A review of functional neuroimaging studies of vagus nerve stimulation (VNS)

Journal of Psychiatric Research 37 (2003) 443–455 www.elsevier.com/locate/jpsychires

Review

A review of functional neuroimaging studies of vagus nerve stimulation (VNS) Jeong-Ho Chaea,b, Ziad Nahasa, Mikhail Lomareva,c, Stewart Denslowa, Jeffrey P. Lorberbauma, Daryl E. Bohninga, Mark S. Georgea,d,* a

The Center for Advanced Imaging Research and Brain Stimulation Laboratory, Medical University of South Carolina, 171 Ashley Avenue, Charleston, SC 29425, USA b Department of Psychiatry, The Catholic University of Korea, Seoul, South Korea c Institute of the Human Brain, St. Petersburg, Russia d The Ralph H. Johnson Veterans Hospital, Charleston, South Carolina, USA Received 9 May 2002; received in revised form 30 April 2003; accepted 9 May 2003

Abstract Vagus nerve stimulation (VNS) is a new method for preventing and treating seizures, and shows promise as a potential new antidepressant. The mechanisms of action of VNS are still unknown, although the afferent direct and secondary connections of the vagus nerve are well established and are the most likely route of VNS brain effects. Over the past several years, many groups have used functional brain imaging to better understand VNS effects on the brain. Since these studies differ somewhat in their methodologies, findings and conclusions, at first glance, this literature may appear inconsistent. Although disagreement exists regarding the specific locations and the direction of brain activation, the differences across studies are largely due to different methods, and the results are not entirely inconsistent. We provide an overview of these functional imaging studies of VNS. PET (positron emission tomography) and SPECT (single photon emission computed tomography) studies have implicated several brain areas affected by VNS, without being able to define the key structures consistently and immediately activated by VNS. BOLD (blood oxygen level dependent) fMRI (functional magnetic resonance imaging), with its relatively high spatio-temporal resolution, performed during VNS, can reveal the location and level of the brain’s immediate response to VNS. As a whole, these studies demonstrate that VNS causes immediate and longer-term changes in brain regions with vagus innervations and which have been implicated in neuropsychiatric disorders. These include the thalamus, cerebellum, orbitofrontal cortex, limbic system, hypothalamus, and medulla. Functional neuroimaging studies have the potential to provide greater insight into the brain circuitry behind the activity of VNS. # 2003 Elsevier Ltd. All rights reserved. Keywords: Vagus nerve stimulation (VNS); Functional neuroimaging; Positron emission tomography (PET); Single-photon emission computed tomography (SPECT); Functional magnetic resonance imaging (fMRI)

Contents 1. Introduction ............................................................................................................................................................................... 444 1.1. Vagus nerve stimulation .................................................................................................................................................... 444 1.2. The role of neuroimaging in understanding VNS effects...................................................................................................444 2. Positron emission tomography (PET) ........................................................................................................................................ 447 2.1. Epilepsy ............................................................................................................................................................................. 447 2.2. Depression ......................................................................................................................................................................... 448

* Corresponding author. Present address: Brain Stimulation Laboratory, Institute of Psychiatry, 67 President Street, Room 502 North, Charleston, SC 29425, USA. Tel.: +1-843-876-5142. E-mail address: [email protected] (M.S. George). 0022-3956/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0022-3956(03)00074-8

444

J.-H. Chae et al. / Journal of Psychiatric Research 37 (2003) 443–455

3. Single photon emission computed tomography (SPECT) ..........................................................................................................448 3.1. Epilepsy ............................................................................................................................................................................. 448 3.2. Depression ......................................................................................................................................................................... 449 4. Blood oxygen level dependent functional MRI (BOLD fMRI) .................................................................................................449 4.1. General consideration........................................................................................................................................................ 449 4.2. Epilepsy ............................................................................................................................................................................. 449 4.3. Depression ......................................................................................................................................................................... 450 5. Brain regions implicated in the neuroimaging data of VNS ......................................................................................................450 6. Conclusions ................................................................................................................................................................................ 452 Acknowledgements.......................................................................................................................................................................... 454 References ....................................................................................................................................................................................... 454

1. Introduction 1.1. Vagus nerve stimulation Although scientists have long been interested in whether and how stimulation of cranial nerves might produce changes in higher cortex (Bailey & Bremer, 1938; Dell & Olson, 1951; MacLean, 1990), it was not until the mid-1980s that electrical stimulation of the vagus nerve was developed as a potential therapy (Zabara, 1985a,b; for review see George et al., 2000c). In 1985, the ability of vagus nerve stimulation (VNS) to abort seizures was proven in canine studies (Zabara, 1985a,b). Although the general term of ‘VNS’ refers to any technique used to stimulate the vagus nerve, for practically all recent human studies, VNS refers to the stimulation of the left cervical vagus nerve using a commercially available implantable device, called the VNS Therapy System (Cyberonics, Houston, TX). A first human implant of a VNS generator was performed in 1988 (Wheless et al., 2001), and now more than 16,000 patients worldwide with refractory epilepsy are treated with VNS (Schachter, 2002). The knowledge of the afferent vagus connections to many of the brain regions implicated in neuropsychiatric disorders has invited theoretical considerations for potential applications of VNS in psychiatry (George et al., 2000a–c). An open study of VNS in 59 outpatients with refractory depression reported a 31% response rate after 10 weeks of VNS (Rush et al., 2000; Sackeim et al., 2001). Most recently, an open naturalistic follow-up study showed that the response rate was sustained (from 40 to 46%) and the remission rate increased (from 17 to 29%) with one-year-long VNS treatment (Marangell et al., 2002). Although the preliminary results of a recently completed multisite, controlled, double-blind trial of VNS in depression failed to demonstrate acute efficacy

differences greater than sham, the longer-term response rates still appear encouraging, and VNS may have promise as a novel neuropsychiatric treatment in depression and other disorders (Elger et al., 2000). 1.2. The role of neuroimaging in understanding VNS effects Neuroimaging technologies provide unparalleled opportunities for elucidating the anatomic correlates of neuropsychiatric disorders. Functional imaging tools such as positron emission tomography (PET), single photon emission computed tomography (SPECT), and functional magnetic resonance imaging (fMRI) have enabled in vivo characterization of neurophysiological correlates of normal and pathologic emotional states (Devous, 1995; Drevets, 2000; Bullmore & Suckling, 2001). These imaging tools differ widely in many respects (see Table 1; Toga & Mazziotta, 1996). One of the most important differences has to do with the length of time to perform each technique. For example, fMRI can image the brain every 3 s (or faster), while a fluorodeoxyglucose (FDG) PET scan sums activity over 20–30 min. While faster imaging methods, such as fMRI, can demonstrate the more immediate effects of VNS (several seconds), slower imaging tools, such as SPECT and PET (George, et al., 1991; Gjedde, et al., 2001), can demonstrate the intermediate effects averaged over several minutes, as well as longer-term changes associated with constant VNS over time. The time resolution of the technique relative to the VNS device parameters is important to consider. Since, metabolism and blood flow studies using PET and SPECT are limited by relatively slow time resolution, fMRI may be better suited for assessing the acute changes by VNS (Rosen, et al., 1994). On the other hand, the fMRI method is limited by its sensitivity to vasoactive

Table 1 Different characteristics of functional neuroimaging tools used in studies of vagus nerve stimulation Physiologic model

Internal source of gamma radiation emitted from a gamma-emitting radiopharmaceutical that has been administered to the subject

6–9 mm often decreased to 12–15 mm

Time resolution Single acquisition

Minimum inter-acquisition interval

1 min

2h

Advantages

Disadvantages

Allow for assessment of cerebral blood flow and metabolism

Only relative flow

Less expensive

Relatively low resolution Limits with exposure to radiation

Injection away from camera Possible separate injection within a day PET (positron emission tomography)

Emitting positron from atoms of the positron-emitting isotope, attached to the probe molecule-detection of annihilation with electrons with positron.

4–5 mm often decreased to 8–9 mm

10 s

15 O: 1–5 min 18FDG: 20-30 min.

Allow for assessment of cerebral blood flow and metabolism

Expensive

Superior spatial resolution

Needs nearby cyclotron Limits with exposure to radiation

Possible absolute data

fMRI (functional magnetic resonance imaging)

Blood oxygenation level-dependent (BOLD): dynamic measurement of concentration changes in oxyhemoglobin.

1–2 mm

2–3 s

2–3 s

High temporal spatial resolution

Extremely sensitive to head movement

Lack of radioactivity

No receptor ligand studies like PET and SPECT Contraindication in irremovable magnetic device

Can be repeated multiple times

J.-H. Chae et al. / Journal of Psychiatric Research 37 (2003) 443–455

SPECT (single photon emission computed tomography)

Spatial resolution

Performed on common MRI machine

445

446

Table 2 Differences in confounding variables across neuroimaging studies in implanted patients with vagus nerve stimulation (VNS) (only included when available data were reported) Studies

Numbers of subject

Pool of subjects

Imaging tool

Scanning protocol

VNS parameter

Method of data analysis

Remark Method of data analysis Statistical Parametric Mapping (SPM) t-statistics on a pixel by pixel bases Volume of interest

Two subjects had seizure activity during scan Two subjects had previous epilepsy surgery First study for acute VNS effects

Final image resolution =12 mm

Pixel by pixel SPM 96

First PET study in depression. Scans were done while the stimulator was in its regular cycle (only 10% on of 10 min scanning were active) Limit of single-day, split dose activation protocol Basically same study with Vonck et al. (2000) A split-dose activation design 63% of the scanning time, the VNS was off

Garnett et al. (1992)

5

Epilepsy

H2(15)O PET

3 VNS on, 3 VNS off

–

FWHM=5 mm

Ko et al. (1996)

3

Intractable partial seizure (E04 protocol)

H2(15)O PET

6 VNS on, 6 VNS off

2 mA 30 Hz, 60 s

Final image resolution =15 mm FWHM

Henry et al. (1998a)

10

Refractory complex partial seizure (E05 protocol)

H2(15)O PET

3 baseline 3 VNS on

Final resolution of the filtered image: 11.8 mm FWHM

Conway et al (submitted for publication)

7

Treatment resistant depression

18

F-fluorodeoxyglucose (FDG) PET

Baseline 10 weeks follow up

High stimulation group (mean 0.5 mA, 500 ms, 30 Hz, 30 s on, 5 min off) Low stimulation group (mean 0.85 mA, 130 ms, 1Hz, 30 s on, 180 min off) 0.5 mA (SD 0.14), 18.6HZ (SD 3.8), 356 ms (SD 180.0), 30 s on, 5 min off

Van Laere et al. (2000)

12

Refractory epilepsy

(99m)Tc-ethyl cysteinate dimmer SPECT

1 baseline 1 VNS on

0.25–0.5 mA, 500 ms, 30 Hz, 30 s on, 10 min off

Transaxial resolution of 7.8 mm FWHM

Volume of interest Voxel by voxel SPM96

Ring et al. (2000)

7

Treatment resistant epilepsy

(99m)Tc-hexamethylpropylenamine oxime (HMPAO) SPECT

1 VNS rapid cycling versus 1 off

0.5–2.0 mA, 500 ms, 30 Hz, 7 s on, 12 s off

FWHM=10.9 mm

Liu et al. (2001)

6

Epilepsy

BOLD fMRI

3 runs of 4 cycles

30 s on, 66 s off

FOV=24 cm, 64
64 matrix size, 28 slices with 5 mm slice thickness

Hermes Multimodality image analysis package Region of interest SPM99

Bohning et al. (2001)

9

Treatment resistant depression

BOLD fMRI

10 cycles Comparing with 440 Hz tone

Lomarev et al. (2002)

6

Treatment resistant depression

BOLD fMRI

10 cycles Comparing with 440 Hz tone

Chae et al. (2002)

6

Treatment resistant depression

BOLD fMRI

10 cycles Comparing with 440 Hz tone

Mean 0.54 mA, 20Hz, 500 ms, 7 s on, 108 s off Mean 0.79 mA, 5 and 20 Hz, 500 ms, 7 s on, 108 s off 20 Hz, 500 ms, 7 s on, 108 s off

FOV=27 cm, 64
64 matrix size, 15 slices with 8 mm slice thickness FOV=27 cm, 64
64 matrix size, 15 slices with 8 mm slice thickness FOV=25.6 cm, 128
128 matrix size, 15 slices with 8 mm slice thickness

Pixel by pixel SPM96 Voxel by voxel SPM96 Voxel by voxel SPM96

Synchronization of VNS and MRI by patient’s subject sensation First computerized synchronization of VNS and MRI Partial overlap of subjects with Bohning et al. (2001)’s study First analysis of serial scanning over 5 times. Comparison between initiation and 2 weeks VNS

J.-H. Chae et al. / Journal of Psychiatric Research 37 (2003) 443–455

Acquisition parameters

J.-H. Chae et al. / Journal of Psychiatric Research 37 (2003) 443–455

changes, and is technically difficult to perform with current technology. The various imaging techniques also differ as well in their spatial resolution, and in their methods of image acquisition (Gsell et al., 2000). All of these differences (temporal and spatial resolution; nature of the neuronal signal; relative versus absolute measures) are potentially important in understanding the various neuroimaging findings investigating the effects of VNS. In addition to these general and non-specific fundamental differences across the various imaging tools, functional imaging studies involving VNS are vulnerable to specific potential confounds. Since VNS intermittently stimulates the cervical segment of the vagus nerve, stimulator-on and stimulator-off scans would be substantially different even in the same subject. Additionally, the therapeutic effects of VNS often increase over months or longer in some individuals (Salinsky et al., 1996). Therefore, it would not be surprising that VNS-induced changes in the brain differ between acute and chronic states of VNS. That is, there are likely immediate effects of the cervical vagus stimulation, and there are likely longerterm effects of the VNS as a therapy. A rapidly growing literature now exists of VNS combined with functional neuroimaging. These studies differ somewhat in their methodologies, findings and conclusions (Table 2). This article systematically summarizes, synthesizes and hopefully simplifies these neuroimaging studies in VNS, and highlights common findings as well as key differences.

2. Positron emission tomography (PET) 2.1. Epilepsy So far, at least three groups have used PET to study the effects of VNS on the brain in patients with epilepsy. Garnett et al. (1992) used H15 2 O PET to demonstrate that left cervical VNS in epilepsy caused increased blood flow in the ipsilateral anterior thalamus and the cingulate gyrus as a group. However it should be noted that two of their five subjects had seizures during image acquisition. Since brain activity and blood flow increase over threefold during a seizure (Kuhl et al., 1980), the ictal CBF alterations may have influenced the groupaveraged measurements more than did VNS itself in this small study. Ko and colleagues (1996) reported that left VNS activated blood flow in the right thalamus, right posterior temporal cortex, left putamen, and left inferior cerebellum, on averaged H15 2 O PET data in three subjects. VNS (2 mA at 30 Hz) was activated for 60 s concurrent with the injection of the isotope. Importantly, their results may have been influenced by prior epilepsy surgery, with right anterior temporal lobectomy in one

447

case and left frontal resection in another of the three subjects (Helmstaedter and Elger, 1998). Moreover, in both of these pioneering studies, subjects were scanned after months or years of chronic VNS therapy. Thus, these studies reveal both the acute effects of the VNS during the time of image acquisition, and the potential confound of long-term effects of VNS over time. In a landmark paper in this area, Henry and colleagues at Emory University found blood flow changes in PET studies of VNS in larger samples (Henry et al., 1998a,b, 1999b). Initially, they studied the acute effects of VNS on regional CBF (rCBF) in 10 epilepsy patients who had H15 2 O PET scans before receiving VNS and again after VNS initiation, (Henry et al., 1998a). The first three scans were performed without VNS, less than 2 h before VNS was started. The other three PET scans were acquired during VNS, within 20 h after VNS was started. At the start of each active scan, the preprogrammed magnet was used to trigger a 30-s train of VNS, timed so that the bolus of H15 2 O reached the brain at the onset of stimulation. The VNS was delivered at high levels (500 ms, 30 Hz, 30 s on, 5 min off, mean 0.5 mA) in 5 patients and low levels (130 ms, 1 Hz, 30 s on, 180 min off, mean 0.85 mA) in a different five patients. The high stimulation group had significant blood flow increases in the rostral and dorsal medulla oblongata. In both groups, VNS caused increased activity in the right thalamus, right postcentral gyrus, and bilateral inferior cerebellum. VNS-induced blood flow alterations were also observed bilaterally in the hypothalami, as well as the anterior insula. The high-stimulation group had significant blood flow increases in the bilateral orbitofrontal gyri, right entorhinal cortex, and right temporal pole, which did not occur in the low-stimulation group. Both groups had significant bilateral decreases in the

Fig. 1. Vagus nerve stimulation (VNS)-induced regional cerebral blood flow alteration by positron emission tomography (PET). Volumes of blood flow increases (in yellow) in the brainstem, thalamus, hypothalamus and decreases (in blue) in the medial temporal cortex, and hippocampus, (Adapted from Henry et al., 1998a).

448

J.-H. Chae et al. / Journal of Psychiatric Research 37 (2003) 443–455

amygdala, hippocamus, and posterior cingulate gyrus (Fig. 1). It is of some interest that the amygdala and hippocampus had reduced activity during VNS because these regions often are involved early in complex partial seizures. The authors suggested that the VNS-induced decreases might have been a reflection of the anticonvulsant effects of the device, with lower sustaining repetitive ictal firing in these regions. Importantly, this study was perhaps the first to show a VNS dose effect, as they also found that the high-stimulation group had larger volumes of activation and deactivation sites than did the low-stimulation group. The Emory investigators restudied the same patients after 3 months of VNS treatment (Henry et al., 1998b; Henry, 2000). Chronic VNS-activation PET detected increased blood flow in many regions that had CBF increases on the initial studies, including the dorsal-rostral medulla (high-stimulation group only), inferior cerebellar hemispheres bilaterally, cerebellar vermis (low stimulation group only), thalami, hypothalami, inferior parietal lobules bilaterally, and the right postcentral gyrus. Compared with the acute VNS study that found CBF decreases in bilateral hippocampus, amygdala, and cingulate and increases in the bilateral insula, the chronic VNS-activation studies did not detect significantly altered CBF in these regions. Although several explanations for the differences between the acute and chronic VNS-activation PET studies are possible, it is reasonable to suspect that differences may reflect the brain’s adaptation to chronic VNS (Henry, 2000). Henry et al. (1999b) further used PET to study these patients, by comparing VNS-induced rCBF changes at the initiation of therapy with changes in seizure frequency during 3 months of VNS therapy, to determine whether any acutely changed regions may be more likely to mediate the antiseizure actions of VNS. Using a volume of interest analysis, only the bilateral thalami showed significant associations of rCBF change with seizure frequency change. Thus they concluded that increased thalamic activity initially might mediate the later anticonvulsant effects of VNS (Henry, 2000). 2.2. Depression Conway et al. (submitted for publication) first scanned 18F-fluorodeoxyglucose (FDG) PET in seven depressed patients enrolled in a multi-center, doubleblind, randomized, prospective study of VNS for treatment-resistant depression at baseline and again after 10 weeks of VNS therapy. Compared with baseline, metabolic activity after 10 weeks of VNS was significantly higher in the bilateral orbitofrontal gyrus, left amygdala and parahippocampal gyrus, bilateral thalamus, left insula, and right cingulate gyrus. Areas of decreased activity included the bilateral cerebellum and right fusiform gyrus. This study examined metabolic changes

that occurred after 10 weeks of VNS therapy, during the regularly programmed cycle of stimulation. This methodology thus represents a combination of acute VNS stimulation at the second scan, and chronic blood flow changes associated with VNS over 10 weeks of therapy. This study has several limitations in addition to the unmatched on/off design, including the lack of a yoked and scanned depressed group receiving sham VNS. Another group at Minnesota has serially PET scanned subjects in this same multi-site trial. Results are still pending.

3. Single photon emission computed tomography (SPECT) 3.1. Epilepsy In direct and puzzling contrast to the findings in PET studies, where increased thalamic activity was found with VNS, several SPECT examinations in epilepsy showed that VNS was associated with relatively decreased thalamic activity (Ring et al., 2000; Van Laere et al., 2000; Vonck et al., 2000). Researchers in Belgium investigated the acute effects of VNS in 12 patients with complex partial seizures by means of a perfusion activation study with 99mTc-ethyl cysteinate dimer (ECD: Neurolite) SPECT (Van Laere et al., 2000; Vonck et al., 2000). The first scan was performed 10 min after the bolus of Neurolite (with VNS off). For the second scan later the same day, the VNS was switched on from 0.25 to 0.5 mA, depending on individual tolerance, at a frequency of 30 Hz, pulsewidth of 500 ms, for a duration of 30 s on, 10 min off. The second dose was injected at the end of the 30 s of VNS. Since Neurolite likely images brain activity over the minute following injection (George et al., 1992), this study shows the immediate off effect of VNS. The voxelby-voxel analysis revealed decreased activity in the left thalamus following VNS. They suggested that acute VNS might reduce seizure onset or propagation through inhibition of the thalamic relay center. In direct contrast to the Henry studies (Henry et al., 1998a,b; 1999a,b) where the image was taken during the VNS on cycle, in this SPECT study the image was taken just after the VNS was stopped. Although other differences in subjects and techniques may affect the results, it is thus not surprising that they found the opposite effect in thalamic activity compared to the PET studies. Ring and colleagues examined 99mTc-hexamethylpropylenamine oxime (HMPAO) SPECT in seven subjects with epilepsy who had been receiving VNS over 6 months (Ring et al., 2000). All subjects received stimulation at 30 Hz with a pulse-width of 500 ms. During the active condition, stimulation was delivered in a ‘rapid cycling’ fashion with output current of the range of 0.5–

J.-H. Chae et al. / Journal of Psychiatric Research 37 (2003) 443–455

2.0 mA. The authors demonstrated that rapid cycling VNS was associated with relatively decreased activity in bilateral medial thalamic regions by a region of interest analysis. Interestingly as well, their rapid cycling ‘on VNS’ image was still weighted heavily to the off time of VNS. That is, the SPECT image likely reflects summed activity of about a minute. During that time, the VNS was on for 7 s, and off for 12 s, which repeated. Thus, for 63% of the scan acquisition time, the VNS was off, between activations. Further, in direct contrast with previous SPECT studies where patients were scanned before and just after initial switching on of the VNS device, the subjects of Ring’s study had been undergoing regular stimulation for at least 6 months. Thus, these subjects likely had chronic plasticity changes due to VNS treatment, which might have affected the final image. 3.2. Depression Devous et al. (2001) performed the first study using SPECT to investigate VNS effects in depression. They scanned five depressed patients at baseline, postimplantation (before stimulation) and after 10 weeks of treatment of VNS and 6 age- and gender-matched controls (not implanted). Preliminary analyses of whole group images for the 10 weeks of treatment data relative to either control condition demonstrated bilateral thalamic activation, greater on the left. In this small sample, when compared to baseline, VNS responders (N=2) showed an increase in anterior cingulate, anterior thalamus, and anterotemporal rCBF and a decrease in right temporal rCBF. In contrast, non-responders (N=3) showed few changes. The small sample sizes in these comparisons preclude firm conclusions. The divergent findings between the SPECT and PET data may be associated with time differences (George et al., 1991; Henry, 2000). Additionally, the differences in direction of brain activation may be related to differences between the study groups including duration of VNS treatment prior to scanning, the stimulation paradigm, the nature of diseases, concomitant and previous medications, or subjects’ individual response to stimulation.

4. Blood oxygen level dependent functional MRI (BOLD fMRI) 4.1. General consideration BOLD fMRI, with its relatively high spatio-temporal resolution, performed during VNS, can reveal the exact location and level of the brain’s immediate response to VNS. Further, because it does not require the injection

449

of radioactive tracers, fMRI permits repeated and longterm studies within individuals to assess whether VNS has different regional brain effects with continued use. However several technical hurdles hindered the initial applications of fMRI to study patients implanted with the NCP system. There were initial safety concerns for brain MRI in the presence of coiled wires over the vagus nerve and the pulse generator, which included the possibility of magnetic induction of heat-generating currents over the nerve and of reprogramming or other magnetic effects on the generator. These issues have been shown not to be problems, with appropriate use of head coils to minimize potential heating (Bohning et al., 2001). In addition, the internal on–off switch for the NCP generator has a magnetic control, which makes fMRI studies difficult. This switch is designed for patients to temporarily control the VNS device such as providing on-demand stimulation, inhibiting stimulation, or resetting or testing the generator, by applying a magnet over the generator to activate a switch (Cyberonics, 2000). Thus, the pulse generator often is automatically switched off when exposed to the high strength, static field of the MRI system. In order to study the stimulation response in the brain with fMRI, a technique to assure the device was switched to on was developed by three groups working independently. For example, Maniker et al. (2000) described how a pulse generator placed with the electrode inputs parallel to the long axis of the body was not deactivated by the magnetic field of the MRI. In a series of VNS fMRI studies at the Medical University of South Carolina (MUSC), the generator returns to ‘on’ in about 70% of patients when placed in the MRI scanner if the device is implanted properly in a special alignment with the MRI field (Bohning et al., 2001; Lomarev et al., 2002). 4.2. Epilepsy The first fMRI study of VNS in epilepsy was reported by Sucholeiki (1999). In this pilot study, 4 patients with epilepsy received VNS at 0.25 mA for 30 s on and 30 s off cycling. The length of time ranged from 6 to 18 min of cycling depending on patient tolerance. BOLD-fMRI showed activation in both frontal lobes and the contralateral post-central gyrus. Independently, investigators in New Jersey showed the feasibility of fMRI scans in six VNS implanted epilepsy patients. The scan started 30 s after the patient reported the sensation of the VNS. The entire run consisted of 66 s at baseline followed by four cycles of 30 s on and 66 s off periods. Three runs were collected for each patient. They found that the temporal, angular, and supramarginal gyrus, and parietal, occipital, insular lobes were activated, and the cingulate gyrus was deactivated (Liu et al., 2001).

450

J.-H. Chae et al. / Journal of Psychiatric Research 37 (2003) 443–455

4.3. Depression The first computer-synchronized BOLD fMRI feasibility study in depressed patients involved scans in nine subjects with depression who, on average, had the device implanted for 10.1 months (Bohning et al., 2001). The computer-synchronized method developed at MUSC uses a computer to detect the signal from the implanted stimulator and then to synchronize fMRI image acquisition with the preprogrammed regular firing of the VNS generator. In the first study using this method, immediately before the scan, the patient’s VNS device was reprogrammed to a 7 s on, 108 s off stimulation cycle. The VNS frequency setting was 20 Hz, the pulse-width was 500 ms, and the current settings, which were kept at the patient’s treatment level setting, ranged from 0.5 to 1.25 mA (mean 0.54). Data were acquired at rest, with the VNS device on for 7 s, and also, for comparison, while the patient listened to a 440 Hz tone for 7 s, which served as an internal reference for comparing the VNS response. These results showed that acute VNS, for only 7 s in these patients who had been receiving chronic VNS, activated many brain regions including the orbitofrontal and parieto-occipital cortex bilaterally, the left temporal cortex, the hypothalamus, and the left amygdala. In the next study in a largely independent cohort of six depressed patients, the effects of two different frequencies of VNS were compared (Lomarev et al., 2002). Data were acquired in two separate randomized backto-back sessions. The stimulation frequency was set at either 5 or 20 Hz with the same paradigm as in the previous VNS/fMRI study. The time since the start of their VNS varied from 8 to 19 months (mean 14.2 months). The pulse width was 500 ms; the current settings were left at the patient’s treatment level setting, and ranged from 0.25 to 1.25 mA (mean 0.79) and they were on diverse chronic VNS therapeutic settings. There was robust activity during 20 Hz VNS, with no significant activation during 5 Hz. Compared to during 5 Hz, 20 Hz VNS produced more activity changes from rest in regions similar to the previous MUSC fMRI study. Brain regions activated by hearing a tone were also greater when VNS was intermittently being applied at 20 Hz than at 5 Hz. In addition to this immediate effect of VNS on regional brain activity, this study suggested further that VNS at different frequencies likely had frequency and/or dose dependent modulatory effects on other brain activities such as hearing a tone. In addition, 17 adults enrolled in the recently completed double blind trial of VNS in treatment resistant depression have been longitudinally scanned five times over 6 months using fMRI. First, Chae et al (2002) analyzed the VNS effect compared to rest at the time of the first VNS exposure (initiation) and then following 2 weeks of VNS treatment. A group analysis of the initial

scans at the time of VNS initiation in six subjects showed that VNS increased BOLD signals in the prefrontal gyri, caudate nuclei, temporal and parietal lobes and the cerebellum, consistent with the prior MUSC studies (Fig. 2). A group analysis in seven subjects after 2 weeks of VNS treatment showed activation of similar regions, with some shifts. A formal comparison of the changes over the 2 weeks showed increasing BOLD activity in the frontal and temporal lobes, and with decreasing activity in the occipital lobe and cerebellum (Fig. 3). Further formal analysis of this serial study is currently underway. Future work with these data will address whether changes on an initial VNS/fMRI scan are predictive of clinical response (Nahas et al., 2001). Additionally, they allow one to address whether longitudinal changes occur over the initial course of VNS treatment in depression.

5. Brain regions implicated in the neuroimaging data of VNS As discussed above, the neuroimaging studies of VNS effects have differed in many respects. Thus, looked at superficially and without adequate understanding of the source of the imaging signal, the literature disagrees regarding the specific locations and the direction of activation caused by VNS (Table 3). However, incorporating information about the different techniques used across the studies suggests that there may be some agreement across the studies regarding the neuroanatomic structures influenced by VNS. However, the high degree of clinical heterogeneity, combined with the diverse array of various imaging techniques, makes it difficult to draw many meaningful conclusions. Vagal afferents traverse the brainstem in the solitary tract, with terminating synapses located mainly in the nuclei of the dorsal medullary complex of the vagus. Among medullary structures, the nucleus of the tractus solitarius (NTS) receives the greatest number of vagal afferent synapses. The NTS projects to a wide variety of structures within the posterior fossa including all of the other nuclei of the dorsal medullary complex, the parabrachial nucleus and other pontine nuclei, and the vermis and inferior portions of the cerebellar hemispheres (Henry, 2001). The parabrachial nucleus projects to several structures including the hypothalamus, the thalamus, the amygdala, the anterior insular, and infralimbic cortex, lateral prefrontal cortex, and other cortical regions. (Van Bockstaele et al., 1999). Through its projection to the amygdala, the NTS gains access to amygdala-hippocampus-entorhinal cortex pathways of the limbic system. In addition, the NTS projects to the locus ceruleus, which provides widespread noradrenergic innervation of the entire cortex, and to raphe nuclei providing widespread serotonergic innervation of the brain (Henry, 2001).

J.-H. Chae et al. / Journal of Psychiatric Research 37 (2003) 443–455

One of the most robust findings in neuroimaging studies with VNS-treated patients is the involvement of the thalami—the major site of processing of somatic sensation (Nicolelis & Shuler, 2001) and of regulating cortical activity (Sur & Leamey, 2001). On the one hand, activation of the thalamus with VNS is not surprising because most patients in many studies (but not all) reported feeling the cervical tingling sensations during VNS. Since the postcentral gyrus is the sensory strip (Iwamura, 1998), one would expect to find activation in this area during VNS. Beyond this simple sensory association, numerous thalamic nuclei contain thalamocortical relay neurons (Alexander et al., 1986; Nicolelis & Shuler, 2001), which as a group drive the entire cortex and all subcortical structures. Specific thalamocortical relay neurons transmit preprocessed sensory information to higher levels, and preprocessed motor commands to lower levels (Henry, 2000). These thalamic neurons also synchronize cortical rhythms among many other thalamic processes that initiate or modulate cortical activities (Steriade et al., 1993). There are multiple potential pathophysiological VNS anticonvulsant mechanisms of action, including VNS induced altered activity in the reticular activating system, the central autonomic network, the limbic system, and the diffuse noradrenergic projection system (for review see Henry, 2002). However, one of the most popular theories of VNS anticonvulsant mechanisms is that the thalami are regions that might generate active processes to prevent seizure onset, or to limit propagation of seizure

451

activities. In support of this thalamic anticonvulsant mechanism, all studies using PET and SPECT currently reported have noted changes in thalamic activity by VNS in epilepsy patients, although the direction of the changes vary (Garnett et al., 1992; Ko et al., 1996; Henry et al., 1998a, b; Van Laere et al., 2000; Vonck et al., 2000; Ring et al., 2000). These studies are thus softly consistent with, but do not prove, the theory that VNS exerts anticonvulsant effects by modulating thalamic activity, which then modulates the cortex through thalamocortical connections (Sur & Leamey, 2001). By contrast, although Lomarev et al. (2002) reported relatively weak activation in the thalamus, thalamic involvement has not been prominent in VNS implanted subjects imaged with fMRI who had either depression (Bohning et al., 2001; Chae et al., 2002), or epilepsy. (Sucholeiki, 1999; Liu et al., 2001). This difference in thalamic activation in PET/SPECT versus fMRI studies may derive from time resolution differences, with fMRI showing a shorter duration image with the VNS device totally on. However this is not the only plausible explanation for the discrepant effects on thalamic effects. Other explanations include differences in the subjects, VNS settings, scanning protocols and the limits of the current fMRI technology for assessing subcortical structures. The paradoxes and inconsistencies in these initial studies require further study. The finding of VNS induced-activation of the cerebellum is consistent with the known anatomy of projections of the vagus nerve to the NTS and NTS

Fig. 2. Vagus nerve stimulation (VNS)-induced regional activity using functional magnetic resonance imaging (fMRI) in a subject with depression at the time of the acute initial start of VNS treatment (Adapted from Chae et al., 2002). VNS increased BOLD signals are shown in the prefrontal gyri, caudate nuclei, temporal and parietal lobes and the cerebellum. (P <0.01, extent P <0.05 for display).

452

J.-H. Chae et al. / Journal of Psychiatric Research 37 (2003) 443–455

projections to the inferior portions of the cerebellum (Xu & Frazier, 1995). These imaging findings are also consistent with a report describing improved cerebellar function following VNS (Smith et al., 2000). Finally, several VNS imaging studies, particularly those using fMRI, have found changes in the orbitofrontal cortex, anterior temporal poles, insula, and the hypothalamus. A number of studies in depression have consistently indicated abnormal metabolism of dorsal prefrontal cortex, cingulate, and less consistently, other paralimbic regions such as orbitofrontal cortex, insula, and anterior temporal cortex (Brody et al., 2001; Ebert et al., 1994; George et al., 1997; Goodwin et al., 1993; Kimbrell et al., 2002; Mayberg, 1997; Wu et al., 1999). Thus, these VNS activation patterns suggest that VNS may be affecting areas of the brain involved in depression. In summary, incoming afferent connections of the vagus nerve provide direct projections to many of the modulatory regions that have long been implicated in neuropsychiatric disorders. These connections provide an understanding of how VNS might be a portal to the brainstem and connected limbic and cortical regions.

ongoing VNS) have shown different results, suggesting that there are dynamic brain changes associated with VNS. On the technical level, a variety of imaging machines and techniques have been used, as well as a host of image data analysis programs. Brain images are extensively processed through multiple stages of analysis, and are highly dependent on machine type and the underlying analytic assumptions. The VNS functional imaging literature disagrees regarding the specific locations and the direction of activation caused by VNS, with only a soft core of regions commonly activated. The high degree of clinical heterogeneity, combined with the diverse array of

6. Conclusions A fundamental limitation of VNS at present is the lack of understanding of the definite functional anatomy of VNS as modified and controlled by its use parameters (intensity, pulse-width, frequency, duty cycle). A better understanding of this regional neurobiology would allow for a more hypothesis-driven approach to VNS clinical trials, particularly with respect to dosing strategies (George et al., 2002). Therefore, the key question is whether VNS, applied with different use parameters, might be selectively ‘‘targeted’’ to modify different brain regions, with attendant ‘‘focusing’’ of behavioral effects. The most straightforward method for addressing this question is to combine VNS with functional neuroimaging with higher temporal and spatial resolution. Animal studies with VNS may also help inform the neurobiology of VNS with respect to the use parameters (frequency, intensity, pulse width and duty cycle). However, there are many confounding factors in the VNS functional imaging literature. By and large, sample sizes in the VNS imaging studies have been small, often under 20 subjects, leading to a strong likelihood of both missed findings (type II errors), and falsely positive findings (type I errors). These issues are compounded by the clinically heterogeneous samples who also differed with respect to medications, course of illness, and treatment response. Moreover, since VNS is a relatively invasive procedure, it is difficult to acquire data from healthy control subjects. It also should be noted that VNS studies performed acutely (within 24 h after VNS is initiated) or chronically (after months or years of

Fig. 3. Differences of vagus nerve stimulation (VNS)-induced regional activity between initiation and 2 weeks treatment of VNS using functional magnetic resonance imaging (fMRI) in subjects with depression [increasing activity in the frontal and temporal lobes (a), and with decreasing activity in the occipital lobe and cerebellum (b)]. Group data in six subjects at the first exposure and in seven subjects after 2 weeks treatment of VNS (P<0.01, extent P <0.05).

Table 3 Vagus nerve stimulation-induced regional activity alterations in various studies PET Ko et al. (1996)

Henry et al. (1998a)

Epilepsy

Epilepsy

Epilepsy High

Low

Hypothalamus Medulla Cerebellar hemisphere Frontal lobe Superior frontal gyrus Cingulate gyrus Orbitofrontal gyrus Postcentral gyrus Entorhinal cortex Temporal pole Temporal lobe Insula Amygdala Hippocampus Parahippocampal gyrus Parietal lobe Occipital lobe Putamen

Lt Rt Lt Rt Bilat

" " " " " "Bilat

" " " " – "Bilat

" "

"Lt. Inf

FMRI

Henry et al. (1998b)

Conway et al. (submitted for publication)

Ring et al. (2000)

Van Laere et al. (2000)/ Vonck et al. (2000)

Devous et al. (2001)

Sucholeiki, (1999)

Liu et al. (2001)

Bohning et al. (2001)/ Lomarev et al. (2002)

Epilepsy

Depression

Epilepsy

Epilepsy

Depression

Epilepsy

Epilepsy

Depression

" " " " "High "Bilat, Low

"

# #

#

" " "

# "Bilat

"Ant, Lt Bilat Lt Rt Lt Rt Lt Rt Lt Rt

#Post " – " – " – "

#Post – – " # – – –

# "

"Rt. "

# "

"

"

"Post

"

"

# "Ant, Bilat #Bilat #Bilat

"Ant, Bilat #Bilat #Bilat

"Lt. "Lt.

" "Lt. #Rt

" "

#Lt

" "

"Bilat "Bilat

J.-H. Chae et al. / Journal of Psychiatric Research 37 (2003) 443–455

Garnett et al. (1992)

Structure Thalamus

SPECT

"Lt

PET, positron emission tomography; SPECT, single photon emission computed tomography; fMRI, functional magnetic resonance imaging. Bilat, bilateral; Rt, right; Lt, left; Inf, inferior; Ant, anterior; Post, posterior; Lat, lateral; High, high VNS stimulation group only; Low, low VNS stimulation group only; Mid, mid line.

453

454

J.-H. Chae et al. / Journal of Psychiatric Research 37 (2003) 443–455

various imaging techniques, makes it difficult to draw many meaningful generalized conclusions at this point. Continued work is needed to understand this most interesting new method of stimulating the brain.

Acknowledgements Funded in part by grants from the Dana Foundation and Cyberonics. The authors would like to thank William Buras (Cyberonics), James Russell and Burke Barrett (formerly of Cyberonics) and Drs. Thomas Henry, and Howard Ring, for helpful comments about this review.

References Alexander GE, DeLong MR, Strick PL. Parallel organization of functionally segregated circuits linking basal ganglia and cortex. Annual Review of Neuroscience 1986;9:357–81. Bailey P, Bremer F. A sensory cortical representation of the vagus nerve. Journal of Neurophysiology 1938:405–12. Bohning DE, Lomarev MP, Denslow S, Nahas Z, Shastri A, George MS. Feasibility of vagus nerve stimulation-synchronized blood oxygenation level-dependent functional MRI. Investigative Radiology 2001;36:470–9. Brody AL, Saxena S, Mandelkern MA, Fairbanks LA, Ho ML, Baxter LR. Brain metabolic changes associated with symptom factor improvement in major depressive disorder. Biological Psychiatry 2001;50:171–8. Bullmore ET, Suckling J. Functional magnetic resonance imaging. International Review of Psychiatry 2001;13:24–33. Chae JH, Nahas Z, Lomarev M, Denslow S, Bohning DE, George MS. Changes in brain activity over the first 2 weeks of vagus nerve stimulation in depression. XXIIIrd. Collegium Internaltionale Neuro-Psychopharmacologicum Congress Final Programme. 2002; P.4.W.009. Conway CR, Chibnall JT, Fletcher JW, Filla-Taylor JA, Grossberg GT, Li X, et al. Vagus nerve stimulation increases libic/parlimbic activitiy in treatment-resistant depressed patients. Biol. Psychiatry, (submitted for publication). Cyberonics Inc. Physician’s manual for the NeuroCybernetic prosthesis (NCP) system. Pulse Generator, models 100 and 101. Houston; 2000. Dell P, Olson R. Projections ‘‘secondaires’’ mesencephaliques, diencephaliques et amygdaliennes des afferences viscerales vagales. Comptes rendus des seances de la Societe de biologie et de ses filiales 1951;145:1088–91. Devous MD. SPECT functional brain imaging. Technical considerations. Journal of Neuroimaging 1995;1(S):2–13. Devous MD. Effects of VNS on regional cerebral blood flow in depressed subjects. Vagus Nerve Stimulation (VNS) for TreatmentResistant Depression. Satellite Symposium in conjunction with the 7th World Congress of Biological Psychiatry. Berlin, Germany, 2001. Drevets WC. Neuroimaging studies of mood disorders. Biological Psychiatry 2000;48:813–29. Ebert D, Feistel H, Barocka A, Kaschka W. Increased limbic blood flow and total sleep deprivation in major depression with melancholia. Psychiatry Research 1994;55:101–9. Elger G, Hoppe C, Falkai P, Rush AJ, Elger CE. Vagus nerve stimulation

is associated with mood improvements in epilepsy patients. Epilepsy Research 2000;42:203–10. Garnett ES, Nahmias C, Scheffel A, Firnau G, Upton AR. Regional cerebral blood flow in man manipulated by direct vagal stimulation. Pacing and Clinical Electrophysiology 1992;15:1579–80. George MS, Ketter TA, Parekh PI, Rosinsky N, Ring HA, Pazzaglia PJ, Marangell LB, Callahan AM, Post RM. Blunted left cingulate activation in mood disorder subjects during a response interference task (the Stroop). Journal of Neuropsychiatry and Clinical Neuroscience 1997;9:55–63. George MS, Nahas Z, Bohning DE, Lomarev M, Denslow S, Osenbach R, Ballenger JC. Vagus nerve stimulation: a new form of therapeutic brain stimulation. CNS Spectrum 2000a;5:2–11. George MS, Ring HA, Costa DC, Ell PJ, Kouris K, Jarritt P. Neuroactivation and Neuroimaging with SPET. London: Springer-Verlag; 1991. George MS, Ring HA, Costa DC, Kouris K, Ell PJ. Demonstration of human motor cortex activation using SPECT. Journal of Neural Transmmission, General Section 1992;87:231–6. George MS, Sackeim HA, Marangell LB, Husain MM, Nahas Z, Lisanby SH, Ballenger JC, Rush J. Vagus nerve stimulation. A potential therapy for resistant depression? Psychiatric Clinic of North America 2000b;23:757–83. George MS, Sackeim HA, Rush AJ, Marangell LB, Nahas Z, Husain MM, Lisanby S, Burt T, Goldman J, Ballenger JC. Vagus nerve stimulation: a new tool for brain research and therapy. Biological Psychiatry 2000c;47:287–95. George MS, Nahas Z, Bohning DE, Kozel FA, Anderson B, Chae JH, Lomarev M, Denslow S, Li X, Mu C. Vagus nerve stimulation therapy: a research update. Neurology 2002;59(S4):56–61. Gjedde A, Hansen SB. Physiological imaging of the brain with PET. New York: Academic Press; 2001. Goodwin GM, Austin MP, Dougall N, Ross M, Murray C, O’Carroll RE, Moffoot A, Prentice N, Ebmeier KP. State changes in brain activity shown by the uptake of 99mTc-exametazime with single photon emission tomography in major depression before and after treatment. Journal of Affective Disorder 1993;29:243–53. Gsell W, De Sadeleer C, Marchalant Y, MacKenzie ET, Schumann P, Dauphin F. The use of cerebral blood flow as an index of neuronal activity in functional neuroimaging: experimental and pathophysiological considerations. Journal of Chemical Neuroanatomy 2000;20: 215–24. Helmstaedter C, Elger CE. Functional plasticity after left anterior temporal lobectomy: reconstitution and compensation of verbal memory functions. Epilepsia 1998;39:399–406. Henry TR. Functional imaging studies of epilepsy therapies. Advances in Neurology 2000;83:305–17. Henry TR. Anatomical, experimental, and mechanistic investigations. In: Schachter SC, Schmidt D, editors. Vagus nerve stimulation. London: Martin Dunitz; 2001. p. 1–29. Henry TR. Therapeutic mechanism of vagus nerve stimulation. Neurology 2002;59:3–14. Henry TR, Bakay RA, Votaw JR, Pennell PB, Epstein CM, Faber TL, Grafton ST, Hoffman JM. Brain blood flow alterations induced by therapeutic vagus nerve stimulation in partial epilepsy: I. Acute effects at high and low levels of stimulation. Epilepsia 1998a;39:983– 90. Henry TR, Davis MR, Votaw JR, Bakay RA, Pennell PB, Epstein CM, Hoffman JM, DeLong MR. Vagus nerve stimulation does not induce significant changes in striatal fluoro-Dopa uptake. Annals of Neurology 1999a;46:477. Henry TR, Votaw JR, Bakay RA, Pennell PB, Epstein CM, Faber TL, Grafton ST, Hoffman JM. Vagus nerve stimulation-induced cerebral blood flow changes differ in acute and chronic therapy of complex partial seizure. Epilepsia 1998;39(S6):92. Henry TR, Votaw JR, Pennell PB, Epstein CM, Bakay RA, Faber TL, Grafton ST, Hoffman JM. Acute blood flow changes and efficacy of

J.-H. Chae et al. / Journal of Psychiatric Research 37 (2003) 443–455 vagus nerve stimulation in partial epilepsy. Neurology 1999;52: 1166–73. Iwamura Y. Hierarchical somatosensory processing. Current Opinion of Neurobiology 1998;8:522–8. Kimbrell TA, Ketter TA, George MS, Little JT, Benson BE, Willis MW, Herscovitch P, Post RM. Regional cerebral glucose utilization in patients with a range of severities of unipolar depression. Biological Psychiatry 2002;51:237–52. Ko D, Heck C, Grafton S, Apuzzo ML, Couldwell WT, Chen T, Day JD, Zelman V, Smith T, DeGiorgio CM. Vagus nerve stimulation activates central nervous system structures in epileptic patients during PET H(15) 2 O blood flow imaging. Neurosurgery 1996;39:426–31. Kuhl DE, Engel Jr J, Phelps ME, Selin C. Epileptic patterns of local cerebral metabolism and perfusion in humans determined by emission computed tomography of 18FDG and 13NH3. Annals of Neurology 1980;8:348–60. Liu WC, Moser K, Kalnin A, Marks D (2001): Vagus Nerve Stimulation in patients: a BOLD fMRI study. 7th Annual Meeting of the Organization for Human Brain Mapping. Brighton, UK. Abstract No. 810. Lomarev M, Denslow S, Nahas Z, Chae JH, George MS, Bohining DE. Vagus Nerve Stimulation (VNS) Synchronized BOLD fMRI suggests that VNS in depressed adults has frequency and/or dose dependent effects. Journal of Psychiatric Research 2002;36:219–27. Mayberg HS. Limbic-cortical dysregulation: a proposed model of depression. Journal of Neuropsychiatry and Clinical Neuroscience 1997;9:471–81. MacLean PD. The triune brain in evolution: role in paleocerebral functions. New York: Plenum Press; 1990. Maniker A, Liu WC, Marks D, Moser K, Kalnin A. Positioning of vagal nerve stimulators: technical note. Surgical Neurology 2000;53: 178–81. Marangell LB, Rush AJ, George MS, Sackeim HA, Johnson CR, Husain MM, Nahas Z, Lisanby SH. Vagus nerve stimulation (VNS) for major depressive episodes: one year outcomes. Biological Psychiatry 2002;51:280–7. Nahas Z, Denslow S, Lomarev M, Beale M, Kozel A, Kellner C, Bohning DE, George MS. Interleaved vagus nerve stimulation (VNS) and fMRI in depression: a new investigative technology. 7th World Congress of Biological Psychiatry Berlin, Germany 2001;P008-24. Nicolelis MA, Shuler M. Thalamocortical and corticocortical interactions in the somatosensory system. Progress in Brain Research 2001; 130:90–110. Ring HA, White S, Costa DC, Pottinger R, Dick JP, Koeze T, Sutcliffe J. A SPECT study of the effect of vagal nerve stimulation on thalamic activity in patients with epilepsy. Seizure 2000;9:380–4. Rosen BR, Belliveau JW. Functional neuroimaging. In: Kucharczyk J, Moseley M, Barkovich AJ, editors. Magnetic resonance neuroimaging. Boca Raton, FL: CRC Press; 1994. Rush AJ, George MS, Sackeim HA, Marangell LB, Husain MM,

455

Giller C, Nahas Z, Haines S, Simpson Jr RK, Goodman R. Vagus nerve stimulation (VNS) for treatment-resistant depressions: a multicenter study. Biological Psychiatry 2000;47:276–86. Sackeim HA, Rush AJ, George MS, Marangell LB, Husain MM, Nahas Z, Johnson CR, Seidman S, Giller C, Haines S, Simpson Jr RK, Goodman RR. Vagus nerve stimulation (VNS) for treatmentresistant depression: efficacy, side effects, and predictors of outcome. Neuropsychopharmacology 2001;25:713–28. Salinsky MC, Uthman BM, Ristanovic RK, Wernicke JF, Tarver WB. Vagus nerve stimulation for the treatment of medically intractable seizures. Results of a 1-year open-extension trial. Vagus Nerve Stimulation Study Group Archives of Neurology 1996;53:1176–80. Schachter SC. Vagus nerve stimulation therapy; five years after FDA approval. Neurology 2002;59(S):15–29. Smith B, Shatz R, Elisevich K, Bespalova IN, Burmeister M. Effects of vagus nerve stimulation on progressive myoclonus epilepsy of Unverricht-Lundborg type. Epilepsia 2000;41:1046–8. Steriade M, McCormick DA, Sejnowski TJ. Thalamocortical oscillations in the sleeping and aroused brain. Science 1993;262:679–85. Sucholeiki R. FMRI in patients with vagus nerve stimulation. Epilepsia, 40:(Supp7) 1999:181–2. Sur M, Leamey CA. Development and plasticity of cortical areas and networks. Nature Reviews Neuroscience 2001;2:251–62. Toga AW, Mazziotta JC. Brain mapping: the methods. San Diego: Academic Press; 1996. Van Bockstaele EJ, Peoples J, Valentino RJ. Anatomic basis for differential regulation of the rostrolateral peri-locus coeruleus region by limbic afferents. Biological Psychiatry 1999;46:1352–63. Van Laere K, Vonck K, Boon P, Brans B, Vandekerckhove T, Dierckx R. Vagus nerve stimulation in refractory epilepsy: SPECT activation study. Journal of Nuclear Medicine 2000;41:1145–54. Vonck K, Boon P, Van Laere K, D’Have M, Vandekerckhove T, O’Connor S, Brans B, Dierckx R, De Reuck J. Acute single photon emission computed tomographic study of vagus nerve stimulation in refractory epilepsy. Epilepsia 2000;41:601–9. Wheless JW, Baumgartner J, Ghanbari C. Vagus nerve stimulation and the ketogenic diet. Neurologic Clinics 2001;19:371–407. Wu J, Buchsbaum MS, Gillin JC, Tang C, Cadwell S, Wiegand M, Najafi A, Klein E, Hazen K, Bunney Jr WE, Fallon JH, Keator D. Prediction of antidepressant effects of sleep deprivation by metabolic rates in the ventral anterior cingulate and medial prefrontal cortex. American Journal of Psychiatry 1999;156:1149–58. Xu F, Frazier DT. Medullary respiratory neuronal activity modulated by stimulation of the fastigial nucleus of the cerebellum. Brain Research 1995;705:53–64. Zabara J. Peripheral control of hypersynchronous discharge in epilepsy. Electroencephalography and Clinical Neurophysiology 1985a;61(S):S162. Zabara J. Time course of seizure control to brief, repetitieve stimuli. Epilepsia 1985b;26:518.