Ketamine infusion for patients receiving extracorporeal membrane oxygenation support: a case series

The use of ketamine infusion for sedation/analgesia in patients receiving extracorporeal membrane oxygenation (ECMO) therapy has not been described. The aims of this retrospective cohort study were to explore whether ketamine infusion for patients requiring ECMO therapy was associated with altered RASS scores, decreased concurrent sedative or opioid use, or with changes in vasopressor requirements. All patients on ECMO who received ketamine infusions in addition to sedative and/or opioid infusions between December 2013 and October 2014 at Barnes-Jewish Hospital in St. Louis were retrospectively identified. Patient characteristics and process of care data were collected. A total of 26 ECMO patients receiving ketamine infusion were identified. The median (inter quartile range [range]) age was 40 years (30-52 [25-66]) with 62% male. The median starting infusion rate of ketamine was 50 mg/hr (30-50 [6-150]) and it was continued for a median duration of 9 days (4-14 [0.2-21]). Prior to ketamine, 14/26 patients were receiving vasopressor infusions to maintain hemodynamic stability. Ketamine initiation was associated with a decrease in vasopressor requirement in 11/26 patients within two hours, and 0/26 required an increase (p<0.001). All patients were receiving sedative and/or opioid infusions at the time of ketamine initiation; 9/26 had a decrease in these infusions within two hours of ketamine initiation, and 1/26 had an increase (p=0.02; odds ratio for decrease to increase = 9; 95% CI, 1.14 to 71.04). The median (IQR[range]) RASS score 24 hours before ketamine initiation was -4 (-3 to -5, [0 to -5]) and after ketamine was -4 (-3 to -4 [-1 to -5]) ( P = 0.614). Ketamine infusion can be used as an adjunctive sedative agent in patients receiving ECMO and may decrease concurrent sedative and/or opioid infusions without altering RASS scores. The hemodynamic effects of ketamine may provide the benefit of decreasing vasopressor requirements.

Continuous infusion of sedatives and opioids are commonly employed to alleviate pain, agitation, and anxiety in critically ill patients admitted to the intensive care unit 1-2 . Significant variability in sedation practices exist nationwide in the United States 3 , and efforts have been made to investigate systematic approaches to sedation management in order to improve patient outcomes 4 . As the adverse effects of over-sedation and prolonged sedation have become more apparent (e.g. longer duration of mechanical ventilation, longer ICU length of stay, increased mortality) 5-7 , the paradigm has shifted towards a "more is less" strategy 8 . Keeping critically ill patients less sedated and more interactive is supported by recent clinical practice guidelines on pain, agitation, and delirium 9 . However, severe agitation is common in critically ill patients and may have many underlying causes; agitation in such patients is associated with prolonged ventilator and ICU days as well as higher rates of self-extubation 10 . It is therefore important to determine the optimum level of sedation for every individual critically ill patient. Patients requiring extracorporeal membrane oxygenation (ECMO) therapy typically require sedation during the acute phase of their illness or when they are agitated such that they are at risk for selfharm, such as displacement of ECMO tubing. However, patients receiving ECMO support are challenging to sedate, often require escalating doses of sedatives and opioids to maintain an appropriate, safe level of sedation 11 . The challenges often relate to the dynamic alterations in pharmacokinetic (PK) and pharmacodynamic (PD) properties of commonly used sedatives and opioids in the setting of ECMO. Hemodilution, protein binding changes, changes in blood flow especially through the lungs, sequestration of medications by the ECMO circuit, binding of drugs to the oxygenator membrane, and organ function changes may impact the clinical effects of sedative and analgesic medications in this patient population 10-19 .
Ketamine is potentially a promising complementary sedative agent in the setting of ECMO in that it has hypnotic, analgesic, and amnesic properties owing to its activity at a variety of receptors. Ketamine is an antagonist of glutamate receptors as well as N-methyl-D-aspartate (NMDA) receptors and it has activity at all opioid receptors, though with different affinities and activities at those receptors. Ketamine also antagonizes nicotinic acetylcholine receptors noncompetitively and profoundly inhibits muscarinic receptors in addition to inhibiting L-type calcium channels 20 . Ketamine's lack of respiratory depression, bronchodilating properties, in addition to generally being well tolerated hemodynamically (despite some myocardial depressant properties) theoretically make it an attractive choice for sedation and analgesia 21 . Recent clinical practice guidelines on pain, agitation, and delirium 9 mention the use of ketamine as a possible adjunct for non-neuropathic pain management and suggest that nonbenzodiazepine sedatives such as propofol and dexmedetomidine may be preferred over benzodiazepine sedation. They do not provide recommendations for ketamine use as a continuous infusion. To our knowledge, no studies have compared clinical outcomes in ICU patients sedated with ketamine infusion to other agents. Ketamine may provide an alternative or adjunct option in difficult to sedate patients. As critical drug shortages continue to impact the selection of available drugs (e.g. propofol shortage 22 ), it is important to review all relevant agents available for clinical use. The purpose of this study was to describe the use of ketamine infusion for patients receiving concurrent sedation/analgesia on ECMO support. The specific aims were to determine whether ketamine altered Richmond Agitation Sedation Scale (RASS) scores or decreased concurrent sedative, opioid, or vasopressor requirements.

Methods
The Washington University School of Medicine Human Research Protection Office and the Protocol Review and Monitoring Committee approved this study (IRB # 201409142). This case series was conducted at Barnes-Jewish Hospital, a 1,250 bed urban teaching hospital affiliated with the Washington University School of Medicine in St. Louis, MO, between December 2013 and October 2014. All patients on ECMO support receiving ketamine infusion in addition to sedative and/or opioid infusions during the study period were retrospectively identified via an informatics query. There were no exclusion criteria. Patient characteristics, medical and surgical history, as well as process of care data were collected from electronic medical records. All ECMO patients are cared for in a single cardiothoracic intensive care unit. Doses and durations of all infusions of sedatives, opioids and vasopressor agents documented as administered during ketamine administration were collected. All available RASS scores charted 24 hours before and after ketamine initiation were collected on each patient. Statistical analysis was completed by using the SPSS software, version 18.0 (SPSS, Inc., Chicago, IL). Descriptive statistics were used to analyze the data collected. The median RASS scores as well as the median norepinephrine infusion rates (in patients who were receiving norepinephrine) before and after ketamine initiation were compared with a Wilcoxon signed-rank test. The McNemar test was used to assess whether or not a significant proportion of patients had a clinically meaningful increase or decrease in the dose of sedative, analgesic or vasopressor infusions two hours after ketamine initiation. Based on consensus of the investigators, the following minimum changes were arbitrarily pre-specified as clinically meaningful for sedative, analgesic and vasopressor infusions: propofol, 10 mcg/kg/min; midazolam, 1 mg/hour; dexmedetomidine, 0.2 mcg/kg/hour; fentanyl, 25 mcg/hour; norepinephrine 0.02 mcg/kg/min; and vasopressin 0.02 units/min.   duration of 9 days (4-14 [0.2-21]). Of note, 6 patients were receiving concurrent neuromuscular blocking agents during ketamine infusion (Table 2).

Discussion
ECMO has repeatedly shown to complicate sedation management and optimization in the ICU as the PK/PD of sedatives and opioids become altered 11,12,14-18 . Lorazepam, midazolam, diazepam, propofol, and morphine have all been shown to be sequestered by ECMO circuits to some extent 11,17,18 , posing a challenge when faced with trying to provide appropriate sedation/analgesia. In an ex vivo study, lipophilic medications such as fentanyl and midazolam have been shown to sequester in the ECMO circuit resulting in significant loss, while morphine remained relatively stable 18 . A similar finding was shown by Shekar and collegaues 11 who assessed sedation requirements in 30 patients receiving ECMO support. Significant increases in dosing requirements of midazolam and morphine were noted, with venovenous ECMO patients receiving higher sedative doses overall than patients on venoarterial ECMO. The effect of ECMO on the PK/PD and other drug properties of ketamine in patients on ECMO has not been studied.
Ketamine causes dissociation of the thalamus from the limbic cortex, resulting in patients resembling a cataleptic state. Patients may be unable to respond to sensory stimulation, may have nystagmus, and are able to conserve laryngeal and corneal reflexes 21 . Ketamine has been shown to decrease opioid consumption in surgical patients 23,24 , and is typically used as a one-time administration for analgesia, anesthesia induction, or procedural sedation. In addition to these indications, in some of the ICUs at our institution ketamine infusion has been added for patients who have been difficult to sedate to a target RASS. The use of ketamine infusion has particularly been adopted at our institution for patients requiring ECMO therapy, as these patients have anecdotally been difficult to sedate optimally. However, the safety and efficacy of ketamine for these patients has not been established.
The favorable hemodynamic and PK/PD profile that ketamine provides in patients not requiring ECMO therapy makes ketamine theoretically an attractive sedative agent generally for use in the ICU. The onset of action after IV administration occurs within 30 seconds, with a maximum effect in about 1 minute. The distribution half-life is 5-10 minutes while the elimination half-life is 2-3 hours, as ketamine undergoes extensive hepatic uptake and is primarily excreted in the urine 21,25 . There are no dosing adjustments provided by the manufacturer in the presence of hepatic or renal dysfunction, however the optimal dose and duration of ketamine infusion for sedation remains unknown. Interestingly, in recent years neuroprotection, immune modulating, and antidepressant properties have also been suggested for ketamine 26-28 .
Ketamine's lack of respiratory depression, bronchodilation properties and ability to increase blood pressure, heart rate, and cardiac output (although it does act as a myocardial depressant) further add to the attractive profile of this agent 21,29 . Interestingly, a decrease in vasopressor requirements with ketamine initiation was observed in many of the patients in this series. Although charted RASS scores before and during ketamine infusion did not change, a decrease in sedatives and/or opioids within two hours of ketamine initiation was more common than an increase in sedatives and/or opioids.
This study has several limitations. The small sample size, lack of blinding and retrospective design limit our ability to make causal inferences. Although the majority of patients receiving vasopressors had a meaningful decrease in their requirements, we cannot exclude the fact that a decrease in vasopressors may have occurred unrelated to ketamine infusion. The observational design also limits our ability to determine why certain drugs were chosen concurrently or why one was decreased prior to another. Although RASS values were collected, we were unable to determine the specified RASS goal for each patient. A RASS goal of 0 to -2 is targeted in most ICU patients, however 6 patients in this series were receiving neuromuscular blocking agents, rendering RASS values uninterpretable, while other patients received escalating doses of sedatives/opioids despite RASS values < -2. Furthermore, the use of a RASS score to monitor and guide sedation with ketamine has not been validated.
To our knowledge, this is the first study evaluating the use of ketamine infusion in patients on ECMO support. This study demonstrated that ketamine infusion can be used as an adjunctive agent in patients receiving ECMO with possible sedative/opioid sparing effects. The cardiovascular properties may provide the hemodynamic benefit of reducing vasopressor requirements. The lack of understanding related to optimizing pharmacotherapy in patients on ECMO has led to ongoing research, aimed at improving patient care 13,19,30 . All appropriate drugs should be considered in the absence of clear standards or guidelines for analgesia/sedation in patients receiving ECMO support. More rigorous study is warranted to further investigate the suitability of ketamine for sedation in patients receiving ECMO support.

Introduction
The introduction provides a clear and concise summary of the clinical background and relevance of the study. The proposed hypothesis is clinically very relevant and plausible. The specific aims of the study are stated clearly.

Methods
The description of the methods is completely clear. Appropriate methodology was applied. To assess the influence of ketamine on sedative and vasopressor requirements, the authors arbitrarily chose, by consensus, definitions of " clinically meaningful minimal changes in medication after the start of ketamine ". This is a reasonable approach. The statistical analysis is clearly described and is appropriate. infusion

Results
We have two comments about the presentation of the results. The first is that table 2 is somewhat confusing. It is not clear what the "maximum" and "minimum" infusion rates refer to. We suggest that the authors clearly state the time period over which these maximum and minimum infusion rates were determined. If these rates are not the rates after starting ketamine, then we suggest that the authors present the median (IQR, range) infusion rates during some (arbitrary) time period after starting ketamine. Our second comment about the results concerns the consistency of units used in presentation of sedative and vasopressor infusion rates. In the methods the authors state arbitrarily chosen definitions of clinically meaningful changes in medication. These thresholds are defined in mcg/kg/min units for each of the 1 2 1 2 meaningful changes in medication. These thresholds are defined in mcg/kg/min units for each of the drugs. In the text and in table 2 of the results, the infusion rates of fentanyl and midazolam are reported in different units (mcg/min). Thereafter the authors only report the numbers of patients in which there was a clinically relevant change. To more easily give the reader an idea of the magnitude of the relative changes in infusion rate of fentanyl and midazolam, we suggest that for these drugs the before and ketamine infusion rates are also presented in mcg/kg/min.

Discussion
Discussion is very well written and informative. The study limitations are well recognised. The conclusions are supported by the data.