Ketamine restriction correlates with reduced cholestatic liver injury and improved outcomes in critically ill patients with burn injury

Background & Aims Ketamine-associated cholestatic liver injury is reported in patients with severe burn injury, but its association with patient outcome is unclear. We investigated the relationship between ketamine exposure, cholestatic liver injury, and outcome of critically ill patients with burn injury. Methods In a retrospective study, patients with severe burn injury were analysed across two periods: unrestricted ketamine prescription (ketamine-liberal) and capped ketamine dosage (ketamine-restricted). The primary endpoint was cholestatic liver injury, and the secondary endpoint was 3-month mortality. Binary logistic regression models and the revised electronic causality assessment method were used to measure the strength of associations and causality assessment, respectively. Results Of 279 patients (median age 51 [IQR 31–67] years; 63.1% men; burned surface area 28.5%, IQR 20–45%), 155 (56%) were in the ketamine-liberal group, and 124 (44%) were in the ketamine-restricted group, with comparable clinical characteristics, except for ketamine exposure (median doses 265.0 [IQR 0–8,021] mg and 20 [IQR 0–105] mg, respectively; p <0.001). A dose- and time-dependent relationship was observed between ketamine exposure and cholestatic liver injury. Ketamine restriction was associated with a reduced risk of cholestatic liver injury (adjusted odds ratio 0.16, 95% CI 0.04–0.50; p = 0.003) and with a higher probability of 3-month survival (p = 0.035). The revised electronic causality assessment method indicated that ketamine was probably and possibly the cause of cholestatic liver injury for 14 and 10 patients, respectively. Cholangitis was not observed in the ketamine-restricted group. In propensity-matched patients, the risk of 3-month mortality was higher (adjusted odds ratio 9.92, 95% CI 2.76–39.05; p = 0.001) in patients with cholestatic liver injury and ketamine exposure ≥10,000 mg. Other sedative drugs were not associated with liver and patient outcome. Conclusions In this cohort, ketamine restriction was associated with less cholestatic liver injury and reduced 3-month mortality. Impact and implications In a cohort of 279 critically ill patients with burn injury, ketamine was associated with a risk of liver bile duct toxicity. The risk was found to be dependent on both the dosage and duration of ketamine use. A restriction policy of ketamine prescription was associated with a risk reduction of liver injury and 3-month mortality. These findings have implications for the analgesia and sedation of critically ill patients with ketamine, with higher doses raising safety concerns.


Introduction
Ketamine is an intravenous hypnotic agent used in critically ill patients for rapid sequence induction 1 ; for the management of acute and chronic pain, 2,3 including patients with severe burn injury 4 ; and for maintenance sedation of patients with acute respiratory distress syndrome. 5rolonged ketamine use has been associated with organ injuries, including ulcerative cystitis 6 and hepatic toxicities, such as cholestatic liver injury, cholangitis, and sclerosing cholangitis. 7In 2017 and 2018, two pharmacovigilance alerts were released by the French National Agency for Medicines and Health Products Safety (ANSM): one on severe liver injuries following prolonged ketamine administration in patients with burn injury 8,9 and one on severe cholestatic liver injury, uro-nephrological injuries (including acute kidney injury), and endocrinological disturbances in patients or street users with prolonged/chronic ketamine exposure. 8,9e previously reported a 10% rate of cholestatic liver injury in a 2012-2015 retrospective cohort of patients with severe burn injury. 10Following the ANSM alert, we modified our ketamine prescription policy, and we observed a reduction in the incidence of cholestatic liver injury. 11These observations suggested a dosedependent drug-induced cholestatic liver injury.We therefore     explored the relationship between ketamine exposure, cholestatic liver injury, and 3-month mortality in our cohort.

Setting
We conducted a retrospective, single-centre, cohort study in consecutive patients admitted to the Burn Intensive Care Unit (ICU) of Saint Louis Hospital (Assistance Publique-Hôpitaux de Paris, AP-HP, Paris, France) between December 2014 (start of electronic prescriptions) and June 2019.Patients were treated according to our local management protocols. 12The study was approved by the local ethics committee (Comite de protection des personnes IV, St-Louis hospital; Institutional review board 00003835, protocol 2013/17NICB).

Data sources
The data sources were the electronic records available (Diane, Bow Medical, France), the medical charts, the medical prescription database, and the biological data warehouse of the institution.Patient-level data check was performed for all patients by a senior investigator.

Patients
Inclusion criteria were adult patients with at least one of the following: total burned surface area > − 20%, full thickness burned surface area > − 10%, and mechanical ventilation or vasopressor administration during the first 48 h after burn injury.Patients with the following criteria were excluded from the analysis: no liver function test performed during the burn ICU stay, and uncharted sedative and analgesic drugs prescription.

Outcome measures
4][15] We also considered other definitions for cholestatic liver injury, including serum ALP level > − 2 × ULN and serum GGT level > − 1 × ULN based on the drug-induced liver injury (DILI) definition 16 and grade 3 or higher serum ALP elevation (> − 5 × ULN) based on the Common Terminology Criteria for Adverse Events (CTCAE). 17The term cholangitis refers to prolonged inflammation and/or infection of the bile ducts.To assess the imputability of ketamine on cholestatic liver injury occurrence for each individual, we used the revised electronic causality assessment method (RECAM) (http://gihep.com/dili-recam/). 18,19The secondary outcome was 3-month mortality.

Exposures
We defined two time periods, according to ketamine prescription modalities 8,9 : a ketamine-liberal period, from December 2014 to end of March 2017, when ketamine prescription was 'liberally' used for maintenance sedation (> − 1 mg/kg/h); and a ketamine-restricted period, from April 2017 to June 2019, when ketamine was used only as a second-line co-analgesic drug with a capped dose (<0.015 mg/kg/h) and not as a sedative agent.Other anaesthetic drug exposures, including total i.v.midazolam and sufentanil, were used as inner controls.Other exposures were patient demographics; burn characteristics, including inhalation injury; severity of illness scores, including the abbreviated burn severity index (ABSI; a score that ranges from 0 to 18, with higher scores indicating a greater probability of death after the burn injury), the Simplified Acute Physiology Score II (SAPS II; a score that ranges from 0 to 163, with higher scores indicating greater severity of illness), 20 and the Sequential Organ Failure Assessment (SOFA; a score that ranges from 0 to 24 with higher scores indicating more severe organ failure) 21 ; initial crystalloid and norepinephrine administrations; critical care level (number of surgical procedures and parenteral nutrition); and organ failure, including acute kidney injury according to the Kidney Disease Improving Global Outcomes (KDIGO) criteria, 22 acute respiratory distress syndrome (ARDS) according to the Berlin definition, 23  Fig. 1.Probabilty of cholestatic liver injury according to the number of days of ketamine infusion and dose exposure.Probabilities were computed using binary logistic regression according to the restrictive cubic spline method, using four knots.

Statistical analysis
Associations were computed using backward stepwise binary logistic regression models.Variables with nominal two-tailed p values less than 0.1 were entered into the multivariate model, except for variables with obvious multicollinearity.Probabilities of cholestatic liver injury and 3-month mortality by ketamine dose were assessed by binary logistic regression according to the restrictive cubic spline method, using four knots.The knots were determined to have homogenous population distribution.As 38% of the patients did not receive any ketamine, the population could not be cut into quartiles, and we divided the population into four groups, namely, 38%, 38%, 12%, and 12% of the population.To address confounding by indication of ketamine and other source of bias arising from observational data, we estimated a full propensity score matching, without replacement, using the Matchit and Optmatch packages. 25,26Propensity scores were estimated using logistic regression of the likelihood of total ketamine doses > − 1,000 mg, or not, on severity of illness and organ failures (see Fig. S1).We used the 1,000 mg threshold for the two ketamine exposure periods because the risk of cholestatic liver injury increased beyond this threshold (see Fig. 1

Characteristics of patients
Of the 885 patients included in the study, 279 (median age 51 [IQR 31-67] years; 63.1% men) were eligible for analysis: 155 (56%) during the ketamine-liberal period and 124 (44%) during the ketamine-restricted period (see Fig. 2).The majority of patients (96.1%) were admitted after thermal burn injuries, with a median burned body surface area of 28.5%, and one-third (32.6%) had inhalation injury.Patient characteristics by period are outlined in Table 1, demonstrating no statistically significant differences between the two groups in initial severity scores (SAPS II, ABSI, and SOFA), initial resuscitation procedures, and organ failures during the ICU stay, including duration of vasopressor infusion and mechanical ventilation, acute kidney injury, renal replacement therapy, acute respiratory syndrome, sepsis, and septic shock.Liver tests at admission were also comparable between the two periods.
Fig. 3 displays the evolution of liver tests during the ICU stay by ketamine period, showing a progressive elevation of serum ALP, GGT, and total bilirubin in the ketamine-liberal period and not (p <0.001) in the ketamine-restricted period.The evolution of liver tests by year is depicted in Fig. S2, indicating that, except for serum GGT, liver tests remained generally stable during the study period.

Ketamine and risk of cholestatic liver injury
Cholestatic liver injury was associated with severity of illness; critical care level, including number of surgical procedures (p <0.001) and parenteral nutrition; total sedative drugs doses (p <0.001); ketamine-liberal period (p = 0.003); 3-month mortality (p = 0.001), and not 28-day mortality (p = 0.2).Other characteristics are detailed in Table 2.In multivariate analysis (Table 3), the ketamine-restricted period was associated with a lower risk of cholestatic liver injury (adjusted odds ratio [AOR] 0.11, 95% CI 0.02-0.44;p = 0.003).Other associated factors were inhalation injury (p = 0.008), number of surgical procedures (p = 0.018), parenteral nutrition (p = 0.039), renal replacement therapy (p = 0.002), and sepsis (p = 0.017).ARDS was not an independent risk factor for cholestatic liver injury.In the subgroup of patients with ARDS (n = 65; 23.2%), cholestatic liver injury (p = 0.008), acute kidney injury (p = 0.014), and 3-month mortality (p = 0.004) were electively associated with the ketamine-liberal period (see Table S1).There was a linear relationship between the duration of ketamine exposure and the risk of cholestatic liver injury (Fig. 1, left).The association between cholestatic liver injury and ketamine dose was not linear but increased with total ketamine doses > − 1,000 mg (Fig. 1, right).There was no relationship between total midazolam and sufentanil exposures and cholestatic liver injury.Total sufentanil and midazolam doses were not independent risk factors of cholestatic liver injury.Similar results were obtained with other definitions of cholestasis (not shown).

Causality assessment between ketamine and cholestatic liver injury
The medical records of all patients with cholestatic liver injury are documented in Table S1.A total of 19 (54%) patients with cholestatic liver injury progressed to overt cholangitis, characterised by grade 3 (> − 5 N) ALP elevation (n = 16), unexplained prolonged cholestasis (n = 11), biliary sepsis (n = 3), and one case of multiple liver abscess.In addition, four patients had progressive bile duct strictures and dilatations.Magnetic resonance cholangiopancreatography during the ICU stay revealed biliary casts and figures of sclerosing cholangitis in one case (Fig. S3).All patients with cholangitis, except one, were in the ketamine-liberal group.The median total ketamine doses for patients with and without cholangitis were 17,219 and 50 mg, respectively.The single ketamine-restricted patient with common bile duct dilatation had chronic hepatitis C and was undergoing opioid substitutive therapy.Sepsis preceded cholestatic liver injury for 26 (74%) patients, with a median delay of 5 days.
Only two patients experienced very early cholestatic liver injury after burn injury (1 day), and two patients without ketamine had cholestatic liver injury without any apparent risk factor.A post hoc RECAM analysis of all cholestatic liver injury cases indicated that ketamine was most likely the cause for 3 patients, probably the cause for 11, and possibly the cause for 10 cases.
During the ketamine-liberal period, ketamine was unlikely the cause of cholestatic liver injury for five patients, three of whom did not receive ketamine.One patient received ketamine after the onset of cholestatic liver injury, and one patient's liver function temporarily improved while still receiving ketamine.

Summary of the main results
We report a dose-and time-dependent relationship between ketamine and the risk of cholestatic liver injury in critically ill patients with burn injury.The risk increased for ketamine exposures > − 1,000 mg and was maximal for exposures > − 10,000 mg.By contrast, we did not observe any time-or dose-dependent relationship between cholestatic liver injury and midazolam or sufentanil, two drugs commonly given to critically ill patients for sedation and analgesia.A reduction in the prescription policy of ketamine reduced (by 80%) the risk of cholestatic liver injury, and cholangitis, including severe forms of sclerosing cholangitis such as progressive sclerosing cholangitis, and was associated with lower (60%) 3-month mortality.The causality assessment, including patient-level check, and propensity scores matching, suggested that ketamine contributed to cholestatic liver injury in this cohort.

Added value of the study
This study is the first to report a connection between a reduction in ketamine dose exposure and improved liver and patient outcomes.2][33] The link between ketamine-associated cholestatic liver injury and 3month mortality is supported by reports of liver-related deaths in patients with COVID-19 exposed to high ketamine doses, 31,34 emphasising the importance of considering hepatic dysfunction, including cholestatic liver injury, in critical care settings. 35,36Our study's observation of a dose-dependent relationship between ketamine exposure and cholestatic liver injury is consistent with animal models 37 and previous reports of a dose-effect relationship between long-term ketamine infusion and rising total bilirubin levels. 33he study is also first to provide evidence of a time-and dosedependent relationship between ketamine and cholestatic liver injury.In addition, it stands out as the first study to thoroughly assess causality for each event and use propensity score matching in the context of ketamine-induced cholestatic liver injury.

Meaning of the study
In this cohort, we observed both a linear relationship and an exponential relationship between the duration of ketamine exposure and total ketamine drug exposure, and the occurrence of cholestatic liver injury.Notably, it took only 2 and 5 days to reach 1,000 and 10,000 mg of total ketamine drug exposure, respectively. 10The risk of cholestasis, regardless of jaundice, was similar in patients under both ketamine-restricted and ketamine-liberal conditions, suggesting that biliary tract injury is common in patients with severe burn injury, regardless of ketamine use. 10   7 JHEP Reports 2024 vol.6 j 100950 Burn-associated biliary injury likely involves a systemic inflammatory response, shock-induced bile duct ischaemia, sepsis, and beta-lactam. 38There may be a modification of microsomal cytochrome P450 metabolism after burn injury, which could favour ketamine toxicity. 39ur findings suggest that ketamine acts as an additive factor of liver injury in patients with burn-associated biliary injury.Cholangitis, including progressive sclerosing cholangitis, did not occur in patients exposed to low doses of ketamine.
Ketamine is biotransformed in the liver with multiple metabolites.The most important pathway involves N-demethylation of ketamine to norketamine, a water-insoluble by-product, by cytochrome P450 (CYP 3A4) in the liver.Norketamine is then hydroxylated and conjugated to water-soluble compounds that are excreted in the urine. 40Norketamine has been found in the bile and urine after fatal ketamine poisoning. 41Ketamineassociated hepatobiliary injuries are thought to be the consequence of a direct effect of ketamine or toxic intermediates on the biliary epithelial cell. 42lthough liver injury is associated with ICU patient mortality, 43 ketamine may also have favoured patients' mortality by increasing the risk of acute kidney injury. 44The association between cholestatic liver injury and renal replacement therapy in our cohort was consistent with the idea of an accumulation of ketamine or toxic, hydrophobic by-products in the liver and the kidney. 42rengths and weaknesses of the study One strength of our study is the before and after comparison of two different periods, which helped limit the risk of unobserved confounding.Patients were compatible between the two periods, and the overall liver tests remained unchanged during the study.Medical practice did not change during the two study periods, with the exception of ketamine prescriptions. 12We also normalised the analyses on sufentanil and midazolam prescription, two drugs without known liver toxicity, to reduce the risk of unobserved confounding.A limitation could be the relatively small effective population, although the sample size was comparable with other reports on ketamine toxicity. 7,33To address this limitation, we used a 'full matching' method with propensity Patient were matched (full matching) on the probability of receiving ketamine dose >1,000 mg according to severity of illness (ABSI and SAPS II scores), and organ failure (acute kidney injury, acute respiratory syndrome, and septic shock).Risks were computed with multivariate logistic regression models.Cholestatic liver injury corresponded to an increase of ALP > − 1.5 × ULN with a concomitant increase of serum GGT > − 3 × ULN and TBIL >ULN.The SAPS II ranges from 0 to 163, with higher scores indicating greater severity of illness.The ABSI ranges from 0 to 18, with higher scores indicating a greater probability of death after the burn injury.ABSI, abbreviated burn severity index; ALP, alkaline phosphatase; TBIL, total bilirubin; GGT, gamma glutamyl transferase; Inf, infinite; SAPS II, Simplified Acute Physiology Score II; ULN, upper limit of normal.We defined two time periods, according to ketamine prescription modalities: a ketamine-liberal period, from December 2014 to end of March 2017, when ketamine prescription was 'liberally' used for maintenance sedation (> − 1 mg/ kg/h), and a ketamine-restricted period, from April 2017 to June 2019, when ketamine was only used as a second line co-analgesic drug with a capped dose (<0.015 mg/kg/h).
score matching, which has the advantage of retaining the whole population by assigning a weight to each patient.The generalisability of our findings may be constrained as they could apply solely to patients with severe burn injuries.It is noteworthy that ketamine toxicity has consistently been reported after chronic administration or misuse, and recent reports of liver-related deaths in patients with COVID-19 exposed to ketamine 31 suggest that our results may be transposable to other conditions.Ketamine may have been overlooked in studies reporting on critical care sclerosing cholangitis. 34

Conclusions
In a population at risk for liver injury, high doses of ketamine increased the risk of cholestatic liver injury, cholangitis, and mortality, and a ketamine prescription restriction policy improved patient outcome.Ketamine should be used with caution in critical care patients. 45Liver test monitoring is mandatory for ketamine dose > − 1,000 mg.Ketamine should be considered as a potential culprit when investigating causes of cholangitis (or of abnormal liver tests) in critically ill patients.Note: ALP stands for alkaline phosphatase; ALT for alanine aminotransferase; AST for aspartate aminotransferase; BILT for total bilirubin; and GGT for gamma-glutamyl transferase.

Fig. 4 .
Fig.4.Probability of 3-month survival by ketamine restriction period.We defined two time periods, according to ketamine prescription modalities: a ketamine-liberal period, from December 2014 to end of March 2017, when ketamine prescription was 'liberally' used for maintenance sedation (> − 1 mg/ kg/h), and a ketamine-restricted period, from April 2017 to June 2019, when ketamine was only used as a second line co-analgesic drug with a capped dose (<0.015 mg/kg/h).

Fig. S2 :
Fig. S2: Liver test evolution over the study period

Table 1 .
Characteristics of patients by time period.
).All statistical tests were based on two-tailed p values, with p <0.05 considered to indicate statistical significance.Missing data were not imputed.All analyses were performed using R statistical software (R 4.2.2GUI 1.79 Big Sur ARM build [8160]).

Table 2 .
Characteristics of patients by cholestatic liver injury.Cholestatic liver injury was serum ALP > − 1.5 × ULN with GGT > − 3 × ULN and TBIL >1 × ULN.The SAPS II ranges from 0 to 163, with higher scores indicating greater severity of illness.The ABSI ranges from 0 to 18, with higher scores indicating a greater probability of death after burn injury.The SOFA ranges from 0 to 24, with higher scores indicating more severe organ failure.ABSI, abbreviated burn severity index; ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; TBIL, total bilirubin; GGT, gamma glutamyl transferase; ICU, intensive care unit; Inf, infinite; SAPS II, Simplified Acute Physiology Score II; SOFA, Sequential Organ Failure Assessment; ULN, upper limit of normal.
* Data are presented as n (%) or median (IQR).† Pearson's Chi-squared test, the Wilcoxon rank sum test, or Fisher's exact test.

Table 3 .
Adjusted probabilities of cholestatic liver injury after severe burn injury.

Table 5 .
Adjusted probabilities for 3-month mortality in propensitymatched patients.

Table S2 : Characteristics of patients by day-90 mortality
Cholestasis was serum ALP ≥ 1.5 x ULN with GGT ≥ 3 x ULN; cholestatic liver injury was serum ALP ≥ 1.5x ULN with GGT ≥ 3 x ULN and BILT > 1 x ULN; DILI cholestasis was serum ALP level ≥ 2 x ULN and serum GGT level ≥ 1 x ULN; Grade 3 or higher ALP elevation was serum ALP ≥ 5 x ULN.The SAPS II ranges from 0 to 163, with higher scores indicating greater severity of illness.The ABSI ranges from 0 to 18, with higher scores indicating a greater probability of death after burn injury.The SOFA ranges from 0 to 24 with higher scores indicate more severe organ failure.

Table S3 : Characteristics of ARDS patients by time period
Cholestasis was serum ALP ≥ 1.5 x ULN with GGT ≥ 3 x ULN; cholestatic liver injury was serum ALP ≥ 1.5 x ULN with GGT ≥ 3 x ULN and BILT > 1 x ULN; DILI cholestasis was serum ALP level ≥ 2 x ULN and serum GGT level ≥ 1 x ULN; Grade 3 or higher ALP elevation was serum ALP ≥ 5 x ULN.The SAPS II ranges from 0 to 163, with higher scores indicating greater severity of illness.The ABSI ranges from 0 to 18, with higher scores indicating a greater probability of death after burn injury.The SOFA ranges from 0 to 24 with higher scores indicate more severe organ failure.ABSI: abbreviated burn severity index; ALP: alkaline phosphatase; ALT: alanine aminotransferase ; AST: aspartate aminotransferase; BILT: total bilirubin; DILI: Drug induced liver injury; GGT: gamma-glutamyl transferase; Inf= Infinite; SAPS II: simplified acute physiology score II; SOFA: Sequential Organ Failure Assessment score; ULN : upper limit of normal 1 n (%); Median (25%-75%) 2 Pearson's Chi-squared test; Wilcoxon rank sum exact test; Wilcoxon rank sum test; Fisher's exact test