa; at the last follow-up, off levodopa and with the stimulation on at optimized settings (Table 1). We report the optimal contact combination and stimulation settings at the latest follow-up programming session (Table 1). One patient (Patient 7) was referred from and performed programming at another institution and therefore we cannot provide related data (UPDRS-III scores, optimal stimulation settings).

Table 1.

Clinical patient information.

Patient (Age, Gender)	Disease Duration (Years)	Preoperative UPDRS-III score (Off)	Preoperative I UPDRS-III score (On)	Postoperative UPDRS-III score (Off)	Follow-up UPDRS-III score (Off)	Follow-up (months)	Change in score at last follow-up (percent change)	Optimal contacts at follow-up	Optimal settings at follow-up	Stimulation side during fMRI
Patient 1 (60Y, M)	8	41	27	40	24	6	41%	Left 1-C+ Right 9-C+	Left 3.15V 60us 130 Hz Right 1.90V 60us 130 Hz	Left
Patient 2 (70Y, F)	11	21	7	21	8	2	62%	Left 0-C+	Left 1.65V 60us 130Hz	Left
Patient 3 (46Y, M)	14	37	2	26	8	13	78%	Left 2-C+ Right 10-C+	Left 2.90V 60us 130Hz Right 2.90V 60us 130Hz	Left
Patient 4 (62Y, M)	10	32	18	21	16	16	50%	Left 2-C+ Right 9-10+	Left 3.40V 60us 130Hz Right 2.40V 60us 130Hz	Left
Patient 5 (78Y, F)	1	27	26	19	21	5	22%	Left 0+1-2-3+ Right 8+9-10-11+	Left 2.50V 60us 130Hz Right 2.40V 60us 130Hz	Left
Patient 6 (61Y, M)	12	21	5	26	19	11	10%	Left 3-C+ Right 11-C+	Left 3.70V 60us 130Hz Right 3.75V 60us 130Hz	Left
Patient 8 (54Y, F)	14	26	3	33	51	1	−96%	Left 3-C+ Right 11-10+	Left 2.00V 90us 130Hz Right 2.10V 90us 130Hz	Right
Patient 9 (57Y, F)	12	31	2	31	26	3	16%	Left 3-C+ Right 11-C+	Left 2.20V 90us 130Hz Right 2.20V 90us 130Hz	Right
Patient 10 (60Y, F)	16	35	16	34	28	2	20%	Left 3-C+ Right 11-C+	Left 2.40V 60us 130Hz Right 2.40V 60us 130Hz	Right
Total (N=9) mean [±SD]	11±4	30±7	12±10	28±7	22±13	7±5	22±50%

Abbreviations: SD, standard deviation; UPDRS, Unified Parkinson’s Disease Rating Scale.

Functional Imaging

During DBS lead implantation surgery, a percutaneous lead extension was connected to the DBS electrode and tunneled through the skin, a second percutaneous lead extension was connected and extended out of skin. The head frame was removed and the patient was moved to the intraoperative MRI (intra-OR-MRI) for the fMRI studies (Figure 1A). A custom extension wire was connected to the lead extension and then to a handheld pulse generator (DualScreen 3628 Medtronic. Medtronic, Inc., Minneapolis, MN) located outside the scan room.

Studies were conducted in a 1.5 T intra-OR-MRI scanner (General Electric Healthcare, Wakasha, WI, 16x, Signa software). For all sequences, a manufacturers standard transmit/receive RF head coil was used (GE Healthcare,1.5-T Quad Head Coil, model 46-28211862). fMRI was acquired using a two-dimensional GRE-EPI: TR/TE: 3000/27, flip angle: 90, FOV: 22 cm x 22 cm, matrix: 64 x 64, slice thickness 3.0 mm with 1.5 mm gap for 5 patients and slice thickness 3.5 mm with 0 mm gap thickness for the remaining 5 patients. For each acquisition, 135 volumes (the first 5 volumes were discarded for scanner equilibrium) were acquired using a block paradigm with five 6 second stimulation periods (two volumes) alternated with six 60 second rest periods (20 volumes) for a total scan time of 6 minutes 45 seconds per scan.³⁴ Initial DBS effect was tested during an awake DBS surgery procedure, testing 0(−)–3(+) contact stimulation with a handheld pulse generator prior to the fMRI scan. Patients showed initial symptom improvement in the operating room using 2V amplitude, 90μs pulse width, 130–185Hz frequency. Five patients underwent the fMRI study on the day of the pulse generator implantation under general anesthesia; the other five patients on the day of lead implantation conducting awake fMRI. Following the fMRI, the second externalized percutaneous lead extension was removed.

Data Analysis and Statistics

The fMRI data were subjected to a standard pre-processing steps implemented in Brain Voyager QX software (Maastricht, the Netherlands), including slice scan time correction, 3D motion correction, temporal filtering (High-pass: Fourier basis set- 5 cycles, and low-pass: Gaussian filter-FWHM 3.1 sec), and spatial smoothing (Gaussian filter with FWHM: 2.0 pixel size). FMRI data were then co-registered to the anatomical MP-RAGE images using corresponding points-based alignment and normalized to the Talairach brain coordinate system. For right-sided stimulation data was mirrored to corresponding voxels on the left side of the brain in order to be analyzed together with left-sided stimulation data. Functional activation maps (t-maps) were generated using a double-gamma hemodynamic response function (onset 0 s, time to response peak 5 s, time to undershoot peak 15 s) representing the block design for each voxel. Group analyses were computed using a fixed effects analysis to concatenate the data from all subjects after registration of voxels in Talairach space and thus integrate the data from multiple subjects into a single general linear model (GLM) analysis (Brain masking applied to reduce the number of voxels for GLM).

To correct for multiple comparisons and exclude false positive voxels, we applied False Discovery Rate (FDR<.001). In addition to and separate from the FDR, we applied the Bonferroni correction (p<.001) to the original data. The brain areas that survived Bonferroni correction are listed in Table 2. While FDR are widely accepted correction method, Bonferroni correction allows for a more stringent analysis than FDR, in that it considers the number of multiple comparisons, and based on this number corrects the p-value required for each voxel to reach the desired significance level.³⁷

Table 2.

Areas of significant brain activation: regions of interest and corresponding Brodmann area, function, t-score and Talairach coordinates of the voxel with the peak t-score, size. Areas with increased BOLD-signal after Bonferroni correction (p<.001) are highlighted.

Anatomical Location	Functional Location	BA	Talairach Coordinates (x, y, z)	Size (mm3)	Max t-Score
Premotor cortex, Supplementary motor area (I)	Motor	6	−11, −27, 68	10138	8.82*
Primary motor cortex (I)	Motor	4	−10, −27, 68	5991	8.75*
Precuneus (I)	Somatosensory association	7	2, −83, 33	5464	8.25*
Occipital (C)	Visual	18,19	3, −83, 33	7986	8.21*
Posterior cingulate (I)	Limbic	30	−4, −64, 14	3506	8.02*
Precuneus (C)	Somatosensory association	7	3, −67, 51	3679	7.76*
Occipital (I)	Visual	17,19	−14, −81, 12	10164	7.65*
Primary Somatosensory cortex (I)	Somatosensory	3	−11, −40, 63	4859	7.40*
Cingulate gyrus (I)	Limbic		−4, −25, 52	3850	6.79*
Premotor cortex (C)	Motor	6	34, 6, 46	2274	6.67*
Cerebellum (C)	Motor		3, −52, −5	903	6.46*
Pedunculopontine nucleus (I)	Motor		−11, −28, −9	1617	6.11*
Parahippocampal gyrus (I)	Limbic	28,35,36	−16, −12, −25	888	6.07
Caudate (I)	Motor		−25, −12, 35	121	5.79
Supramarginal gyrus (C)	Visuospatial	40	35, −47, 38	1383	5.77
Insula (C)	Limbic	13	33, −18, 26	399	5.39
Thalamus (I)	Motor		−10, −10, 9	528	5.34
Inferior frontal gyrus (I)	Language	45,47	−56, 8, −1	990	5.25
Superior parietal lobule (I)	Somatosensory association	7	−32, −60, 53	349	5.24
Posterior cingulate (C)	Limbic	29	3, −51, 12	709	5.23
Middle frontal gyrus (I)	Executive	10	−34, 37, 16	1021	5.03
Superior parietal lobule (C)	Somatosensory association	7	21, −59, 40	640	5.03
Cerebellum (I)	Motor		−21, −32, −23	80	4.90
Anterior cingulate (I)	Limbic	32	−8, 37, 19	355	4.89
Parahippocampal gyrus (C)	Limbic	36	24, −44, −17/41, −36, −6	80/40	4.87/−4.50
Inferior temporal gyrus (I)	Language, Visual	20,37	−64, −42, −23	574	4.86
Angular gyrus (I)	Language	39	−31, −64, 23	145	4.57
Middle temporal gyrus (I)	Auditory, Language	21	−45, −5, −16	73	4.55
Cingulate gyrus (C)	Limbic	31	12, −41, 34	111	4.54
Pons (C)			6, −34, −21	24	4.50
Postcentral gyrus (C)	Language,	3,43	51, −17, 22	53	4.44
Angular gyrus (C)	Visuospatial	39	36, −45, 7	42	4.40
Primary motor cortex (C)	Motor	4	17, −23, 50	25	4.26
Middle temporal gyrus (C)	Auditory	22	49, −44, −1	25	4.22
Supramarginal gyrus (I)	Language	40	−36, −54, 28/−56, −44, 20	29/110	4.20/−5.22
Orbitofrontal (C)	Associative	11	22, 41, −15	5	4.20
Inferior frontal gyrus (C)	Language	44	50, 0, 12	13	4.18
Superior temporal gyrus (C)	Cognition, Limbic	38	48, 13, −2	4	4.10
Dorsolateral prefrontal cortex (I)	Executive	46	−40, 45, 8	4	4.09
Fusiform gyrus(C)	Visual	37	34, −58, −5	8	4.08
Superior temporal gyrus (I)	Language	22	−43, −37, −1	46	−4.53

Abbreviations: BA: Brodmann Area; C: Contralateral; I: Ipsilateral

^*Areas showing with Bonferroni correction P < .001

Cortex Based Alignment

A cortex based alignment was performed, segmenting the gray/white matter boundary of the hemispheres using an automatic segmentation process based on intensity histograms. The cortices from individual subjects were then aligned through transformation into spherical representations and non-rigid alignment to the group average in an iterative coarse-to-fine process. After cortex-based alignment, group analysis as described above was repeated in this new standard surface space. To exclude false positive voxels and correct for multiple comparisons, we only considered surface voxels with a significance level FDR<.001.

Results

Phantom Testing

Prior to application in patients, heat induction at the DBS electrode contacts was measured in a head and trunk phantom during GE-EPI and 3D MP-RAGE sequences. Both sequences generated heating well below 1°C. Expectedly, the RF-heating of the DBS lead tips during the GRE-EPI fMRI sequence (SAR=.016W/kg; ΔT=.12±.01°C; Figure 1C) was lower than that during the 3D MP-RAGE sequence (SAR=.064W/kg; ΔT=.27±.01°C; Figure 1C). Neither GRE-EPI nor MP-RAGE sequence produced predicted SAR values in excess of the vendor-specified .1W/kg safety SAR threshold to prevent adverse effects during MRI of patients with an implanted DBS system.³⁸

STN DBS evokes BOLD signal activation in motor and non-motor circuitry

To characterize the circuitry affected by STN DBS (2V 185Hz 90μs pulse width in 10 subjects, 2V 130Hz 90 μs) in patients with PD, 10 patients underwent intraoperative fMRI during DBS. BOLD signal activation was detected in several regions of the sensorimotor network, including ipsilateral primary motor and somatosensory cortices, bilateral premotor cortex, ipsilateral supplementary motor area (SMA), thalamus (dorsal medial nucleus), caudate, and peduncolopontine nucleus, and bilateral cerebellum (Figure 2A). Maximum t-scores were detected within ipsilateral premotor cortex/SMA (t=8.82) and primary motor cortex (t=8.75). Furthermore, activation was detected in limbic and associative areas, and in the bilateral precuneus and cuneus regions was observed (Figure 2A; Table 2).

Figure 2. BOLD signal activation with STN DBS (2V 130–185Hz 90μs) for PD.

A) Areas of activation with unilateral STN stimulation at 2V 130–185Hz 90μs (n = 10) for PD. Slice Locations are presented in Talairach coordinates. Significant activation (FDR<.001) was observed in bilateral premotor and primary motor cortices, precuneus, occipital lobes, cerebellum, and anterior and posterior cingulate. Activation of ipsilateral thalamus, pedunculopontine nucleus, parahippocampal gyrus, and hippocampus, and contralateral insula were also observed. B) The average time courses for five regions of interest were plotted as average percent change in BOLD signal from baseline vs. time (one scan is equal to TR = 3 seconds) using ten frames (30 seconds) prior to stimulation (yellow box) as the baseline.

To study the time course of DBS-evoked fMRI BOLD activation in each of three functionally-defined ROIs (SMA, premotor/primary motor and somatosensory association), an event-related averaging analysis was performed. The resulting time courses revealed peak BOLD signal percent change between .25–.40% in each of the ROIs within 5 frames (15 seconds) of stimulation onset (Figure 2B).

Cortex Based Alignment

A cortex-based alignment was performed to better resolve the overlapping cortical ROIs in the midline. This alignment revealed more distinct regions of activation in ipsilateral SMA, primary motor cortex, premotor cortex, primary somatosensory cortex, anterior cingulate cortex, thalamus, PPN and visual cortex. Contralateral posterior cingulate cortex, precuneus, and visual cortex showed activation (Figure 3).

Figure 3. Sensorimotor BOLD signal activation with cortex-based alignment.

Cortical areas of significant BOLD activation resolved by cortex based analysis projected on inflated representations of the dorsal (A) and medial (B) surfaces of the brain. Areas of activation included: bilateral supplementary motor and occipital lobes; ipsilateral primary motor, and primary and secondary somatosensory cortices, thalamus, anterior cingulate gyrus, and pedunculopontine nucleus, and contralateral precuneus (FDR <.001).

Anesthesia vs. Awake State

To assess the effect of anesthesia on the patterns of DBS-induced fMRI activation, patients were sub-divided into those who were studied under anesthesia (n=5) versus those who were studied awake (n=5). Interestingly, as shown in Figure 4, the areas of activation were similar regardless of anesthesia state, but the signal strength was much stronger and in the awake state, significant at the FDR<.01 in the awake state as compared to the FDR<.05 level in the anesthesia state (Figure 4B).

Figure 4. Awake vs Anesthetized.

Comparison of BOLD activation during DBS conducted in the anesthetized (n=5) vs. awake state (n=5). The signal strength was much stronger and in the awake state, significant at the FDR<.01 as compared to FDR<.05 level in the anesthetized state.

Discussion

In this study, we aimed to implement intra-OR-fMRI for the investigation of DBS mechanisms in PD patients. To date, only a few groups have performed fMRI during STN DBS, due to the potential risks associated with MRI acquisition in patients with DBS implants, which include local tissue heating in the brain.³⁹

MRI safety during DBS

There is large variability between measured RF-heating reported by different groups,^39–41 suggesting lack of generalizability of these findings and underscoring the importance of performing site-specific safety testing experiments.³⁶ The present study was conducted in accordance with the manufacturer’s DBS-MRI safety guidelines.⁴² We performed a phantom test reporting maximum temperature increases below the 1°C safety threshold for this study, which is consistent with our previous large animal in vivo safety study report.⁴⁰ Average head SAR values of less than 0.02W/kg were recorded during the fMRI study in all the patients, and an MR physicist with expertise in MRI for patients with implanted electronic devices was present during all sessions.

fMRI for DBS

Previous STN DBS functional network studies strongly suggest that STN DBS exerts distributed effects throughout motor network structures, including premotor and motor cortices, thalamus, and contralateral cerebellum. However, reports of DBS-evoked modulation of non-motor networks have been somewhat variable, likely due to differences between studies in patient numbers and disease-state characteristics, as well as experimental methods.⁴³

Jech et al. were the first to report the network effects of STN DBS using fMRI in three patients.²⁸ Stimulation-evoked BOLD signal increases were observed in ipsilateral motor areas including thalamus, globus pallidus, and premotor cortex. A similarly-designed experiment in a single subject corroborated these findings,²⁹ showing BOLD increases in motor regions (premotor, SMA, primary motor cortex, cerebellum, putamen, and thalamus), while also detecting changes in non-motor areas (mediodorsal thalamus, parietal lobe, bilateral parahippocampal gyri, posterior cingulate). In a series of five patients, STN DBS-evoked motor activation (ipsilateral basal ganglia and motor cortex, and contralateral cerebellum) with 3T fMRI, in line with previous studies.³² The effect of DBS on non-motor, in addition to motor networks, corroborates the results of previous PET studies, which have reported DBS-evoked activations in primary motor cortex, SMA, premotor cortex, prefrontal cortex, cerebellum, thalamus, basal ganglia, and cingulate cortex.^44–49

The effect of DBS on resting state fMRI networks has also been examining the effective connectivity in twelve PD patients.¹⁷ Their findings support the role of STN stimulation in modulating the basal ganglia direct and indirect pathways, the hyperdirect pathway between the STN, the cortex, the cortico-striatal and thalamo-cortical pathways. Together, these studies suggest that the clinically effective stimulation was found to alter inter-regional coupling within the motor cortico-striato-thalamo-cortical loop, concordant with the current hypothesis of DBS functioning.^{13–15,17,50,51}

Our findings, both in large animals^33,34 and in the present study, reports STN stimulation affects in both ipsilateral and contralateral motor cortices, a phenomenon previously reported in clinical,⁵² and partially explainable by antidromic activation of STN bilateral afferents from the motor cortices. Stimulation-induced antidromic and orthodromic activation of regions functionally connected to the STN is legitimated also by results from other groups adopting neurophysiological and functional imaging techniques, supporting the hypothesis that modulation of the STN activity may lead to normalization of neural activity in areas such as the primary motor, premotor, and SMA, that are known to be hypo-functioning in PD patients.^{16,46,53–56}

In our study, to further test the statistical significance of the observed motor network activation, we applied a more stringent Bonferroni correction (<0.001). The BOLD signal increase was still significant mainly in the motor network including ipsilateral primary motor, premotor, SMA, PPN, cingulate gyrus, and in the contralateral cerebellum. However, we did not observe equally preserved non-motor network activation, indicating that the effect of STN DBS on these areas presents higher inter-subject variability. In addition, to minimize the effect that the inter-subject anatomical variability in gyri and sulci has on normalization in multi-subject analysis, we performed a cortex-based alignment. This approach confirmed segregated activation in the primary motor and premotor cortices, SMA, and primary somatosensory areas by STN DBS.

In addition to cortical activity, the PPN is one of the areas detected in both our human and large animal series.^33,34 The PPN is known to have an important role in motor functions and gait control, and to have reciprocal connections with the STN.⁵⁷ These elements suggest that the PPN may be an important mediator of DBS STN efficacy, and support the exploration of this site as a possible stimulation target for motor disorders.^58–60

Our findings from ten patients are consistent not only with the ones from our animal studies, but also with results from previous human case reports and series, even though characterized by different experimental settings. Importantly, the ROIs detected in our series include all of those that were partially reported in the described previous studies. Functional circuitry effect of DBS might vary as a function of targeting accuracy, individual disease-state variance, and stimulation parameter programming.⁶¹ Since most of the ROIs that we detected in our study have been sparsely described in previous reports, we believe that the larger population number increased the power of the analysis, bringing many ROIs to higher significance levels.

In addition to the prominent motor circuitry activation, activation of cognitive and emotional circuitry was observed, including ipsilateral dorsal anterior cingulate, orbitofrontal, and insula, and contralateral parahippocampus. This is consistent in patients with more ventral STN DBS,²⁹ as well as with the anatomical connectivity of ventromedial STN.⁶² This activation may be of particular interest in light of several reports of adverse cognitive and emotional effects, reporting decrease verbal fluency, worsening of apathy and thought disorders, with reports of suicidality and hypomania.^62–67 Finally, insula shown to be affected by STN DBS, is being revisited as the central hub for processing relevant information related to the state of the body as well as cognitive and mood states in PD.⁶⁸

Limitations

FMRI acquisition on the day of electrode implantation (n=5) may have been confounded by micro-lesion effects.⁶⁹ Additionally, as seen in Supplementary Figure 1, while the DBS electrode produced minimal artifact, the extension wire connector produced prominent susceptibility artifact. This might have created signal loss in the temporal lobe, ventral sensory cortices and parietal lobe. Finally, our study is limited to seeing short-term effect of DBS, as there are possible long-term synaptic plasticity effects reported.⁷⁰

Conclusions

Functional neuroimaging, including fMRI and PET, has an important role in studying DBS mechanisms, due to its wide clinical availability and ability to assess global functional neural activity. While PET provides molecular specificity and safe application in the context of implanted electrical devices, its disadvantage is the limited flexibility in experimental designs, due to the radioisotope washout time and the consequent limit of testing limited number of DBS parameter per day. FMRI, on the other hand, allows experimenters to test differences among varying DBS stimulating contacts and stimulation parameters in a single relatively short session.

This study has demonstrated BOLD signal activation of both motor and non-motor circuitry, whereas the motor network activation showed more consistency throughout the subjects compared to non-motor network activity with STN DBS in PD using intra-OR-fMRI. This technique allows for a relatively quick scan time during which stimulation parameters can be dynamically manipulated, and therefore holds potential interest to provide insight into the therapeutic outcome and/or adverse effects of during DBS surgery. This may become increasingly important as DBS indications continually expand to include treatment of non-motor disorders, in which therapeutic outcome is more subjective and difficult to measure.

Supplementary Material

supplement. Supplementary Figure 1. EPI distortion due to the DBS electrode and Extension connector.

DBS electrode image distortion. (A) gradient echo (GRE) echo-planar imaging (EPI) raw image showing DBS electrode distortion. The DBS electrode creates a susceptibility artifact of ~5 mm or less in diameter (1.4 voxel). (B) The extension wire connector (Medtronic DBS Extension Model 37086), connecting the DBS lead to the IPG, that is tunneled under the skin on the back side of head is creating a greater susceptibility artifact of ~ 30 mm in diameter (8.6 voxels)

Abbreviations

DBS

deep brain stimulation

PD

Parkinson’s disease

STN

subthalamic nucleus

PPN

pedunculopontine nucleus

NHP

non-human primate

PET

positron emission tomography

fMRI

functional magnetic resonance imaging

GE-EPI

gradient echo-echo planar imaging

RF

radio frequency

SAR

specific absorption rate

FWHM

full width at half maximum

SMA

supplementary motor area