Pre‐Emptive Initiation of Continuous Renal Replacement Therapy in Patients With Acute Liver Failure With Cerebral Edema Improves Outcomes: A Randomized Controlled Trial

ABSTRACT

Background and Aim

An early initiation of continuous renal replacement therapy (CRRT) for ammonia reduction has shown a reduction in deaths due to cerebral edema (CE) in acute liver failure (ALF); however, there are no controlled trials assessing the same.

Methods

We performed an open‐label pilot randomized controlled trial (RCT) on ALF patients with CE. Patients underwent therapeutic plasma‐exchange (PLEX) by centrifugal apheresis after initial resuscitation. Group 1 received CRRT initiation within the first 12 h, while Group 2, CRRT was initiated for PLEX non‐responders. The primary endpoint was 28‐day survival.

Results

Patients aged 28.11 ± 10.10 years, 56.67% viral, 69% hyperacute were randomized. At day 28, 46% died and 11% underwent liver transplant. There were a total of 31 protocol violations, significantly more in group 2 (55.6% vs. 13.3%; p < 0.001). On ITT and piece‐wise exponential regression analysis, pre‐emptive CRRT was associated with a significantly lower hazard of death during the first 7 days (2.2% vs. 17.8% Hazard ratio [HR] 0.12, 95% confidence interval [CI] 0.02–0.96) with comparable survival at 28‐days. Notably, higher 28‐day mortality was observed on PP analysis in group 2 (80% vs. 46%, HR 3.10, 95% CI 1.57–6.12) with a significantly higher reduction in ammonia and improvement in mean arterial pressure and lower number of sessions of TPE. Each hour delay in CRRT was associated with increased mortality (HR 1.01, 1.00–1.02).

Conclusion

A pre‐emptive initiation of CRRT in ALF patients is synergistic to TPE, reduces early deaths by rapid improvement in hemodynamics, ammonia and cerebral edema (NCT04991259).

Time to CRRT initiation, absence of sepsis, and lower SOFA score were determinants of improved 28‐day survival CRRT was synergistic to plasma‐exchange in reducing early deaths in patients with ALF and cerebral edema.

Abbreviations‐

ADAMTS13

A Disintegrin and Metalloproteinase with Thrombospondin Motif‐13 (ADAMTS‐13)

ALF

acute liver failure

CI

confidence interval

CRRT

continuous renal replacement therapy

CT

computed tomography

DAMPS

damage‐associated molecular patterns (DAMPs)

ELISA

Enzyme linked immunosorbent assay

HMGB1

High mobility group box protein‐1

HR

hazard ratio

IL‐6

interleukin‐6 (IL‐6)

ITT

intention‐to‐treat analysis

KCH

King's College Hospital Criteria

MAP

mean arterial pressure

ONSD

optic nerve sheath diameter

PAMPS

pathogen‐associated molecular patterns

PLEX

Therapeutic plasma‐exchange

PP

Per‐protocol

RCT

RCT‐randomized controlled trial

sHR

subdistribution hazard ratio

SIRS

systemic inflammatory response syndrome

SNOSE

sequentially numbered opaque sealed envelopes

SOFA

sequential organ failure scores

SVPE

standard‐volume plasma exchange

vWF

von‐Willebrand factor

1. Introduction

Patients with acute liver failure (ALF) have jaundice and coagulopathy, followed by hepatic encephalopathy, which is very often associated with cerebral edema [1]. The main drivers of cerebral edema include systemic inflammation, cerebral perfusion, and hyperammonemia [1, 2]. Immune dysfunction secondary to severe systemic inflammation which rapidly progresses to functional immunoparesis is a hallmark of the development of secondary organ dysfunction and sepsis in ALF patients [1, 3]. The systemic inflammation is driven by damage‐associated molecular patterns (DAMPs) released by the dying or necrotic hepatocytes and pathogen‐associated molecular patterns (PAMPs) by the gut dysbiosis and bacterial translocation released by the ongoing liver injury. The brain is the most frequent secondary organ that gets involved in the severe systemic inflammation and the unabated liver injury that results in ammonia accumulation [4]. The perfusion pressure of the brain is also affected by the hemodynamic alterations that are driven by sepsis and associated vasodilatation secondary to the release of inflammatory mediators, cytokines, and nitric oxide. Altogether, the accumulation of neurotoxins like ammonia, DAMPs, endotoxin, inflammatory cytokines, and lactate requires an extracorporeal liver support system for clearing these cytotoxins to enable an environment that is conducive to liver regeneration [5].

Therapeutic plasma‐exchange (PLEX) with its ability to clear all the protein or water‐soluble toxins and replenish the coagulation factors is the most promising strategy for managing ALF patients [6]. Two randomized controlled trials have demonstrated the efficacy of PLEX in improving outcomes of ALF patients [7, 8]. The large multicentric trial by Larsen and colleagues demonstrated the efficacy of high‐volume PLEX in ALF patients [7]. Smaller randomized controlled trial showed the efficacy of standard‐volume plasma exchange (SVPE) in improving transplant‐free survival, in patients predominantly with hyperacute presentation, viral aetiology and with cerebral edema [8]. However, multiple other studies, retrospective and cohort, including a recent systematic review, demonstrated the benefit of use of continuous renal replacement therapy (CRRT) in ALF patients [9, 10]. The benefits of CRRT are related to clearance of ammonia, cytokines, and therefore improvement in cerebral edema. Currently, there are retrospective series; no randomized controlled trials evaluating the timing of initiation of CRRT in ALF patients, and whether this could be synergistic to TPE.

We hypothesized that a pre‐emptive initiation of continuous renal replacement therapy would ameliorate the cytokine storm, improve ammonia clearance, reduce cerebral edema, and overall outcomes in patients of ALF compared to standard initiation. Our primary objective was to evaluate the outcomes on 28‐day transplant‐free survival and secondary objectives were to compare the improvement in cerebral edema (ammonia and optic nerve sheath diameter), duration of mechanical ventilation and intensive care unit stay, impact on arterial lactate, systemic inflammatory response syndrome, systemic hemodynamics, sequential organ failure scores (SOFA) scores, adverse events of therapy and to evaluate the effect on DAMPS, pro‐inflammatory cytokines, endothelial functions wherever possible in a subset of patients.

2. Patients and Methods

The study was conducted at a tertiary care liver hospital in Delhi from October 2021 to May 2024. All patients with confirmed ALF admitted to the Liver Intensive Care Unit (L‐ICU) were screened and enrolled. Written informed consent was obtained from the next of kin. The institutional ethics committee approved the study protocol (IEC/2021/86/NA0) which was registered with ClinicalTrial.gov (identifier: NCT02718079). All authors had access to the study data and had reviewed and approved the final manuscript.

2.1. Participants

Consecutive patients of ALF, defined as jaundice and coagulopathy complicated by encephalopathy within 4 weeks of the onset of jaundice without underlying chronic liver disease, were screened for enrollment into the study. Patients with documented cerebral edema on CT‐scan and arterial ammonia > 150 μg/dL were randomized to the intervention. Patients aged < 12 or more than 75 years and extremely moribund patients with active sepsis, disseminated intravascular coagulation, or severe hemodynamic instability despite norepinephrine dose > 0.15 μg/kg/min were excluded. Patients with post‐resection, malignancy, pregnancy‐related liver failure, coma of non‐hepatic origin, those meeting emergency criteria for immediate initiation of dialysis at the time of randomization, or transferred from other hospitals who were already on renal replacement therapy were excluded. The non‐availability of a living donor at the time of randomization, those having any contraindication to PLEX, and patients with a lack of informed consent were also excluded.

2.2. Randomization

Randomization was performed by the clinical trial co‐ordinator, and allocation concealment was performed by the sequentially numbered opaque sealed envelopes (SNOSE) technique. Patients in both groups received standard medical treatment along with standard‐volume plasma exchange.

2.3. Standard Medical Treatment

All patients were managed in the L‐ICU (Details in Supporting Information).

2.3.1. Liver Transplantation (LT)

As a protocol, after initial stabilization, all ALF patients with cerebral edema are assessed for the need of LT and counselled for a donor evaluation [11]. The indication for LT was according to KCH meeting a single criterium of an international normalized ratio (INR) > 6.5 or meeting 3 out of 5 poor prognostic factors—age < 10 or > 40 years, jaundice‐to‐encephalopathy (J‐to‐E) time duration > 7 days, total bilirubin value > 17.5 mg/dL, Non A, Non B aetiology, and INR ≥ 3.5. Patients with an available living donor who meet the KCH criteria are considered for living donor liver transplantation, while those without a suitable donor are listed for deceased donor transplantation. For other patients—those unresponsive to standard medical treatment, without an available donor, or whose families request additional time—we consider extracorporeal therapies (PLEX/CRRT) with dynamic monitoring until either clinical recovery or liver transplantation can be achieved. Because deceased donation is sparse in our part of the country, the family is counselled about the need of a voluntary healthy donor from the family, if possible. The contraindication of LT included patients with uncontrolled sepsis, rapidly escalating need of vasopressors, evidence of compromised brain stem function, especially fixed and dilated pupils, invasive fungal infection, severe irreversible cardiopulmonary disease, advanced age, and extrahepatic malignancy.

2.4. Protocol of Standard Volume Plasma‐Exchange (SVPE)

The first session was performed within 12–24 h of admission. The duration of each exchange varied from 3 to 4 h and was 1.5–2.1 times the patient's plasma volume, with the rate of plasma flow of 25–30 mL/min. Fresh frozen plasma‐was used as the replacement fluid. The sessions were performed on consecutive days until the desired response was achieved [8].

2.5. Interventions

2.5.1. Group 1 (Pre‐Emptive CRRT)

CRRT was initiated within the first 12 h of intensive care unit admission in patients meeting the inclusion criteria.

2.5.2. Group 2 (Standard Initiation)

Patients received CRRT for standard renal indications or in patients with non‐response in cerebral edema after at least two consecutive sessions of TPE.

Patients were followed for the entire duration of the ICU stay with monitoring of optic nerve sheath diameter (ONSD), ammonia, INR, lactate, mean arterial pressure, arterial blood gas, and other relevant parameters.

2.5.3. Continuous Renal Replacement Therapy

Continuous venovenous haemodiafiltration (CVVHDF) using Prismaflex (Gambro, Sweden) with the target effluent flow rate: 25–35 mL/kg/h, and blood flow rates of 100–180 mL/h were performed. The dose was adjusted to facilitate ammonia clearance, \ hemodynamic status and the response to therapy. All patients underwent CRRT with Prismaflex M100 Set: 0.9 m² Vantive AN69 HF hollow fibre: Acrylonitrile and sodium methallyl sulfonate copolymer. In both groups, a proportion of patients underwent CRRT with oXiris membrane which is a three‐layered membrane and features a modified AN69 membrane with polyethyleneimine (PEI) for endotoxin adsorption and a pre‐grafted heparin layer to reduce thrombogenicity, aiding in circuit patency. No anticoagulation was used for CRRT; however, citrate anticoagulation was used only if there was early filter clotting within12–24 h.

2.6. Measurement of Inflammatory Cytokines, Endothelial Markers, and High Mobility Group Box Protein‐1 (HMGB‐1)

We also aimed to study the inflammatory cytokines interleukin‐6 (IL‐6), interleukin‐beta (IL‐1β), IL‐10, HMGB1, von‐Willebrand factor, and A Disintegrin and Metalloproteinase with Thrombospondin Motif‐13 (ADAMTS‐13) in a subset of patients at randomization and 24 h compared between the two groups by Enzyme linked immunosorbent assay (ELISA).

2.6.1. Sample‐Size Calculation and Statistical Methods

The study was initiated to compare the difference in transplant‐free survival in patients initiated with preemptive CRRT compared to standard initiation. Both groups received SVPE and standard medical treatment. Due to the lack of data and limited RCTs, the trial was designed as a pilot randomized controlled trial (RCT) with an aim to enrol a minimum of 30 patients in each group. The study was registered on ClinicalTrials.gov (NCT02718079). Although the registry currently lists an enrollment of 40 participants, recruitment continued under the same protocol and eligibility criteria, and a total of 90 participants were ultimately enrolled in the study. The trial enrolled more participants than initially planned, as additional patients became available over time. Continuous variables were analysed using an independent Student's t‐test or Mann–Whitney test for parametric or non‐parametric data, respectively. Categorical variables were analysed using Fisher's Exact test or the Chi‐square test between the two randomized groups of patients. Intention‐to‐treat analysis and per‐protocol analysis were performed. Kaplan–Meier curves were created, and Cox‐regression was used for survival analysis. Competing risk‐survival analysis was performed considering liver transplant as a competing event. Considering early mortality in ALF is often driven by cerebral edema, whereas later outcomes reflect sepsis or transplant, to better characterize this pattern, we performed an additional time‐interval–based survival analysis by dividing follow‐up into clinically relevant periods (0–7, 8–14, and > 14 days). A piecewise exponential regression model was fitted, assuming constant hazard within each interval while allowing variation between intervals [12]. Treatment effects were estimated using interaction terms between treatment and time intervals, permitting time‐varying effects [13, 14]. Results are presented as interval‐specific hazard ratios with 95% confidence intervals.

3. Results

A total of 313 patients were screened for enrolment in the study, 223 patients were excluded who met the exclusion criteria, and a total of 90 patients, 45 each to the intervention group, were allocated (Figure 1). The demographic, clinical, biochemical features, severity of liver disease, intracranial pressure indices, and neuroimaging findings suggestive of cerebral edema, hemodynamic parameters of the two groups of patients at baseline were similar and are summarized in Table 1. There was a predominance of viral aetiology, of which the majority had hepatitis A in either arm (29 [64.4%] vs. 22 [48.9%]; p = 0.136), and the majority of patients were mechanically ventilated (39 [86.7%] vs. 36 [80%]; p = 0.40), and all had evidence of cerebral edema on computed tomography CT‐imaging. On comparing the two groups at baseline, there were no differences in the requirement of vasopressors (17 [37.8%] vs. 19 [42.2%]; p = 0.68) and proportion meeting King's College Hospital criteria (KCH) (15 [33.3%] vs. 18 [40%]; p = 0.512) in the pre‐emptive versus standard initiation group, respectively. There were no differences in the arterial lactate, ammonia levels, and markers of liver failure severity between the two groups (Table 1).

FIGURE 1.

Study consort of the patient enrolment.

TABLE 1.

Baseline characteristics of the study cohort stratified by the randomization group.

Variable	Total	Randomization group
		Pre‐emptive CRRT (n = 45)	Standard initiation (n = 45)	p
Age (in years)	28.11 ± 10.10	27.71 ± 8.67	28.51 ± 11.44	0.709
Gender (Male: Female)	53:37 (58.89%: 41.11%)	29:16 (64.44%: 35.55%)	24:21 (53.33%: 46.67%)	0.284
Aetiology n (%) Viral Hepatitis A	51 (56.67%)	29 (64.44%)	22 (48.89%)	0.136
Hyperacute presentation	62 (68.89%)	31 (68.89%)	31 (68.89%)	1.000
KCH criteria met; n (%)	33 (36.67%)	15 (33.33%)	18 (40.00%)	0.512
International normalized ratio	3.68 ± 2.45	4.01 ± 2.61	3.35 ± 2.26	0.205
Total bilirubin (mg/dL)	12.96 ± 9.06	10.97 ± 8.07	14.95 ± 9.64	0.037
SOFA score	10.11 ± 3.22	9.96 ± 3.51	10.27 ± 2.93	0.649
Lactate (μmol/L)	4.33 ± 3.75	4.09 ± 3.62	4.57 ± 3.90	0.552
Ammonia (μg/dL)	363.12 ± 186.07	389.29 ± 208.31	336.96 ± 158.87	0.184
Optic nerve sheath diameter (ONSD) (average of right and left eye, in mm)	4.74 ± 0.62	4.83 ± 0.60	4.65 ± 0.63	0.171
Systemic inflammatory response syndrome	67 (74.44%)	32 (71.11%)	35 (77.78%)	0.468
Serum creatinine (mg/dL)	1.49 ± 0.94	1.42 ± 0.77	1.56 ± 1.09	0.490
Time of CRRT from ICU admission (in hrs)	36.73 ± 30.95	18.12 ± 8.21	57.23 ± 33.83	< 0.001
Jaundice to encephalopathy (days)	5.03 ± 4.09	4.96 ± 3.06	5.11 ± 4.95	0.858
Hyperacute presentation	62 (68.89%)	31 (68.89%)	31 (68.89%)	1.000
Mechanical ventilation	75 (83.33%)	39 (86.67%)	36 (80.00%)	0.396
Suspected sepsis	42 (46.67%)	24 (53.33%)	18 (40.00%)	0.205
Total leucocyte counts (~103/L)	13.26 ± 9.10	13.46 ± 10.71	13.06 ± 7.27	0.836
Platelets (~109/L)	173.99 ± 101.58	183.50 ± 110.57	163.74 ± 91.24	0.385
Use of vasopressors	36 (40.00%)	17 (37.78%)	19 (42.22%)	0.667
Pa02/Fi02 ratio	290.19 ± 82.18	302.00 ± 72.60	278.38 ± 90.02	0.174
Mean arterial pressure (mm of Hg)	85.28 ± 13.57	84.87 ± 15.01	85.69 ± 12.11	0.776
C‐reactive protein (mg/dL)	40.46 ± 50.91	36.79 ± 59.54	44.12 ± 40.87	0.498

Note: Data presented as mean ± standard deviation for continuous variables, number (%) for categorical variables.

Abbreviations: CRRT, Continuous renal replacement therapy; Fi02, Fraction of inspired oxygen; KCH, King's college Hospital criteria; ONSD, optic nerve sheath diameter; Pa02, Partial pressure of oxygen; SOFA, sequential organ failure assessment.

The mean number of sessions and the mean plasma exchange volume were similar between the groups ([2.18 ± 1.6 vs. 2.13 ± 1.56; p = 0.894] and [2.76 ± 0.97 vs. 2.46 ± 0.69; p = 0.143], respectively). However, the time to initiate CRRT was significantly shorter in the pre‐emptive group compared to the standard initiation group (8.38 [7.80] vs. 51.97 [30.23] hours; p < 0.001). The initial dose of CRRT was consistent across both groups (22.6 [4.56] vs. 22.51 [4.05] mL/kg; p = 0.92), as were the duration of CRRT (154 [188.85] hours vs. 122 [153.07]; p = 0.89) and the mean number of filters used (3.83 [4.32] vs. 2.94 [2.78]; p = 0.34).

3.1. Primary Outcome

3.1.1. Intention‐To‐Treat Analysis

The primary outcome of the study was to assess the transplant‐free survival on day 28. There were a total of 41 deaths at 28 days, 21 (46.7%) in pre‐emptive CRRT and 20 (44.4%) in standard initiation group. Among the patients who died, most fatalities were attributed to sepsis (19 cases, 46.3%), multiorgan failure with severe cerebral edema (11 cases, 26.8%), or refractory intracranial hypertension (ICH) (11 cases, 26.8%). Those with refractory ICH experienced earlier mortality, with a mean time to death of 6.45 ± 1.97 days, compared to patients who died from sepsis (14.21 ± 6.09 days) or multiorgan failure with severe cerebral edema (14.18 ± 5.91 days) (p < 0.001). A significantly higher proportion of patients in the standard initiation group died due to refractory ICH (9 [22%] vs. 2 [4.8%]; p = 0.015).

On intention‐to‐treat (ITT) analysis, there was no significant difference in 28‐day survival between the two groups (Hazard ratio [HR] 1.22, 95% confidence interval [CI] 0.67–2.23) (Figure 2A). Both groups included patients who underwent living donor liver transplantation (4 [8.8%] vs. 7 [15.5%]). After adjusting for liver transplant as a competing event, the group comparison remained non‐significant for 28‐day survival (subdistribution hazard ratio [sHR] 1.18, 95% CI 0.64–2.17) (Figure 2B).

FIGURE 2.

(A) Kaplan–meier survival curves between the two randomization groups for the 28‐day survival on intention‐to‐treat analysis. (B) Fine and Grey survival curves for the 28‐day survival between the two randomization groups on the intention‐to‐treat analysis.

To explore time‐varying effects, piecewise exponential regression modelling was conducted, revealing that pre‐emptive CRRT was associated with a significantly lower hazard of death during the first 7 days (HR 0.12, 95% CI 0.02–0.96; p = 0.046). This effect diminished over time, as indicated by significant interaction terms reflecting attenuation in later intervals (p < 0.05). The corresponding interval‐specific hazard ratios were approximately 1.23 for days 8–14 and 1.56 for days 15–28, consistent with trends observed in the Kaplan–Meier curves (Figure 2A and Table S1). Overall, the time‐interval analysis indicates that the benefit of pre‐emptive CRRT is most pronounced during the initial 7 days, with subsequent intervals showing reduced effects.

3.1.2. Per‐Protocol Analysis

There were a total of 31 protocol violations, which were significantly more often in the standard initiation group compared to the pre‐emptive group (25 [55.6%] vs. 6 [13.3%]; p < 0.001). Liver transplant was also considered a protocol violation. The other reasons included worsening of cerebral edema, hyperammonemia, or multiorgan dysfunction requiring CRRT initiation. The details of patients who required early CRRT initiation in the standard initiation arm are presented in Table S2. Notably, baseline characteristics were similar between groups across all parameters except serum creatinine, which was significantly elevated (2.19 ± 1.42 vs. 1.32 ± 0.91; p = 0.045) in patients who experienced protocol violations requiring early CRRT initiation in the standard initiation group. Additionally, these patients demonstrated markedly higher ammonia levels following therapeutic plasma exchange (446.95 ± 135.15 vs. 279.95 ± 114.23; p < 0.001).

Other reasons for protocol violations included spontaneous recovery of liver functions after PLEX (6/25 [24%] vs. 1/6 [17%]) and liver transplant (7/25 [28.0%] vs. 4/6 [66.7%]) respectively. On per‐protocol (PP) analysis (n = 59), significantly higher deaths at 28 days were observed in the standard initiation (16/20 [80%] vs. 18/39 [46.2%], HR 3.10, 95% CI 1.59–6.05) compared to the pre‐emptive group, respectively (Figure 3A). On competing risk survival analysis, with liver transplant as a competing risk, the survival was worse for the standard initiation of CRRT (sHR 3.10, 95% CI 1.57–6.12) (Figure 3B). The number of PLEX sessions was also significantly lower in patients in the pre‐emptive group ([2.18 ± 1.68] vs. [3.60 ± 1.96]; p = 0.005) on per‐protocol analysis.

FIGURE 3.

(A) Kaplan–meier survival curves between the two randomization groups for the 28‐day survival on per‐protocol analysis. (B) Fine and Grey survival curves for the 28‐day survival between the two randomization groups on the per‐protocol analysis.

On ITT and PP analysis, the relative risk reduction for pre‐emptive CRRT was 1.05; 95% CI (0.67–1.65) and 0.57; 95% CI (0.38–0.86) with an absolute risk reduction of −0.023 and 0.338 and number needed to treat of 45 and 3 respectively (Table S3).

The dose of CRRT in both groups was targeted based on serum ammonia, hemodynamics, and clinical response. The maximal dose of CRRT was significantly associated with peak ammonia (R = 0.423; p < 0.001), that is, the higher the ammonia, the higher was the administered dose of CRRT (Figure S1). A proportion of patients in both groups underwent oXiris with CRRT versus CRRT alone; 8 (17.78%) in the pre‐emptive group versus 7 (15.56%) in the standard initiation of CRRT group, respectively. The characteristics of patients who underwent oXiris hemofilter versus those who did not are summarized in Table S4. Interestingly, patients who had oXiris had significantly higher SOFA scores, INR, and CRP levels, and lower PaO₂/FiO₂ ratio compared to those who did not.

3.2. Predictors of 28‐Day Mortality

3.2.1. Intention‐To‐Treat (ITT) Analysis

On ITT univariate competing risks survival analysis, higher age (sHR 1.04, 95% CI 1.02–1.06), arterial lactate (sHR 1.09, 95% CI 1.02–1.16), presence of culture‐proven sepsis (sHR 2.63, 95% CI 1.35–5.14), time to CRRT initiation (sHR 1.01, 95% CI 1.00–1.02), SOFA score (sHR 1.12, 95% CI 1.02–1.23), use of vasopressors (sHR 3.11, 95% CI 1.67–5.79), and meeting the KCH criteria (sHR 1.33, 95% CI 1.03–1.73) were significant predictors of 28‐day mortality. The randomization group was not a significant predictor of 28‐day mortality. On multivariable competing risks survival analysis, higher age (sHR 1.04, 95% CI 1.02–1.07), sepsis (sHR 2.66, 95% CI 1.45–4.90), and higher SOFA score (sHR 1.10, 95% CI 1.01–1.21) were independent predictors of 28‐day mortality. The same factors were identified as independent determinants of 28‐day mortality in the Cox‐regression analysis (Table 2 and Table S5).

TABLE 2.

Factors associated with 28‐day survival on competing risk survival analysis—Intention‐to‐treat and per‐protocol analysis.

	Intention‐to‐treat (n = 90)				Per‐protocol‐analysis (n = 59)
	Crude subdistribution hazard ratio	95% confidence intervals	Adjusted subdistribution hazard ratio	95% confidence intervals	Crude subdistribution hazard ratio	95% confidence intervals	Adjusted subdistribution hazard ratio	95% confidence intervals
Variable
Age (in years)	1.04**	1.02–1.06	1.04**	1.02–1.07
Gender (females)	1.74	0.95–3.18			1.03*	1.00–1.06	1.02	0.99–1.04
Etiololgy (hepatitis A)	0.87	0.47–1.60			1.65	0.86–3.16
Total Bilirubin (mg/dL)	1.03*	1.00–1.05			0.71	0.37–1.37
ONSD (right) (in mm)	1.39	0.84–2.30			1.02	1.00–1.05
ONSD (left) (in mm)	1.68	0.96–2.97			1.02	0.58–1.80
ONSD (average) (in mm)	1.57	0.90–2.73			1.18	0.60–2.31
Mean arterial pressure (mm of Hg)	0.98	0.96–1.00			1.09	0.57–2.08
Arterial lactate (μmol/L)	1.09**	1.02–1.16			0.99	0.97–1.01
SOFA	1.12*	1.02–1.23	1.10*	1.01–1.21	1.09*	1.02–1.17
Ammonia (μg/dL)	1.00	1.00–1.00			1.11*	1.00–1.22	1.12*	1.03–1.23
International normalized ratio	1.01	0.90–1.14			1.00	1.00–1.00
Serum creatinine (mg/dL)	1.09	0.78–1.50			1.06	0.93–1.20
KCH criteria met	1.33*	1.03–1.73			1.14	0.81–1.59
Randomization group (Pre‐emptive vs. Standard initiation)	1.22	0.67–2.23	1.00	1.00–1.00	1.29	0.98–1.69
Sepsis (absent) as ref	1.00	1.00–1.00	1.00	1.00–1.00	3.10**	1.59–6.05	2.25*	1.16–4.36
Bacterial	2.63**	1.35–5.14	2.66**	1.45–4.90	1.00	1.00–1.00	1.00	1.00–1.00
Fungal	4.02**	1.52–10.60	3.89**	1.48–10.20	2.95**	1.42–6.10	2.35*	1.12–4.95
Suspected sepsis	2.22*	1.18–4.20			2.16	0.80–5.83	1.79	0.54–5.90
Time to CRRT initiation	1.01*	1.00–1.02			2.13*	1.04–4.38
Use of vasopressors	3.11**	1.67–5.79			4.13**	2.07–8.24

Note: * and ** denote statistical significance at the 5% and 1% levels, respectively.

Abbreviations: CRRT, continuous renal replacement therapy; Fi02, Fraction of inspired oxygen; KCH, King's college Hospital criteria; ONSD, optic nerve sheath diameter; Pa02, partial pressure of oxygen; SOFA, sequential organ failure assessment.

3.2.2. Per‐Protocol (PP) Analysis

Per‐protocol multivariable competing risk survival analysis demonstrated that the randomization group was an independent predictor of mortality (sHR 2.25, 95% CI 1.16–4.36), after adjusting for SOFA score and sepsis status. Notably, delay in initiation of CRRT correlated with poorer outcomes (sHR 2.13, 95% CI 1.04–4.38). Furthermore, Cox regression analysis indicated that the randomization group—standard versus pre‐emptive initiation—was independently associated with increased 28‐day mortality (HR 2.47, 95% CI 1.18–5.15).

3.3. Secondary Outcomes

The mean group differences were comparable with respect to most parameters on the ITT analysis, while a significant reduction in ONSD (−1.01 [95% CI −1.9, −0.23]) was observed between the two groups suggesting a reduction in cerebral edema. Further, on PP analysis, there was a decrease in ammonia −177.18 (95% CI −334.04, −20.32) in the pre‐emptive group. A significant reduction was observed in the SOFA score −3.77 (−6.38, −1.16) in favour of pre‐emptive CRRT (Table 3).

TABLE 3.

Comparison of change in parameters between the two groups.

Measure	Pre‐emptive CRRT (n = 45)	Standard initiation (n = 45)	Difference between changes in pre‐emptive CRRT and standard initiation at day 5 (95% CI)
Intention‐to‐treat analysis (n = 90)
Lactate (μmol/L)	−2.04	−1.59	−0.45 (−2.26, 1.37)
Ammonia (mg/dL)	−244.22	−142.78	−101.44 (−211.78, 8.90)
SOFA	−1.31	0.58	−1.89 (−3.93, 0.15)
MAP (mm of Hg)	−20.76	−17.49	−3.27 (−19.87, 13.33)
ONSD (average of left and right eye in mm)	−1.17	−0.11	−1.01 (−1.90, −0.23)*
Per‐protocol analysis (n = 59)
Lactate (μmol/L)	−2.05	0.05	−2.09 (−4.53, 0.34)
Ammonia (mg/dL)	−254.40	−77.23	−177.18 (−334.04, −20.32)*
SOFA	−1.82	1.95	−3.77 (−6.38, −1.16)**
MAP (mm of Hg)	−21.08	−12.10	−8.98 (−31.93, 13.98)
ONSD (average of left and right eye in mm)	−1.11	−0.42	−0.69 (−1.96, 0.57)

Note: *significant p < 0.05, ** significant p < 0.001.

Abbreviations: CRRT, continuous renal replacement therapy; MAP, mean arterial pressure; ONSD, optic nerve sheath diameter; SOFA, sequential organ failure assessment.

On comparing the two groups at each time point from day 1 to day 5 on both ITT and PP analysis, there was a significant improvement in mean arterial pressure (MAP) by day 3 in the pre‐emptive group, decline in the SOFA and reduction in the ONSD (Figure S3). On per‐protocol analysis, the ammonia levels also showed significant decline in the pre‐emptive group by day 2 (Table 4) The lactate levels did not show difference between the two groups (Figure 4A–E) (Figure S2A–E).

TABLE 4.

Comparison of primary and secondary outcome measures on intention‐to‐treat and per‐protocol analysis.

	Intention to treat (n = 90)			Per protocol (n = 59)
	Pre‐emptive CRRT (n = 45)	Standard initiation (n = 45)	p	Pre‐emptive CRRT (n = 45)	Standard initiation (n = 45)	p
	N = 45	N = 45		N = 39	N = 20
28‐day mortality	21 (47%)	20 (44%)	1.00	18 (46%)	16 (80%)	0.015
Lactate (mmol/L)
Day 1	4.09 ± 3.62	4.57 ± 3.90	0.55	4.13 ± 3.77	4.80 ± 4.00	0.52
Day 2	3.55 ± 3.49	4.40 ± 5.30	0.37	3.68 ± 3.68	4.50 ± 5.83	0.52
Day 3	2.56 ± 2.71	3.04 ± 3.31	0.45	2.62 ± 2.85	4.47 ± 4.08	0.047
Day 4	2.92 ± 3.01	3.55 ± 3.37	0.35	2.88 ± 3.11	4.16 ± 4.09	0.18
Day 5	2.06 ± 2.29	2.98 ± 2.90	0.098	2.08 ± 2.36	4.85 ± 3.26	< 0.001
Mean arterial pressure (mm of Hg)
Day 1	84.87 ± 15.01	85.69 ± 12.11	0.78	84.69 ± 15.09	83.80 ± 11.02	0.82
Day 2	85.29 ± 17.67	85.20 ± 9.53	0.98	84.69 ± 18.21	83.65 ± 10.11	0.81
Day 3	96.62 ± 17.46	87.04 ± 15.26	0.007	96.82 ± 17.69	80.95 ± 14.92	0.001
Day 4	80.33 ± 32.41	74.64 ± 33.88	0.42	80.62 ± 31.77	74.00 ± 32.98	0.46
Day 5	64.11 ± 44.40	68.20 ± 37.80	0.64	63.62 ± 43.79	71.70 ± 37.67	0.49
SOFA score
Day 1	9.96 ± 3.51	10.27 ± 2.93	0.65	10.36 ± 3.48	10.60 ± 3.35	0.80
Day 2	9.44 ± 3.45	11.16 ± 3.27	0.018	9.67 ± 3.47	11.50 ± 3.75	0.067
Day 3	9.49 ± 4.10	11.29 ± 3.40	0.026	9.67 ± 4.14	12.65 ± 2.76	0.005
Day 4	10.27 ± 4.42	10.47 ± 4.04	0.82	10.26 ± 4.30	12.30 ± 3.66	0.075
Day 5	8.64 ± 5.05	10.84 ± 4.45	0.031	8.54 ± 5.07	12.55 ± 4.10	0.003
Ammonia (μg/dL)
Day 1	346.54 ± 199.15	388.45 ± 225.22	0.35	350.31 ± 202.20	394.82 ± 252.23	0.47
Day 2	263.68 ± 159.18	328.89 ± 180.73	0.073	262.65 ± 155.26	394.69 ± 188.24	0.006
Day 3	270.02 ± 194.50	316.75 ± 261.48	0.34	268.77 ± 200.27	307.95 ± 146.92	0.44
Day 4	292.10 ± 288.50	281.34 ± 262.74	0.85	300.74 ± 305.67	297.39 ± 174.09	0.96
Day 5	145.07 ± 188.37	194.18 ± 188.20	0.22	137.66 ± 196.23	240.81 ± 221.34	0.072
ONSD (in mm) Average of left and right eye
Day 1	4.83 ± 0.60	4.65 ± 0.63	0.171	4.86 ± 0.60	4.80 ± 0.71	0.75
Day 2	4.74 ± 0.53	4.86 ± 0.65	0.35	4.78 ± 0.53	5.00 ± 0.80	0.20
Day 3	4.58 ± 0.46	4.92 ± 0.57	0.003	4.60 ± 0.47	5.14 ± 0.60	< 0.001
Day 4	4.39 ± 1.06	4.86 ± 1.25	0.058	4.55 ± 0.75	4.90 ± 1.83	0.31
Day 5	3.67 ± 2.06	4.57 ± 1.59	0.023	3.75 ± 2.02	4.38 ± 2.32	0.28
Systemic inflammatory response syndrome (SIRS)
Day 1	32 (71.1%)	35 (77.8%)	0.47	28 (71.8%)	16 (80.0%)	0.49
Day 2	7 (15.6%)	21 (46.7%)	0.001	6 (15.4%)	10 (50.0%)	0.005
Day 3	2 (4.4%)	15 (33.3%)	< 0.001	2 (5.1%)	14 (70.0%)	< 0.001
Day 4	14 (31.1%)	10 (22.2%)	0.34	12 (30.8%)	8 (40.0%)	0.48
Day 5	7 (15.6%)	8 (17.8%)	0.78	7 (17.9%)	8 (40.0%)	0.066

Abbreviations: CRRT, continuous renal replacement therapy; MAP, mean arterial pressure; ONSD, optic nerve sheath diameter; SOFA, sequential organ failure assessment.

FIGURE 4.

Comparison of day 1 to day 5 parameters in each randomization group depicted by paired‐T test (A–E). (A) Arterial Lactate (B) Mean arterial pressure (C) Sequential organ failure assessment (D) Ammonia levels (E) The average optic nerve‐sheath diameter F. Proportion of patients with systemic inflammatory response syndrome from day 1 to day 5 stratified by the randomization group.

3.4. Systemic Inflammatory Response Syndrome (SIRS)

The two groups were comparable at day 1 (32 [71.1%] vs. 35 [77.8%]; p = 0.47) while by day 2 and day 3 a significant reduction in SIRS was observed in the pre‐emptive CRRT group (7 [15.6%] vs. 21 [46.7%]; p = 0.001) and (2 [4.4%] vs. 15 [33.3%]; p < 0.001) while by day 4 and day 5 this was comparable between the groups (14 [31.1%] vs. 10 [22.2%]; p = 0.34) and (7 [15.6%] vs. 8 (17.8%); p = 0.78). Similar results were observed on the PP analysis (Figure 4F and Figure S2F).

3.5. Duration of Mechanical Ventilation and the Hospital Stay

The duration of mechanical ventilation (11.47 ± 5.01 vs. 9.36 ± 6.05; p = 0.091) and hospital stay (13.41 ± 11.18 vs. 13.89 ± 13.20; p = 0.86) were also not different between the pre‐emptive versus standard initiation groups, respectively. Similar results were observed in the PP analysis with comparable days of mechanical ventilation (10.91 ± 4.07 vs. 9.50 ± 5.79; p = 0.30) and hospital stay (12.00 ± 9.04 vs. 12.2 ± 13.72; p = 0.95).

3.6. Adverse Events

The adverse events were also comparable between the two groups with patients developing hypotension (1 [2.2%] vs. 1 [2.2%]), hypertension (1 [2.2%] vs. 0 [0%]), tachycardia (4 [8.8%] vs. 0 [0%]), chills (0 [0%] vs. 1 [2.2%]), pruritus (0 [0%] vs. 1 [2.2%]) in pre‐emptive versus standard initiation groups respectively. Interestingly, sepsis was the most common adverse event noted in (15 [33%] vs. 22 [50%]) in the pre‐emptive versus standard initiation groups respectively. Hypotension after CRRT initiation was observed in (1 [2.2%] vs. 0 [0%]), hypothermia (4 [8.9%] vs. 2 [4.4%]), and filter clotting (2 [4.4%] vs. 0 [0%]) in the pre‐emptive versus standard initiation groups, respectively.

3.7. Impact on Pro and Anti‐Inflammatory Cytokines, Damage‐Associated Molecular Patterns and Markers of Endothelial Function

A comparison of the different cytokines (IL‐6, IL‐10, IL‐1β) showed comparable levels between the two groups at baseline. At 24 h, interestingly, showed a significant lower level of IL‐6 in the pre‐emptive group compared to the standard initiation. The rest of the cytokines were comparable. The levels of HMGB1 and similarly the ADAMTS‐13 and vWF were comparable at baseline and at 24 h (Table S6).

4. Discussion

In the present pilot randomized controlled trial (RCT) conducted among patients with acute liver failure (ALF) and cerebral edema, all participants received plasma exchange (PLEX). A pre‐emptive approach to CRRT did not demonstrate a survival advantage over standard initiation of CRRT based on intention‐to‐treat (ITT) analysis. Mortality benefit was observed only after excluding 31 protocol violations (34% of the cohort), which primarily occurred in the standard initiation arm during per‐protocol analysis. These violations were notably disproportionate (55.6% in the standard group vs. 13.3% in the pre‐emptive group), with over 80% in the standard group attributed to worsening hyperammonemia or cerebral edema, with or without multiorgan dysfunction. This asymmetry indicates that maintaining the delayed CRRT strategy proved challenging in clinical practice. Of note, piece‐wise exponential regression analysis demonstrated pre‐emptive CRRT was associated with a reduction in early mortality during the first week. Most of these deaths were due to raised ICH. While the survival differences were not evident in ITT analysis for 28‐day mortality, early CRRT initiation combined with PLEX resulted in more rapid improvement in ammonia levels, optic nerve sheath diameter, hemodynamic parameters, and SOFA scores by day 3. Furthermore, patients in the pre‐emptive group required significantly fewer PLEX sessions according to per‐protocol analysis. Higher age, SOFA scores, and sepsis were identified as other independent predictors of 28‐day mortality in all patients, as determined by the ITT analysis. The RCT was designed as a pilot trial and enrolled more participants than initially planned, as additional patients became available. The authors assessed this as appropriate, as an additional sample size would increase precision in estimates, adjusting also for the higher number of protocol violations. The majority of patients included had a viral aetiology of ALF and had a hyperacute presentation. Almost 80% of these patients were mechanically ventilated. One‐third met the KCH criteria and were on vasopressors, suggesting a very sick cohort of ALF patients requiring extracorporeal therapies in the absence of an available liver transplant.

The current trial is the first pilot RCT exploring the benefits of an early initiation of CRRT in patients with ALF and cerebral edema alongside PLEX. In our study, the majority of patients with ALF had a viral aetiology. Cerebral edema is a significant cause of death in these patients. In fact, in viral‐related ALF, only 10%–40% of patients with cerebral edema survive with the native liver [8]. There are multiple retrospective case series suggesting the benefits of CRRT in patients with ALF, with ammonia being considered a key metabolite driving cerebral edema [2]. CRRT could effectively clear ammonia and improve outcomes in ALF patients. However, the data primarily comes from the Western world and paediatric series, wherein CRRT is used for the management of ALF of paracetamol aetiology and for inborn errors of metabolism [15, 16, 17]. Kidneys are involved in almost 40%–80% of patients with ALF. Approximately 80% of paracetamol aetiology cases have AKI, compared to only 3%–10% of patients with viral‐related ALF. Notably, protocol deviations occurred in the standard initiation arm when CRRT was initiated due to worsening ammonia levels despite ongoing PLEX, particularly in patients with significant renal dysfunction. Simultaneously, six patients in the standard initiation group did not require CRRT, achieving spontaneous recovery with PLEX alone. This underscores the effectiveness of PLEX in improving clinical outcomes. Future studies directly comparing these strategies are warranted.

In the study by the USALF Study Group with paracetamol‐associated ALF, CRRT was independently associated with improved survival [18]. Use of intermittent haemodialysis instead worsened outcomes in ALF patients. However, the timing of initiation was not studied. In a paediatric series from King's College, an early initiation of CRRT, achievement of a 50% reduction in ammonia at day 2, and a high‐volume ultrafiltration were significant determinants of improved outcomes [17]. Similar to this, a retrospective series from New Zealand and Australia demonstrated that an early CRRT, initiated within 6 h, was associated with improved outcomes [14, 15]. In our study, a pre‐emptive initiation of CRRT caused a faster reduction in ammonia. Ammonia is not only a neurotoxin but also instrumental in causing immune dysfunction and infection, causing multiorgan dysfunction in these patients [19]. Interestingly, even though not significant, a higher proportion of patients developed sepsis in the group with standard initiation of CRRT.

Ammonia could be non‐specifically elevated in critically ill patients; therefore, no specific cut‐off could be chosen alone, and we randomized patients based on CT‐evidence of cerebral edema alongside hyperammonemia [19]. In patients diagnosed with ALF, PLEX represents a superior extracorporeal therapy due to its effectiveness in removing large protein‐bound toxins that hemofiltration is unable to eliminate [20]. Additionally, PLEX facilitates the replacement of coagulation factors and demonstrates greater safety compared to other liver dialysis devices [21]. At present, PLEX is recommended as a first‐line intervention in the management of ALF [6]. Evidence supports its positive impact on survival, including efficacy for etiologies such as Wilson's disease, rodenticide poisoning, and acute fatty liver of pregnancy. A systematic review involving patients with dengue‐related ALF found that PLEX, either alone or combined with CRRT, effectively improved clinical outcomes [22]. The current study was designed to investigate whether early initiation of CRRT alongside PLEX offers benefits compared to CRRT only for PLEX non‐responders. Consequently, both groups in our trial received PLEX. The addition of CRRT on day one proved synergistic, resulting in more rapid and pronounced reductions in surrogate markers of cerebral edema—namely, ammonia levels and optic nerve sheath diameter (ONSD)—as well as improvements in SOFA scores, systemic hemodynamics, mitigation of systemic inflammation, and prevention of fatalities due to cerebral edema. However, the trial experienced multiple protocol deviations; specifically, CRRT was started earlier in patients who failed to respond to PLEX and exhibited worsening ammonia levels and non‐invasive indicators of ICH, with or without multiorgan dysfunction, while undergoing PLEX. The CRRT dosage administered in this study was lower than that reported in previous investigations [23], potentially attributable to the inclusion of PLEX and additional hemadsorption strategies for patients experiencing severe hypercytokinemia. When comparing patients with persistent hyperammonemia necessitating protocol deviations in the standard initiation arm to those without such requirements, the former group exhibited more pronounced renal dysfunction. The implementation of PLEX, coupled with a predominance of viral aetiology of ALF in our cohort contrasting with Western populations, where paracetamol‐associated cases and kidney involvement are more prevalent, may account for the reduced treatment intensity. Additionally, patients with hyperacute ALF often experience a swift recovery, necessitating the early termination of extracorporeal liver support therapies and reduction in the dose of CRRT. Given that ammonia measurements may be affected by multiple variables, the CRRT dosage was determined using multiple other indicators—including hemodynamic data, alternative markers of cerebral edema, lactate levels, and overall clinical response—rather than relying solely on serum ammonia concentrations. The CRRT was discontinued prematurely for patients being taken up for liver transplantation, or not performed for patients who had spontaneous recovery with PLEX. Interestingly, the pre‐emptive initiation of CRRT also led to a reduction in the number of PLEX sessions. This is possibly due to a sustained decrease in systemic inflammation and the clearance of middle molecules and cytokines. The small proportion of patients who had oXiris hemofilter and CRRT, showed significantly higher SOFA scores, lower PaO₂/FiO₂ ratio and higher C‐reactive protein at baseline. The field of blood purification is advancing rapidly in the management of critically ill patients; however, limited data exist in the context of ALF patients. The AN69 membrane in oXiris is more efficient in cytokine removal including endotoxin [24]. More studies are required to explore the benefits of oXiris with CRRT in patients with ALF compared to CRRT alone. In another extensive retrospective multicentric study on PLEX, a lack of survival benefit was demonstrated despite the fact that PLEX could improve hemodynamics [25]. Unfortunately, the PLEX was performed quite late in the disease course, wherein the majority of patients already met the KCH criteria and had hemodynamic instability. Further, the data on CRRT were not provided in this study. In our study, all patients underwent PLEX within the first 12–24 h after initial resuscitation and stabilization, and only one‐third of the patients either met the KCH criteria or required vasopressors. Interestingly, the adverse events, particularly the metabolic abnormalities, the development of transfusion‐associated pulmonary complications, and sepsis, are more often encountered with PLEX, and these could also be ameliorated or prevented by the addition of pre‐emptive CRRT as it also lowers the number of PLEX sessions.

The use of extracorporeal therapies has improved the management of ALF patients in the intensive care units, especially in countries where liver transplants, the ultimate saviour, are not performed in the majority of patients due to different reasons [1, 11]. In a multicentric centre in India, only 3% of patients underwent liver transplant [26]. In our centre, we have a live donor programme, and several challenges are encountered in identifying and working up the live donor [11]. The use of CRRT also has the advantage of not altering the international normalized ratio and is therefore preferable to PLEX for the management of cerebral edema. The analysis suggested clear benefits of CRRT in reduction of early deaths, which are more often driven by cerebral edema in this population. The support of the failing organs, amelioration of systemic inflammation and reduction in cerebral edema are the key benefits of the therapy, which was observed in the reduction of early deaths while these patients are waiting for a liver transplant. Interestingly, the comparison of most cytokines and markers of endothelial function and HMGB1 was not different between the two groups at 24 h, barring the significant reduction in IL‐6 levels. Possibly because both groups received PLEX, and therefore, additional benefits of CRRT on the cytokines were not observed. However, reduction in IL‐6 correlates with a rapid reduction in systemic inflammation that was seen by combining CRRT with PLEX in the pre‐emptive group.

Our study is not without limitations. The first and foremost is the single‐centre pilot RCT with a small sample size. The majority of patients had viral aetiology and hyperacute presentation, while none had paracetamol as the aetiology of ALF. Many patients had protocol violations, and therefore, the results of ITT analysis did not show the benefit of pre‐emptive CRRT initiation. In a real‐world setting, performing a randomized trial on ALF is the biggest challenge considering the rarity, heterogeneity, and rapid clinical worsening. Our centre, with a dedicated liver ICU, caters to a very large number of ALF patients. Protocol violations were allowed wherever the clinical condition of the patient changed, which was not conducive to the protocolized approach in the trial. Also, our study could not justify the selected CRRT intensity nor analysed the delivered dose versus ammonia reduction. Multiple factors could influence hyperammonemia in critically ill. High ammonia is a prognostic marker; however, despite ammonia reduction, patients may succumb to sepsis or multiorgan dysfunction. The analysis of patients who failed to show response to PLEX especially the mechanistic basis, and dose–response relationships should be explored in future studies. The data coming from even though a pilot RCT may provide robust evidence on pre‐emptive CRRT initiation in ALF patients.

In summary, the results of the current RCT showed that the pre‐emptive strategy of CRRT compared to standard initiation improves outcomes of ALF patients by reducing early deaths due to cerebral edema by causing a faster amelioration of systemic inflammation, better ammonia reduction, and improvement in systemic hemodynamics and SOFA scores.

Author Contributions

Meenu Bajpai: writing – review and editing. Neha Chauhan: data curation. Samba Siva Rao Pasupuleti: formal analysis, validation. Rajendra Prasad Mathur: writing – review and editing. Sherin Thomas: visualization. Prashant Agarwal: data curation. Mohit Prajapati: data curation. Shiv Kumar Sarin: resources. Rakhi Maiwall: conceptualization, visualization, investigation, writing – original draft. Manya Prasad: writing – review and editing.

Funding

This experimental work was supported by the Indian Council of Medical Research, 09/7882/SGP‐2023.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Relationship of CRRT dose administered with the peak ammonia concentration. CRRT dose administered was significantly associated with peak ammonia (R = 0.423; p < 0.001) that is, higher is the ammonia higher is the CRRT dose administered.

Figure S2: Comparison of day 1 to day 5 parameters in each randomization group depicted by paired‐T test (A–E) on per‐protocol analysis (n = 59) (A) Arterial Lactate (B) Mean arterial pressure (C) Sequential organ failure assessment (D) Ammonia levels (E) The average optic nerve‐sheath diameter (F) Proportion of patients with systemic inflammatory response syndrome from day 1 to day 5 stratified by the randomization group.

Figure S3: Comparison of the different parameters from baseline to day 3 (paired‐T‐test comparisons) on intention‐to‐treat analysis (A) Ammonia levels (B) optic nerve sheath diameter (C) Mean arterial pressure (D) Sequential organ failure assessment (E) Arterial lactate.

Table S1: Piecewise exponential regression model.

Table S2: Comparison of characteristics of patients with early initiation of continuous renal replacement therapy versus those who did not.

Table S3: Characteristics of relative and absolute risk reduction on intention‐to‐treat and per‐protocol analysis.

Table S4: Comparison of patients who in addition underwent OXIRIS membrane versus those who did not.

Table S5: Factors associated with 28‐day survival on Cox‐regression survival analysis‐Intention‐to‐treat and per‐protocol analysis.

Table S6: Comparison of different pro and anti‐inflammatory cytokines at day 1 and 24 h in the two randomization groups.

Maiwall R., Bajpai M., Pasupuleti S. S. R., et al., “Pre‐Emptive Initiation of Continuous Renal Replacement Therapy in Patients With Acute Liver Failure With Cerebral Edema Improves Outcomes: A Randomized Controlled Trial,” Alimentary Pharmacology & Therapeutics 64, no. 6 (2026): 800–814, 10.1111/apt.70794.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.