Altered amplitude of low‐frequency fluctuations and regional homogeneity in drug‐resistant epilepsy patients with vagal nerve stimulators under different current intensity

Abstract Background The mechanisms of vagal nerve stimulation (VNS) for the treatment of drug‐resistant epilepsy (DRE) remain unclear. This study aimed to measure spontaneous brain activity changes caused by VNS in DRE patients using resting‐state functional MRI (rs‐fMRI). Methods The rs‐fMRI scans were performed in 16 DRE patients who underwent VNS surgery. Amplitude of low‐frequency fluctuations (ALFF) and regional homogeneity (ReHo) was generated and examined using paired sample t‐test to compare activity changes at different current intensity stage. The preoperative and postoperative ALFF/ReHo were also compared in eight responders (≥50% reduction of seizure frequency three months after surgery) and eight nonresponders using paired sample t‐test. Results The significant ALFF and ReHo changes were shown in various cortical/subcortical structures in patients under different current intensity. After three months of stimulation, responders exhibited increased ALFF in the right middle cingulate gyrus, left parahippocampal gyrus, and left cerebellum, and increased ReHo in the right postcentral gyrus, left precuneus, left postcentral gyrus, right superior parietal gyrus, right precentral gyrus, and right superior frontal gyrus. Nonresponders exhibited decreased ALFF in the left temporal lobe and right cerebellum, increased ALFF in bilateral brainstem, decreased ReHo in bilateral lingual gyri, and increased ReHo in the right middle frontal gyrus and right anterior cingulate gyrus. Conclusions The spontaneous neural activity changes in DRE patients caused by VNS were in an ongoing process. Increased ALFF/ReHo in frontal cortex, cingulate gyri, precentral/postcentral gyri, parahippocampal gyri, precuneus, parietal cortex, and cerebellum may implicate in VNS‐induced improvement in seizure frequency.


| INTRODUC TI ON
Vagal nerve stimulation (VNS) was approved in 1997 by the US Food and Drug Administration for the treatment of drug-resistant epilepsy (DRE) patients with partial seizures who are ≥12 years of age. 1 Then, the age at implantation was extended to patients ≥4 years of age in 2017. 1 In China, VNS was reported to be viable for DRE patients between 6 and 60 years of age. 2 Since the first operation in 1988, more than 100 000 VNS stimulators have been implanted around the world. 3 VNS has also been used for patients with various seizure types and epilepsy syndromes, including in children <4 years old. 4 As an adjunctive therapy, numerous studies have reported that VNS can effectively reduce seizure frequency. 4 After either two or three-five years of stimulation, approximately 50% and 60% of epilepsy patients, respectively, were reported to achieve a ≥50% seizure reduction. 5,6 Commonly employed stimulation parameters are as follows: between 1.5 and 2.25 mA, 20-30 Hz, 250-500 μs, on time 30 s, off time 3-5 min, which mainly based on doctors' experience. 7 It has been reported that the current intensity is an important factor of VNS outcome, and VNS can cause seizure reduction in a current range of 0.25-2 mA. 8,9 A study on the relationship between clinical outcome and current intensity suggested that the high intensity of stimulation corresponds to a better outcome, although some other authors considered no significant correlation between them. 10 In order to avoid laryngeal complications which started to appear when patients reached the output current of 1 mA, the device designed to produce electrical stimulation up to 3.5 mA was usually used at levels 1-2 mA. [11][12][13] Until now, owing to unclear therapeutic mechanisms, no surgical prediction criteria or parameter schedule has been proposed.
Resting-state functional MRI (rs-fMRI) was first proposed by Biswal to explore brain activity in 1995. 14 The amplitude of low-frequency fluctuation (ALFF) is a measure of rs-fMRI which calculates the amplitude of each voxel in local brain regions in the frequency range of 0.01-0.08 Hz. 15 And the regional homogeneity (ReHo) is another measure to evaluate the similarity of the time series of a given voxel to those of its nearest neighbors in a voxel-wise way based on Kendall's coefficient concordance (KCC). 16 Both above two methods are commonly used in the rs-fMRI study on spontaneous regional brain activity. In this study, we hypothesized that the increased current intensity participated in the process of seizure frequency reduction and selected ALFF and ReHo to analyze the changes of aberrant intrinsic brain activity in DRE patients with vagal nerve stimulators. 16  were enrolled in this study. All patients were without other neurological disorders and previous surgical history, and underwent a detailed preoperative consultation in Beijing epilepsy consultation center.

| Clinical data
Considering the long-term (interictal and ictal) video-EEG, standard MRI, positron emission tomography, magnetoencephalography, and clinical manifestations, an optimal VNS surgery plan was worked out.
The seizure types, epilepsy types, and epilepsy syndromes were classified following the 2017 new International League Against Epilepsy Classification of the Epilepsies. The primary outcome measure was response rate defined using the equation: (seizures/ month on VNS -baseline seizures/month)/(baseline seizures/ month) × 100%. The baseline period was three months before surgery and VNS period at every current intensity stage was one month.
Responders were defined as those experiencing a seizure frequency reduction of ≥50% compared with the mean seizure frequency before implantation.
The VNS implantations were performed by two neurosurgeons according to standard procedures. 17 Then, all patients followed an uniform VNS parameter schedule: Continuous electrical stimulation was commenced with 0.5 mA (current intensity), 30 Hz (signal frequency), 250 μs (pulse width), 30 s/5 minutes (on time/ off time) two days following the implantation; the current was gradually increased by 0.5 mA at standard intervals (every month) until it reached 1.5 mA intensity. So, when patients were followed up every month, the stimulation parameter was 0.5 mA, 30 Hz, 250 μs, 30 s/5 minutes at one month after surgery, 1.0 mA, 30 Hz, 250 μs, 30 s/ 5 minutes at two months after surgery, and 1.5 mA, 30 Hz, 250 μs, 30 s/ 5 minutes at three months after surgery. All subjects signed the informed consent.

| Image processing and statistical analysis
The rs-fMRI scans with a transmit/receive head coil were performed at one day before operation and one, two, three months after operation in 16 patients with DRE, respectively. There were no changes in the types or dosages of antiepileptic drugs during the time between preoperative and postoperative scan.
The process for the rs-fMRI scan followed the VNS Therapy Manual (VNS Therapy Physicians' Manual, LivaNova, Inc). The output current settings of the device were adjusted to 0 mA. The MRI examination was performed using a 3.0T scanner (Philips, Achieva TX). The BOLD fMRI sequence was single-shot echo-planar imag-

| Patients' characteristics
There were ten males and six females (age range: 5-41 years).

| Changes of ALFF in 16 patients with different current intensity
Amplitude of low-frequency fluctuations was generated from 16 patients with different VNS current intensity before and during 3 months after the operation (0, 0.5, 1.0, 1.5 mA). A paired sample t-test was used to compare the ALFF data between groups.
The brain regions with significant differences were displayed on standard brain templates (P < .001, GRF correction) ( Figure 1A-C).

| Changes of ReHo in 16 patients with different current intensity
ReHo was generated from 16 patients with different VNS current intensity before and during 3 months after the operation (0, 0.5, 1.0, 1.5 mA). A comparison of ReHo data before and after the operation was performed using a paired sample t-test.
The brain regions with significant differences were displayed on standard brain templates (P < .001, GRF correction) (

| Changes in ALFF before and after the operation in responders and nonresponders
Amplitude of low-frequency fluctuations was generated from 16 patients with a VNS stimulator who underwent an rs-fMRI scan before and at three months after the operation. Patients were divided into two groups: responders (n = 8) and nonresponders (n = 8). The ALFF generated from each group was analyzed before and after the operation. A paired sample t-test was used to compare the ALFF data before and after the operation.
There was significant difference in ALFF in various brain regions using the standard brain templates (P < .001, GRF correction). After three months of stimulation, the ALFF values increased in the right middle cingulate gyrus, left parahippocampal gyrus, and left cerebellum in responders ( Figure 3A), while ALFF decreased in the left temporal lobe and right cerebellum, and increased in the bilateral brainstem in nonresponders ( Figure 3B). The brain areas with significantly different ALFF values before and after the operation are listed in Table 4.

| Changes in ReHo before and after the operation in responders and nonresponders
ReHo was generated from 16 patients with a VNS stimulator who underwent an rs-fMRI scan before and at 3 months after the operation. A comparison of ReHo data before and after the operation was performed on each group with imaging data using a paired sample t-test (P < .001, GRF correction). The brain regions with significant differences were displayed on standard brain templates.  Table 5.

| D ISCUSS I ON
DRE was defined as that more than four seizures per month occurred after more than two years of regular antiepileptic drugs and reaching the maximum dose that patients can tolerate. 19 Despite rapid advances in effective drug treatments and surgical techniques, 15%-40% of epilepsy patients remain unable to effectively control seizures. 20 Neuromodulation is an optimal treatment for patients for whom resection surgery is unsuitable and can effectively reduce the frequency and severity of seizures. 21 With more than 20 years of development, VNS has become the most widely used operation for patients with DRE. 22 More than 50% of patients can achieve ≥50% F I G U R E 2 Changes of ReHo in 16 patients with different current intensity. The ReHo was generated from 16 patients with different VNS current intensity before and during three months after the operation (0, 0.5, 1.0, 1.5 mA). A paired sample t-test was used to compare the ReHo data between groups. The brain regions with significant differences were displayed on standard brain templates (P < .001, GRF correction Numerous studies have examined the changes in the brain after VNS. The vagal-locus coeruleus-hippocampal noradrenergic pathway is thought to represent a key mechanism in VNS. 24 For example, VNS was reported to cause significant changes in blood flow in the thalamus, hypothalamus, hippocampus, and amygdala. 25 VNS may play an antiepileptic role by altering synaptic activity in the thalamus. 26 Imaging studies have also shown that VNS can cause changes in activity in the insula, amygdala, hippocampus, parahippocampus, thalamus, and cerebellum. 27 Further, thalamic activation measured by BOLD fMRI was associated with improved VNS treatment response in patients with seizures. 28 Thalamic connections to the anterior cingulate and insular cortices are stronger in VNS responders. 29

F I G U R E 3
Changes of ALFF before and after operation in responders and nonresponders. A, In responders after three months of stimulation, the ALFF values increased in right middle cingulate gyrus, left parahippocampal gyrus, and left cerebellum in responders using the standard brain templates (P < .001, GRF correction). B, In nonresponders after three months of stimulation, ALFF decreased in the left temporal lobe and right cerebellum, and increased in the bilateral brainstem in nonresponders using the standard brain templates (P < .001, GRF correction)

TA B L E 4 Brain areas with significantly different ALFF values before and after surgery (p t-test, Bonferroni correction)
Further, the default mode network, salience network, and ascending reticular activating system were reported to have close associations with the generation and propagation of epileptic activity. 30,31 The ALFF and ReHo provide useful tools in rs-fMRI for the study of epilepsy. 32 In the present study, we defined the responder rate as ≥50% seizure frequency improvement from baseline. 33 There were no changes in the types or dosages of antiepileptic drugs during the time between two rs-fMRI scans. ALFF and ReHo were chosen as methods to analyze the characteristics of slow wave oscillations and functional activity in local brain regions.
This rs-fMRI study was made to explore aberrant intrinsic brain activity in epilepsy patients with different current intensity. Our   [34][35][36] In addition, our findings also support that 3.0T of the present study are as follows; relatively small case number, short stimulation period, and no analysis of epileptic network connections. Additionally, there was no wash-out period before new stimulation period as these patients have frequent seizures. It is worth to make future study with larger number of cases and longer VNS duration to explore the activity changes of brain networks caused by VNS in epilepsy patients.

This work was supported by the Beijing Postdoctoral Research
Foundation (ZZ2019-15).

CO N FLI C T O F I NTE R E S T
The authors declare no financial or commercial conflict of interest.