Continuous renal replacement therapy in acute liver failure

Abstract

Acute liver failure (ALF) is a devastating condition that primarily affects young adults. This often-lethal condition involves a rapid loss of hepatic function, that then leads to multiple organ failure. The accumulation of numerous toxins, especially ammonia, causes cerebral oedema and intracranial hypertension. Continuous renal replacement therapy (CRRT) is increasingly recognized as having a key role in ammonia removal in ALF and current evidence suggesting that timing of initiation, dose, and duration of therapy may influence survival. In addition to this important role of toxin clearance, CRRT helps with other complications of ALF such as acid-base balance, prevention of fever and management of fluid balance. As such, we propose that CRRT in ALF should be viewed as an “metabolic-toxin-fluid” therapy as much as a treatment for renal failure. In this review article we will explore the mechanisms of benefit, indications and evidence to support this concept of CRRT in ALF.

Key Words: Acute liver failure; Continuous renal replacement therapy; Hyperammonaemia; Cerebral oedema; Acute liver injury

Core Tip: The early initiation of continuous renal replacement therapy in patients admitted to the intensive care unit with acute liver failure is safe and provides a range of benefits that are likely neuroprotective. These include control of hyperammonaemia, prevention of fever, control of acidaemia, control of fluid balance and control of electrolyte derangement. Continuous renal replacement therapy can be started in all acute liver failure patients intubated for hepatic encephalopathy and should not be delayed until evidence of renal failure becomes apparent.

INTRODUCTION

Acute liver failure (ALF) is a rare condition with a high mortality and is characterized by rapid loss of liver function leading to multiple organ failure. Many definitions have been proposed over the years, but the original definition by Trey and Davidson[1] highlights the key elements of encephalopathy, coagulopathy (commonly assessed by prolongation of pro-thrombin time or the international normalized ration), and the absence of pre-existing liver disease. The absence of chronic liver disease differentiates of ALF from decompensated chronic liver disease, hepato-renal syndrome, and acute-on-chronic liver failure, all of which have different management priorities[2-4].

The causes of ALF are diverse (Table 1), with considerable regional differences in cause. In English speaking countries acetaminophen is the largest single cause while hepatotropic viruses dominate in Africa and Asia[4]. It is important that an extensive effort is made to establish the cause of ALF, as some have specific therapies (such as N-acetyl cysteine in paracetamol overdose; antiviral therapy for viral hepatitis; or silibinin for amatoxin ALF) and other potential etiologies may be contra-indications for liver transplantation (such as malignant infiltration, hemophagocytic lymphangitis, or ischemic hepatitis secondary to advanced cardiogenic shock).

Table 1 Causes of acute liver failure.

Aetiology	Examples
Metabolic	Wilsons disease, fatty liver disease of pregnancy
Viral	HAV, HBV, HDV, HEV, HSV, EBV, CMV
Drugs	Paracetamol, aspirin, niacin, MDMA
Toxins	Amanita mushroom, yellow phosphorous
Vascular	Budd-Chiari syndrome

HAV: Hepatitis A virus; HBV: Hepatitis B virus; HDV: Hepatitis D virus; HEV: Hepatitis E virus; HSV: Herpes simplex virus; EBV: Epstein-Barr virus; CMV: Cytomegalovirus; MDMA: 3,4-methylenedioxymethamphetamine.

The length of time from the onset of jaundice to the development of hepatic encephalopathy is important. All classification systems are based upon dividing ALF patients into hyperacute ALF (< 7 days), acute ALF (7-21 days), and sub-acute ALF (> 2 days), albeit with subtle variation in terminology[5]. This approach assists in identifying the cause, the presenting illness’s characteristics and severity, development of extra-hepatic failures (hyperacute > acute > sub-acute), and mortality in the absence of liver transplantation (sub-acute > acute > hyperacute)[6].

BENEFICIAL EFFECTS OF CONTINUOUS RENAL REPLACEMENT THERAPY IN ALF

ALF frequently presents with multi-organ failure and management priorities largely centre on providing high quality supportive care (Figure 1). Many of the complications in ALF have either underlying causes or consequences, that can be addressed by the timely provision of continuous renal replacement therapy (CRRT) (Table 2). The presence and impact of these challenges has seen the conceptual evolution of CRRT from providing support based upon traditional renal-centric indications to a view of CRRT providing ALF specific “metabolic-toxin-fluid balance” management.

Figure 1  Impact of acute liver failure.

Table 2 Beneficial effects of continuous renal replacement therapy.

Metabolic	Toxin	Fluid
Correction of electrolyte imbalance	Ammonia removal	Maintenance of euvolemia
Maintenance of normothermia	Potential removal of additional inflammatory mediators	-
Control of acid-base balance	-	-
Prevention of hyponatremia	-	-

INDICATIONS FOR CRRT IN ALF

Acute kidney injury

Acute kidney injury (AKI) is a frequent complication of ALF and an important indication for renal replacement therapy (RRT)[7,8]. Our understanding of the incidence and impact of AKI in ALF has matured in conjunction with the concurrent refinement of AKI diagnostic criteria, specifically the risk, injury, failure, loss of kidney function, and end-stage kidney and acute kidney injury network classifications in the wider critical care populations. The largest study to date exploring the incidence of AKI in ALF was published by Tujios et al[9] from the American Liver Failure Study Group where, over the period of 1998-2010 and using the acute kidney injury network criteria, 70% of ALF patients met any criteria for AKI, and of these 30% required renal replacement therapy. Similarly, a study from Australia with a smaller population cohort, using the Kidney Disease: Improving Global Outcomes criteria demonstrated that 70% patients met criteria for AKI. Of these, the utilization of RRT was significantly higher at over 80% and 23% of the patients who received CRRT did not meet any traditional criteria for AKI, suggesting clinicians commenced CRRT for other reasons[10].

The pathophysiology of AKI in ALF is complex and often includes direct renal injury from the same insult that caused hepatic injury. Direct nephrotoxicity has been implicated in patients with paracetamol toxicity, via processes parallel to the hepatotoxic N-acetyl-p-benzoquinone imine pathway[11], and in amanita mushroom poisoning as a direct effect of amanita toxin[12]. The secondary effect of sustained multi-organ dysfunction also plays a significant role in the development of AKI in ALF. Patients with ALF manifest a shocked state that is similar in many ways to severe sepsis and is likely to share some of the same microcirculatory and macrocirculatory pathological processes. Early hepatocyte necrosis is associated with release of pathogen-associated molecular patterns and damage-associated molecular patterns that trigger an innate immune system response via recognition by toll-like receptors and a pro-inflammatory response ensues. This disordered inflammatory response result in renal microcirculation shunting as well as direct oxidative stress and mitochondrial dysfunction in renal tubular cells that is similar to sepsis related AKI[13,14].

Other possible contributing mechanisms for ALF associated AKI impact the macrocirculation. Systemic inflammation, vasodilation and hypovolemia from commonly associated gastrointestinal disturbances, coupled with variable cardiac depression related to ALF can lead to renal hypoperfusion and associated ischemic kidney injury[8]. Whilst the literature is sparse, it appears that age, requirement for vasopressor therapy, systemic inflammation, concurrent sepsis, and paracetamol-induced ALF, are associated with the development of AKI[9].

Hyperammonaemia

Severe cerebral oedema and associated raised intracranial hypertension are a feared and potentially fatal complication of ALF. The pathophysiology of cerebral oedema involves complex interactions between toxic metabolites, direct neurotoxins, cerebral hyperemia, and systemic inflammation[15,16]. Hyperammonaemia is a major contributor to neurological injury in ALF. Ammonia mostly originates from intestinal protein metabolism and is transported via the portal vein to the liver where it is normally detoxified to water and soluble urea by the urea cycle. In the setting of significant hepatic dysfunction however, detoxification of ammonia fails and it enters the systemic circulation. The consequences of sudden and severe hyperammonaemia are most marked upon cerebral astrocytes, where the elevated ammonia concentration causes an increase in intracellular glutamine, mitochondrial dysfunction, and astrocyte swelling. Cytotoxic cerebral oedema ensues which, coupled with vasogenic oedema from inflammation mediated cerebral hyperaemia results in raised intracranial pressure that can cause tonsillar herniation and death[17,18]. Ammonia levels greater than 150 μmol/L are associated with high grade encephalopathy, cerebral oedema, and a significantly increased risk of death[17,18].

Ammonia has several properties that promote its removal by CRRT. It is a small (17 kilodalton), water soluble, non-protein bound molecule; properties attractive for clearance by standard dialysers/membranes in both intermittent hemodialysis and CRRT[19,20]. The largest evidence for the role of RRT in ammonia clearance comes from the paediatric literature, where in hyperammonaemia secondary to in-born errors of metabolism, CRRT has been shown to be effective in lowering ammonia concentration to safe levels[19,21,22]. In these studies, the clearance of ammonia was shown to range widely between 4 mL/minute/m² and 257 mL/minute/m²[22], a likely consequence of variation in the reported modality’s blood flow rate, dialysate rate and the concurrent use of non-extracorporeal ammonia lowering interventions. In paediatric critical care, CRRT for this indication has a well-established role in international guidelines and is widely recommended[21,23].

In the adult critically ill population, specifically ALF, there have been few studies looking at the clearance of ammonia. Despite the important differences between ALF and in-born errors of metabolism, the available evidence suggests a similar efficacy of ammonia clearance. Slack et al[24] looked at a mixed population of ALF, decompensated chronic liver disease and post-operative patients, who were hyperammonaemic, finding greater ammonia clearance with high intensity continuous veno-venous haemofiltration (CVVH) 90 mL/kg/hour compared with low intensity CVVH 35 mL/kg/hour. The authors concluded that ammonia clearance was related closely related to effluent flow rate, although the reported clearance rate was greater than the effluent rate, suggesting methodological challenges may have affected reported clearance values[24].

Other studies have looked at changes in ammonia levels in response to the CRRT. Warrillow et al[25] retrospectively reviewed 45 ALF patients with high grade encephalopathy and ammonia > 150 μmol/L. In this population, all patients received CRRT as either CVVHDF or CVVH, with a median effluent dose of 43 mL/kg/hour, and a median duration of CRRT of 75-hours. The authors found a significant reduction in ammonia levels over time, and that the reduction of ammonia levels was associated with the cumulative dose of CRRT rather than the intensity of therapy, (r = 0.299; P = 0.03)[25].

These findings have further supported by a large multi-center review from the America Liver Failure Study Group by Cardoso et al[26]. The authors retrospectively looked at 340 ALF patients who received RRT and had their ammonia status recorded. When they compared CRRT, intermittent dialysis, and no RRT; CRRT between the days 1-3 resulted in significant greater decrease in ammonia compared to no RRT (37.9% vs 18.6%; P < 0.007), but no significant difference was evident between intermittent and continuous modalities. However, this study is limited by the small number of CRRT patients (n = 56/340, 17%) and the absence of data regarding total duration of CRRT therapy, intensity of CRRT therapy, or modality of CRRT[26]. These papers suggest that the early aggressive initiation of CRRT is effective in facilitating ammonia clearance and may improve outcomes for patients with elevated ammonia levels and hepatic encephalopathy.

One challenge that is yet to be answered is the potential impact of the different CRRT modalities on hyperammonaemia. Whilst ammonia can be removed by all modalities it is unclear whether (CVVH - based on pressure gradient over the membrane), continuous veno-venous haemodiaylsis (CVVHD - based on concentration gradient across the membrane), or continuous veno-venous haemodiafiltration (CVVHDF - based on a combination of both modalities), offer any specific benefits. Based on ammonia’s chemical properties, it is possible that a purely diffusive technique, such as continuous veno-venous haemodiaylsis, may be more effective than other techniques using pre-dilution fluid, such as CVVH[27]. Alternatively, CVVHDF by offering combination of both may provide additional clearance of toxic small and middle-sized molecules, in addition to ammonia removal[28]. Evidence to guide decision making in ALF may be extrapolated from the studies above. Slack et al[24] used CVVH for all their patients, whereas in Warrillow et al[25] 42/54 (78%) used CVVHDF and 12/54 (22%) used CVVH, with no reported clearance difference between the modalities[24,25]. Looking outside the ALF population, Fisher et al[29] looked at differing CRRT modalities in a small population of mild hyperammonaemic patients with decompensated chronic liver disease, median ammonia 95 (interquartile range: 70-99) μmol/L. In this study, Fisher et al[29] found using a fixed 3000 mL/hour effluent rate, no significant difference in ammonia clearance between CVVH, CVVHD, CVVHDF; 22 mL/minute (19-25) 21 mL/minute (17-28) 19 mL/minute (13-25) 0.79[29]. Given the absence of robust evidence to date, the choice of modality should be based upon local factors such staff familiarity, skill-mix and patient safety, with a pragmatic decision making aiming to maximise CRRT duration by prolonging circuit life as effectively as possible.

Another area of the uncertainty in the literature is the optimal intensity of CRRT for ammonia clearance. Slack et al[24] when comparing high volume exchange (90 mL/kg/hour) and low (35 mL/kg/hour) found a significantly higher clearance rate. Warrillow et al[25] reported that despite variation in effluent dose from 2000 mL/hour (24%), 3000 mL/hour (35%), 4000 mL/hour (33%), and 5000 mL/hour (9%) for a median dose 43 mL/kg/hour, there was a trend towards significance relationship between dose and decreasing ammonia levels (Spearman’s rank correlation coefficient r_s = 0.249; P = 0.07), however this was significantly related to duration of therapy over days one and two (r_s = 0.159; P = 0.26, rather than the intensity[24,25]. The current evidence would suggest that higher doses of CRRT therapy facilitate greater ammonia clearance, but this needs to be balanced by therapy duration. In their scoping review of ammonia clearance, Naorungroj et al[22], found that in the 13 studies (11 paediatric and 2 adult) ammonia clearance was weakly associated with effluent dose r_s = 0.584; P < 0.001.

Given the importance of cerebral oedema and raised intracranial pressure in early ALF mortality, and the key role of ammonia in its pathogenesis, aggressive lowering of ammonia levels by CRRT is an attractive therapeutic option. However, given the frequency of liver transplantation as the definitive treatment option, demonstrating survival benefit from this approach is challenging. Warrillow et al[25] reported a 65% transplant free survival to hospital discharge, although there was no direct comparator group as the non-CRRT group had lower ammonia levels and lower illness severity scores. In a similar cohort, although looking specifically at patients with ammonia concentration greater than 150 mmol/L, Chaba et al[30] found that day 7 transplant free survival was associated with CRRT, using both CVVH and CVVHDF, with a median dose of 54 mL/kg/hour, (hazard ratio = 0.67, confidence interval: 0.46-0.98)[30]. Given this study is the only study looking specifically at patients with an ammonia level that correlates with a significantly increased risk of cerebral oedema, it is interesting that the rapid correction of ammonia was reported with impressive survival rates. It may be that the utility of CRRT at higher doses may be more effective in the early phase of the disease due to differing ammonia dynamics and gradient. This possibility has been identified by Warrillow et al[25] who found a greater proportional decline in ammonia levels in patients with ammonia greater 150 mmol/L, than those less 150 mmol/L, for the same effluent dose.

The largest study to date attempting to answer this question has come from Cardosa and the America Liver Failure Study Group. Across 880 patients, when corrected for year of admission, age, aetiology, and disease severity, they found that CRRT was associated with 21-day transplant free survival, with an odds ratio of 0.47 (0.26-0.82) P = 0.008[26]. Although over this time there was a significance increase in use of other supporting therapies such as N-acetylcysteine, intravenous antibiotics and a preference for CRRT over intermittent hemodialysis. Of note is the finding that intermittent haemodialysis was associated with increased mortality, despite efficacy in ammonia clearance. One possible explanation is that the significant haemodynamic and fluid fluctuations that characterise intermittent techniques can potentially precipitate cerebral oedema and elevate intra-cranial pressure, directly leading to serious neurological complications[31].

Acid-base, electrolyte, and temperature control

ALF is also frequently accompanied by profound metabolic and electrolyte imbalance. Acidosis in ALF is usually multifactorial in nature related to systemic shock, impaired hepatic lactate clearance, and kidney dysfunction[32]. The importance of these factors is highlighted by their inclusion in the King’s College Listing for Liver Transplant Criteria in both paracetamol and non-paracetamol ALF. Hyponatraemia is known to be detrimental in the neurosurgical/traumatic brain injured population where it is associated with poor outcomes[33]; and should be carefully avoided in ALF patients. Further adding to the metabolic chaos characterizing ALF is the impact of temperature and positive fluid balance on cerebral oedema and raised intracranial pressure. Whilst a randomized multicenter trial looking at prophylactic hypothermia (33 degrees vs 36 degrees) failed to demonstrate a survival benefit, extrapolation from other neurological populations suggests that the avoidance of fever may be beneficial with patients at risk of cerebral oedema[34]. Additionally, a positive fluid balance has been shown to be associated with poorer outcomes in both the neurocritical ill, and the wider critically ill multi-organ failure populations; both circumstances that are applicable to patients with ALF[35,36]. CRRT has the potential to alleviate many of these potential complications in ALF patients. The use of lactate-free buffers containing 30-32 mmol/L of bicarbonate can rapidly restore acid-base homeostasis and reverse negative impact of profound acidosis of systemic shock states. CRRT can provide sodium balance, either in conjunction with the administration of hypertonic saline or independently[37], and turning off the blood warmer provides a simple method to maintain normothermia. Fluid is balance is best controlled continuously via CRRT in the setting of potential fluid shifts and evolving multi-organ dysfunction. The combination of these additional benefits of CRRT further suggest that the early aggressive institution of CRRT in ALF patients should occur independently of the traditional triggers for CRRT in critically ill patients.

CONCLUSION

Over the last decade, the evidence supporting the early role of CRRT in ALF has evolved significantly. Cumulatively, this evidence suggests that CRRT should be viewed as a “metabolic-toxin-fluid” intervention, with early initiation aiming to minimize the impact of hepatic dysfunction by ammonia level control, acid-base and electrolyte abnormalities, maintenance of normothermia, and strict regulation of fluid balance. As such, CRRT has a unique and specific role in ALF that extends considerably beyond the traditional “renal” indications where the goals are fluid management and solute clearance. Further research is required to determine the optimal CRRT dose, duration and modality in its use in ALF.

ACKNOWLEDGEMENTS

We gratefully acknowledge the past support and inspiration of our great friend and mentor, Professor Rinaldo Bellomo.