Hypothesis: Cortical spreading depolarizations are inhibited by the NMDA receptor antagonist Ketamine Aim 1: To demonstrate, in a group of patients with acute severe brain injury requiring surgery including traumatic brain injury and aneurysmal subarachnoid hemorrhage, whether use of continuous infusion of ketamine decreases frequency of occurrence of cortical spreading depolarizations.

Cortical spreading depolarizations (CSD) are massive events which recently have been observed in many types of acute brain injury and likely lead to expansion of injury. These "brain tsunamis" are unlike any other type of brain electrical event (such as seizures or normal neuronal transmission) in that they progress very slowly across the surface of the brain (2-5mm/minute) and involve near complete depolarization of the neurons and other cells. The only similar event in neurophysiology is an anoxic depolarization, which is the final wave of loss of cell function preceding death in cells suffering severe, irreversible hypoxia or ischemia(1). In the case of CSD, the cell is able to recover function, however, at an enormous metabolic expense. Massive amounts of energy substrate (ATP, glucose, oxygen) as well as the delivery system to bring these substrates (blood flow) are required to restore the normal ionic gradient of the cell membrane and cell function. Because of this loss of function of cells, normal electrocortical (ECog) activity is transiently lost, resulting in a depression of the high frequency cortical activity, which is why the phenomenon is also frequently referred to as "cortical spreading depression." CSD has been definitively documented to occur after many types of acute brain injury including ischemic stroke, aneurysmal subarachnoid hemorrhage, intracerebral hemorrhage, and severe traumatic brain injury(2,3). The true incidence is, for the time being, unknown, in that the measurement technique requires placement of a small cortical electrode at the time of a surgical procedure. This limits the region of measurement to relatively small area in patients undergoing surgery, however even in this very small sample, the incidence of delayed SD after brain injury ranges from 53-88%(4). Efforts are underway to attempt to measure CSD less invasively(5) or non-invasively(6,7), however these techniques are currently under exploration and do not have the robust reliability of the cortical electrode system.

Mounting human data coupled with extensive animal data supports the assertion that CSD is not only a marker in response to severe brain injury, but in fact, plays a causal role in injury propagation(8). Animal data is fairly definitive in this assertion, in that CSD can be studied in uninjured brain and inducing CSD leads to neuronal death, particularly with repeated events. Note the progressive loss of brain electrical activity with repeated CSD in the figure to the right. In animal models, CSD clearly leads to expansion of injury, particularly in ischemic stroke models. Human data is unavoidably observational to this point, however by observing multiple physiologic modalities, the deleterious effects become clear. A spectrum of local blood flow responses to CSD have been observed, ranging from a wave of hyperemia (termed the normal hemodynamic response) to a wave of ischemia (termed the inverse hemodynamic response(9, 10)). The factors that determine the response likely have to do with the availability of substrate (glucose, oxygen) and delivery (blood flow) coupled with the baseline metabolic state of the tissue (depressed metabolic state may be more resistant). When the inverse hemodynamic response is observed, an associated wave of tissue hypoxia is observed, which becomes linearly more hypoxic with repeated CSD in a short interval(11). Brain metabolism also can be measured during CSD, and consistent metabolic challenge is noted, with increased micro dialysis lactate and decreased glucose(12). In the case of repeated events, this glucose depletion becomes progressive due to inadequate time for the tissue to recover between these massive events leading to progressive ischemia(12).

From a clinical perspective, the metabolic data can support a deleterious effect, but the effect on clinical outcome is critical in determining if the events are relevant as a potential target for therapy. The occurrence and severity of CSD has been closely linked to both development of new stroke as well as clinical outcome in both retrospective and prospective series. In subarachnoid hemorrhage, Dreier(13) reported a direct association with clinical delayed ischemic neurologic defect (DIND) and the presence of a cluster of SD. Furthermore, in this small series, the patients who went on to develop stroke had markedly longer periods of depression, indicating inability of the tissue to recover from the event compared to patients without delayed stroke. The most extensive clinical outcome data is from traumatic brain injury (TBI)(14,15) where the presence of any CSD showed a non-significant trend toward predicting worse outcome, however CSD occurring in already dysfunctional tissue (termed isoelectric spreading depolarization or ISD) was stronger predictor of clinical outcome than a composite score of most standard variables through to predict outcome (OR 7.58 (95%CI 2.64-21.8) for ISD compared to 1.76(95%CI 1.26-2.46) for the composite prognostic score)(15).

This mounting observational data as to the deleterious effects of CSD has led to increased excitement regarding CSD as a novel target for prevention of delayed injury after diverse types of acute brain injury(16). The optimal target or agent has not been defined, but there are promising animal data supporting a wide variety of agents, primarily targeting NMDAVR, as this is thought to be an important factor in propagation of SD(17). Initial clinical case reports of the effect of ketamine being used as sedation in patients with severe TBI(18) led to a larger scale effort to retrospectively study the various anesthetics used for standard clinical care on the frequency of CSD in monitored patients(19). Using only the sedation medications for which there were >1000 cumulative hours of ECog recording while on that medication, the effects of propofol, fentanyl, midazolam, ketamine, morphine, and sufentanyl were examined. The study found a consistent effect of ketamine in decreased probability of CSD/h per patient. This was nearly linearly dose dependent, and importantly, in multivariate analysis, ketamine still emerged as having a significant effect on decreasing both occurrence of CSD as well as the occurrence of the more deleterious clusters of CSD(19).

Though ongoing observational data is still clearly needed to better characterize the susceptibility and effects of CSD, in order to move toward trial of CSD directed therapy, a prospective trial of the effect of ketamine on the occurrence of CSD is necessary to confirm these retrospective observations and establish the precedent for future therapeutic trials. The SAKS trial will provide important confirmatory pilot data to direct the implementation of future trials.

This is a prospective, randomized, controlled, multiple cross-over trial evaluating the efficacy of ketamine in the suppression of CSDs. This multiple crossover design was chosen in order to be able to develop preliminary data which could guide implementation of future multicenter trials. Because of the significant variability between patients, a study randomized by patients would be subject to a large amount of potential bias. Because factors such as time of day or hospital day also are known to affect CSD, a brief crossover period of 6 hours was chosen. The study will be registered with clinicaltrials.gov prior to enrollment of patients. Patients with severe traumatic brain injury or subarachnoid hemorrhage who fit the inclusion/exclusion criteria will be approached by either research coordinators or study investigators who will consent the LAR for the study prior to clinically indicated craniotomy. It is not expected that patients will be able to independently consent given the severity of the condition, however, if the patient is conscious, attempts will be made to discuss the study with him or her as well.

The patient's surgical procedure will be carried out as planned. The only alteration of the surgical procedure will be the placement of a subdural electrode strip (1x6 cortical strip: Integra: Plainsboro, NJ) on the brain cortex adjacent to the operative site at the end of the procedure. These strips are standard, FDA approved, disposable, pre-sterilized devices used routinely for epilepsy monitoring. In addition, the investigators have used these strips as part of our post-injury IRB approved protocol (10-159) for many days after surgery. The cortical strip (plus a dermal reference electrode on the mastoid or apex of the skull) will be monitored with a Moberg CNS monitor. (Moberg Research, Ampler, PA).The Moberg monitor is a modified version of a standard clinical use multiparametric monitoring system shown below which was cleared by the FDA in 2008. The only difference is the ECog amplifier, which allows for direct full frequency spectrum DC recording.

Upon arrival, post-operatively, to the Neurosciences Intensive Care Unit, the patient will have randomization completed via online randomization program. Randomization will be to allocate patients to either of two groups: 1) Ketamine first or 2) Propofol/other first. No secondary randomization criteria are thought to be necessary given the small sample size for this pilot trial. Initiation of the protocoled sedation regimen will begin on the next hour divisible by 6 (i.e. 06:00, 12:00, 18:00, 24:00). The randomization will determine which sedative to start, and after that the ketamine and propofol/other infusions will be alternated every 6 hours on the above schedules.

Dosages of these sedating medications will not be standardized, but rather titrated to clinical effect. The clinical effect will be determined by the attending intensivist based on the patient's clinical needs. This level of sedation will be communicated to nursing via the Riker Sedation-Agitation Score(20). A minimal dose of ketamine (0.1mg/min or 6mg/hr) will be infused during the ketamine periods, which is lower than required to induce sedation. No minimal sedation requirements will exist for the propofol or other regimen period. This will be done to test the effect of ketamine (which is hypothesized to affect frequency of SD) compared to other sedations regimens (which are not thought to affect SD.) Each period of adjustment of the sedation regimen will be treated as a "spontaneous breathing trial" which is a common standard of care procedure for nursing which involves holding sedation to determine a patient's neurologic exam and respiratory ability with subsequent titration back to appropriate clinical effect. These sedation breaks are very common in the ICU and titration to the desired clinical effect will be performed with the appropriate drug per the standard ICU nursing protocols. In the event that the patient no longer needs invasive positive pressure ventilation prior to discontinuation of ] neuromonitoring, propofol/other sedation intervals will not have mandatory sedative infusions, however, ketamine intervals will have a basal dose of 0.1mg/min (6mg/hr). 'The sedation protocol will end when the strip is removed. This is determined by the patients critical care needs. The strip is checked daily for function as well as any sign of problem such as leak of CSF. Once other critical care monitoring is discontinued (such as ventricular drains and invasive monitoring) the strip will be removed. Other endpoints will include any sign of CSF leak, adverse event reported, or treating intensivist does not think alternating sedation is safe.

During the sedation protocol, cortical electroencephalographic monitoring with the cortical electrodes will be continuously recording. Other physiologic data obtained clinically (including, but not limited to, vital signs, arterial wave forms, laboratory values, video EEG) will be subject to review and data collection for correlation with occurrence of SD. This data is obtained as part of standard of care and stored in a departmental server in an anonymous fashion. Clinical video EEG will be obtained on the majority of patients (if not all patients) as part of standard multimodal monitoring. This video will be reviewed to look for any external stimuli that might induce cortical spreading depressions.

Inclusion Criteria:

• GCS <8
• SAH or severe traumatic brain injury requiring craniotomy
• Consent obtainable (via legal representative)
• Ictus (bleed or injury) within 48 hours of enrollment
• Clinically appropriate for multimodality monitoring

Exclusion Criteria:

• Anticipated survival <48 hours
• No craniotomy
• Infratentorial craniotomy only•Unable to obtain consent
• Absence of clinically used multimodality monitoring
• Prisoners
• Pregnant