Not for Reprint 1 N

Authors: Please review all authors’ names, academic degrees, affiliations, and contributions MS# 11-1488, author pages (in the Disclosure) for spelling and accuracy.

This article contains some figures that are displayed in color DOI: 10.3171/2012.4.JNS111488 on­lineNot but in black-and-white in the print edition.for Reprint Long-term behavioral, electrophysiological, and neurochemical monitoring of the safety of an experimental antiepileptic implant, the muscimol-delivering Subdural Pharmacotherapy Device in monkeys Laboratory investigation

Nandor Ludvig, M.D., Ph.D.,1 Hai M. Tang, M.D.,1 Shirn L. Baptiste, B.S.,1 Geza Medveczky, M.S.,1 Jonathan K. Vaynberg, B.S.,1 Jacqueline Vazquez-DeRose, Ph.D.,2 Dimitre G. Stefanov, Ph.D.,3 Orrin Devinsky, M.D.,1 Jacqueline A. French, M.D.,1 Chad Carlson, M.D.,1 and Ruben I. Kuzniecky, M.D.1 1Department of Neurology, Comprehensive Epilepsy Center, NYU Langone Medical Center/School of Medicine, New York, New York; 2Center for Neurosciences, SRI International, Menlo Park, California; and 3Scientific Computing Center, SUNY Downstate Medical Center, Brooklyn, New York

Object. The authors evaluated the extent to which the Subdural Pharmacotherapy Device (SPD), chronically implanted over the frontal cortex to perform periodic, localized muscimol-delivery/CSF removal cycles, affects over- all behavior, motor performance, electroencephalography (EEG) activity, and blood and CSF neurochemistry in macaque monkeys. Methods. Two monkeys were used to adjust methodology and 4 monkeys were subjected to comprehensive test- ing. Prior to surgery, the animals’ behavior in a large test chamber was monitored, and the motor skills required to remove food pellets from food ports located on the walls of the chamber were determined. The monkeys underwent implantation of the subdural and extracranial SPD units. The subdural unit, a silicone strip integrating EEG electrodes and fluid-exchange ports, was positioned over the right frontal cortex. The control unit included a battery-powered, microprocessor-regulated dual minipump and radiofrequency module secured to the cranium. After implantation, the SPD automatically performed periodic saline or muscimol (1.0 mM) deliveries at 12-hour intervals, alternating with local CSF removals at 6-hour intervals. The antiepileptic efficacy of this muscimol concentration was verified by demonstrating its ability to prevent focal acetylcholine-induced seizures. During SPD treatment, the monkeys’ behavior and motor performance were again monitored, and the power spectrum of their radiofrequency-transmitted EEG recordings was analyzed. Serum and CSF muscimol levels were measured with high-performance liquid chro- matography electrochemical detection, and CSF protein levels were measured with turbidimetry. Results. The SPD was well tolerated in all monkeys for up to 11 months. The behavioral study revealed that during both saline and muscimol SPD treatment, the monkeys could achieve the maximum motor performance of 40 food-pellet removals per session, as before surgery. The EEG study showed that local EEG power spectra were not affected by muscimol treatment with SPD. The neurochemical study demonstrated that the administration of 1.0 mM muscimol into the neocortical subarachnoid space led to no detectable levels of this compound in the blood and cisternal CSF, as measured 1–125 minutes after delivery. Total protein levels were within the normal range in the cis- ternal CSF, but protein levels in the cortical-site CSF were significantly higher than normal: 361± 81.6 mg/dl. Abrupt discontinuation of 3-month, periodic, subdural muscimol treatments induced withdrawal seizures, which could be completely prevented by gradually tapering off the subdural muscimol concentration from 1.0 mM to 0.12–0.03 mM over a period of 2 weeks. The monkeys’ general health and weight were maintained. Infection occurred only in one monkey 9 months after surgery. Conclusions. Long-term, periodic, transmeningeal muscimol delivery with the SPD is essentially a safe pro- cedure. If further improved and successfully adapted for use in humans, the SPD can be used for the treatment of intractable focal neocortical epilepsy affecting approximately 150,000 patients in the US. (http://thejns.org/doi/abs/10.3171/2012.4.JNS111488)

Key Words • subdural pharmacotherapy • muscimol • macaque monkey • frontal cortex • cerebrospinal fluid • electroencephalography • epilepsy

rugs delivered through the cranial subdural/sub- arachnoid space on the pia mater can diffuse into Abbreviations used in this paper: ACh = acetylcholine; AED = 10,26 antiepileptic drug; CCD = charged-coupled device; EEG = electro- the underlying cerebral cortical tissue. We encephalography; HPLC-ED = high-performance liquid chroma- Dand others have demonstrated that when this “transmen- tography electrochemical detection; RF = radiofrequency; SPD = ingeal” route of drug delivery is used, focal neocortical Subdural Pharmacotherapy Device. seizures can be prevented and/or terminated.9,15,21,23,39,42 J Neurosurg / Month Day, 2012Not for Reprint 1 N. Ludvig et al.

Despite its therapeutic potential, this phenomenon has nication module, regulates fluid movement and electro- not yet been translated into a treatment for intractable fo- physiological data acquisition through the subdural strip. cal neocortical epilepsy, a condition that affects approxi- According to our plan, this control unit will ultimately mately 150,000 people in the US.6,7 One of the remaining be a fully internalized system, implantable as a single, roadblocks to moving this therapy forward is the deter- encased device, subcutaneously, in the patient’s chest or mination of the safety of localized AED delivery into the abdomen. However, for the present study the control unit cerebral cortex through the subdural/subarachnoid space. was secured to the cranium in monkeys to allow us to To assess this, we implanted monkeys with an experi- readily replace the battery, refill the minipump, or empty mental drug-delivery apparatus, the Subdural Pharmaco- the CSF reservoir, when necessary. therapy Device, and used this apparatus to deliver mus- The purpose of this study was to evaluate the impact cimol into the frontal cortex, transmeningeally, for more of long-term use of the muscimol-delivering SPD on over- than 6 months. all safety, occurrence of adverse events, behavior, motor Muscimol is a GABAA receptor agonist occurring functions, EEG activity, and, to a limited extent, blood in the mushroom Amanita muscaria.11,18 When delivered and CSF neurochemistry in macaque monkeys. Nonhu- systemically, muscimol exerts CNS side effects, prohib- man primates were used to obtain information as relevant iting its use for therapeutic purposes.11,18,41 However, the to human conditions as possible. We selected the frontal single delivery of 1.0–2.5 mM muscimol into the rat or cortex for muscimol delivery. Frontal cortical epilepsy is monkey neocortex via the transmeningeal route can exert typically associated with the most treatment-resistant sei- more powerful antiepileptic effects than clinically used zures in humans.7,37 Should the SPD be proved safe and AEDs2 and can leave the animal’s behavior intact.21 These effective, it will likely be used first for the treatment of favorable pharmacological properties of muscimol, along this type of intractable focal epilepsy. Accordingly, this with its water solubility and stability in solution,21 made preclinical work has been focused on frontal cortical– this compound the drug of choice for the SPD. However, SPD interactions. the safety of long-term transmeningeal muscimol deliv- ery into the neocortex has been unknown. The concentration of muscimol for this safety study Methods was 1.0 mM. To verify that this is a relevant, antiepilep- Animals tic concentration, we tested its potential to prevent focal, frontal cortical ACh-induced seizures. High, nonphysi- Six adult male bonnet macaques (Macaca radiata) ological concentrations of ACh are known to cause sei- were used. Their weights over the course of the study zures in the neocortex,14,16 through the suppression of are listed in Table 1. The drug-delivery system became M-current, a muscarinic-receptor-modulated K+ channel clogged in Monkey 1, as regular CSF removal had not also involved in benign familial neonatal convulsions.19,38 yet been implemented in this animal. Monkey 6 under- We selected this seizure model because it produces re- went implantation a month before the writing of this versible seizures with predictable onsets and progression report. Thus, comprehensive testing was performed in and is thus appropriate for intracranial AED tests. Acetyl- Monkeys 2–5. All experimental procedures adhered to choline-induced neocortical seizures develop rapidly and the standards detailed in the Guide for the Care and Use can last for an hour or longer unless treated locally with of Laboratory Animals (Institute for Laboratory Animal muscimol or other AEDs.21,25 Local saline and artificial Research, National Research Council, Washington, DC, CSF are ineffective in changing the course of these sei- National Academy Press, 1996). The described protocol zures.21,25 was approved by the Institutional Animal Care and Use The SPD, based on US patent 6,497,699,22 was de- Committees of NYU Langone Medical Center/School of veloped at NYU Langone Medical Center/School of Medicine and that of SUNY Downstate Medical Center. Medicine for the treatment of intractable focal neocorti- To promote interactions between the monkeys, their 123 cal epilepsy. Detailed descriptions of the concept21 and × 91 × 82–cm (height by width by length) housing cages, hardware25 of the device, as well as preliminary data on compliant with USDA regulations, were placed close to its histological impact,25 are available. Essentially, the each other. However, only one monkey was placed in each SPD includes a subdural silicone strip equipped with cage. All behavioral, electrophysiological, and neurophar- EEG recording electrode contacts and fluid exchange macological studies were performed while the monkeys ports as well as a control unit. Each fluid port in the SPD moved freely in a large, 178 × 124 × 124–cm test chamber strip is suitable for both directing a drug solution into the equipped with a CCD camera on the ceiling. Thus, at no subarachnoid space and directing CSF out of the drug- time were the monkeys restrained in a primate chair. exposed area. Local CSF removal is necessary to wash out inflammatory host reaction cells and molecules (for Hardware and Software for the SPD example, fibroblasts and collagen) that otherwise would The subdural strip (DocXS Biomedical) is a 0.8- to clog the fluid ports and block transmeningeal drug diffu- 1.3-mm-thick, custom-made combination of an Ad-Tech sion. The function of the recording electrodes is to pro- subdural EEG electrode and a fluid-port-integrating sili- vide feedback on the effects of the delivered AED and cone rubber strip with a total surface area of 21 × 18 mm. thus help to adjust the effective drug-delivery parameters. The control unit (Cygnus, LLC) comprises a custom- The control unit, a battery-powered, microprocessor- made dual minipump and a controller board for a micro- controlled dual minipump equipped with an RF commu- processor, a minipump driver circuit, operational amplifi- 2 Not for Reprint J Neurosurg / Month Day, 2012 Safety of the Subdural Pharmacotherapy Device in monkeys

TABLE 1: Weights of monkeys during the course of the study scribed.25 Once the adequate depth of anesthesia was achieved, the monkey was placed in a stereotactic appa- Weight (kg) ratus. The scalp was incised, exposing the skull, and the Monkey No. 1st Date* 2nd Date 3rd Date stereotactic manipulator was used to mark the inter­aural line. A right frontoparietal craniotomy was made with a 1 11/17/09: 5.3 3/17/10: 5.1 6/29/10: 5.1 25-mm long and 7-mm short diameter, centered 15 mm 2 3/17/10: 5.1 6/29/10: 5.0 9/17/10: 5.3 to the right of the midline and 8.0 mm posterior from 3 11/9/10: 6.3 12/2/10: 6.8 7/11/11: 6.1 the interaural line.29 The fluid port–integrating subdural 4 11/16/10: 4.4 12/2/10: 4.4 7/11/11: 5.0 strip was slipped under the dura, as shown in Fig. 1 upper, 5 4/12/11: 5.9 7/11/11: 5.8 8/5/11: 6.0 over the frontal cortex. The craniotomy was closed with 6 7/26/11: 10.9 8/5/11: 11.5 not weighed† a 1.5-mm-thick silicone rubber sheet secured to the skull with self-tapping screws and dental cement, with the elec- * Day of surgical SPD implantation. trode wires and fluid tubes protruding from the sealed † Monkey 6 underwent implantation only 1 month before the writing of area (Fig. 1 lower). Anchoring screws and one or more this article. epidural reference electrodes were placed in the skull. In 5 animals a 7.0-mm-diameter left frontoparietal crani- otomy was also made with its center 8.5 mm to the left ers, and a custom-made RF communication module. The of the midline and 8.0 mm posterior from the interaural dual minipump consists of 2 units, each connected to a line. This accommodated a single-contact EEG record- 2.5-ml silicone reservoir. One reservoir is filled with ei- ing strip slipped under the dura over the left motor cor- ther saline or muscimol for delivery and the other serves tex. The craniotomy was closed as on the right side. Next, as a CSF collection reservoir. However, the design of the the electrode wires and the fluid tubes were connected minipump allowed us to use the CSF reservoir unit for to the SPD microcontroller and minipump, respectively, delivery of a seizure-inducing agent, ACh, during phar- and this was followed by covering all externalized com- macological tests. The minipump has been programmed ponents with a rectangular protective cap secured to the to produce a standard flow rate of 120 μl/minute. Dur- skull. The microcontroller and the minipump were con- ing the course of the study, the saline/muscimol reservoir nected to the battery. The minipump initiated the fluid was refilled and the CSF collection reservoir was emptied delivery/CSF removal cycles shown in Fig. 2, with saline biweekly under ketamine sedation (10 mg/kg, intramus- as the delivered fluid. A 2- to 3-minute telemetry EEG re- cularly). Sterile saline and muscimol solutions were used, cording, transmitted by the microcontroller’s RF module with the latter sterilized by flowing through 0.2-μm pore- to an external PC, was obtained to verify the integrity of size Acrodisc syringe filters. The unique function of the the recording system. The protective cap was then closed, SPD software is to control the dual minipump in such a the skin around the cap was approximated with staples, way that it can periodically alternate saline or drug deliv- and Neosporin was applied to the wound. Finally, anes- ery with CSF removals (US patent, serial no. 12/868/890, thesia was discontinued and the animal was transferred pending). back to his home cage. Postoperative care lasted for 2 weeks, as described,25 followed by another 2- to 4-week Experimental Design period without systemic drug treatment. During this pe- riod the monkeys were in their home cages. In Phase 1 we collected baseline (control) behav- ioral and neurochemical data before surgery. Phase 2 in- Establishment of Antiepileptic Efficacy cluded the surgical implantation of the SPD and a 1- to After 1–2 months of periodic subdural saline deliv- 1.5-month postoperative recovery period. Phase 3, lasting ery by the SPD, the operation of the minipump was tem- for 1–2 months, was dedicated to monitoring behavioral porarily suspended in Monkeys 2–5. Each monkey was performance and EEG activity during the SPD’s periodic sedated with 10 mg/kg intramuscular ketamine, and the subdural saline delivery. This produced an additional set protective cap was opened to access the pump. The CSF of control data. The device was then switched to peri- reservoir of the minipump was removed and replaced odic muscimol delivery, initiating Phase 4, which served with a reservoir containing 300 mM ACh. The saline res- to establish the antiepileptic efficacy of 1.0 mM subdu- ervoir of the pump was also removed and replaced with ral muscimol. Phase 5 was used to monitor the effects a reservoir containing 1.0 mM muscimol (muscimol HBr, of long-term (1- to 4-month) subdural treatment with 1.0 Sigma-Aldrich). The minipump was then reconfigured to mM muscimol on the behavior, EEG activity, and blood/ allow the delivery of ACh to induce seizures. Ten to 15 CSF neurochemistry. The sixth phase was not included in hours after the monkey’s full recovery from the ketamine the original study design. This phase was added once the effect, the animal was placed in the test chamber and te- effects of abrupt discontinuation of long-term muscimol lemetry EEG was started. After a 2-minute baseline re- treatment were observed, prompting us to devise a pre- cording, the minipump was remotely instructed to deliver vention strategy. 60–120 μl ACh solution through the subdural fluid port– integrating strip for focal seizure induction. The detected Surgical Implantation of the SPD seizures were scored as follows: 1, focal EEG seizures; Implantation of the SPD in the 6 monkeys was 2, focal EEG seizures associated with contralateral con- performed after induction of general anesthesia, as de- vulsions; 3, generalized EEG seizures; and 4, generalized J Neurosurg / Month Day, 2012Not for Reprint 3 N. Ludvig et al.

Behavioral Studies The behavioral studies were performed in Monkeys 3–5 before surgery (Phase 1), after surgery during peri- odic saline delivery by the SPD (Phase 3), and during the subsequent muscimol delivery by the device (Phase 4). This design served to determine whether SPD implanta- tion and/or the muscimol delivery protocol induced mo- tor deficits. Motor performance was monitored as follows. The animal was placed in the test chamber. In addition to the CCD camera, the chamber was also equipped with 8 food ports located on the walls at different spatial posi- tions. Four of these ports were baited with food pellets, while the other food ports were empty. The task for the monkey was to explore the chamber (Fig. 3), climb to the food ports, and remove the hidden pellets from the 4 baited ports. This required unimpaired locomotion and hand skills. Each session consisted of 10 trials, with each trial allowing the monkey 5 minutes to remove the food pellets from all 4 baited ports. The next session was prompted by either the removal of all food pellets or the start of the 5th minute. The food pellets were delivered, and their effective removal by the monkey was detected, by a custom-made apparatus and software similar to what we used previously for squirrel monkeys.27 Thus, reach- es into baited ports were counted for each session. This also yielded maximum food-pellet removals per session and average food-pellet removals per session values. To achieve maximum motor performance, the monkey had to remove 40 food pellets during each session. Using the CCD camera, the animals’ behavior was monitored on the screen of the PC. In most motor performance sessions, 20-to 60-second representative segments were stored for subsequent examination. For these functions, Pinnacle Dazzle Video Creator Plus USB video capture device and Pinnacle Studio 11 software were used. The video clip Fig . 1. Top view of the exposed cranium of Monkey 6 during surgical attached to this paper for online viewing was made with implantation of the SPD. Upper: Photograph taken at the moment when the electrode- and fluid port–integrating strip was slipped under this system. Monitoring the monkeys’ behavior in the test the dura to reach the frontal cortical target site. The broken line indi- chamber continued after the motor performance tests un- cates the interaural line and the arrow shows the location of the epidural til the end of this study. reference electrode. d = dura; e = electrode-contact; f = fluid port; m = midline. Lower: The same area 10 minutes later, after the subdural Electroencephalography Studies strip was moved 1.5 cm toward the right frontal bone and the craniotomy The EEG studies, obtained in all monkeys, served to was closed with the silicone rubber sheet (arrow). Dental acrylic sealed the gap between the sheet and the bone. determine whether the implantation and long-term use of the subdural drug-delivery strip induced local neuro- physiological abnormalities (for example, focal slowing, EEG seizures associated with contralateral or bilateral voltage depression, and interictal spikes) that might not convulsions. During Score 4 seizures, 60–120 μl musci- be reflected in altered behavior. Acquisition of the EEG mol was delivered via the minipump in an attempt to stop studies was performed by the SPD microcontroller, and the ongoing seizure. After this experiment, the ACh so- the data were wirelessly transmitted by the SPD’s RF lution was removed, the CSF reservoir was reconnected, (Bluetooth) module to the PC, which was also used for and the original configuration of the minipump was re- behavioral data collection. These EEG recordings were stored, and periodic muscimol delivery/CSF removal cy- collected while the monkey performed the behavioral cles commenced (Fig. 2). One to 2 months later, the ACh task. Recording Channels 1–3 represented EEG activity experiment was repeated, with the ACh solution delivered at the ipsilateral (right), frontal cortical implantation site in between 2 consecutive muscimol applications with the of the fluid port–integrating strip, while Recording Chan- SPD pump. This tested whether periodic subdural treat- nel 4 represented EEG activity at the contralateral cortex ment with 1.0 mM muscimol can prevent ACh-induced (Fig. 4A). The contralateral cortex was implanted with focal frontal cortical seizures and thus has antiepileptic only a single-contact subdural EEG electrode in Monkeys properties. In each monkey, the same ACh dose was used 1–5. Reference recordings with ×2000 gain, 0.1- to 70- during periodic saline treatment and during periodic Hz bandwidth, and 250-Hz analog-to-digital conversion muscimol treatment by the SPD. sampling rate per channel were used, with epidural screw 4 Not for Reprint J Neurosurg / Month Day, 2012 Safety of the Subdural Pharmacotherapy Device in monkeys

Fig . 2. Temporal pattern of the fluid-delivery/fluid-removal cycles executed by the SPD dual minipump for up to 11 months. As indicated, one unit of the minipump directed either saline or muscimol into the frontal cortical subdural/subarachnoid space at 12-hour intervals, whereas the other pump unit removed CSF from the same site at 6-hour intervals to wash out accumulat- ing products of host tissue reaction. Inflow and outflow volumes were adjusted to keep the total volume of the treated subdural/ subarachnoid compartment constant. The extra flushing cycle, implemented for each 24-hour period, served to ensure adequate cleansing of the treated area. h = hours. or strip electrodes as reference electrodes. The EEG data were received with a Bluetooth transceiver, fed into the same PC that collected the behavioral data, and pro- cessed with custom-made software. This software per- formed 4-channel EEG acquisition, display, storage, and replay. The stored EEG files were both replayed for vi- sual examination and subjected to Fast Fourier transform analysis to determine the power spectrum of the record- ings, as described.21,23 For this analysis, 20-second EEG epochs were selected. The spectral intensities (power) of delta- (0.1–3.9 Hz), theta- (4.0–7.9 Hz), alpha- (8–11.9 Hz), beta- (12–29.9 Hz), and gamma- (30–70 Hz) frequency bands were calculated for these epochs. Only those re- cording segments were used during which the monkey 1) moved around in the test chamber and 2) did not eat. This assured that the EEG segments selected for power spec- trum analysis were generated during similar behavioral states and were not confounded with chewing artifacts. Ipsilateral and contralateral recordings during periodic saline delivery and during periodic muscimol delivery were analyzed. The EEG data obtained in Monkeys 2–5 were processed for power spectrum analysis. Neurochemical Studies These studies included measurements of muscimol and total protein content in serum and CSF samples ob- tained in 4 monkeys. In Monkeys 2–4, samples were tak- en in the 1st week of periodic muscimol delivery, whereas in Monkey 5 the samples were taken in the 3rd week of periodic muscimol delivery. For each serum sample, 1 ml blood was collected from the saphenous vein, allowed to clot for 10 minutes, and then centrifuged to generate Fig . 3. Photograph of Monkey 4, taken while the animal explored the sample for analysis. Collection times for the serum the test chamber. Note the white protective cap on the animal’s head. This cap protected all externalized SPD components, including the mi- samples were 1, 10, 30, 60, and 120 minutes after deliv- nipump, the microcontroller, the RF communication module, and the ery. The CSF samples included 0.4-ml samples collected battery, and could be opened to provide access to these components from the cerebellomedullary cistern, with a percutaneous for maintenance. (Adaptation and encasement of these externalized tap 5 and 125 minutes after subdural muscimol delivery. SPD components for implantation in the chest or abdomen of human In addition, 0.2-ml fluid samples were collected from the patients have yet to be accomplished.) J Neurosurg / Month Day, 2012Not for Reprint 5 N. Ludvig et al.

Fig . 4. Relationship between a single (120-μl, 1.0-mM) subdural muscimol delivery with the SPD during ongoing seizure activ- ity and the levels of muscimol in the blood, cisternal CSF, and implantation-site CSF. A: Electroencephalography recordings transmitted from the frontal cortex of the implanted monkey before focal seizure induction. B: Acetylcholine-induced focal EEG seizure activity, localized to the ipsilateral ACh-delivery site (Channels [Ch.] 1–3). C: Termination of the focal seizure follow- ing the delivery of 1.0 mM muscimol into the seizure focus. D–F: High-performance liquid chromatography electrochemical detection chromatograms of serum, cisternal CSF, and implantation-site CSF samples, respectively, collected in this experiment 35–45 minutes after muscimol delivery. Muscimol (32 μg/ml) is detectable only in the subdural implantation/delivery site (F). site of the fluid port–integrating subdural strip with the isocratic pump (Dionex Corp.). Muscimol was detected minipump, as the last step of the sampling protocol. Dur- by precolumn derivatization using a working solution of ing these procedures, the monkeys were sedated by intra- 2.2 mM O-phthalaldehyde (Pickering Labs) and 0.8 mM muscular injection of 10 mg/kg ketamine. The samples 2-mercaptoethanol (Sigma-Aldrich) mixed by automation were stored at -70°C until shipment. with the sample at 10°C, 2 minutes prior to injection into Total protein levels in the CSF samples were deter- the HPLC. Separation was achieved through a reversed- mined at BioReliance, with standard turbidimetry meth- phase C18 column by Shiseido (Capcell MG, 4.6-mm ods using the COBAS Integra 400 plus system. inner diameter × 75-mm length, 3-μm pore size, with a Muscimol measurements were performed as follows. C18 guard column attached) and electrically detected on Serum and CSF samples were stored at -80°C until pro- a CouloChem III (Dionex Corp.) at the following poten- cessing time. Prior to HPLC-ED, the samples were thawed tials: E1, +250 mV; E2, +550 mV; and Guard, +650 mV at on ice, 5% perchloric acid was added to the sample, and 45°C. Calibration curves were constructed using Chrome- the mixture was centrifuged at 4000 rpm for 5 minutes at leon 6.8.0 software (Dionex Corp.), and the concentration 4°C. A 20-μl portion of the resulting supernatant was re- (μg/ml) of muscimol for each sample was determined by moved and analyzed for muscimol content by HPLC-ED. comparing the peak area against the external standards. Muscimol serum and CSF samples (20 μl) were separated Serum and plasma samples were run in several batches, by HPLC-ED, consistent with isocratic separation meth- and new standard curves were generated with each cor- ods used for amino acid analyses.8,30,34 The mobile phase responding sample batch for analysis. The detection limit consisted of 100 mM Na2HPO4, 22% MEOH, and 3.5% of the assay was 0.05 μg/ml muscimol. acetonitrile, pH 6.75, filtered through a 0.2-μm nylon membrane, degassed with 99% He for 20 minutes, and set Statistical Analysis to a flow rate of 0.7 ml/minute on a U3000 biocompatible The dependent variables for statistical analysis were: 6 Not for Reprint J Neurosurg / Month Day, 2012 Safety of the Subdural Pharmacotherapy Device in monkeys

1) reaches into baited port counts, as obtained in the be- seizures were terminated within 4–14 minutes follow- havioral sessions, and 2) spectral intensity (power) values, ing the delivery of 60–120 μl of 1.0 mM muscimol into as obtained in the EEG recordings. Natural log transfor- the acute ACh seizure focus (Fig. 4C). Consistent with mation was used for processing the EEG power values our prior power spectrum analyses,21,25 the detected EEG due to their skewed distribution. Both data sets were ana- seizures were characterized with robust increases in the lyzed with SAS software (version 9.2, SAS Institute), us- power of delta-, theta-, alpha-, and beta-frequency bands. ing linear mixed models (ANOVA with random effects). This phenomenon of EEG power increase was used to For modeling the motor performance data, we used fixed- capture the evolvement of electrographic seizures in the effects variables for 1) the 3 experimental phases corre- course of long (≥ 30-minute) recording sessions (Fig. 5 sponding to presurgery, postsurgery saline treatment, and upper). When periodic saline delivery with the SPD was subsequent muscimol treatment, and 2) the temporal or- switched to the periodic delivery of 1.0 mM muscimol for der of the sessions within each phase. Random effects for 3 weeks, focal ACh seizures were fully prevented. This animal within each phase were used in the modeling. For was reflected in the lack of power increase in the EEG modeling of the EEG power data, we used fixed-effects waves (Fig. 5 lower) and in the monkeys’ normal behav- variables for 1) the 5 examined frequency bands, 2) the 2 ior. Establishment of the seizure-preventing, thus antiepi- recording sites corresponding to Channels 1 and 4, and leptic, efficacy of a 1.0-mM subdural muscimol dose was 3) the 2 treatment modalities corresponding to subdural verified in all monkeys subjected to this test (Monkeys saline and subdural muscimol administrations. Random 2–5), justifying the use of this muscimol concentration in effects for animal and frequency bands within an ani- the subsequent safety studies. mal were used in the modeling. We used a heterogeneous variance model for both data sets. Separate variance error Behavioral and Motor Functions terms were considered for the different frequency bands in the EEG power data set and for the different phases in The behavioral task was performed by all tested the motor performance data set. monkeys before and after surgery, indicating their main- tained motivation to explore. Table 2 summarizes the main motor performance data. The animals’ ability to Results remove all 40 food pellets from the baited ports within a session (maximum food-pellet removal per session) was General Observations unchanged throughout the study, regardless of whether All implant-treated monkeys maintained their weight this motor test was conducted before or after SPD implan- over the course of the study—neither significant weight tation. After surgery, the minimum food-pellet removal loss nor significant weight gain was observed (Table 1). per session value decreased during saline treatment but Monkey 1, whose minipump became clogged in the 2nd approached the presurgical level during the subsequent postoperative month as no regular CSF removal proce- muscimol treatment. The mean food-pellet removal per dure was yet implemented, carried the SPD for 9 more session values remained similar throughout the study, months without infection or deterioration of health. Mon- varying between 37.9 ± 0.8 (mean ± SEM) and 39.9 ± key 2 remained healthy throughout his examination dur- 0.1, except in the postsurgical saline-delivery phase in ing 6 months of subdural saline delivery and 1 month of Monkey 4, which yielded a value of 32.0 ± 2.4 (Table subdural muscimol delivery. Monkey 3 carried the fully 2). Statistical analysis revealed that the mean number functioning device and for 8 months underwent more of reaches into baited ports in the 3 monkeys was 39.75 than 30 minipump-refilling procedures without compli- (95% CI 39.6–39.9) in the 12th session of the presurgical cations. In the 9th month Monkey 3 developed signs of phase and 38.64 (95% CI 37.4–39.8) in the 12th session infection, including lethargy, and was killed. Monkeys 4, of the postsurgical muscimol-delivery phase. The differ- 5, and 6 were healthy and in ongoing experiments at the ence was not statistically different (t75 = 1.82, p = 0.07). A time of the writing of this report, 10 months, 5 months, video clip, presented in the online version of this report, and 1 month after SPD implantation, respectively. Infec- shows a left-handed monkey moving around in the test tion around the externalized SPD components did not chamber and skillfully removing food pellets from the develop. Such neurological symptoms as tremor, ataxia, baited food ports (Video 1). paresis, or paralysis were absent in all animals. Daily Video 1. Behavior of Monkey 3, a left-handed animal, monitoring of the movement pattern (Fig. 3), social inter- implanted with the SPD over the right frontal cortex. This actions, and external features (for example, fur and eyes) recording was made 8 months after SPD implantation and 6 months after the start of periodic (12-hour) subdural muscimol of the implanted monkeys confirmed that the animals re- deliveries alternating with periodic (6-hour) local CSF remov- mained healthy during the entire study period (with the als. The monkey approaches only the 4 baited food ports and exception of the aforementioned case in Monkey 3). skillfully removes the piece of food hidden in each of these ports. Note the normal locomotion and use of the left hand Antiepileptic Efficacy of the SPD without any impairment. Click here to view with Media Player. Click here to view with Quicktime. Delivering 300 mM ACh into the site of the subdu- ral fluid–integrating strip consistently induced focal EEG seizures (Fig. 4B). These were Score 1 seizures, which Electroencephalography Activity in the Subdural generalized within 4–8 minutes, producing contralateral Implantation Sites clonic convulsions (Score 4 seizures). The ACh-induced Wireless EEG recordings from the frontal cortical J Neurosurg / Month Day, 2012Not for Reprint 7 N. Ludvig et al.

Fig . 5. Demonstration of the seizure-preventing potency of muscimol (60 μl, 1.0 mM) delivered by the SPD into the frontal cortical seizure focus in Monkey 5. Changes in the power of 4- to 8-Hz EEG waves are shown during 30-minute (1800-second) recording sessions. Upper: Changes in power before and after ACh application in the condition where the SPD performed pe- riodic, subdural saline delivery into the frontal cortex. Note the ACh-induced robust power increase related to EEG seizure activi- ty, and the return to baseline power about 3 minutes after the SPD minipump delivered muscimol into the seizure focus. Lower: Repeat of the same experiment in the condition in which the SPD was switched from periodic saline delivery to periodic muscimol delivery. This treatment completely prevented the development of ACh-induced seizures, as reflected in the lack of increase in the power of the 4- to 8-Hz band. Corresponding maximum power values of the 4- to 8-Hz band are indicated on each panel. drug-delivery/CSF-removal site were collected from all Results of statistical analysis of the power of delta, theta, 5 monkeys for up to 11 months. We recorded no focal alpha, beta, and gamma waves in the ipsilateral saline/ slowing, EEG flattening, loss of recording via 1 or more drug-delivery site and the contralateral cortex, during channels, or movement-related artifacts. Slight hemi- both subdural saline and subdural muscimol delivery, are spheric electrographic differences, captured in the ipsi- presented in Table 3. As expected, the mean power values and contralateral recording channels (Fig. 4A) occurred. of the 5 frequency bands differed significantly, regardless

TABLE 2: Motor performance measures in 3 monkeys before and after SPD implantation

Motor Performance Measures Experimental Phase Monkey 3 Monkey 4 Monkey 5 max/min food-pellet removals/session preop 40/38 40/35 40/35 postop: SPD periodically delivers saline 40/12 40/17 40/30 after periodic saline treatment: SPD 40/25 40/29 40/35 periodically delivers muscimol mean ± SEM food-pellet removals/session preop 39.9 ± 0.1 39.8 ± 0.2 39.6 ± 0.2 postop: SPD periodically delivers saline 36.5 ± 2.3 32.0 ± 2.4 38.5 ± 0.9 after periodic saline treatment: SPD 38.5 ± 1.0 37.9 ± 0.8 39.5 ± 0.4 periodically delivers muscimol 8 Not for Reprint J Neurosurg / Month Day, 2012 Safety of the Subdural Pharmacotherapy Device in monkeys

TABLE 3: Power of the 5 examined EEG bands in the subdural treatment conditions and recordings sites*

EEG Band Subdural Treatment Delta Theta Alpha Beta Gamma saline delivery site (a) 136.3 ± 16.5 28.1 ± 4.2 29.1 ± 7.6 32.7 ± 4.9 11.4 ± 2.9 contralat site (b) 107.7 ± 13.6 19.8 ± 1.3 21.5 ± 3.6 30.7 ± 3.2 9.6 ± 2.1 muscimol delivery site (c) 160.1 ± 21.1 28.3 ± 4.1 24.3 ± 4.2 34.2 ± 6.4 10.4 ± 2.8 contralat site (d) 99.2 ± 21.1 21.6 ± 1.7 20.9 ± 3.9 34.6 ± 4.2 11.3 ± 2.2 p values a vs b 0.225 0.203 0.508 0.699 0.698 a vs c 0.464 0.812 0.83 0.939 0.89 c vs d 0.009† 0.327 0.626 0.452 0.177

* Raw data (mean ± SEM) are shown. Probability values, derived from a linear mixed model using log-transformed power values, were not adjusted for multiple testing. † Difference due to the larger delta band in the delivery site than in the contralateral site.

of recording site or treatment (F4,12 = 22.4, p < 0.001). livery (implantation) site in the 3 animals revealed mus- The F tests yielded no significant main effect for either cimol concentrations of 42.2, 27.1, and 21.5 μg/ml (30.7 recording site (F1,205 = 1.272, p = 0.261) or treatment (F4,205 ± 6.1 μg/ml), respectively (Table 4). Figure 4F shows = 0.313, p = 0.576), but did reveal that there was a signifi- the chromatogram of a CSF sample collected from the cant interaction between EEG frequency and recording implantation site following localized muscimol delivery site (F4,205 = 2.548, p = 0.041). The source of this differ- during an ongoing ACh seizure. Only these samples from ence was the larger delta band in the delivery site than in the implantation site contained muscimol in measurable the contralateral site (Table 3). Interaction between fre- concentrations. In summary, muscimol concentrations in quency and treatment (F1,205 = 0.159, p = 0.959), and that the systemic blood and CSF circulation after a single sub- between recording site and treatment (F1,205 = 0.024, p = dural muscimol delivery with the SPD were below the 0.877), were not significant. detection level in 30 of 32 samples collected in 4 mon- keys, regardless of whether these sample collections took Muscimol and Protein Concentrations in Blood and CSF place 1 week or 3 weeks after the start of periodic sub- dural muscimol treatment or whether the samples were Retention time and peak for muscimol were identi- collected in seizure-free or postseizure periods. fied in samples containing 1.0 mM muscimol dissolved in The total protein level in the systemic CSF circula- saline at a pH of 7.2 to 7.4 (Fig. 6A). Analysis of blood (se- tion, as measured in the cisternal CSF samples, was in rum) samples collected in Monkeys 3 and 4 before SPD the normal range, 30.6 ± 11.7 mg/dl. However, a signifi- implantation, and thus before any exposure to muscimol, cantly increased total protein concentration, 361.4 ± 81.6 confirmed the lack of muscimol peak in these samples mg/dl was detected in the CSF samples collected from (Fig. 6B). Analysis of serum samples collected 1, 5, 10, the subdural implantation/drug-delivery site (p = 0.035, 30, 60, and 120 minutes after the single subdural delivery paired t-test). of 40 ml of 1.0 mM muscimol showed that the levels of this drug were below the detection limit of 0.05 μg/ml in 13 of the 15 samples (Fig. 6C, Table 4). Similar data were Withdrawal Seizures and Their Prevention found in Monkey 2, used for methodological adjustments: No spontaneous epileptiform EEG or behavioral no muscimol was detected in the serum at any time after signs occurred in Monkeys 1 and 2. Following termina- the single application of 40 μl of 1.0 mM subdural mus- tion of subdural muscimol delivery, withdrawal seizures cimol. Examination of muscimol concentrations in cister- developed in Monkeys 3 and 4. These seizures occurred nal CSF samples consistently showed that the molecule within a day after abrupt discontinuation of 3 months of could not be detected in this medium either, regardless periodic subdural muscimol delivery (Fig. 7). However, a of whether the samples were taken 5 or 125 minutes after single administration of 1.0 mM muscimol with the SPD subdural muscimol delivery (Table 4). The same results promptly stopped the seizures and normalized both EEG were obtained with cisternal CSF samples from Monkey activity and behavior (Fig. 7). When we subsequently 2. When serum and CSF samples were collected 35–45 reduced the concentration of the applied muscimol in a minutes after the subdural application of muscimol to at- gradual fashion, from 1.0 mM to 0.12–0.03 mM over a tempt to stop an ongoing seizure in Monkeys 3, 4, and 5, period of 2 weeks, discontinuation of muscimol delivery muscimol was again absent in the collected blood and cis- did not lead to epileptiform EEG or behavioral events ternal CSF samples (Fig. 4D and E). However, analysis of (Fig. 7). Once withdrawal seizures were prevented with the CSF collected from the subdural ACh/muscimol-de- this protocol, epileptiform EEG or behavioral signs did J Neurosurg / Month Day, 2012Not for Reprint 9 N. Ludvig et al.

for the first time, the most basic issues regarding the safety of long-term drug delivery into the cerebral cortex transmeningeally, via the subdural/subarachnoid space. The device used for this study was the SPD, a nonrespon- sive, automatic “hybrid neuroprosthesis,” designed to pe- riodically administer muscimol into neocortical seizure foci to prevent seizures. Treatment-resistant epilepsy has a high morbidity and mortality rate, approaching 10% over a decade in some cohorts.13 Focal drug delivery of- fers multiple advantages in these patients, including the ability to use drugs that may not be systemically tolerable or able to cross the blood-brain barrier, delivery of high concentrations of the drug in the affected neocortical area, and minimal or no drug levels in systemic tissue or other subcortical and cortical brain areas in which a drug would cause adverse effects. Functioning SPD Implants Carried by Nonhuman Primates Without Adverse Effects on Health for Nearly 1 Year The SPD can be safely implanted and used in ma- caque monkeys for nearly 1 year. The animals maintained their presurgical weight throughout the study, indicating that their eating habits and general health were not af- fected by SPD implantation and use. Signs of pain in nonhuman primates, such as refusal of food and water, sitting in a crouched posture, or lack of motivation to per- form behavioral tasks, were absent. The ability to study the monkeys’ behavioral performance without difficulty for months after implantation indicates the safety of this treatment. Infection occurred in 1 of 6 monkeys after 8 months of involvement in the experiments. We are aware of the risk of infection caused by repeated refilling of the drug-delivery pump reservoir through externalized tub- ing. This is a problem that can be eliminated once the device is fully implanted in the body and refilled percu- taneously. Frontal Cortical Muscimol Delivery With the SPD Causes no Local EEG Abnormalities and Leaves Behavior and Motor Functions Intact A key finding was that delivery of 1.0 mM muscimol into the frontal cortical subdural/subarachnoid space, al- Fig . 6. High-performance liquid chromatography electrochemical ternating with periodic local CSF removals, did not af- detection chromatograms of the 1.0-mM muscimol solution used for fect local EEG activity and did not interfere with normal subdural delivery by the SPD (A), that of a serum sample collected from behavior. The power spectrum of frontal cortical EEG a monkey before surgery (B), and that of a serum sample collected 120 waves during periodic muscimol treatment remained in minutes after the extra delivery of 40 μl of 1.0 mM muscimol in a SPD- implanted monkey, 3 weeks after the start of periodic subdural musci- the physiological range, and the behavior-induced modu- mol delivery by the device (C). Note the muscimol peak in A (122 μg/ml; lation of EEG activity was unaffected by localized mus- arrow) and the lack of this peak at the same retention time (indicated cimol administration. The sole alteration of slightly de- with the broken line) in chromatograms B and C. creased delta power in the contralateral site was not due to muscimol but to the implant itself, possibly due to pres- sure by the roughly 1-mm thick subdural strip. This effect not reemerge; the monkeys’ EEG activity and behavior is unlikely in humans in whom the subdural/subarach- returned to normal, baseline patterns (Fig. 4A; Video 1). noid space is wider. During surgery, the drug-delivery/ CSF-removal strip was placed over the right motor cortex. Discussion Accordingly, significant damages in these areas produced by implantation, muscimol delivery, and/or local CSF re- The use of intracranial drug-delivery devices for moval should have been associated with signs of motor treatment of localized brain disorders is a promising neu- deficit. No such impairments were observed: Video 1 (on- rosurgical strategy that has emerged over the past 10–15 line version) proves this conclusion. After surgery, we did years.4,15,17,20,21,23,24,31,33,35,36,39 The present study addressed, observe a temporary, slight decrease in the food-pellet 10 Not for Reprint J Neurosurg / Month Day, 2012 Safety of the Subdural Pharmacotherapy Device in monkeys

TABLE 4: Muscimol concentrations in blood and CSF samples in monkeys subjected to subdural muscimol delivery by the SPD*

Sampled Time From Muscimol De- ­ Muscimol Concentration (μg/ml) Sampling Condition Medium livery to Sampling (mins) Monkey 3 Monkey 4 Monkey 5 samples collected after 40 μl of 195 μg/ml blood 1 <0.05 <0.05 <0.05 (1.0 mM) muscimol delivered via SPD, w/o blood 10 <0.05 0.45 <0.05 preceding seizure induction blood 30 <0.05 0.30 <0.05 blood 60 <0.05 <0.05 <0.05 blood 120 <0.05 <0.05 <0.05 cisternal CSF 5 <0.05 <0.05 <0.05 cisternal CSF 125 <0.05 <0.05 <0.05 samples collected after 60–120 μl of 195 μg/ blood 35–45 <0.05 <0.05 <0.05 ml (1.0 mM) muscimol delivered via SPD to cisternal CSF 35–45 <0.05 <0.05 <0.05 stop an ongoing focal seizure delivery site CSF 35–45 42.2 27.1 21.5

* The HPLC-ED data generated with the same methodology are included in the table. removals per session. We think this was not a reflection of distribution could lead to greater CNS toxicity (for ex- a motor deficit but, rather, was related to the monkeys’ re- ample, effects on hypothalamus or brainstem) or system- habituation of the test chamber, which they did not visit ic toxicity. Leakage could result from hardware failure, during the 1- to 1.5-month postoperative period. very high drug concentrations, and neurobiological fac- tors, such as increased blood-brain barrier permeability Negligible Risk of Systemic Muscimol Exposure During after seizures.32 The risk of leakage can be determined SPD Treatment by monitoring cisternal CSF and serum drug levels dur- A critical component for successful SPD drug deliv- ing subdural treatment. Our measures, using HPLC-ED, ery to discrete neocortical areas is preventing the drug revealed undetectable serum and cisternal CSF muscimol from leaking beyond local cortex through more wide- concentrations at various times after subdural delivery of spread CSF or vascular circulations. More widespread 1.0 mM of muscimol. Measurements were made both in

Fig . 7. Withdrawal seizures and their prevention in 2 SPD-implanted monkeys. Abrupt discontinuation after 97 and 99 days of subdural delivery of 1.0 mM muscimol (musc.) in Monkeys 3 and 4, respectively, induced Score 4 (generalized) seizures within 10 hours. Restoration of the original muscimol delivery schedule promptly stopped these seizures and normalized both EEG and behavior. Subsequent gradual reduction of the original muscimol concentration to 0.03 mM (Monkey 3; upper) or 0.12 mM (Monkey 4; lower) prevented the development of withdrawal seizures after the discontinuation of muscimol treatment. J Neurosurg / Month Day, 2012Not for Reprint 11 N. Ludvig et al. seizure-free conditions and after focal seizure induction modifying the SPD software so that it can instruct the followed by subdural muscimol delivery. When muscimol minipump to start to deliver gradually decreasing doses was measured in serum and cisternal CSF 35–45 min- of muscimol once 90% of the drug reservoir is emptied. utes after a focal seizure, the drug levels were below the This should prevent the risk of increased seizure suscep- detection threshold. As expected, muscimol levels in the tibility related to the abrupt discontinuation of chronic subdural implantation-site CSF were detected after SPD subdural muscimol treatment. delivery. Muscimol metabolites were not monitored in this study, and thus we cannot exclude with certainty that Potential Adaption of the SPD for Safe and Effective such metabolic products accumulated in the systemic cir- Prevention of Otherwise Untreatable Focal Neocortical culation. However, HPLC-ED assays of serum samples Seizures in Humans collected from the monkeys before any muscimol expo- This study is the first among several from our labora- sure (Fig. 6B) yielded the same chromatography peaks tory that will hopefully advance the SPD toward clinical as those 120 minutes after muscimol exposure (Fig. 6C). trials. Several other pieces of data are yet to be obtained, Since no new peaks were observed, it is likely that no including information on whether transmeningeal musci- muscimol metabolites were present in the blood after mol penetrates all layers of the treated neocortex, whether muscimol delivery with the SPD, as muscimol in blood is long-term subdural muscimol treatment induces toler- completely metabolized within 120 minutes.3 The HPLC- ance, whether the subdurally implanted device causes ED data of this study are consistent with the very low clinically relevant histological alterations in the inter- doses of muscimol applied with our subdural delivery faced neocortical tissue, and whether subdural muscimol method. Thus, the delivery of 40 μl of 1.0 mM muscimol treatment is as effective in controlling chronic (for exam- HBr (MW 195.0) equals a 7.8-μg dose of this compound. ple, alumina gel–induced) neocortical seizure foci such as acute foci in the present study. These questions need to Periodic Removal of Local Inflammatory Host Reaction be answered. Also, the SPD should be further miniatur- Products Critical for Maintaining SPD Functions ized and properly encased to be fully implantable in the All foreign bodies can induce an inflammatory reac- human body, and pathological studies are required to de- tion in the brain, including recording electrodes, stimu- termine risks for systemic complications and teratogenic lators, and other devices, without infection.5,12,43 Despite and/or carcinogenic side effects. these reactions, intracranial shunts, reservoirs, and elec- Nevertheless, this study did prove that essentially trodes have been safely and effectively used for decades. the muscimol-delivering, frontal cortical SPD is safe Local inflammatory processes pose a specific problem and effective in primates for at least about a year without for chronic drug-delivery implants because proliferating hardware failure, muscimol leakage, or the induction of fibroblasts, gliosis, aggregating extracellular matrix pro- behavioral, motor, or neurophysiological abnormalities. teins, and other processes can clog the system and impede Thus, based on these pieces of evidence, it appears that drug diffusion into the target area. Side effects of antiin- SPD treatment has the potential to provide seizure pre- flammatory drugs limit their long-term use. We devised vention to at least a subclass of patients who suffer from a strategy to reduce the effects of the inflammatory re- otherwise treatment-resistant, focal neocortical epilepsy sponse: to periodically wash out local inflammatory cells who are not candidates for resective surgery. The present and molecules from the fluid exchange ports. This strat- device is a nonresponsive hybrid neuroprosthesis that de- egy was effective in the present study to prevent clogging livers the seizure-controlling drug periodically, in an au- and to allow adequate transmeningeal drug diffusion into tomatic, not on-demand, fashion. We view the automatic the cortical tissue with neuropharmacological responses. SPD as a first-generation intracranial pharmacotherapy We found no adverse effects of this procedure. device, a basic platform on which responsive, on-demand drug-delivery systems can be built in the future for the Withdrawal Seizures After Abrupt Discontinuation of treatment of other subclasses of focal epilepsy. Long-Term Subdural Muscimol Treatment: Preventable Side Effect Technical and Methodological Novelties for General The detection of withdrawal seizures after the abrupt Neurosurgical Practice and Research cessation of subdural muscimol after 3 months of treat- The wireless EEG recordings we obtained during be- ment confirms the paramount importance of long-term >( havior and convulsions were free of artifacts due to cable 2-month) behavioral and electrophysiological monitoring movements. Such artifacts often bedevil cable-based, di- in studies involving intracranial drug-delivery implants. agnostic intracranial recordings. The described battery- This phenomenon was not observed in prior animal stud- powered microcontroller-RF communication apparatus ies with muscimol,2,9,20,21,41,42 likely because these studies could be readily adapted for human use, and such an ap- did not extend monitoring beyond 2–3 weeks. Important- paratus could be placed in the postoperative head wrap ly, gradual reduction of the muscimol dose prior to cessa- without difficulty to transmit data for 24-hour video/EEG tion prevented withdrawal. Thus, while increased seizure recordings. This method would improve diagnosis and al- susceptibility does seem to develop upon abrupt discon- low the patients to move freely after the implantation of tinuation of chronic subdural muscimol, similarly to the strip, grid, or depth electrodes. increased seizure susceptibility that follows the abrupt The protective cap on the monkey’s head was criti- discontinuation of oral AEDs,1,28 this potential side effect cal in avoiding infection around the externalized implant can be controlled. As a preemptive strategy, we are now components and thus performing wireless EEG record- 12 Not for Reprint J Neurosurg / Month Day, 2012 Safety of the Subdural Pharmacotherapy Device in monkeys ing and remote-controlled intracranial drug delivery for authors. Approved the final version of the manuscript on behalf of all nearly 1 year, in freely behaving monkeys, without the authors: Ludvig. Statistical analysis: Ludvig, Stefanov. Administra- need of chair restraint. This methodology can signifi- tive/tech­ ­nical/material support: Ludvig, Kuzniecky. Study supervi- sion: Lud­vig. Principal investigator: Ludvig. Construction of the cantly expand the possibilities of neurosurgery studies in SPD for the study: Medveczky. nonhuman primates in a humane way. Finally, this paper describes for the first time a meth- od for measuring muscimol levels from blood or CSF Acknowledgments samples with HPLC-ED. This method offers better sen- The authors thank Jean Charchaflieh, M.D., for his contribution sitivity than HPLC-ultraviolet, it is less costly than gas to the anesthesia protocol; Madhav Paranjpe, D.V.M., for analyzing chromatography and mass spectrometry, and it yields re- the protein content of CSF samples at BioReliance; Sandor Toth, liable data. As such, it is ideally suited for larger-scale M.S., for helping to maintain the experimental setup; Carol A. muscimol assays in expanded preclinical tests and future Novotney, D.V.M., for her guidance in monitoring the health status clinical trials. of the animals; Motria Mishko, Pharm.D., for preparing the sterile drug solutions; and Paul A. Ullucci, M.A., and Ray Smith for help- ing to adjust the muscimol HPLC method. The authors are grateful Conclusions to John G. Kral, M.D., Ph.D., and Werner K. Doyle, M.D., for their helpful suggestions during the course of this study. Administrative The major findings of this study were as follows. 1) assistance by Anjanette Burns and Heather Wall, as well as techni- The SPD can be readily implanted in monkeys, and the cal contributions by Elizabeth Rivera, Kerel Ferguson, and Allison animals tolerate this device well for nearly 1 year. 2) Pe- Maurice to the surgical procedures are appreciated. Photographs for riodic local CSF removal from the treated neocortical Fig. 1 were taken by Ernest Cuni. The authors thank Donald Cum- subarachnoid space can effectively wash out inflamma- mings, Kenneth DeFreitas, Neal Grant, and Mark Corbin for their tory host reaction cells and proteins, preventing clogging conscientious care for the monkeys throughout this project. and allowing adequate transmeningeal drug diffusion. 3) Periodic transmeningeal delivery of 1.0 mM muscimol into the frontal cortex exerts local antiepileptic effects. References 4) Periodic frontal cortical muscimol applications, alter- 1. Andersen NB, Alving J, Beniczky S: Effect of medication nating with local CSF removals, can be maintained for withdrawal on the interictal epileptiform EEG discharges in many months without affecting motor functions or caus- presurgical evaluation. Seizure 19:137–139, 2010 ing EEG abnormalities. 5) Delivering 1.0 mM muscimol 2. Baptiste SL, Tang HM, Kuzniecky RI, Devinsky O, French into the neocortical subarachnoid space leads to no de- JA, Ludvig N: Comparison of the antiepileptic properties of tectable levels of this compound in the systemic CSF and transmeningeally delivered muscimol, lidocaine, midazolam, pentobarbital and GABA, in rats. Neurosci Lett 469:421– cardiovascular circulation. 6) Like abrupt discontinuation 424, 2010 of chronic oral AED therapy, abrupt discontinuation of 3. Baraldi M, Grandison L, Guidotti A: Distribution and me- 3-month long, periodic subdural muscimol treatments tabolism of muscimol in the brain and other tissues of the rat. with the SPD induces increased seizure susceptibility, a Neuropharmacology 18:57–62, 1979 potential side effect that can be prevented by gradually 4. Bennewitz MF, Saltzman WM: Nanotechnology for delivery tapering off the subdural muscimol concentration. Based of drugs to the brain for epilepsy. Neurotherapeutics 6:323– on these findings in nonhuman primates, we conclude 336, 2009 that long-term, periodic, transmeningeal muscimol deliv- 5. Bruni JE, Del Bigio MR: Reaction of periventricular tissue ery with the SPD is essentially a safe procedure. If the in the rat fourth ventricle to chronically placed shunt tubing implants. Neurosurgery 19:337–345, 1986 hardware and software of the device, as well as its surgi- 6. Callaghan BC, Anand K, Hesdorffer D, Hauser WA, French cal implantation and drug-delivery protocol, are further JA: Likelihood of seizure remission in an adult population improved and successfully adapted for human use, the with refractory epilepsy. Ann Neurol 62:382–389, 2007 SPD may be used for the treatment of intractable focal 7. Cascino GD: When drugs and surgery don’t work. Epilepsia neocortical epilepsy affecting about 150,000 patients in 49 (9 Suppl 9):79–84, 2008 the US. 8. Clarke G, O’Mahony S, Malone G, Dinan TG: An isocratic high performance liquid chromatography method for the de- Disclosure termination of GABA and glutamate in discrete regions of the rodent brain. J Neurosci Methods 160:223–230, 2007 This work was supported by the Epilepsy Research Foundation 9. Collins RC: Anticonvulsant effects of muscimol. Neurology (grant no. 140929 to N.L.) and funds from Finding A Cure for Epi- 30:575–581, 1980 lepsy & Seizure (FACES). Dr. Ludvig holds, with Lorant Kovacs, 10. Cornblath DR, Ferguson JH: Distribution of radioactivity US patent no. 6,497,699, on which the SPD is based. This patent is from topically applied [H3]acetylcholine in relation to sei- not yet licensed, and Dr. Ludvig is not associated with any business zure. Exp Neurol 50:495–504, 1976 entity, nor did he receive any payment or promise of payment during 11. DeFeudis FV: Physiological and behavioral studies with mus- the course of this study other than his academic salary. Mr. Medvec- cimol. Neurochem Res 5:1047–1068, 1980 zky is a consultant for Cygnus, LLC. Dr. Devinsky is a patent holder 12. Del Bigio MR: Biological reactions to cerebrospinal fluid at NYU Medical Center. shunt devices: a review of the cellular pathology. Neurosur- Author contributions to the study and manuscript preparation gery 42:319–326, 1998 include the following. Conception and design: Ludvig, Devinsky, 13. Devinsky O: Sudden, unexpected death in epilepsy. N Engl J French, Carlson, Kuzniecky. Acquisition of data: Ludvig, Tang, Med 365:1801–1811, 2011 Baptiste, Vaynberg, Vazquez-DeRose. Analysis and interpretation 14. Dichter MA, Ayala GF: Cellular mechanisms of epilepsy: a of data: Ludvig. Drafting the article: Ludvig. Critically revising the status report. Science 237:157–164, 1987 article: all authors. Reviewed submitted version of manuscript: all 15. Eder HG, Jones DB, Fisher RS: Local perfusion of diazepam J Neurosurg / Month Day, 2012Not for Reprint 13 N. Ludvig et al.

attenuates interictal and ictal events in the bicuculline model 31. Nilsen KE, Cock HR: Focal treatment for refractory epilepsy: of epilepsy in rats. Epilepsia 38:516–521, 1997 hope for the future? Brain Res Brain Res Rev 44:141–153, 16. Ferguson JH, Cornblath DR: Acetylcholine epilepsy: relation- 2004 ship of surface concentration, chronicity of denervation, and 32. Oby E, Janigro D: The blood-brain barrier and epilepsy. Epi- focus size. Exp Neurol 46:302–314, 1975 lepsia 47:1761–1774, 2006 17. Fisher RS, Ho J: Potential new methods for antiepileptic drug 33. Rogawski MA: Convection-enhanced delivery in the treat- delivery. CNS Drugs 16:579–593, 2002 ment of epilepsy. Neurotherapeutics 6:344–351, 2009 18. Halpern JH: Hallucinogens and dissociative agents naturally 34. Rowley HL, Martin KF, Marsden CA: Determination of in growing in the United States. Pharmacol Ther 102:131–138, vivo amino acid neurotransmitters by high-performance liq- 2004 uid chromatography with o-phthalaldehyde-sulphite derivati- 19. Hamilton SE, Loose MD, Qi M, Levey AI, Hille B, McKnight sation. J Neurosci Methods 57:93–99, 1995 GS, et al: Disruption of the m1 receptor gene ablates mus- 35. Sanftner LM, Sommer JM, Suzuki BM, Smith PH, Vijay S, carinic receptor-dependent M current regulation and seizure Vargas JA, et al: AAV2-mediated gene delivery to monkey pu- activity in mice. Proc Natl Acad Sci U S A 94:13311–13316, tamen: evaluation of an infusion device and delivery param­ 1997 eters. Exp Neurol 194:476–483, 2005 20. Heiss JD, Walbridge S, Morrison P, Hampton RR, Sato S, 36. Sawyer AJ, Piepmeier JM, Saltzman WM: New methods for Vortmeyer A, et al: Local distribution and toxicity of pro- direct delivery of chemotherapy for treating brain tumors. longed hippocampal infusion of muscimol. J Neurosurg 103: Yale J Biol Med 79:141–152, 2006 1035–1045, 2005 37. Silfvenius H: Extratemporal cortical excisions for epilepsy, in 21. Ludvig N, Baptiste SL, Tang HM, Medveczky G, von Gizycki Oxbury J, Polkey CE, Duchowny M (eds): Intractable Focal H, Charchaflieh J, et al: Localized transmeningeal muscimol Epilepsy. New York: WB Saunders, 2000, pp 697–714 prevents neocortical seizures in rats and nonhuman primates: 38. Stafstrom CE: Epilepsy: a review of selected clinical syn- therapeutic implications. Epilepsia 50:678–693, 2009 dromes and advances in basic science. J Cereb Blood Flow 22. Ludvig N, Kovacs L, inventors: Hybrid neuroprosthesis for the M e ­t a b 26:983–1004, 2006 treatment of brain disorders. US Patent 6,497,699, 2002 39. Stein AG, Eder HG, Blum DE, Drachev A, Fisher RS: An au- 23. Ludvig N, Kuzniecky RI, Baptiste SL, John JE, von Gizycki tomated drug delivery system for focal epilepsy. Epilepsy Res H, Doyle WK, et al: Epidural pentobarbital delivery can pre- 39:103–114, 2000 vent locally induced neocortical seizures in rats: the prospect 40. Tamminga CA, Crayton JW, Chase TN: Muscimol: GABA of transmeningeal pharmacotherapy for intractable focal epi- agonist therapy in schizophrenia. Am J Psychiatry 135:746– lepsy. Epilepsia 47:1792–1802, 2006 747, 1978 24. Ludvig N, Medveczky G, French JA, Carlson C, Devinsky 41. Tang HM, French JA, Kuzniecky RI, Devinsky O, Ludvig N: O, Kuzniecky RI: Evolution and prospects for intracranial Periodic transmeningeal muscimol maintains its antiepileptic pharmacotherapy for refractory epilepsies: the subdural hy- efficacy over three weeks without inducing tolerance, in rats. brid neuroprosthesis. Epilepsy Res Treat, 2010 (http://www. Neurosci Lett 494:135–138, 2011 hindawi.com/journals/ert/2010/725696.html) [Accessed April 42. Yokoi I, Kabuto H, Mori A: Effects of GABAmimetics on 12, 2012] electrocorticographic spike discharges induced by guanidino- 25. Ludvig N, Medveczky G, Rizzolo R, Tang HM, Baptiste SL, ethanesulfonic acid (amidino-taurine) in the rat. Neurochem Res 18: Doyle WK, et al: An implantable triple-function device for 533–538, 1993 local drug delivery, cerebrospinal fluid removal and EEG re- 43. Yoshor D, Bosking WH, Lega BC, Sun P, Maunsell JHR: Lo- cal cortical function after uncomplicated subdural electrode cording in the cranial subarachnoid space of primates. J Neu- implantation. Laboratory investigation. J Neurosurg 108: rosci Methods 203:275–283, 2012 139–144, 2008 26. Ludvig N, Sheffield LG, Tang HM, Baptiste SL, Devinsky O, Kuzniecky RI: Histological evidence for drug diffusion across the cerebral meninges into the underlying neocortex in Manuscript submitted August 24, 2011. rats. Brain Res 1188:228–232, 2008 Accepted April 9, 2012. 27. Ludvig N, Tang HM, Eichenbaum H, Gohil BC: Spatial mem- Preliminary data were presented in abstract form at the American ory performance of freely-moving squirrel monkeys. Behav Epilepsy Society Meeting, San Antonio, TX, December 6, 2010. Brain Res 140:175–183, 2003 Please include this information when citing this paper: pub- 28. Marciani MG, Gotman J, Andermann F, Olivier A: Patterns lished online Month Day, 2012; DOI: 10.3171/2012.4.JNS111488. of seizure activation after withdrawal of antiepileptic medica- Supplemental online information: tion. Neurology 35:1537–1543, 1985 Video: http://mfile.akamai.com/21490/wmv/digitalwbc.download. 29. Martin RF, Bowden DM: Template Atlas of the Primate akamai.com/21492/wm.digitalsource-na-regional/jns11-1488_ Brain. Seattle: Primate Information Center, University of video_.asx (Media Player). Washington, 1997 http://mfile.akamai.com/21488/mov/digitalwbc.download. 30. Monge-Acuña AA, Fornaguera-Trías J: A high performance akamai.com/21492/qt.digitalsource-global/jns11-1488_video.mov liquid chromatography method with electrochemical detec- (Quicktime). tion of gamma-aminobutyric acid, glutamate and glutamine Address correspondence to: Nandor Ludvig, M.D., Ph.D., NYU in rat brain homogenates. J Neurosci Methods 183:176–181, Comprehensive Epilepsy Center, 223 East 34th Street, New York, 2009 New York 10016. email: [email protected].

14 Not for Reprint J Neurosurg / Month Day, 2012