Real-time Closed-loop Control in a Rodent Model of Medically Induced Coma Using Burst Suppression ShiNung Ching, Ph.D.,* Max Y. Liberman, B.Sc.,† Jessica J. Chemali, B.Sc.,† M. Brandon Westover, M.D., Ph.D.,‡ Jonathan D. Kenny,† Ken Solt, M.D.,§ Patrick L. Purdon, Ph.D.,║ Emery N. Brown, M.D., Ph.D.# ABSTRACT What We Already Know about This Topic • Medically induced coma with burst suppression is used to Background: A medically induced coma is an anesthetic treat status epilepticus and provide cerebral protection after brain injury. Defining a closed-loop anesthesia delivery system state of profound brain inactivation created to treat status for this purpose would be an efficient and new approach. epilepticus and to provide cerebral protection after traumatic brain injuries. The authors hypothesized that a closed-loop anesthetic delivery system could automatically and precisely control the electroencephalogram state of burst suppression What This Article Tells Us That Is New and efficiently maintain a medically induced coma. • A closed-loop anesthesia delivery system using a computer- controlled infusion of propofol can achieve a reliable and ac- curate real-time control of burst suppression in rats. * Research Fellow, Department of Anaesthesia, Harvard Medi- cal School, Boston, Massachusetts; Research Fellow, Department of Anesthesia, Critical Care, and Pain Medicine, Massachusetts General Hospital, Boston, Massachusetts; Research Affiliate, Department of Methods: In six rats, the authors implemented a closed- Brain and Cognitive Sciences, Massachusetts Institute of Technol- loop anesthetic delivery system for propofol consisting of: ogy, Cambridge, Massachusetts. † Research Assistant, Department of Anesthesia, Critical Care, and Pain Medicine, Massachusetts General a computer-controlled pump infusion, a two-compartment Hospital. ‡ Instructor, Department of Neurology, Harvard Medical pharmacokinetics model defining propofol’s electroencepha- School; Assistant in Neurology, Department of Neurology, Massachu- logram effects, the burst-suppression probability algorithm setts General Hospital.§ Assistant Professor, Department of Anaes- thesia, Harvard Medical School; Assistant Anesthetist, Department of to compute in real time from the electroencephalogram the Anesthesia, Critical Care, and Pain Medicine, Massachusetts General brain’s burst-suppression state, an online parameter-estima- Hospital; Research Affiliate, Department of Brain and Cognitive Sci- tion procedure and a proportional-integral controller. In the ences, Massachusetts Institute of Technology. ║ Instructor, Depart- ment of Anaesthesia, Harvard Medical School; Instructor, Department control experiment each rat was randomly assigned to one of of Anesthesia, Critical Care, and Pain Medicine, Massachusetts Gen- the six burst-suppression probability target trajectories con- eral Hospital; Research Affiliate, Department of Brain and Cognitive structed by permuting the burst-suppression probability lev- Sciences, Massachusetts Institute of Technology. # Warren M. Zapol Professor of Anaesthesia, Department of Anaesthesia, Harvard Medi- els of 0.4, 0.65, and 0.9 with linear transitions between levels. cal School; Anesthetist, Department of Anesthesia, Critical Care, and Results: In each animal the controller maintained approxi- Pain Medicine, Massachusetts General Hospital; Professor of Com- mately 60 min of tight, real-time control of burst suppression putational Neuroscience, Edward Hood Taplin Professor of Medical Engineering, Institute for Medical Engineering and Sciences, Depart- by tracking each burst-suppression probability target level ment of Brain and Cognitive Sciences, Harvard-MIT Health Sciences for 15 min and two between-level transitions for 5–10 min. and Technology Program, Massachusetts Institute of Technology. The posterior probability that the closed-loop anesthetic Received from the Department of Anesthesia, Critical Care, and delivery system was reliable across all levels was 0.94 (95% Pain Medicine, Massachusetts General Hospital, Boston, Massachusetts. Submitted for publication January 10, 2013. Accepted for publication CI, 0.77–1.00; n = 18) and that the system was accurate May 10, 2013. Emery N. Brown, Patrick L. Purdon, ShiNung Ching, across all levels was 1.00 (95% CI, 0.84–1.00; n = 18). Max Y. Liberman, Jessica J. Chemali, and Ken Solt have applied for a Conclusion: The findings of this study establish the feasi- patent on the CLAD system presented in the article. Support was pro- vided by grants DP1-OD003646 (to Dr. Brown), ROI-GM10498 (to Dr. bility of using a closed-loop anesthetic delivery systems to Brown), DP2-OD006454 (to Dr. Purdon), and K08-GM094394 (to Dr. achieve in real time reliable and accurate control of burst Solt) from the National Institutes of Health, Bethesda, Maryland, and suppression in rodents and suggest a paradigm to precisely the Department of Anesthesia, Critical Care, and Pain Medicine, Mas- sachusetts General Hospital, Boston, Massachusetts. Dr. Ching holds a control medically induced coma in patients. Career Award at the Scientific Interface (1010625) from the Burroughs- Wellcome Fund, Durham, North Carolina. The first four authors con- EDICALLY induced coma is an anesthetic state of tributed equally to this work. profound unconsciousness and brain inactivation Address correspondence to Dr. Brown: Department of Anesthe- M sia, Critical Care, and Pain Medicine, Massachusetts General Hospital, created to treat status epilepticus and to facilitate recovery 55 Fruit Street, Gray-Bigelow 444, Boston, Massachusetts 02114. enb@ after traumatic brain injuries.1–3 When treating status epilep- neurostat.mit.edu. Information on purchasing reprints may be found at ticus, a hypnotic, such as propofol or a barbiturate, is used to www.anesthesiology.org or on the masthead page at the beginning of 2,3 this issue. Anesthesiology’s articles are made freely accessible to all read- directly inhibit seizure activity. After a brain injury these ers, for personal use only, 6 months from the cover date of the issue. drugs are administered to provide brain protection by reduc- 1 Copyright © 2013, the American Society of Anesthesiologists, Inc. Lippincott ing cerebral blood flow and metabolism. In both cases the Williams & Wilkins. Anesthesiology 2013; 119:848-60 Anesthesiology, V 119 • No 4 848 October 2013 Downloaded From: http://anesthesiology.pubs.asahq.org/pdfaccess.ashx?url=/data/journals/jasa/930989/ on 04/03/2018 PERIOPERATIVE MEDICINE anesthetic is titrated to achieve a specific clinical target that Materials and Methods indicates a state of large-scale brain inactivation. A standard Animal Care and Use approach is to monitor the patient’s brain activity continu- These animal studies were approved by the Subcommittee ously with an electroencephalogram and use a specified level on Research Animal Care, Massachusetts General Hospi- of burst suppression as the target. Burst suppression is an tal, Boston, Massachusetts. Six male Sprague–Dawley rats electroencephalogram pattern indicating a state of highly (Charles River Laboratories, Wilmington, MA) weighing reduced electrical and metabolic activity during which 377–460 g were used for these studies. Animals were kept periods of electrical bursts alternate with isoelectric periods on a standard day–night cycle (lights on at 7:00 AM, and off termed suppressions.4–6 at 7:00 PM), and all experiments were performed during the No established guidelines exist for specifying the level of day. We use rats as our experimental system because they are 27,28 burst suppression required for a medically induced coma. A an established model for study of burst suppression. target level is chosen, and control of that level is managed Instrumentation and Preparation by continually monitoring the electroencephalogram and Extradural electroencephalogram electrodes were surgically manually adjusting the drug infusion rate. A common goal of implanted at least 7 days before experimentation as previ- medically induced coma is maintaining a reduction in brain ously described.29,30 Briefly, general anesthesia was induced activity for 24 h or more, periods significantly longer than any and maintained with isoflurane. A microdrill (Patterson Den- human operator can maintain tight control. Defining a pre- tal Supply Inc., Wilmington, MA) was used to make four cise, quantitative target level of burst suppression and design- holes at the following stereotactic coordinates: A0L0, A6L3, ing a closed-loop anesthetic delivery (CLAD) system for A6L-3, and A10L2 relative to the lambda.29,30 An electrode maintaining that target would be a more efficient approach. with mounting screw and socket (Plastics One, Roanoke, Closed-loop anesthetic delivery systems for control of VA) was screwed into each hole, and the sockets were inserted unconsciousness and sedation have been extensively stud- in a pedestal (Plastics One). The screws, sockets, and ped- ied.7–26 Although no CLAD system has been designed to man- estal were all permanently fixed with dental acrylic cement, age medical coma in humans, Vijn and Sneyd27 implemented and the animal underwent a minimum recovery period of 7 a CLAD system to test new anesthetics in rodents using as days. The potential difference between electrodes A0L0 and the control signal the burst-suppression ratio; the fraction of A6L-3 (left somatosensory cortex) was recorded. The signal time per 15 s that the electroencephalogram is suppressed. For was referenced to A10L2 and recorded using