Neuromodulation in

s are only included if they are not published later on in full article format. Abbreviations: IED,interictal epileptiform discharges; VNS, vagus nerve stimulation; 3-MPA, 3-mercaptoproprionate;PTZ, pentylenetetrazol; MES, maximal electroshock seizure; SWDs, spike and wave discharges; ADD,afterdischarge duration. Zanchetti et al. (1952) were among the first to experiment with the anti-convulsiveproperties of VNS. They showed that interictal spikes, produced by topical application ofstrychnine to the cerebral cortex of the cat, were blocked by low-intensity VNS (output General discussionChapter 6 151current: 1-2 V, pulse duration: 0.5 ms and frequency 50 Hz). Similarly, Stoica and Tudor(1967, 1968) found that low-intensity VNS (output current: 1-4 V, pulse duration: 0.3 ms andfrequency: 30 Hz) decreased the frequency of spiking activity by about 34% induced bystrychnine applied topically to the coronal, anterior sigmoid, suprasylvian or medial marginalgyri of the cat. Spread to the contralateral side was also inhibited by VNS. Stoica and Tudor(1968) stated that the vagal afferent projections to these areas are diffuse and non-specific andthat a different number of fibers project to each area. On the contrary, they also noted anincrease in interictal spike frequency when high-intensity stimulation of the vagus nerve wasperformed (Stoica and Tudor, 1968). Two decades later, Woodbury and Woodbury (1990,1991) reported that acute high-intensity stimulation at frequencies greater than 4 Hzprevented or reduced chemically (3-mercaptoproprionate, 3-MPA and pentylenetetrazole,PTZ) induced clonic convulsions in anesthetized rats and electrically (maximal electroshockseizure, MES) induced clonic and tonic clonic seizures in awake rats. This effect wasattributed directly to the fraction of C fibers stimulated. It was observed that VNS shortened,but did not shut down a chemically induced seizure once it had begun. Optimal stimulationparameters were described as: output current 0.2-0.5 mA/mm2 of nerve cross-section, pulseduration 0.5-1 ms and frequency 10-20 Hz. In 1992, Zabara published his work in manuscriptform on the anti-convulsant effect of VNS in epileptic dogs. He found that high-intensityVNS could interrupt or terminate strychnine-induced seizures and PTZ-induced tremors inanesthetized dogs. Inhibitory effects of VNS persisted for a considerable time aftertermination of stimulation. It was suggested that small unmyelinated nerve fibers must bestimulated and that the optimum parameters were: output current 10-20 V (~3-20 mA), pulseduration 0.2 ms and frequency 20-30 Hz. McLachlan (1993) found that low-intensity VNSnot only suppressed PTZ-induced seizures (within 3 s after seizure onset), but also reduced(by 33%) focal interictal spikes produced by penicillin in anesthetized rats. Interictal spikescould already been suppressed at 0.2 mA; however, higher currents > 0.8 mA (pulse duration:0.5 ms and frequencies 20-50 Hz) produced a more consistent response and persisted up to 20s after stimulation. Vagal stimulation caused no change in seizure duration in high-dose PTZinduced seizures, which was explained by a more intense ictal involvement of the brainstemreticular system. Takaya et al. (1996) found that VNS pretreatment (output current: 1 mA,pulse duration: 0.5 ms and frequency: 30 Hz continuously for 0.1 or 60 min or intermittently -30 s/5 min on/off for 60 min) induced a sustained anti-convulsive effect on PTZ-inducedseizures in awake animals, which efficacy was dependent on the cumulative stimulusduration. Sunderam et al., (2001) reported the effect of left VNS (output current: 1.4 mA,pulse duration: 0.5 ms, frequency: 20 Hz and duty cycle: 30 s/3 min on/off) and left sciaticnerve stimulation on 3-MPA-induced seizures using a highly complicated study design. VNSonly had a small anti-seizure effect, which the authors ascribed to a hemodynamicallyGeneral discussionChapter 6 152induced deficit in energy substrates. In a chronic model of spontaneous absence epilepsynamely GAERS, we showed a transient increase in seizure duration following acute VNS(output current: 3 V, pulse duration: 0.5 ms and frequency: 30 Hz), which disappeared in asub-acute setting (Dedeurwaerdere et al., 2004). However, when VNS was applied at higherintensities (5-10 V) perceptible for the animals, the typical spike and wave discharges(SWDs) were shut down immediately (Dedeurwaerdere et al., 2005a). In amygdala kindledFast rats (genetic strain of seizure prone rats), acute VNS (output current: 0.5 mA, pulseduration: 0.5 ms, frequency: 30 Hz for 60 s) showed clear anti-convulsant effects in a subsetof animals when applied immediately after the kindling stimulus. However, when VNS (dutycycle: 30 s/1.1 min on/off) was applied for two hours before the kindling pulse, generalizedconvulsions were prolonged.Only a few studies have assessed the effect of chronic VNS in animal models ofepilepsy (Lockard et al., 1990; Munana et al., 2002; Dedeurwaerdere et al., 2005a). Apossible reason for this is that long-term research on VNS in laboratory animals like rats hassome practical implications. Firstly, chronic studies are laborious and time consuming,because they require long-term monitoring of the animal and subsequent analysis of all data.Secondly, there are also technical aspects, which make chronic VNS stimulation difficult toapply in small animals. To deliver chronic VNS, a pulse generator has to be implanted orVNS has to be delivered via flexible leads connected to an external stimulator. Because theCyberonics stimulating device is rather big to implant in a rat, we chose for the latter option.We have constructed a long-term video EEG monitoring unit equipped with spring coveredflexible leads (to avoid gnawing at the leads by the rat), a spring (allows up and downmovements of the rat) and a communicator or swivel (allows the rat to walk around freely inthe cage). The lead ends of the cuff-electrode have to be easily accessible, to connect the ratto the VNS pulse generator without difficulties. In addition, these lead ends have to be safelysecured in a way that the animals cannot pull or move the leads. Therefore, we have fixatedthe lead ends together with the EEG electrodes with dental acrylic on the head of the rat.Finally, VNS necessitates implantation of a cuff-electrode, which is well tolerated by theanimals, does not induce nerve damage and stays functional over prolonged periods. In ourlaboratory, such a stimulation electrode is available now (Dedeurwaerdere et al., 2005b).A study in rhesus monkeys showed that chronic VNS was feasible and that epilepticprocesses are influenced (Lockard et al., 1990). However, at that time, the authors stated thatsafety and efficacy of the procedure were still in question. Munana et al. (2002) evaluated theuse of VNS (output current: 0.25-1 mA, pulse duration: 0.5 ms, frequency: 30 Hz and on/offcycle: 30 s/ 5 min) as a treatment for refractory epilepsy in dogs. No significant difference inseizure frequency, duration or severity was detected between overall 13-week treatment andcontrol periods. However, during the final four weeks of the treatment period, a significant General discussionChapter 6 153decrease in mean seizure frequency (34.4%) was found, compared with the control group.This may reflect an increase in the efficacy of VNS over time. When chronic VNS (outputcurrent: 1.5 mA, pulse duration: 0.5 ms, frequency: 30 Hz and duty cycle: 60 s/12 s on/off)was applied during one week in GAERS, the decrease in SWDs did not significantly differfrom the control group (Dedeurwaerdere et al., 2005a). It can be hypothesized that a longerperiod of VNS or earlier intervention during life might be required to affect an alreadyestablished absence epilepsy syndrome.The effect of VNS on epileptogenesis has been investigated in the amygdala kindlingmodel of temporal lobe epilepsy. During the kindling process seizure severity and durationgradually progress. It offers the advantage that seizures can be elicited at will and that itallows detailed study of the events associated with the epileptogenic process. In cats, it wasfound that VNS trains (output current: 1.2-2 mA, pulse duration: 0.5 ms and frequency: 30 Hzfor 1 min) before and four times after the kindling stimulus interfered with epileptogenesis(Fernandez-Guardiola et al., 1999). These findings were confirmed in a previous abstract ofNaritoku and Mikels (1997) in rats, who showed that VNS before each kindling stimulusdelays the completion of kindling. However, we could not reproduce these observations usinga genetic seizure prone rat strain. Kindling rate was not delayed by VNS (output current: 0.5mA, pulse duration: 0.5 ms, frequency: 30 Hz and duty cycle: 30 s/1.1 min on/off) and duringthe stage-5 convulsions evidence of hyperexcitability was even noted in the VNS group(Dedeurwaerdere et al, submitted). It is conceivable that VNS cannot interfere withepileptogenesis in individuals with a strong genetic predisposition to develop epilepsy, likethe Fast rats in our study. Moreover, kindling might involve different mechanisms in Fast ratscompared to healthy animals. It is noted that in this study we saw that some VNS-treated rats(n= 3) showed delayed kindling rates compared to controls (n= 3) when the piriform cortex,rather than the amygdala, served as the kindling focus (S. Dedeurwaerdere, personalobservations, 2004). Although, kindling rate in the piriform cortex is faster than in theamygdala (McIntyre et al., 1999), VNS rats did not display stage-5 seizures unless VNS wasterminated (by lead fracture). Therefore, the origin of epileptogenesis and seizure generationcould also be an influencing factor whether VNS can act on kindling development in Fastrats.It is impossible to extract an ideal set of stimulation parameters from these studies.Most studies use frequencies between 20 and 30 Hz and a pulse duration of 0.5 ms. However,the probably most crucial parameter namely the output current is difficult to evaluate. Theactual current delivered to the nerve is depending on the construction of the electrode.Different electrode types have been used and in addition, studies have been performed inanesthetized animals as well as in awake animals. In human, output current is increased tilltolerability level. General discussionChapter 6 154Zabara (1992) observed that right or left vagal stimulation is equally effective incontrolling motor seizures, but bilateral vagal stimulation produced no measurable greatereffect than did unilateral stimulation. Krahl et al. (2003) confirmed that right-sided VNS wasjust as effective as left-sided VNS in reducing the severity of PTZ-induced seizures. They,however, speculated that stimulation of both vagus nerves may affect a larger portion of thebrain than unilateral VNS. This was the case for trigeminal nerve stimulation (Fanselow et al.,2000), though was not observed by Zabara in earlier VNS studies using anesthetized dogs.Stimulation of the right or left nerve of Hering (ninth cranial nerve) can also successfullycontrol focal seizure activity, whereas stimulation of the twelfth cranial nerve fails to suppressseizure activity (Patwardhan et al., 2002). 6.2.2 Emerging indications for VNS treatment Vagal stimulation may affect numerous brain structures involved in the regulation ofmood and cognition (Schachter, 2004). Several studies report improved alertness, behaviorand mood following VNS independent of changes in seizure frequency (Elger et al., 2000;Harden et al., 2000; Harden, 2001; Kossoff and Pyzik, 2004). Accordingly, VNS has beenexplored as a treatment for depression (Rush et al., 2000; Sackeim et al., 2001a; Sackeim etal., 2001b; Sjogren et al., 2002). Also in animal studies, VNS has shown to be an effectiveantidepressant in the forced-swim test in rats (Krahl et al., 2004).An open-label pilot study suggested a positive effect on cognition of VNS treatmentfor patients with Alzheimer’s disease (Sjogren et al., 2002). Clark et al. (1995) hypothesizedthat vagal afferents affect central operations involved in the modulation of memoryconsolidation processes. In rats and humans, they reported that VNS was able to enhancememory storage when applied during memory consolidation, which was dependent on aninverted-U shaped function of output current (Clark et al., 1995; Clark et al., 1998; Clark etal., 1999). In contrast, chronic VNS application in a clinical setting has not been found toaffect cognitive performance in patients with epilepsy since standardized tests have notidentified systematic positive or negative changes in attention, motor function, short-termmemory, learning and memory or executive functions following chronic VNS (Hoppe et al.,2001; Dodrill and Morris, 2001). Therapeutic agents such as AEDs, on the contrary, oftenexhibit mild to serious effects on cognition (Ortinski and Meador, 2004). In Fast rats, agenetic strain with learning impairment, we found that VNS stimulation was devoid ofcognitive side effects in the Morris water maze, which is used to investigate spatial memoryin rats. In the Alzheimer study and the studies of Clark et al. (1995, 1998, 1999) intermediatestimulation parameters were used, whereas our study and those performed in patients withepilepsy (Hoppe et al., 2001; Dodrill and Morris, 2001) used high stimulation parameters (up General discussionChapter 6 155to tolerability level), which are believed to have anti-epileptic effects. Therefore, differentstimulation parameters may result in a different therapeutic effect.Obesity is the most significant health problem facing westernized societies being theprimary risk factor for diabetes and obstructive sleep apnea (Roslin and Kurlan, 2001). Inaddition, it increases the risks for heart disease, pulmonary disease, infertility, osteoarthritis,cholecystitis and several major cancers, including breast and colon cancers (Roslin andKurlan, 2001). It is estimated that 50 to 60% of the population is obese or overweight ofwhom 5 to 6% are considered morbidly obese (Roslin and Kurlan, 2001). Surgical proceduresfor morbid obesity are becoming more common because of long-term successful results(Mason, 1992). However, optimal treatment for morbid obesity is not yet at hand (Roslin andKurlan, 2001).Weight reduction has been reported in patients treated with VNS (Burneo et al., 2002;Kneedy-Cayem et al., 2002). Vagal afferents play a predominant role in the regulation of foodintake (Schwartz, 2000). They conduct signals from the stomach to the nucleus of the solitarytract (NST) carrying information about the size and chemical composition of a meal, which istransmitted by other specific connections to the satiety center in the hypothalamus and thecollateral ventro-medial nucleus (Laskiewicz et al., 2004). A pilot-study in humans has beenset up with subdiaphragmatic stimulation of the vagus nerve. This study was based on thefinding in dogs that chronic bilateral VNS induced a decrease in weight, which was suggestedto result from changes in the central nervous system and a secondary alteration of food intakeby VNS.In a study in GAERS, we found a significant weight reduction after two weeks ofchronic VNS (Dedeurwaerdere et al., 2003). In addition, we found that VNS treatmentprevented the gradual increase in body weight observed in control rats throughout thekindling process. A basic association between amygdala kindled seizures and weight gain inrats has been documented (Loscher et al., 2003; Bhatt et al., 2004). Interestingly, theamygdala has been suggested as a region where interactions between gustatory and vagalinput take place (Han et al., 2003). This reduction in kindling-induced weight gain in VNSrats was associated with reduced food intake compared to control kindled rats. As in thekindling study the Fast rats only received two hours of VNS a day, it is possible that morecontinuous VNS would result not only in reduced weight gain, but also in reduced bodyweight as in GAERS. We are the first to report altered body weight in animal studiesevaluating the efficacy of VNS on epilepsy likely because previous studies have primarilyapplied VNS acutely over very short time spans. In addition, our studies showed that inanimals, weight can be regulated by chronic unilateral left VNS applied through the vagusnerve in the neck. General discussionChapter 6 1566.2.3 Mechanism of action of VNS Desynchronization of the background EEG by stimulation of vagal C fibersThe idea of using VNS to suppress epileptic seizures arose from early animalexperiments in which background EEG was desynchronized by VNS. As seizures result fromhypersynchronous firing of a group of neurons, it was hypothesized that stimuli that produceddesynchronization of the EEG may have anti-epileptic properties.Experiments on “encéphale isolé” cats showed that stimulation of the vagus reducedthe amplitude of the EEG background, produced EEG desynchronization and blocked sleepspindles during slow wave sleep (Zanchetti et al., 1952). However, these effects were notobtained in animals without spindling and the effect was similar to that of other alertingstimuli (nose rubbing, whistling). Forty years later, unilateral and bilateral VNS in catsinduced, within less than 5 s, changes in the pattern and periodicity of EEG spindles,associated with depressed background rhythms or rhythmic EEG activities (Balzamo et al.,1990).In apparent contrast, several studies observed VNS induced synchronization of theEEG. As early as 1938, Bailey and Bremer reported that VNS in the cat elicited synchronizedactivity on the orbital-frontal cortex EEG. Rojas (1964) showed that direct current stimulation(15-60 s) of the vagus nerve first produced EEG desynchronization with subsequentsynchronization. Also Puizillout and Foutz (1974) reported that VNS producedsynchronization of the EEG. McLachlan (1993) observed that background rhythms were notaltered by VNS in the majority of the animals, however, in a few animals a slight increase insynchrony was noted after low-intensity and low-frequency (2-3 Hz) stimulation (McLachlan,1993).Stimulation of the solitary tract, which receives projections of the vagus nerve, at lowfrequencies (1-16 Hz) produced EEG synchronization, whereas high-frequency (> 30 Hz)stimulation results in EEG desynchronization from a sleep spindle like background alsocomparable to that observed with tactile or auditory stimuli (Magnes et al., 1961). Neithersynchronization nor desynchronization could be elicited when the EEG background was thatof arousal. Chase and colleagues (1966) performed more detailed experiments and showed acomplex relation between VNS and EEG rhythmicity associated with the activation ofspecific nerve fiber types of the vagus. In encéphale isolé cat preparations, stimulation of thevagus at frequencies above 20 Hz and at intensities greater than 3 V produced EEGdesynchronization or rather a blockage from spindle like activity. At frequencies above 70 Hzand at intensities less than 3 V, assuming that these parameters only activate myelinatedfibers, vagal stimulation produced EEG synchronization. However, the changes elicited wereoften inconsistent and varied with the background activity. In a study in humans, both General discussionChapter 6 157synchronization and desynchronization of the EEG was present during long-term VNSutilizing standard stimulation parameters, which was most prominent in patients with muchepileptiform activity (Koo, 2001).The vagus nerve mainly consists of unmyelinated C fibers (65-80%) and a smallerportion myelinated A and B fibers (Rutecki, 1990). It was assumed that high-intensity andhigh-frequency (20-70 Hz) stimuli induce desynchronization resulting from the activation ofunmyelinated C fibers, whereas, low-intensity high-frequency stimulation inducessynchronization by activating myelinated A and B fibers only. However, this supposition ismost likely oversimplified, because several studies contradict and the distinction betweenlowversus high-intensity and lowversus high-frequency stimulation is often vague.In the study of Takaya et al. (1996), using awake and freely moving animals, noobvious VNS-induced changes in background EEG activity were observed. In addition, earlystudies in humans revealed that VNS induces little if any effect on EEG background rhythms.(Hammond et al., 1992a; Salinsky and Burchiel, 1993). Hence, it was hypothesized that acutedesynchronization of background EEG activity is not a prominent feature of VNS whenadministered during physiologic wakefulness and sleep, nor does it explain the anti-convulsant effect of VNS. In these studies C fibers were probably not affected and this couldbe a reason why desynchronization of the EEG was not observed. Indeed, effectivetherapeutic stimulation parameters appear to be subthreshold for vagal C fibers in humans asthere are no clinical reports of autonomic side effects (significant gastrointestinal, cardiac orrespiratory effects), which would arise from C fiber stimulation (Krahl et al., 2001). Thesefindings were supported by studies in humans who performed intraoperative nerve potentialrecordings (Koo et al., 2001; Evans et al., 2004). Their results showed that predominantly Aand B fibers and not C fibers are activated by the stimulation output currents used in humans.Therefore, C fiber activation cannot fully explain the MOA of VNS. Moreover, Krahl et al.(2001) showed that selective destruction of capsaicin-sensitive C fibers did not affect the anti-convulsive effects of VNS on PTZ-induced seizures in rats. General discussionChapter 6 158Anatomical and functional projections of the vagus nerve into the cerebral hemispheresThe vagus nerve is a mixed cranial nerve that consists of ~80% afferent fibers and20% efferent fibers parasympathically innervating heart, lungs and gastrointestinal tract(Figure 1). Mechanistic VNS research generally presumes that VNS exerts its effect throughthe induction of action potentials in the afferent fibers of the left vagus nerve (Zagon et al.,1999; Henry, 2002). Indeed, the anatomical diffuse projections of the vagus nerve in thecerebral hemispheres would support a broad effect for VNS on neural excitability (Rutecki,1990) (Figure 1).Moreover, vagal stimulation produces evoked potentials (EPs) which have beenrecorded from several brain regions. In the parietal association cortex, VNS evoked slowhyperpolarization in the pyramidal neurons (Zagon and Kemeny, 2000). Thishyperpolarization of the membrane could reduce the excitability of these neurons. In theseanesthetized rats, stimulus intensities that predominantly activated myelinated fibers (less than200 μA) were more effective in inducing long-lasting inhibitory effects than higher stimulusintensities that activated unmyelinated vagal afferents. Furthermore, stimulation of the cervicalvagus produced EPs in the cerebral cortex of several animal models (O’Brien et al., 1971; Caret al., 1975) and human (Tougas et al. 1993). In the cat thalamus, low-intensity vagal nervestimuli facilitated cellular firing in the reticular nuclei and depressed cellular firing in theventro-postero-medial nucleus of the thalamus (Car et al., 1975). Higher intensity vagalstimulation increased firing frequency and duration of discharges in both nuclei. EPs have alsobeen measured in the brainstem (Bennett et al., 1988), cerebellum (Dell and Olsen, 1951;Hennemann and Rubia, 1978) and the hippocampus (Serkov and Bratus, 1970) of animals. Functional imaging of the effect of VNSNaritoku et al. (1995) demonstrated expression of c-fos (reflecting high neuronalactivity) immunoreactivity in several brain regions of anesthetized rats after VNS. Thesebrain regios included limbic structures, thalamic nuclei and brainstem noradrenergic nucleiimportant in the control of seizures (Naritoku et al., 1995) and supports the idea that the anti-epileptic action of VNS involves pathways that project from brainstem to forebrain.Non-invasive mapping of anatomical sites of increased or decreased synaptic activitycan be performed with functional imaging. The influence of VNS on brain activity has beendemonstrated by imaging studies using positron emission tomography (PET), single photonemission computed tomography (SPECT) or functional magnetic resonance imaging (fMRI)techniques (Henry et al., 1998; Vonck et al., 2000; Narayanan et al., 2002; Sucholeiki et al.,2002). General discussionChapter 6 159Figure 14: Efferent and afferent projections of the vagus nerve. A. The vagus nerve hadefferent projections towards the pharynx, lungs, heart and gastrointestinal tract. B.Anatomical diffuse projections of the vagus nerve to cerebral hemisphere structures adaptedfrom Rutecki et al. (1990). General discussionChapter 6 160These imaging studies found changes on both sides of the brain by unilateral left VNSand pointed out a key role for the thalamus and medial temporal lobe structures in the MOAof VNS. However, there is no consensus on other activated structures neither on the type ofchanges (inhibition or excitation). This discrepancy can be attributed to a number ofconfounding factors such as imaging techniques used (PET, SPECT, fMRI), tracer andcontrast agents, scanning protocols, stimulation parameters, medication regimes, course of theillness and treatment response. Heterogeneity of relatively small patient samples is difficult toavoid. In addition, data gathering from healthy subjects is impossible for ethical reasons dueto the invasiveness of VNS.We found that changes in glucose metabolism due to VNS in healthy rats can bemonitored using small animal PET, avoiding confounding factors as described above(Dedeurwaerdere et al., 2005c). In this animal study, results have been observed consistentwith several previous imaging studies in epileptic patients (Vonck et al., 2000; Van Laere etal., 2002; Henry et al., 1998; Henry et al., 2004). During acute VNS, we found bilateraldecreases in the glucose metabolism of the hippocampi (tendency in the right hippocampus),which are known to be often involved in the generation of complex partial seizures intemporal lobe epilepsy. A decreased metabolism in this region may reflect the anti-convulsantactivity of acute VNS. More specifically, in patients, acute VNS can be triggered at seizureonset by a magnet to interfere with the seizure. In line with the long-term study of Henry et al.(2004), these changes in hippocampi were absent after chronic VNS. It is believed that thesedifferences between acute and chronic VNS reflect the brains adaptation to chronic VNS. Atthis moment we cannot explain these changes after chronic VNS treatment observed inimaging studies. Further study will be required to determine why the acute effects of VNS onbrain blood flow differ from VNS effects after months or years of stimulation. Henry (2002)suggested that intrasubject differences between acute and chronic VNS activation scans mayreveal processes of adaptation to chronic VNS, and that some of them may have an anti-epileptic effect. These adaptation processes might underlie the finding that therapeutic VNSoften causes gradual improved seizure control over several months of treatment (Vonck,2003). In this case, ‘plasticity’ instead of ‘adaptation’ could be a more suitable term as'adaptation' would in fact be expected to suppress the effect of VNS and therefore reduce theanti-convulsive effect with timeWe also observed that acute VNS bilaterally increased glucose uptake in the olfactorybulbs. Evoked potential and unit activity recording techniques have revealed the existence ofa vagus nerve-olfactory bulb pathway (Garcia-Diaz et al., 1984). Also in patients, VNS hasbeen found to influence and modulate the processing of olfactory information (Kirchner et al.,2004). This neurophysiological study reported even a positive correlation between activationof the olfactory bulb and therapeutic benefit. As the olfactory bulb is on the contrary very General discussionChapter 6 161small in humans, it is conceivable that changes in metabolism of this structure were notmeasurable in human imaging studies.Chronic VNS induced an increase in metabolism in the left medulla oblongata(Dedeurwaerdere et al., 2005c). Changes in dorsal-rostral medulla during chronic VNS havealso been reported in an initial study of Henry et al. (2000). These findings correspond to theknown anatomical connections and projections of the vagus nerve to the brainstem tracti andnuclei.In addition, we found a significantly decreased left/right ratio in striatum and atendency towards a decrease in glucose uptake in the left striatum after chronic VNS(Dedeurwaerdere et al., 2005c). This brain structure has been acknowledged to play animportant role in the control of seizures (Deransart and Depaulis, 2002). Ictal hyperperfusionof the striatum was shown using single photon emission computed tomography (SPECT) inpatients with temporal lobe epilepsy (Semah, 2002). The striatum receives indirectinnervation from the vagus nerve via the medial reticular formation of the medulla whichprojects to the intralaminar nuclei of thalamus (having extremely widespread projections tocerebral cortex and the striatum) (Henry, 2002). Moreover, in a PET imaging study inhumans, changes in glucose metabolism have been found in the putamen (Ko et al., 1996) andcaudate nucleus (Vonck, 2003) due to VNS.Finally, we were not able to detect previously reported changes in thalamus (Henry etal., 1998; Vonck et al., 2000; Chae et al., 2003; Henry et al., 2004) and hypothalamus (Henryet al., 1998; Henry et al., 2004) (for review see Chae et al., 2003). The negative findings inthese regions therefore do not exclude the presence of smaller or more localized changes.Further study will require higher-resolution acquisition and more advanced analysis. Imagingstudies in patients of our group showed a decreased tracer uptake in the left thalamus duringchronic VNS (Vonck et al., 2000; Van Laere et al., 2002; Vonck, 2003). This maydemonstrate that VNS induces a chronic inhibitory state in this key structure for seizurespread, which is manifested by a decrease in blood flow. In line with this, it has been quotedthat VNS could result in hemodynamically-induced hypoperfusion creating a relative deficitof energetic substrates (Sunderam et al., 2001). Changes in neurotransmission induced by VNSNaritoku et al. (1992) found that chronic VNS prolongs the cervicomedullary tothalamocortical potential interval during somatosensory evoked potential studies. It wasconcluded that chronic VNS does alter neuronal networks outside of the brain stem vagussystem and that alteration of forebrain neurotransmission might be a primary mechanism ofVNS. General discussionChapter 6 162Woodbury and Woodbury hypothesized that VNS appears to act via release of largequantities of the inhibitory mediators GABA and glycine throughout large volumes of thebrain, because it counteracts PTZ, 3-MPA (decreases efficacy of all inhibitory synapses) andstrychnine induced seizures (Woodbury and Woodbury, 1991). In addition, GABA-mediatedinhibition is implicated in the anti-epileptic effect of NST stimulation in rats, which is theprincipal target of VNS (Walker et al., 1999).Despite these experimental observations, a direct or indirect role of GABA in theanti-epileptic effect of VNS in humans remains to be demonstrated. In humans, the effect ofVNS has been examined on the concentration of amino acids and neurotransmitters incerebrospinal fluid (CSF) samples and no clear effect on GABA concentrations wasdetermined (Hammond et al., 1992b; Ben-Menachem et al., 1995). Nevertheless, significantincreases were seen in homovanillic acid (metabolite of dopamine), 5-hydroxyindoleaceticacid (metabolite of serotonin) and ethanolamine levels. Decreases were found in the level ofaspartate. Although serotoninergic as well as dopaminergic systems have been found to haveanti-convulsant effects in animal and human studies of various types of epilepsies, it remainsto be clarified whether these findings are epiphenomena or findings directly related to VNS(Hammond et al., 1992b).The integrity of an additional neurotransmitter system, namely norepinephrinereleased from the locus coeruleus, appears to play an important role in seizure suppression(Krahl et al., 1998). VNS results in a significant increase in the discharge rate of locuscoeruleus neurons in rats (Groves et al., 2005). Bilateral lesions of the locus coeruleus in ratsprevented the seizure-suppressing effects of VNS (Krahl et al., 1998). The authors claimedthat lesioning the locus coeruleus blocks the anti-convulsant effects of VNS by preventingVNS-induced norepinephrine release either globally or in some specific brain sites. VNS alsoresults in a long-lasting (greater than 80 min) increase in norepinephrine efflux in thebasolateral amygdala (Hassert et al., 2004). Noradrenergic projections to the amygdala arisefrom the locus coeruleus, which is the largest population of noradrenergic neurons in the brainand which receives projections from the NST, thus could be modulated by the vagus (VanBockstaele et al., 1999). It is hypothesized that part of the mechanism of kindling may be adecrease in the effects of norepinephrine (Loscher et al., 2003). Interestingly, we found thatVNS could suppress generalized convulsions evoked by amygdala kindling in some animals(Dedeurwaerdere et al., submitted). Moreover, electrical stimulation of the locus coeruleussuppresses epileptiform activity produced by stimulation of the amygdala (Jimenez-Rivera etal., 1987). Taken together, these data suggest that activation of the locus coeruleus by VNSmight be a significant factor for the attenuation of seizures.The decrease of norepinephrine in various forebrain regions (including the stimulatedamygdala) due to kindling could, besides the induction of seizures, also be a possible reason General discussionChapter 6 163for kindling-induced weight gain (Loscher et al., 2003). Drugs to treat obesity includenoradrenergic drugs that reduce food intake (Bray, 2000). This could be of possible relevanceto the finding that VNS reduced kindling-induced weight gain in Fast rats, which wasassociated with a reduced food intake compared to control kindled rats (Dedeurwaerdere etal., submitted). In conclusion, an additional role of VNS-induced changes in norepinephrinein the amygdala could include the regulation body weight and decreased food intake.The effects of chronic VNS on satiety and food intake could suggest an additionalanti-convulsant MOA of VNS by caloric restriction. Diets have been used for the treatment ofintractable childhood epilepsy since the 1920s and are re-emerging as a treatment option(Kossoff, 2004), although the MOA of these diets may be related to induced ketosis. Acute and chronic VNSMost likely there is not one MOA of VNS, but rather a summation of severalphysiological phenomena attributing to the overall effect of VNS. This is also apparent in thedifferences between acute and chronic VNS. Acute and chronic VNS may act via two distinctmechanisms of action which is supported by imaging studies in humans (Vonck et al., 2000;Van Laere et al., 2002).Acute VNS could be mediated by a possible mechanism of non-specific arousal bystimulation of somatic sensory pathways (Rutecki, 1990; McLachlan, 1993; Fanselow et al.,2000). McLachlan et al. (1993) observed that similar reduction in spike frequency by VNSwas obtained by thermal stimulation of the tail. He hypothesized that VNS influencesepileptiform activity by a nonspecific mechanism mediated through the reticular activatingsystem. When acute VNS was applied during absence seizures in GAERS at high intensity,noticeable for the animals, SWDs were stopped (Dedeurwaerdere et al. 2005a). Indeedabsence seizures are aborted by unexpected sensory stimuli (e.g. noise, touching the animal)in GAERS (Danober et al., 1998). On the other hand, when VNS was applied at lowerintensity, not perceptible by the animal, seizures were not interrupted and even prolonged(Dedeurwaerdere et al., 2004, 2005a). We also found that VNS reduced amygdala evokedseizures (Dedeurwaerdere et al. submitted). However, it has been observed that kindling-induced seizures can be stopped by sensory stimulation (K. Gilby, pers. com.). Moreover,kindled seizures can be blocked by an antecedent footshock (Pinel et al., 1973). PTZ-inducedseizures can also be reduced by trigeminal and Hering nerve stimulation (Fanselow et al.,2000; Patwardhan et al., 2002). In humans, it has been demonstrated that sensory stimulationcan suppress focal spikes and absence seizures (Ricci et al., 1972; Rajna and Lona, 1989). Inpatients with VNS, output current is typically programmed higher in the magnet mode toobtain this arousal effect. Acute VNS could likely be mediated by nonspecific arousal,although other or additional mechanisms of action cannot be excluded. General discussionChapter 6 164Chronic VNS is administered in an intermittent way and it appears that seizuresoccurring during the VNS off-time are also affected. This intermittent way of stimulation isinsufficient to explain the reduction of seizures on the basis of abortive effects alone andsuggests a true preventative or so-called anti-epileptic effect of VNS (Vonck, 2003).Processes that mediate VNS-induced sustained changes are unknown, but the persistence ofthe anti-convulsant effects suggests that VNS induces long-term changes in neuronal activity(Takaya et al., 1996). Naritoku et al. (1995) demonstrated expression of c-fosimmunoreactivity in several brain regions important in epileptogenesis by VNS. C-fos is aprotein that signals transcription of other genes. Therefore, the anti-epileptic effect of VNSmay be mediated through transcriptional events resulting into intermediate or long-termchanges. It is reasonable that chronic VNS induces permanent changes in networkconnections as the efficacy of VNS generally increases after long-term treatment.In relation to the improving long-term effect of VNS it has been reasoned out that ifepileptic fits can induce more epileptic fits as is the case in kindling, then reducing thenumber of epileptic events (and perhaps also subclinical ones) by VNS could reduce the riskfor more epileptic events and thus 'cure' the epilepsy. This mechanism could be seen as thereverse of kindling. Firstly, Gowers’ (1881) famous quotation that “seizures beget seizures” isstill under debate (Theodore and Wasterlain, 1999; McIntosh and Berkovic, 2005). Indeed,experimental work in animals provided support that even single seizures are harmful(Bengzon et al., 1997), which may initiate or facilitate a cascade of events leading to anepileptogenic lesion (Briellmann et al., 2005). However, in patients inconsistent data exist.Moreover, human studies suggest that epilepsy is usually not a progressive disorder althoughsome epilepsy syndromes (e.g. progressive myoclonic epilepsies, epileptic encephalopathies)do have a progressive intractable course (Berg and Shinnar, 1997). Anyway, it is probably notby preventing the seizures themselves that epilepsy could be cured, but rather by treating theunderlying cause of the seizures (Berg and Shinnar, 1997). Also our own study in youngGAERS may be relevant to this discussion. Although GAERS displayed fewer seizuresduring development of the absence epilepsy because of the chronic LEV treatment, they alldeveloped a similar epilepsy profile as the control group after terminating the treatment(Dedeurwaerdere et al., 2005d). Nevertheless, the hypothesis that VNS unkindles epilepsy hasbeen tested by Dasheiff et al., (2001). They hypothesized that the longer and more frequent aperson has seizures, the longer it will take to unkindle them. Therefore, they have accessedthe database of two large clinical trials and multiple types of analyses could not support theirhypothesis. Hence they stated that VNS does not unkindle seizures. Moreover, responders toVNS treatment do have seizure relapse after battery failure. Hence, at this moment it does notseem that VNS is curing epilepsy. General discussionChapter 6 1656.2.4 Conclusion Left-sided VNS is an effective treatment for patients with complex partial seizureswith or without secondary generalization who show insufficient response to anti-convulsantdrugs and are unsuited for neurosurgical treatment. In addition it is a well-tolerated and safeform of neurostimulation. Indeed, we found that VNS does not induce cognitive side effectsin Fast rats.Remarkably, VNS could as well prolong as shorten SWDs in GAERS depending onthe output current used. This phenomenon was also observed on amygdala kindling inducedseizures in Fast rats. In this study, VNS application before the induction of a seizureprolonged seizures while VNS applied after the kindling pulse could prevent seizures. Weshould also note that the anti-epileptic effect of VNS on amygdala kindled seizures was onlypresent in a small subgroup of animals. Briefly, VNS can induce both synchronization anddesynchronization and can result in both excitation and inhibition as supported by our studiesin a model of generalized primary epilepsy (GAERS) and a model of partial epilepsy(amygdala kindling).Despite several discrepancies in the VNS literature, we can state that VNS activatesand inhibits multiple brain regions thereby affecting several neurotransmitter systems. Morespecifically, in our micro-PET study we found decreased metabolism in the left hippocampusand left striatum and increases in glucose metabolism in olfactory bulbs and left medullaoblongata.Most likely there is not one MOA of VNS, but rather a summation of severalphysiological phenomena attributing to the overall effect of VNS. The major MOAsunderlying the desired anti-seizure actions of VNS may be mediated by its diffuse projectionsin the cerebral hemispheres inducing i) a transient decrease in synaptic activities in theamygdala, hippocampus and other components of the limbic system, ii) an intermittentincrease in release of norepinephrine in the cerebrum or iii) a change in synaptic activities inthe thalamus and thalamocortical projection pathways bilaterally, leading to increased arousaland possibly to decreased synchrony of synaptic activities between and within corticalregions. Perhaps, these separate mechanisms of VNS can be fine tuned by combining specificstimulation parameters and stimulation protocols resulting in an optimal outcome.There is clear evidence that acute and chronic VNS differ in MOA. Whereas the anti-convulsive effect of acute VNS could be due to an arousal effect, results from differentstudies in humans and animals are increasingly supporting the idea that the anti-epilepticeffect of chronic VNS is based on long-term modulatory changes in synaptic transmissionwithin certain neuronal network. Hence, it takes several months for these changes to be fullyinstalled and expressed. General discussionChapter 6 166Besides its anti-epileptic effect, it is not surprising that VNS, which has numerousprojections to the brain, could influence several other functions like mood, memory, feedingbehavior. In our studies, effect on body weight induced by VNS was indicated in two animalmodels of epilepsy namely GAERS and amygdala kindled Fast rats. 6.2.5 Future perspectives Vagus nerve stimulation has been used since 1988 and at present almost 30 000patients are being treated with VNS worldwide. Experience and knowledge about VNS israpidly increasing, however several questions remain unclear. VNS is used in generalized andpartial epilepsy (Ben-Menachem, 2002), but still responder groups have not been clearlyidentified. Currently 30% of patients treated with VNS will not benefit from VNS. A betterunderstanding of VNS could improve seizure outcome by identifying specific epilepsysyndromes or types of epilepsy that respond well to VNS on the one hand or by optimizingstimulation parameters on the other hand. Future animal research is therefore crucial.VNS efficacy in animals has primarily been assessed in acute models utilizingstimulation protocols in close relation to the time of seizure onset. These studies have pointedout a series of stimulation parameters, although there is no agreement whether highor low-stimulation parameters should be used and which fiber groups are affected by suchstimulation. Unfortunately, an adapted electrode including the measuring capability of actionpotentials of the different nerve fibers induced by VNS, is not yet available for the vagusnerve of rats and previous attempts were faced with lots of difficulties. However, thecapability of the VNS electrode to selectively elicit different fibers can be first tested in asetup using anesthetized animals, avoiding the practical difficulties of using freely movinganimals. After distinguishing sets of stimulation parameters (output current, pulse duration,frequency) that are activating different fibers, their anti-epileptic properties can be tested in achronic model of epilepsy, like the kindling model in which seizures can be induced at will.The efficacy of chronic VNS (several weeks of stimulation) should be investigated indifferent animal models of epilepsy that are closely imitating human epilepsy such asGAERS, kindling model, SE model, concussion model and stroke models. This may provideclues on the identification of responder groups. However, responders to VNS therapy areperhaps not associated with a specific type of epilepsy. In a model of temporal lobe epilepsy,VNS benefited only a subpopulation of the amygdala kindled rats while other rats appearedrelatively unaffected. Further research directed towards identification of essential criteria thatleads to success for VNS application is warranted. Maybe specific genes can be identifiedwhich correlate with the degree of benefit from VNS.Presently, there are only a few guesses of which neurotransmitters contribute to theaction of VNS. Further research should determine which neurotransmitters (e.g. GABA, General discussionChapter 6 167glutamate, norepinephrine, serotonin, dopamine) are involved in the acute and chronic effectsof VNS. This could be realized by means of microdialysis investigating changes inneurotransmitters in several brain structures. Also small animal PET could be performedusing specific tracers during different stimulation protocols. As the resolution of small animalPET is limited, several structures like the nucleus coeruleus cannot be visualized. Therefore,refining research using autoradiography and fMRI can be performed to investigate inhibitionand excitation of anatomically small regions.Besides its anti-epileptic effect, VNS could also influence several other functions likemood, memory, feeding behavior. However, it is conceivable that different stimulationparameters can result in a different effect. Hence, stimulation parameters probably must beoptimized for each target group.Clearly, the complexities of VNS treatment should be further investigated in order tooptimize treatment in patients with refractory epilepsy. General discussionChapter 6