Effect of continuous renal replacement therapy on outcome in paediatric acute liver failure

Dr Akash DEEP (corresponding author)MD, FRCPCH akash.deep@nhs.netKing’s College Hospital, Denmark Hill, London SE5 9RS

Dr Claire E STEWARTMBBS, BScclairematthews@doctors.org.ukKing’s College Hospital, Denmark Hill, London SE5 9RS

Prof Anil DHAWANMD, FRCPCHanil.dhawan@nhs.netKing’s College Hospital, Denmark Hill, London SE5 9RS

Dr Abdel DOUIRIBSc, MSc, PhD, FHEAabdel.douiri@kcl.ac.ukDepartment of Primary Care and Public Health Sciences & NIHR Biomedical Research Centre at Guy’s and St Thomas’ NHS Foundation Trust. King's College London, UK

Institution address:King’s College HospitalDenmark Hill London SE5 9RSUK

Address for reprints: PICU, King’s College Hospital, Denmark Hill, London, SE5 9RS, UK

Financial support:No funding was received for this project and as such the authors have nothing to declare

Acknowledgements:

We wish to thank Dr Palaniswamy Karthikeyan who helped us in the final version of the manuscript. Dr. Abdel Douiri acknowledges financial support from the National Institute for Health Research (NIHR) Biomedical Research and from the NIHR

1

Collaboration for Leadership in Applied Health Research and Care South London at King's College Hospital NHS Foundation Trust though not for this project. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health

Key words: Acute liver failure ; children; renal replacement therapy, outcomes

2

ABSTRACT

Objective: To establish the effect of continuous renal replacement therapy (CRRT)

on outcome in paediatric acute liver failure (PALF).

Design: Retrospective cohort study.

Setting: 16 bed paediatric intensive care unit (PICU) in a university affiliated tertiary

care hospital and specialist liver centre.

Patients: All children (0-18years) admitted to PICU with PALF between Jan 2003-

Dec 2013.

Interventions: Children with PALF were managed according to a set protocol. The

guidelines for CRRT in PALF were changed in 2011 following preliminary results to

indicate the earlier use of CRRT for both renal dysfunction and detoxification.

Measurements and Main Results: Of 165 children admitted with PALF, 136 met

the inclusion criteria and 45 of these received CRRT prior to transplantation or

recovery. Of the children managed with CRRT 26 (58%) survived; 19 were

successfully bridged to liver transplantation and 7 spontaneously recovered. Cox

proportional hazards regression model clearly showed reducing hyperammonaemia

by 48hrs after initiating CRRT significantly improved survival (HR, 1.04; 95% CI,

1.013-1.073; p=0.004). On average, for every 10% decrease in ammonia from

baseline at 48 hours, the likelihood of survival increased by 50%. Time to initiate

CRRT from PICU admission was lower in survivors compared to non-survivors (HR,

3

0.96; 95% CI, 0.916-1.007; p= 0.095). Change in practice to initiate early and high

dose CRRT led to increased survival with maximum effect being visible in the first 14

days (HR 3; 95% CI, 1.0-10.3; p=0.063). Amongst children with PALF who did not

receive a liver transplant, use of CRRT significantly improved survival (HR, 4; 95%

CI, 1.5-11.6; p=0.006).

Conclusion: CRRT can be used successfully in critically ill children with PALF to

provide stability and bridge to transplantation. Inability to reduce ammonia by 48

hours confers poor prognosis. CRRT should be considered at an early stage to help

prevent further deterioration and buy time for potential spontaneous recovery or

bridge to liver transplantation.

4

INTRODUCTION

Paediatric acute liver failure (PALF) is a rare but often fatal disorder.

Although some children spontaneously recover, mortality remains high due to the

high risk of sepsis, cerebral oedema and multi-organ failure that follows the abrupt

loss of hepatic function. Mortality rates of up to 70% have been reported without a

liver transplant, which is the only known curative treatment [1]. However knowing

which children will recover and in how much time, and which ones will rapidly

deteriorate, remains an enigma for intensivists [2]. In the wake of organ scarcity on

one hand and subjecting children to unnecessary transplantation with risks of

surgery and life-long immunosuppression on the other, the role of supportive

therapies to stabilise a child to help facilitate these decisions is an area of growing

importance [3-5].

One of the most significant achievements in intensive care medicine in the

last 40 years has been the development of continuous renal replacement therapy

(CRRT) [6]. CRRT is now the mainstay treatment for managing acute kidney injury

(AKI) in paediatric intensive care units and has significantly reduced mortality [7-9].

AKI is a common multi-factorial complication of PALF that occurs in approximately

55% of all cases [10-12]. Renal dysfunction pre-transplant determines the degree of

renal dysfunction post- transplant. Therefore tackling AKI pre-transplant should

improve the condition of the patient post-transplant. In these patients CRRT has also

been shown to successfully reduce ammonia [13], lactate [14,15] and optimise fluid

balance [16]. However all of these observations have been reported in adult patients.

Recent research on critically ill children with PALF has demonstrated that high

5

ammonia [17,18], high lactate [19] and fluid overload [20] are associated with

increased mortality. In addition to the risk of AKI in PALF, hyperammonaemia

contributes to the development of cerebral oedema, and there is a strong risk of fluid

overload especially after fluid resuscitation due to the systemic inflammatory

response syndrome (SIRS).This raises the question whether the use of CRRT

should be extended to non-renal indications in patients with PALF, as a detoxification

mechanism.

Despite its rapidly increasing uptake, evidence to evaluate whether CRRT is

of justifiable benefit in these situations is currently lacking [21], and there are no

agreed guidelines on when or what dose it should be initiated in PALF. This study

aimed to describe our experience on the use of CRRT in PALF from over a decade

of data.

MATERIALS AND METHOD

Study population

King’s College Hospital is a supra-regional centre for liver referrals in the

United Kingdom and operates one of the largest liver transplantation programmes in

Europe. All children admitted to the PICU at King’s College Hospital that met the

criteria for PALF from January 2003 to December 2013 were entered into the PALF

registry. PALF was defined according to the criteria used by the Paediatric Acute

Liver Failure Study Group [22]: “1) No evidence of pre-existing chronic liver disease

2) Biochemical evidence of acute liver injury (elevation in AST/ALT) within 8 weeks

6

of the onset of symptoms 3) Hepatic-based coagulopathy (PT>15 or INR>1.5 in the

presence of HE, or PT>20 or INR>2 in the absence of HE) within 8 weeks of the

onset of symptoms”. Children with PALF admitted to the PICU for only post

transplant care were excluded. Children who were commenced on CRRT prior to

transplantation or recovery were included in the CRRT treatment group (Figure 1).

Factors affecting survival in children with PALF treated with CRRT were studied. Due

to preliminary findings, the guideline for CRRT in PALF was changed in 2011,

therefore the clinical outcomes in these 2 eras, pre and post 2011, have been

compared.

Data collection

Medical records, laboratory data and observation charts were reviewed for all

patients in the PALF registry. Patient characteristics including age, gender, height,

weight, details and duration of presenting symptoms and signs, time to first medical

contact, time to PICU transfer and pre-referral care were also recorded. Aetiology of

PALF was classified into 7 categories: indeterminate, toxic, infectious, metabolic,

ischaemic, infiltrative and autoimmune.

For clinical observations and laboratory data, admission and peak values

within the first 24 hours were recorded from the unit’s Clinical Information system

(CIS). Paediatric logistic organ dysfunction score (PELOD) and Pedatric Index of

Mortality (PIM2) were calculated for each patient [23]. Timing of listing for liver

transplantation and indication along with transplant date, type and details of any

post-operative complications and the date that the child was removed from the list

7

due to spontaneous recovery or clinical deterioration were included. Duration of

PICU stay, survival on hospital discharge and survival at 6 months post discharge

were also recorded.

For children commenced on CRRT, time from PICU admission to initiation of

CRRT, indication, duration and markers of disease progression at 0, 24 and 48

hours after starting CRRT including arterial ammonia, lactate, mean arterial blood

pressure, percentage fluid overload and creatinine were recorded. Indications for

CRRT were classified into the following eight categories: oligo-anuria, hyperkalaemia

>5.5mmol/L, fluid overload >10%, hyperammoneamia >200mol/L, hepatic

encephalopathy grade >2, hyponatraemia <130meq/L, lactate >2mol/L not

responding to fluid therapy or metabolic acidosis pH<7.1 resistant to fluid therapy.

Importantly not one single indication was considered a sole reason to initiate CRRT;

it was a clinical decision and all contributory factors were recorded. The daily

percentage fluid overload prior to and post initiation of CRRT was calculated using

the formula by Goldstein et al. [24].

Any complications secondary to CRRT including catheter malfunction,

anticoagulation problems, bleeding, thrombosis, infection or shock were also

recorded.

8

Continuous renal replacement therapy

The duty consultant paediatric intensivist determined the requirement for

CRRT. CRRT was commenced according to a local protocol with pre-dilution to

achieve required ammonia and lactate clearances and electrolyte and fluid

homeostasis (Table 1). All children had ultrasound guided venous access via a high

flow double lumen catheter (“Vascath”; Gambro, Stockholm, Sweden) placed either

in the internal jugular, subclavian or the femoral vein. The sizes of the double lumen

venous access catheters used were pre-defined according to body weight - 6.5 Fr

(14%), 8 Fr (28%),10 Fr (15%), 11.5 Fr (32%) and 13.5 Fr (11%). Children <10 kg

were filtered using HF03, 10kg - 50kg using HF07 and >50 kg using HF1200

(membrane surface area HF03 = 0.3 m2, HF07 = 0.7 m2, HF 1200 = 1.2 m2). All the

filters were made of polyethersulfone fiber. Pre-dilution was incorporated in all

filtration episodes using “Accusol 35”, a lactate free electrolyte solution. The net

delivered dose varied and on an average was 8.2±1.1 mls/kg/hour less than the

prescribed dose.

The machine used for CRRT was “Aquarius” (Nikkiso Europe GmbH,

Hannover). Details of CRRT recorded included vascath size and location,

anticoagulation dose, filter life, dose of CRRT and complications. Blood flow rates

ranged from 50 to 250mls/min depending on age (Table 1). Pre-dilution continuous

venous venous haemofiltration (CVVH) was the most commonly used modality.

Anticoagulation used was prostacyclin (2-6ng/kg/min) if the activated clotting time

(ACT) was 160-220 seconds and low dose heparin if ACT <160 seconds unless

contraindicated. If ACT >220 seconds no anticoagulation was used. To facilitate

toxin removal, CRRT dose was sequentially increased to a maximum of 100

mls/kg/hour.

9

Statistical analysis

Statistical data was analyzed with STATA software version 14 (StataCorp,

College Station, Tex). Continuous data was expressed as the mean ± standard

deviation and categorical data as the actual count with percentages unless stated

otherwise. The primary outcome of this study was survival up to 60 days after the

PICU admission with or without liver transplantation: a) death awaiting

transplantation b) death after transplantation c) alive with native liver d) alive with

transplanted organ. Secondary outcome measures included the trend in markers for

disease progression after initiating CRRT. The data was analysed to see what, if

any, variables on admission were associated with mortality and to identify prognostic

markers in children with PALF on CRRT. Univariate analysis was performed to

compare variables between survivors and non-survivors at hospital discharge using

the unpaired t-test for continuous data and Fisher’s exact test for categorical data.

For analysing trends in markers of disease progression, regression lines and box

plots were used and compared by primary outcome. Analysis of variance and

covariance (ANOVA) was used to compare disease severity markers at 0, 24 and 48

hours after initiation of CRRT. Kaplan-Meier curves and Cox model were used to

look at survival of children up to 60 days after admission. Data was adjusted for

severity of illness using the PELOD scoring system. A p value of less than 0.05 was

considered statistically significant.

10

Ethics

Since it was a retrospective analysis of already collected routine data, ethical

approval was not deemed necessary , however as per institutional practice, this

project was registered as service evaluation project - 2592

RESULTS

Baseline characteristics

From January 2003 to December 2013, 136 patients with PALF met the

inclusion criteria of being managed on the unit prior to transplantation or recovery.

45 of these children received CRRT as part of their management (Figure 1). A

comparison between patient demographics and selected physiological and

laboratory values on admission to PICU are shown in the supplementary table.

Amongst the patients who underwent CRRT, the commonest aetiology was

indeterminant (53%) followed by metabolic (20%), and toxic (13%).

Baseline characteristics between survivors and non-survivors treated with

CRRT were compared (Table 2). Non-survivors were significantly younger (p=0.017)

and weighed less (p=0.006). They also spent on average ten days more on the

ward prior to PICU transfer (p=0.051). Toxic (p=0.024) or metabolic (p=0.015)

aetiology was associated with higher mortality. Peak arterial ammonia in the first

twenty-four hours of admission were also significantly higher amongst non-survivors

(p=0.025). All children who died on CRRT were ventilated and on inotropes.

Anticoagulation and complications in CRRT

11

A total of 198 filters were utilised between the 45 patients receiving CRRT

for PALF. Prostacyclin was used in 72% of these filters, unfractionated heparin in

18% filters and no anticoagulant in 10%. Median filter life was 59 hours (44.7 – 60.2)

using prostacyclin, 27 hours (21.1– 33.7) using heparin and 20 hours (14.6 – 25.3)

using no anticoagulation. Safety and efficacy of prostacyclin as an anticoagulant was

acceptable with 1.9 bleeding episodes per 1000 hours of CRRT and hypotension

requiring fluids or vasopressors occurring in less than 10% of filter episodes. The site

of venous access (femoral versus internal jugular) did not contribute to circuit life

among this sub-group. There was 1 episode of haemo-pneumothorax with internal

jugular venous catheter insertion and 3 episodes of vascular access re-wiring due to

thrombus;overall the procedure was well tolerated.

Timing of initiation and indications for CRRT

The median time to initiate CRRT from PICU admission was 27± 6.9 hours

and one in three were started on CRRT within the first eight hours of admission to

the unit. Median time to initiate CRRT in survivors was lower as compared to non-

survivors (15.8 ±3.0 hours versus 32.4 ± 6.9 hours; p =0.023). Thirty patients

(66.7%) had multiple indications for starting CRRT. The most common indications

were oligo-anuria (n=14, 31%) hyperammonaemia (n=13, 29%), hepatic

encephalopathy (n=12, 27%), high lactate (n=10, 22%), fluid overload (n=6, 13%),

resistant metabolic acidosis (n=3, 7%), resistant hyperkalaemia (n=1, 2%) and

hyponatraemia (n=1, 2%). The mean duration of CRRT was 54 hours.

Effect on outcome

12

Overall survival of patients with ALF on CRRT at discharge was 58% (Figure

2). A significant increase in mortality was seen in children who were less than one

year old (HR, 6; 95% CI, 2.0-18.2; p=0.001) (Figure 3). Severity was adjusted for by

the PELOD score as opposed to the PIM2 score as it does not accurately reflect

outcomes in PALF [25].

Twenty-six of the forty-five patients were bridged successfully to either native

organ recovery (n=7) or successful liver transplantation (n=19). In total 24 patients

requiring CRRT underwent a liver transplant and 79% of these survived. Of the 21

who did not, only one third spontaneously recovered. As liver transplantation

interferes with the natural progression of disease, we looked at the effect of CRRT

only in those children with PALF who did not undergo transplantation i.e., either had

spontaneous regeneration of native liver or died without a transplant – those who

received CRRT had a significantly increased chance of survival (HR, 4; 95% CI, 1.5-

11.6; p=0.006) (Figure 4).

Cox proportional hazards regression model adjusting for severity of illness

(PELOD), ammonia at presentation, ability of CRRT to decrease ammonia by 48

hours and time to initiate CRRT from PICU admission clearly shows that in spite of

high ammonia levels in both survivors and non-survivors at initiation of CRRT, a

decrease in ammonia by 48 hours of starting CRRT significantly improved survival

(HR,1.04; 95% CI, 1.013-1.073; p=0.004) . On average, for every 10% decrease in

ammonia from baseline at 48 hours, the likelihood of survival increased by 50%.

Time to initiate CRRT from PICU admission was also found to be lower in survivors

compared to non-survivors , although statistical significance is borderline (HR,0.96;

95% CI 0.916-1.007; p= 0.095). For very 1 hour delay in initiating CRRT, likelihood of

mortality increased by 4%.

13

Efficacy of CRRT

To review the efficacy of CRRT in PALF, daily markers of disease progression

including arterial ammonia, lactate, percentage fluid overload, creatinine and mean

arterial pressure were reviewed using ANOVA at 0, 24, 48 hours after initiation of

CRRT (Table 3). There was a significant reduction in ammonia (p=0.001) and

lactate (p=0.043) within 48 hours of initiating CRRT amongst survivors. Although the

mean level of creatinine and percentage fluid overload also decreased, the

difference was not statistically significant.

Figure 5 is a box plot analysis of the trend in ammonia over time by mortality.

This clearly demonstrates that although ammonia is high at initiation of CRRT in both

survivors and non-survivors, in survivors the ammonia is significantly lower after 48

hours of CRRT while in non-survivors levels remained high (p=<0.001). As

demonstrated by multivariate analysis, inability of ammonia to be removed even after

48 hours of CRRT is a poor prognostic sign (p=0.004).

Although not a controlled variable change, survival in PALF increased by

9% from 65% before the change in guidelines in 2011. Figure 6 shows the trend in

survival pre and post guideline change. The most significant difference was seen in

the first 14 days (HR, 3; 95% CI, 1.0-10.3; p=0.063).

DISCUSSION

14

Despite the increasing use of CRRT in children with acute liver failure, there is

a paucity of data surrounding its effect on outcome and applicability as a

detoxification mechanism. The present study examined an11 year institutional

experience from the PICU at a specialist liver unit to evaluate whether the use of

CRRT in these critically ill children is of justifiable benefit.

Of note, when the prospective paediatric CRRT registry (ppCRRT) was split

into survival rates according to the diagnoses that led to the use of CRRT, children

with liver disease had the lowest survival at 31% compared to the 58% average for

all other diagnoses [26]. In this study almost all children who required CRRT also

required inotropes (91%) and ventilation (100%). This suggests the children who

went on to require CRRT in PALF were much more unstable during their admission

and highlights just how sick these children are. As such it would be

misrepresentative to compare outcomes of the treatment and non-treatment group.

Likewise, it would be unethical to withhold CRRT from a critically ill child due to its

proven success in AKI, a common complication of PALF, and anecdotal and

theoretical evidence for detoxification in PALF for the purpose of a randomised

controlled trial. As such evidence will need to be sought through observational

studies to establish the effect on outcome.

This study found that children frequently had more than one indication for

CRRT prior to commencing therapy. Traditionally CRRT has been used to manage

renal dysfunction which is a common complication of PALF. It is important not to

forget that AKI is often multi-factorial in PALF with hypovolaemia being the most

common precipitating factor setting the stage for acute tubular necrosis [27].

Although renal impairment was the indication for CRRT in a significant number of

15

patients included in this study (oligo-anuria 31%, fluid overload 13% and resistant

hyperkalaemia 2%), more frequently CRRT is being used regardless of the presence

or absence of renal dysfunction as a detoxification mechanism. The potential for

recovery of native organ function coupled with the insufficient donor organ supply

and risk of death or deterioration awaiting a compatible transplant, has focussed

attention on using detoxification mechanisms as liver support devices until recovery

from PALF or bridging to transplantation. As such initiating CRRT in PALF now

serves a dual purpose 1) for managing the AKI and fluid imbalances which frequently

complicate cases, and 2) as a detoxification mechanism for the high ammonia,

lactate and metabolic disturbances which set in.

In this study two of the indications for initiating CRRT were hyperammonemia

>200 micromoles/litre or hepatic encephalopathy greater than grade 2, which was

present in a quarter of children (29% and 27% respectively). It is well established

that when ammonia is not metabolised by the failing liver, it is detoxified to glutamine

in the brain which is osmotically active and this is now thought to be responsible for

astrocyte swelling and cytotoxic oedema seen in PALF [28-31]. Amongst survivors,

CRRT was successful in reducing arterial ammonia (p=<0.001). Interestingly there

was a much greater mortality amongst children whose level of arterial ammonia rose

between the value at admission to PICU to the value at initiation of CRRT. This

suggests that the intensivist should not wait, but intervene early at the first sign of

rising hyperammonemia. Ability to reduce ammonia after 48 hours of starting CRRT

was the most significant factor for survival on multivariate analysis (p=0.004). This is

important as in rapidly progressive PALF, the first 24-48 hours are vital to create a

good milieu to facilitate spontaneous regeneration or prevent deterioration whilst

16

awaiting for a suitable donor organ to become available. It remains to be seen

whether this could be used as a marker of prognosis to help guide decisions

regarding transplantation or prioritisation of scarce donor organs.

Other significant risk factors for mortality in children with PALF treated with

CRRT on admission included age (p=0.017), weight (p=0.006), time on ward prior to

PICU transfer (p=0.051), serum lactate (p=0.003), INR (p=0.062), peak levels of

arterial ammonia in the first 24 hours (p=0.009) and PALF with toxic and metabolic

aetiology. Age less than 1 year lends a significantly worse prognosis to this group of

patients (Figure 3) due to their smaller size, rapidly progressive aetiologies and

increased waiting time for liver transplantation. This implies in the case of rapidly

progressive PALF, early transfer to specialist liver transplant centres is critical

especially in younger children with raised ammonia or INR >4 of toxic or metabolic

aetiology as mortality is significantly higher.

In order to see whether CRRT can alter the natural progression of PALF,

those patients with PALF who did not undergo transplantation and only received

medical interventions were analysed. The survival without transplant was much

higher amongst the children who received CRRT versus those who did not (Figure

4). CRRT in these patients optimises overall milieu including fluid balance and

removal of toxic products to facilitate spontaneous recovery.

Due to reports of increased survival in using CRRT as a detoxification

mechanism and earlier initiation in adult patients with acute liver failure (ALF), and

preliminary findings at our PICU, the PICU guidelines on the indications for CRRT in

PALF were changed in January 2011. In addition to signs of renal dysfunction,

17

CRRT was initiated for hepatic encephalopathy (grade >2) hyperammonaemia (NH3

>150µmol/L & not getting controlled, or an absolute value >200µmol/L) along with

metabolic abnormalities including hyponatremia. Higher doses of CRRT were also

introduced, starting at 60mls/kg/hour and increasing to 100mls/kg/hour if toxin

removal was inadequate, and bigger sized vascaths appropriate to all age groups

were introduced to ensure adequate circuit life and CRRT dose delivery. The effect

of these changes (Figure 7) is seen predominantly in the first 14 days, which is the

time critical period to ensure that the child is in a stable condition either for

spontaneous regeneration or liver transplantation.

Despite its increasing uptake, clear-cut consensus is lacking surrounding

optimal CRRT delivery including time of initiation and dose. There is a growing body

of evidence from adults with AKI that earlier initiation of CRRT has a significant

beneficial impact on survival [32,33]. This study likewise found that delays in

initiation of treatment are associated with higher mortality and suggests the

intensivist should intervene at an early stage to have the best chance of survival.

Regarding optimal doses, Davenport et al. showed improved cardiovascular stability

in ALF using high filtration rates in particular where severe lactic acidosis was

present or vasopressors were required, which applied to 93% of the children with

PALF on CRRT in our group. [34]. In paediatrics recent reports from France and

Tokyo have demonstrated improved beneficial effects in children with PALF

undergoing high volume haemofiltration [35, 36]. As such at King’s College

Hospital, CRRT is started early in the course of PALF patients who are intubated and

ventilated for grade >2 encephalopathy with high arterial ammonia levels, using

hemofiltration as the preferred modality. The dose is then increased as required to

18

achieve effective ammonia clearance due to the increased risk of mortality without a

delta fall as highlighted in this study.

Another area of controversy is the use of anticoagulation. Despite the

prolongation of routine coagulation tests in acute liver failure, Habib et al highlighted

the need for anticoagulation due to the procoagulant state [37]. Anticoagulation is

therefore used in nearly all children with PALF undergoing CRRT in this study, with

prostacyclin being the most frequently used anticoagulant. No complications related

to anticoagulant use were observed. In addition to beneficial effects of prostacyclin

as an anticoagulant, it can help to optimise oxygen delivery and uptake in critically ill

patients which can help improve ultimate prognosis [38]. This study validates the

acceptable safety and efficacy of prostacyclin as an anticoagulant in CRRT for

PALF. Some centres have started using citrate despite initial fears of citrate

accumulation as it is now shown that this can be predicted by the Ca total/Ca ionic

ratio. Though this ratio can rise, equalisation of initial metabolic acidosis is possible

without major disturbances of acid-base and electrolyte status with acceptable filter

lives [39].

This study has its limitations. Firstly, this is a single centre experience with a

relatively small sample size. It is extremely difficult to compare outcomes between

CRRT and non-CRRT groups despite controlling for the severity of illness especially

when the aetiology is diverse, Ideally a propensity score should be used which

controls for the characteristics leading to the decision to use CRRT, however it was

not feasible due to the sample size and variability in characteristics of patients

treated with CRRT versus those without. Therefore it is difficult to make conclusions

19

that outcomes are related to the treatment effect of CRRT only, although there are

significant associations, these must be further tested. However, due to the rarity and

severity of PALF, we feel the results are of value for guiding best practice.

CONCLUSIONS

When decisions regarding risks of surgery and prioritisation of donor organs

are critical, such as in rapidly progressing PALF, the importance of stabilising a child

to help facilitate these decisions is fundamental to survival. This study demonstrates

that early, intensive CRRT can be used successfully in PALF for managing both AKI

and toxin accumulation to provide an environment conducive to regeneration or to

prolong the window of opportunity for successful liver transplantation. Moreover, if

patients are to undergo transplantation, they are presented in a more stable

condition especially with regards to fluid homeostasis which can significantly reduce

mortality in this frequently fatal condition.

20

REFERENCES

1. Bansal S and Dhawan A. Acute liver failure. Indian J Pediatr 2006;73(10):931-4.

2. Shanmugam NP and Dhawan A. Selection criteria for liver transplantation in paediatric acute liver failure: the saga continues. Pediatric Transplantation 2011;15(1):5-6.

3. Stadlbauer V and Jalan R. Acute liver failure: liver support therapies. Curr Opin Crit Care 2007;13(2):215-21.

4. Phua J and Lee KH. Liver support devices. Curr Opin Crit Care 2008; 14(2):208-15.

5. Hoste EA and Dhondt A. Clinical review: use of renal replacement therapies in special groups of ICU patients. Crit Care 2012; 16(1):201.

6. Kramer P, Wigger W, Rieger J, Matthaei D and Scheler F. Arteriovenous haemofiltration: A new and simple method for treatment of over-hydrated patients resistant to diuretics [in German]. Klin Wochenschr 1977;55:1121–1122.

7. Boschee ED, Cave DA, Garros D, Lequier L, Granoski DA, Guerra G and Ryerson L. Indications and outcomes in children receiving renal replacement therapy in pediatric intensive care, Journal of Critical Care 2014;29:37–42.

8. Xue JL, Daniels F, Star RA, Kimmel PL, Eggers PW, Molitoris BA, Himmelfarb J and Collins AJ. Incidence and mortality of acute renal failure in Medicare beneficiaries, 1992 to 2001. J Am Soc Nephrol 2006, 17:1135-1142. 

9. Waikar SS, Curhan GC, Wald R, McCarthy EP and Chertow GM. Declining mortality in patients with acute renal failure, 1988 to 2002. J Am Soc Nephrol 2006, 17:1143-1150. 

10.Mohan N, Raghunathan V, Dhaliwal M and Karkra S. Prevalence of acute kidney injury network (AKIN) in paediatric non acetaminophen acute liver failure (pNA-ALF) & addition of akin staging to king's college criteria (Kcc) in pNA-ALF improves prognostic value. Journal of Clinical and Experimental Hepatology 2013;3(1):S17.

11.Ellis D, Avner ED, Starzl TE. Renal failure in children with hepatic failure undergoing liver transplantation. J Pediatr 1986;108(3):393-398.

12.Moore K. Renal failure in acute liver failure. Eur J Gastroenterol Hepatol. 1999; 11(9):967-75.

21

13.Slack AJ, Auzinger G, Willars C, Dew T, Musto R, Corsilli D, Sherwood R, Wendon JA and Bernal W. Ammonia clearance with haemofiltration in adults with liver disease. Liver Int. 2014;34(1):42-8.

14.Barton IK, Streather CP, Hilton PJ and Bradley RD. Successful treatment of severe lactic acidosis by haemofiltration using a bicarbonate-based replacement fluid. Nephrol Dial Transplant 1991;6:668–670.

15.Kirschbaum B, Galishoff M, Reines HD: Lactic acidosis treated with continuous hemodiafiltration and regional citrate anticoagulation. Crit Care Med 1992;20:349–353.

16.Bouchard J and Mehta RL. Volume management in continuous renal replacement therapy. Semin Dial. 2009;22(2):146-50.

17.Drolz A, Jager B, Wewalka M, Saxa R, Horvatits T, Roedl K, Perkmann T, Zauner C, Kramer L, Ferenci P and Fuhrmann V. Clinical impact of arterial ammonia levels in ICU patients with different liver diseases. Intensive Care Med. 2013; 39(7):1227-37.

18.Bhatia V, Singh R and Acharya SK. Predictive value of arterial ammonia for complications and outcome in acute liver failure. Gut 2006; 55(1):98-104.

19.Bernal W, Donaldson N, Wyncoll D and Wendon J. Blood lactate as an early predictor of outcome in paracetamol-induced acute liver failure: a cohort study. Lancet. 2002; 359(9306):558-63.

20.Sutherland SM, Zappitelli M, Alexander SR, Chua AN, Brophy PD, Bunchman TE, Hackbarth R, Somers MJ, Baum M, Symons JM, Flores FX, Benfield M, Askenazi D, Chand D, Fortenberry JD, Mahan JD, McBryde K, Blowey D and Goldstein SL. Fluid overload and mortality in children receiving continuous renal replacement therapy: the prospective pediatric continuous renal replacement therapy registry. Am J Kidney Dis. 2010; 55(2):316-25.

21.Chevret L, Durand P, Lambert J, Essouri S, Balu L, Devictor D, Tissieres P. High-volume hemofiltration in children with acute liver failure*. Pediatr Crit Care Med. 2014 Sep;15(7):e300-5.

22.Squires RH, Jr., Shneider BL, Bucuvalas J, Alonso E, Sokol RJ and Narkewicz MR. Acute liver failure in children: the first 348 patients in the pediatric acute liver failure study group. J Pediatr 2006;148(5):652-58.

23.Leteurtre S, Duhamel A, Grandbastien B, Lacroix J and Leclerc F. Paediatric logistic organ dysfunction (PELOD) score. Lancet 2006;367(9514):897.

24.Goldstein SL, Currier H and Graf C. Outcome in children receiving continuous venovenous hemofiltration. Pediatrics 2001;107:1309–12.

22

25.Matthews CE, Goonasekera C, Dhawan A, Deep A. Validity of pediatric index of mortality 2 (PIM2) score in pediatric acute liver failure. Crit Care. 2015;18(6):665.

26.Symons JM, Chua AN, Somers MJ, Baum MA, Bunchman TE, Benfield MR, Brophy PD, Blowey D, Fortenberry JD, Chand D, Flores FX, Hackbarth R, Alexander SR, Mahan J, McBryde KD and Goldstein SL. Demographic characteristics of pediatric continuous renal replacement therapy: a report of the prospectice pediatric continuous renal replacement therapy registry. Clin J Am Soc Nephrol. 2007; 2(4):732-8.

27.Albrecht J and Norenberg MD. Glutamine: a Trojan horse in ammonia neurotoxicity. Hepatology 2006; 44: 7888-794.

28. Rao KVR, Reddy PVB, Tong X and Norenberg MD. Brain edema in acute liver failure. Am J Pathol. 2010;176(3):1400–1408.

29.Desjardins P, Du T, Jiang W, Peng L and Butterworth RF. Pathogenesis of hepatic encephalopathy and brain edema in acute liver failure: role of glutamine redefined. Neurochem Int 2012;60(7):690-696.

30.Scott TR, Kronsten VT, Hughes RD and Shawcross DL. Pathophysiology of cerebral oedema in acute liver failure. World J Gastroenterol 2013;19(48): 9240-9255.

31.Tsai MH, Chen YC, Lien JM, Tian YC, Peng YS, Frang JT, Yang C, Tang JH, Chu YY, Chen PC and Wu CS. Haemodynamics and metabolic studies on septic shock in patients with acute liver failure. J Crit Care. 2008; 23(4):468-72

32.Karvellas CJ, Farhat MR, Saiiad I, Mogensen SS, Leung AA, Wald R and Bagshaw SM. A comparison of early versus late initiation of renal replacement therapy in critically ill patients with acute kidney injury: a systematic review and meta-analysis. Crit Care. 2011; 15(1):R72

33.Wang X and Jie Yuan W. Timing of initiation of renal replacement therapy in acute kidney injury: a systematic review and meta-analysis. Ren Fail. 2012; 34(3):396-402.

34.Davenport A, Will EJ and Davidson AM. Improved cardiovascular stability during continuous modes of renal replacement therapy in critically ill patients with acute hepatic and renal failure. Crit Care Med. 1993; 21(3):328-38.

35.Chevret L, Durand P, LambertJ, Essouri S, Balu L, Devictor D, Tissieres P. High-volume hemofiltration in children with acute liver failure. Pediatr Crit Care Med. 2014 Sep;15(7):300-5.

36. Ide K, Muguruma T, Shinohara M, Toida C, Enomoto Y, Matsumoto S, Aoki K, Fukuda A, Sakamoto s, Kasahara M. Continuous Veno-Venous

23

Hemodiafiltration and Plasma Exchange in Infantile Acute Liver Failure. Pediatr Crit Care Med. 2015 Oct;16(8):e268-74.

37.Habib M, Roberts LN, Patel RK, Wendon J, Bernal W, Arya R. Evidence of rebalanced coagulation in acute liver injury and acute liver failure as measured by thrombin generation. Liver Int. 2014 May;34(5):672-8.

38.Bihari D, Smithies M, Gimson A, Tinker J. The effects of vasodilation with prostacyclin on oxygen delivery and uptake in critically ill patients. N Engl J Med. 1987 Aug 13;317(7):397-403.

39. Schultheiss C, Saugel B, Phillip V, Thies P, Noe S, Mayr U, Haller B, Einwachter H, Schmid RM, Huber W. Continuous venovenous hemodialysis with regional citrate anticoagulation in patients with liver failure: a prospective observational study. Crit Care. 2012 Aug 22; 16(4):R162.

24

FIGURE LEGENDS

- Figure 1 Study profile. PALF, paediatric acute liver failure; PICU, paediatric intensive

care unit; CRRT, continuous renal replacement therapy

- Figure 2 Outcome of all children with PALF requiring CRRT. PALF, paediatric acute

liver failure; CRRT, continuous renal replacement therapy; PICU, paediatric intensive care

unit

- Figure 3 Kaplan Meir curve for 60-day survival of children PALF on CRRT. PALF,

paediatric acute liver failure; CRRT, continuous renal replacement therapy

25

- Figure 4 Kaplan Meir curve for 60-day survival of children PALF on CRRT versus

those not on CRRT in non-transplanted group. PALF, paediatric acute liver failure;

CRRT, continuous renal replacement therapy

26

- Figure 5 Box plot of the trend in ammonia level (umol/L) by survival. CRRT,

continuous renal replacement therapy

- Figure 6 Kaplan Meir curve for 60-day survival of children PALF on CRRT. PALF,

paediatric acute liver failure; CRRT, continuous renal replacement therapy

27

TABLES

- Table 1 CRRT set up using CVVH, CRRT, continuous renal replacement therapy;

CVVH, continuous venous venous haemofiltration

Weight (kg) <3.5 3.5-4.9 5-14.9 15-29.9 30-49.9 >50Filter Size HF03 HF03 HF03 HF07+ HF07+ HF07+

Aqualine Size S S S S S Adult

Blood Flow Rate (ml/min) 50 50-80 100 150 200 250

Ultrafiltration Rate (ml/kg/hr) 60 60 60 60 60 3000

max.Priming Volume (ml) 96 96 96 118 159 159

- Table 2 Baseline characteristics of patients treated with CRRT on admission to

PICU PICU, paediatric intensive care unit; PALF, paediatric acute liver failure; MAP,

mean arterial pressure; AST, asparate aminotransferase; INR, international normalised

ratio; CRRT, continuous renal replacement therapy; PIM2, paediatric index of mortality;

LIU, liver injury unit; PELOD, pediatric logistic organ dysfunction score; FO, fluid overload

All CRRT (n=45)

Survivors (n=26)

Non-survivors(n=19)

P-value (survivors vs

non)

Age, months 68±9.8 88±13.4 41±12.1 0.017*Female, n (%) 18 (40) 10 (38) 8 (42) 0.811Male, n (%) 27 (60) 16 (62) 11 (58) 0.811Weight (kg) 23.5±3.9 32.8±5.8 11.9±3.2 0.006*Time on the ward prior to PICU transfer, days 10.7±2.8 6.5±1.2 16.3±6.1 0.051*

Aetiology of PALF: Indeterminate, n (%) 24 (53) 16 (62) 8 (42) 0.206 Toxic, n (%) 6 (13) 6 (23) 0 (0) 0.024* Infectious, n (%) 3 (7) 2 (8) 1 (5) 0.754 Metabolic, n (%) 9 (20) 2 (8) 7 (37) 0.015* Ischaemic, n (%) 1 (2) 0 (0) 1 (5) 0.247 Infiltrative, n (%) 2 (4) 0 (0) 2 (11) 0.086Encephalopathy grade ≥2, n (%) 12 (27) 5 (19) 7 (37) 0.078MAP, mmHg 66±2.4 65±3.3 67±3.8 0.717Urine output, mls/kg/hr 2.6±0.5 2.3±0.6 3.0±1.0 0.552Lactate, mmol/L 5.1±0.7 4.6±0.9 5.8±1.1 0.366Ammonia, umol/L (0hrs) 173±14.9 153±13.7 212±33.1 0.061Ammonia, umol/L (24hrs) 159±17.8 110±11.5 198±24.7 0.009*Ammonia, umol/L (48hrs) 165±34.1 90±11.1 253±54.9 0.009*Bilirubin 283.09±25.8 314±37.3 246.2±31.6 0.200

28

Creatinine 68±9.6 72±10.1 62±18.0 0.593AST, umol/L 2486±547 3075±892 1741±488 0.229INR 4.5±0.2 4.8±0.4 4.1±0.3 0.162Platelets, 109/L 194±20.2 212±28.0 170±28.5 0.403PELOD score 9.6±1.1 8.4±1.3 11.1±1.9 0.228Inotropes, n (%) 41 (91) 23 (88) 19 (100) 0.131Intubated and ventilated, n (%) 45 (100) 26 (100) 19 (100) N/ATime to initiation of CRRT (hrs) 27.0±6.9 15.8±3.0 32.4±6.9 0.023*

- Table 3 Trend in mean ammonia, lactate and percentage fluid overload between

survivors and non-survivors before and after starting CRRT. CRRT, continuous renal

replacement therapy MAP, mean arterial pressure

Survivors (n=26)

Non-survivors (n=19) P-value

Arterial ammonia, umol/L On admission 173±17 201±31 0.402 CRRT (0hrs) 155±14 205±31 0.115 CRRT (24hrs) 110±11 198±25 0.001* CRRT (48hrs) 90±11 253±50 0.001* P value (0 vs 48hrs) 0.001* 0.420Lactate, meq/L On admission 4.6±0.9 5.8±1.0 0.382 CRRT (0hrs) 4.0±0.9 6.8±2.0 0.169 CRRT (24hrs) 3.0±0.6 3.5±0.7 0.590 CRRT (48hrs) 1.6±0.2 3.1±0.8 0.043* P value (0 vs 48hrs) 0.012* 0.042*Fluid overload, % CRRT (0hrs) 5.6±1.7 5.3±1.3 0.896 CRRT (24hrs) 4.9±1.3 3.1±2.1 0.448 CRRT (48hrs) 3.9±1.4 3.1±3.1 0.798 P value (0 vs 48hrs) 0.444 0.517Creatinine, mmol/L CRRT (0hrs) 70±8.8 66±19.0 0.836 CRRT (24hrs) 64±7.7 54±10.7 0.440 CRRT (48hrs) 57±7.5 50±9.1 0.554 P value (0 vs 48hrs) 0.266 0.453MAP, mmHg CRRT (0hrs) 72±3.2 68±5.4 0.504 CRRT (24hrs) 68±4.1 68±3.8 1.000 CRRT (48hrs) 76±4.9 62±8.2 0.129 P value (0 vs 48hrs) 0.498 0.541

- Supplementary table Baseline characteristics of all patients on admission to PICU

PICU, paediatric intensive care unit; PALF, paediatric acute liver failure; MAP, mean

arterial pressure; AST, asparate aminotransferase; INR, international normalised ratio;

CRRT, continuous renal replacement therapy; PIM2, paediatric index of mortality;

PELOD, pediatric logistic organ dysfunction score

29

30

All (n=136)

CRRT (n=45)

Survivors (n=101)

Non-survivors

(n=35)

P-value (survivors vs

non)

Age, months 54±5.7 68±9.8 63±7.0 28±7.4 0.006*Age, months MEDIAN 20 50 27 4Female, n (%) 61 (45) 18 (40) 48 (48) 13 (37) 0.033Male, n (%) 75 (55) 27 (60) 53 (52) 22 (63) 0.033Weight (kg) 20.4±3.0 23.5±3.9 25.9±4.3 9.3±2.0 0.027*Duration of illness prior to admission to hospital, days 7.0±0.9 9.2±1.6 6.4±0.9 8.4±2.2 0.319

Time on the ward prior to PICU transfer, days 7.9±1.2 10.7±2.8 5.6±0.6 13.2±3.7 0.001*

Aetiology of PALF: Indeterminant, n (%) 67 (48) 24 (53) 56 (55) 11 (31) 0.018* Toxic, n (%) 19 (14) 6 (13) 18 (18) 1 (3) 0.026* Infectious, n (%) 20 (15) 3 (7) 10 (10) 10 (29) 0.012* Metabolic, n (%) 21 (16) 9 (20) 11 (11) 10 (29) 0.027* Ischaemic, n (%) 1 (1) 1 (2) 0 (0) 1 (3) 0.229 Infiltrative, n (%) 6 (4) 2 (4) 4 (4) 2 (6) 0.647 Autoimmune, n (%) 2 (2) 0 (0) 2 (2) 0 (0) 1.000Grade of encephalopathy Grade <2, n (%) 89 (65) 23 (51) 69 (68) 20 (57) 0.302 Grade ≥2, n (%) 47 (35) 22 (49) 32 (32) 15 (43) 0.302Heart rate, beats/minute 124±3.5 119±5.2 121±4.3 133±5.4 0.134Respiratory rate, breaths/minute 28±1.4 26±2.1 27±1.4 32±2.9 0.090MAP, mmHg 64±1.9 66±2.5 65±2.4 62±3.1 0.503Urine output, mls/kg/hr 2.5±0.3 2.6±0.5 2.5±3.1 2.4±0.5 0.985Arterial: pH 7.37±0.01 7.38±0.27 7.38±0.02 7.36±0.03 0.603 HCO3-, mmol/L 24.1±0.7 24.2±1.1 23.9±0.7 24.5±1.5 0.686 Lactate, mmol/L 4.7±0.4 5.1±0.7 3.9±0.4 6.3±0.7 0.003* Ammonia, umol/L (admission) 141±10.9 152±20.0 136±13.3 158±19.5 0.387 Ammonia, umol/L (24hr peak) 155±8.9 186±17.1 132±9.6 179±21.8 0.025*Venous: Urea, mmol/L 5.4±0.5 5.1±1.0 5.1±0.4 6.3±1.2 0.222 Creatinine, mmol/L 65±4.5 68±9.6 69±4.6 57±10.5 0.230 AST, umol/L 2877±368 2486±547 2994±488 2612±509 0.666 Bilirubin, umol/L 217±16.8 283±25.8 222±21.2 205±26.2 0.665 INR 4.1±0.3 4.5±0.2 3.7±0.2 5.2±0.7 0.006* Platelets, 109/L 173±11.5 194±20.2 187±13.8 137±19.5 0.057Ionotropes, n (%) 67 (49) 42 (93) 32 (32) 35 (100) <0.001*Intubated and ventilated, n (%) 89 (65%) 45 (100) 54 (53) 35 (100) <0.001*CRRT, n (%) 45 (100) 26 (26) 19 (54) 0.003*Mortality risk score PIM2 53±2.8 54±4.1 39±2.9 54±6.3 0.016* PELOD 7±5.9 10±6.2 5±4.8 12±6.1 0.006*

DECLARATIONS

Conflicts of Interest

All authors declare that the answer to the questions on your competing interest form, www .icmje.org/coi_disclosure.pdf, are all No and therefore have nothing to declare

Financial Disclosure

Dr. Abdel Douiri acknowledges financial support from the National Institute for Health Research (NIHR) Biomedical Research and from the NIHR Collaboration for Leadership in Applied Health Research and Care South London at King's College Hospital NHS Foundation Trust though not for this project. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health

Ethical Statement:

This study has been registered as a service improvement project at King’s College Hospital and all guideline changes were reviewed by the clinical guideline committee prior to implementation

31