Hepatic encephalopathy

This is an altered level of consciousness attributed to the consequences of acute or chronic liver failure. The candidates have seen it a number of times, in lots of different permutations:

  • Question 30 from the second paper of 2023
  • Question 6 from the first paper of 2022
  • Question 10 from the second paper of 2017 (probably the best for revision, as it contains a very detailed college answer)
  • Question 14 from the first paper of 2003
  • Question 1c from the second paper of 2001

Pathogenesis of hepatic encephalopathy

A free access article about hepatic encephalopathy from 2012 is available, detailing the pathological processes which lead to this disturbance of consciousness. Ammonia is only one of the aetiological agents. But, it is certainly the one everyone always thinks of. Copious amounts of blood are being sent for ammonia levels every day. Surely, there must be some reason behind this.

In order to derive some meaning from this seemingly mindless ammonia-lust, one must explore the mechanisms of metabolic derangement which arise within the liver failure patient.

Elevated serum ammonia levels in hepatic failure

The urea cycle is broken, and copious quantities of ammonia are released into the circulation. An excess of ammonia occurs when the hepatocytes are too few, and when their access to portal venous ammonia is impaired.

Increased ammonia causes cerebral oedema

The naive brain of a person suddenly overcome by fulminant hepatic failure will be unprepared for the onslaught of ammonia. Massive amounts of it will be delivered to the astrocytes, and for lack of another coping mechanism they will try to process it into glutamine, stockpiling this glutamine inside their cell bodies. This rapidly becomes counterproductive.

mechanism of increased intracranial pressure in hepatic encephalopathy

The excess stockpile of intracellular glutamine causes a massive osmotic drag, and astrocytes begin to swell. We dont biopsy enough human brains to know by how much they swell, but certainly rat brains marinaded in ammonia revealed astrocytes that were up to 127% larger than they should be. This is bad. They are the most numerous member of the cerebral parenchyma, and contribute significantly to the total mass of brain tissue; naturally any additional mass and volume will result in worsening intracranial pressure. Not only that, but by swelling (and thus occupying more space) the astrocytes increase the distance between capillaries and neurons, with a resulting nutrient diffusion defect. Thus, in a number of ways, the neurons of an acutely encephalopathic patient are screwed.

The influence of hyperammonaemia on blood-brain barrier function

The review article I have linked to reviews the influence of ammonia on blood-brain barrier function, and identifies several functional derangements which are thought to be important in the pathogenesis of hepatic encephalopathy. Firstly, the effect of ammonia on the concentration of intracellular amino acids confuses the mechanisms of amino acid transport, which in turn alters the substrate available for neurotransmitter synthesis - but more on this later. 

The influence of ammonia on neurotransmitter systems

The presence of ammonia in cerebral blood influences the function of transport proteins of the blood-brain barrier. This is largely due to the fact that the detoxification of ammonia produces excess glutamine. In this way, ammonia increases the transport of the large neutral amino acids and aromatic amino acids which rely on the exchange with glutamine.

The transport of these amino acids impairs the inward transport of other amino acids, as they are in competition for the same transport proteins, but lack the boost of an excess countertransport exchange substrate. Thus, "correct" amino acids are not being transported into neurons, and "incorrect" ones are transported in excess. This results in the impairment of neurotransmitter synthesis (specifically, serotonin and dopamine).

The excess of "incorrect" neurotransmitter substrates results in the excess synthesis of "false neurotransmitters" (octopamine and phenylethylamine). These apostates are either inactive or partially active, and wreak havoc on the synapses. Phenylethylamine is similar to amphetamine in its action; octopamine is the normal insect equivalent of noradrenaline, and exerts some degree of excitatory activity by defrauding the human noradrenaline receptors. Needless to say, secreting insect neurotransmitters in your brain is far from ideal for normal human cognitive function.

The influence of ammonia on cerebral metabolic fuel supply

Ammonia affects the transport of glucose and creatine. Animal models of hepatic encephalopathy demonstrate a deactivation of cerebral oxidative glucose metabolism, and activation of cerebral anaerobic glycolysis.

Ammonia seems to inhibit the rate-limiting step of Kreb's cycle, α-ketoglutarate dehydrogenase, which increases the availability of α-ketoglutarate for ammonia detoxification (it can also be used to bind ammonia and form glutamine). This alone decreases the movement of glucose through oxidative phosphorylation. On top of that, ammonia inhibits pyruvate dehydrogenase.

interference of ammonia in cerebral fuel metabolism

With these rudiments of energy metabolism ragged and broken, the brain begins to produce lactate even in the presence of abundant oxygen. However, one can see how this problem can be overcome. One needs to supplement substrates for Krebs' cycle which bypass the inhibited α-ketoglutarate dehydrogenase, and suppy fuel to the rest of the cycle. Branched-chain amino acids such as isoleucine and valine are precisely these sorts of substrates, and the use of these in the nutrition of hepatic encephalopathy patients has been investigated. More on this later.

Diagnosis of hepatic encephalopathy

A good review article is available. It confirms that in fact there is no way you can completely confirm hepatic encephalopathy, but by excluding all the other potential causes of a decreased level of consciousness you can arrive at the conclusion that the most likely diagnosis is hepatic encephalopathy.

The best way to develop a firm impression that your diagnosis is correct is to apply the criteria of clinical trials. A recent round table discussion has attempted to standardise the diagnostic and classification criteria for hepatic encephalopathy, with the aim of simplifying the enrolment into clinical trials, and the cross-comparison of such trials. It builds on an older definition of this condition, which was an international position agreed upon in 1998, and which to this day remains largely unopposed.

The best definition they could agree on was "a spectrum of neuropsychiatric abnormalities seen in patients with liver dysfunction after exclusion of other known brain disease". They then go on to list "stereotypical" features which are associated with hepatic encephalopathy, but which are not diagnostic of it.

  • Asterixis
  • Dysarthria
  • Disordered sleep-wake cycle
  • Long white matter tract signs
  • Decreased level of consciousness, delirium

Ideally, one should also have - at minimum - some sort of liver disease.

The criteria to qualify for hepatic encephalopathy include:

  • Associated severe liver disease
  • The presence of a triggering factor
  • A previous history of similar episodes

Having excluded the other potential causes for these abovementioned signs and symptoms, one should respond to a trial of empiric treatment as a means of confirming hepatic encephalopathy.

Neurological findings in hepatic encephalopathy

The abnormalities in neurological signs are usually bilateral and non-localising, but this is not unifirm, and these patients may present a whole plethora of confusing (and irritatingly inconsistent) neurological examination findings. Attempts to integrate physical signs into a grading system date back to 1979 when Conn and Lieberthal published a great opus on hepatic coma; their contribution persists into the modern era among works such as this 2012 article from Neurochemistry International.

Conn and Lieberthal mentioned hyperreflexia, clonus, extensor posturing and abnormal Babinski reflexes as features of advanced hepatic coma; their original book is not available online, but fortunately other authors have made their suggested grading system available in free online publications (Table 2: Clinical manifestations and severity of hepatic encephalopathy, p.797).

For simplicity of revision, some of this grading scale is reproduced below.

Physical Signs in Hepatic Encephalopathy

Stage I

  • Metabolic tremor
  • Muscular incoordination
  • Impaired handwriting

Stage II

  • Slurred speech
  • Asterixis
  • Hypoactive reflexes
  • Ataxia

Stage III

  • Hyperactive reflexes
  • Nystagmus
  • Babinski’s sign
  • Clonus
  • Rigidity

Stage IV

  • Dilated pupils
  • Opisthotonus
  • Coma

Question 10 from the second paper of 2017 and Question 6 from the first paper of 2022 asked specifically for clinical signs of hepatic encephalopathy, which this table is well suited to answering in spite of the vintage of its publication.

Older classifications

The old Parsons-Smith Scale of Hepatic Encephalopathy (the 1957 version heavily modified by Conn) still makes mention of some reproducible clinical features, but some authors have added Glasgow Coma scores to the mix, which might attract criticism (given that the GCS is not exactly meant for grading of metabolic encephalopathy).

Grade 0 Sublinical- subtle deficits only recognisable with neuropsychometric testing
Grade 1 Shortened atention span, trivial lack of awareness, tremor, incoordination, apraxia
Grade 2 Lethargy, disorientation, asterixis, ataxia, dysarthria. GCS 11-14
Grade 3 Confusion, somnolence, astrixis and ataxia. GCS 8-11
Grade 4 Coma, decerebrate posturing.

Newer classifications

The more recent West Haven Criteria are all but emptied of hard physical signs:

Grade 1 Lack of awareness, short attention span
...difficulty with addition
Grade 2 Lethargy, inappropriate behaviour, mild disorientation
...difficulty with subtraction
Grade 3 Somnolence responsive to verbal stimuli; disorientation, confusion
Grade 4 Coma

Clinical features

These days, there does not seem to be any great attachment to physical examination features among the eminent experts; an increase in deep tendon reflexes is occasionally mentioned, but without great enthusiasm. A French group assessed a prospective case series (2000) and found that "hepatic encephalopathy with focal neurological signs when carefully searched is not uncommon... is reversible, and has no prognostic significance." Subtle signs are also suffering from a loss of professional interest in this era of rapidly available serum rhubarb levels. In 2007, Ortiz et al made an attempt to integrate functional clinical features such as literacy and numeracy into a comprehensive nine-domain grading system, again without evoking much support from the community.

Objective investigations

With such confusion over the clinical features, of course various psychometric and serological tests have been invented to help with this difficult diagnosis. Prakash and Mullen published in Nature in 2010, and their review article in it contains a nice table of the various tests, as well as their advantages and disadvantages. A similar effort is presented by Nabi and Bajaj (2014), with a focus on psychometric testing and neurophysiological diagostic methods (such as EEG). A 2006 review article of the diagnostic methods in hepatic encephalopathy also digresses upon neuroimaging and assessment of neurometabolism.

EEG is particularly interesting, and characteristic features have been described, such as the slowing of posterior alpha waves, followed by a gradual appearance of theta and high-amplitude irregular delta waves as it progresses into coma. However the visual interpretation of EEG loses objectivity to inter-observer variability. Apparently, quantified (digitally analysed) EEG in cirrhosis is of more value, as it can predict overt hepatic encephalopathy and seems generally well correlated with the severity of cirrhosis.


Why else might a middle-aged alcoholic have become progressively unconscious, asks Question 6 from the first paper of 2022? The college examiners asked for four alternative diagnoses. The possibilities are as endless as are the unlisted comorbidities of the aforementioned alcoholic. Potential mimics of hepatic encephalopathy could include:

  • Vascular causes, eg. stroke
  • Infectious causes, eg. sepsis or even a CNS infection
  • Neoplasm of the CNS, which was previously unknown
  • Drug-related decreased level of consciousness, from intoxication of some sort (including iatrogenic, like the overzealous administration of diazepam for alcohol withdrawal) to a drug withdrawal syndrome (eg. a non-convulsive status epilepticus)
  • Something nonmedical, like a psychiatric syndrome involving catatonia
  • Trauma, eg. subdural haematoma from an unreported fall
  • Endocrine causes, eg. hypoglycaemia due to liver disease, hypothyroidism or hypoadrenalism, hyponatremia, 

Hepatic encephalopathy and raised intracranial pressure

Generally speaking, encephalopathy of the chronic liver failure patient does not tend to develop the raised ICP, partly because of alcohol-related degeneration (which leaves a large amount of CSF space unoccupied) and partly because of the chronic process of adaptation (as the astrocytes reabsorb many of the proteins which act as idiogenic osmoles to make room for osmotically active hepatic toxins and glutamine).

So, when do you treat the hepatic encephalopathy patient for increased intracranial pressure, and how? The suggestions have been derived from neurosurgical literature, and are almost entirely analogous to the guidelines for the management of traumatic brain injury. There are several differences in the approach, which are as follows:

  • Grade 3-4 by the Parsons-Smith scale deserves an ICP monitor, IF:
    • the patient is young
    • the presentation is "hyperacute"
    • the serum ammonia is over 150mmol/L (anything over 200 is associated with cerebellar herniation)
  • ICP monitoring should be performed with a parenchymal device, such as a Codman catheter (as the risk of bleeding with these is much less than with an EVD). Some authors recommend covering the coagulopathy with some Factor VIIa just before inserting the monitor.

Differential diagnosis of hepatic encephalopathy

The college love a collapsed drunk. Specifically, they often repeat the same question where an alcoholic has escaped from a rehab facility, and is found by the police three days later in a state of coma, with deranged LFTs, a bilirubin of 500 and a high ammonia. "Why is he unconscious", they ponder. This question has been repeated at least three times:

Thus, the following is a stereotypical list of causes of unconsciousness in a comatose alcoholic. One might wish to use the VINDICATE acronym to organise the differentials.

  • V- Stroke
  • I- Sepsis
  • N- Seizures; hepatic encephalopathy
  • D- Drug intoxication (including alcohol)
  • I -
  • C-
  • A-
  • T- Intracerebral bleed, GI bleed
  • E- hyponatremia, hypoglycaemia


J.S. Bajaj (2010) offers a detailed exploration of the modern management of hepatic encephalopathy. Another excellent review is from Riggio et al (2015). From these, the following approach has been concocted, to easily recall and regurgitate in an exam setting:

Specific management of hepatic encephalopathy

  • Lactulose, or lactose if they are lactose-intolerant
  • Rifaximin
  • Avoidance of hyponatremia
  • Nutritional management:
    • Branched-chain amino acids (BCAAs) and a reduced amount of aromatic amino acids
    • High fiber diet
    • Pro-biotics (though their benefit is unclear)

Management of the precipitating cause

  • Stop GI bleeding (endoscopy, banding, etc)
  • Antibiotics for SBP
  • Correct dehydration
  • Withdraw hepatotoxins

Supportive management of the encephalopathic patient

  1. Support the airway.  
  2. Wean ventilation to spontaneous mode as tolerated.
    Avoid NIV; abdominal distension and a fluctuating level of consciousness will likely result in aspiration. HFNP is ok.
  3. Support haemodynamically;
    noradrenaline +/- terlipressin may be appropriate if hepatorenal syndrome is suspected
    Albumin (20%) is a reasonable resuscitation fluid
  4. Avoid sedation. As needed, use drugs which do not depend on hepatic metabolism (eg. remifentanyl)
  5. Correct electrolyte derangement
  6. Monitor renal function (hepatorenal syndrome)
  7. Ensure BSL is monitored and supplemental glucose is made available
    Ensure thiamine is co-administered with glucose!
    Optimise nutrition (35-40cal/kg/day)
  8. Correct clinically significant anaemia. 
    Address haematinic factor deficiencies.
  9. Antibiotics as appropriate: ceftriaxone may be required if SBP is a real possibility.

In somewhat greater detail, the main points of management are listed below. Rather than trying to group them by purpose (as many agents serve multiple purposes) the list is organised by agent.


In the college answer to Question 10 from the second paper of 2017, they refer to  "non-absorbable disaccharidases" as if they were some sort of enzymes. Though it would be unfair for this severely typo-prone author to be overly critical of the college's minor errors, one needs to point out that lactulose is in fact a non-absorbable disaccharide. In fact, lactulose is the pleb name of 4-O-β-D-galactopyranosyl-D-fructose, an isomerisation product of lactose that has the same molecular formula (C12H22O11). In summary, its positive effects are:

  • Cathartic (increases gut transit time)
  • Decrease ammonia production by changing colonic pH
  • Increase ionisation of ammonia to ammonium (to some minimal degree)
  • Changing the proportion of urease-deficient organisms in the gut (to some minimal degree)

Anyone even remotely interested in lactulose should read Christian Schumann's homage to this substance from 2002; the molecular structure included here is from his article. lactulose molecular structureIt demonstrates how very similar lactose and lactulose are structurally. However, because of its β-1-4-glycosidic bond, lactulose cannot be spit by human gut enzymes , whereas most people are able to metabolise lactose. In fact, taking lactulose is a good way for a normal perosn to test out what it would be like to be lactose-intolerant. Conversely, a lactose-intolerant patient with hepatic encephalopathy could be treated successfully with dairy products (according to the case report by Welsh et al, 1974). 

Cathartic effect: Lactulose remains in the gut lumen, where it has an osmotic laxative effect. Beyond 95-100g of lactulose per day, the metabolic capacity of gut organisms is saturated and any additional lactulose is recovered unchanged in the effluent. The physical evacuation of ammonia-contaning bowel contents must play some role, though it is pretty clear from studies like the polyethylene glycol trials discussed below that the cathartic mechanism of lactulose is probably not responsible for all of its positive effects.

Change of colonic pH: The college answer to Question 10 from the second paper of 2017 mentions that lactulose reduces colonic pH, without really discussing what effect that has. In fact this causes gut bacteria to metabolise fewer amino acids and focus more on carbohydrates. The change in pH occurs because gut organisms tend to metabolise lactulose into lactic acid and acetic acid. Acetate levels in portal blood increase impressively when lactulose is administered.

Ion trapping of ammonium: the college mentions that the low pH facilitates the formation of the nonabsorbable NH+4 from NH3, trapping NH3 in the colon. This is indeed true, but the amount of ammonium recovered from stool increases by such a small margin (1%- Weber et al, 1996) that this cannot possibly be a dominant mechanism. The total stool nitrogen however is massively increased, suggesting that the ammonium (or ammonia) is incorporated into bacteria. 

Inhibition of ammonia synthesis: Most of its therapeutic effects come from the effect on ammonia production  by colonic organisms (which is a major source of ammonia in the human body). The main culprits are Clostridium, Enterobacter and Bacteroides. They are called urea-spiltters or urease-producing organisms, which strictly speaking they are, except this has minimal bearing on their behaviour in the stool where there is minimal urea. Anyway, they do produce ammonia. Giving these organisms a carbohydrate like lactulose tends to railroad their metabolism down a non-urea-producing pathway. Lactulose is a favoured substrate, whereas your dietary amino acids are not .Vince et al (1978) were able to demonstrate this in vitro, using some sort of horrific faecal incubation system, using "freshly passed faeces from three healthy subjects". Their conclusions:

"Presence of a fermentable carbohydrate, lactulose, providing readily available carbon and energy, could reduce the ammonia concentration in the colonic environment in at least two ways: (1) preferential use of lactulose as a carbon and energy source would exert a sparing effect on the metabolism of both exogenous and endogenous aminated compounds with a consequent decrease in the amounts of ammonia liberated as a by-product during this process, and (2) a readily available energy source would presumably encourage assimilation of ammonia, an energy-requiring process which may not occur so readily under conditions of carbon limitation."

This is a good way to describe what happens. The consequences of metabolising much carbohydrate results in the release of much methane, giving rise to "meteorism" and uncontrollable flatulence which Schumann described as a major barrier to the use of lactulose as a commercial food additive. 

Increased ammonia clearance: Lactulose also appears to improve the incorporation of nitrogen into the expanded bacterial mass. Weber et al (1996) reported that the stool nitrogen increased (doubled) with ~ 50g/day of lactulose, while the stool ammonia remained the same (i.e. the bacteria had incorporated the nitrogen into themselves somehow, rather than metabolising stuff to cause its release). Overall, eliminating nitrogen rather than absorbing it as ammonia is a good move for the encephalopathic organism.

Change in the gut microbiome was initially thought to be behind its beneficial effects. It was thought that the abundance of substrate would favour the reproduction of such lactulose-loving species as Lactobacillus, organisms which lack urease. In the presence of much lactulose, it was thought these organisms would reproduce abundantly and outnumber the nasty urea-splitters. However that was never really demonstrated at any level. Elkington (1970) reports that in cirrhotic patients the hepatic encephalopathy improved some weeks before the change in bowel flora. Moreover, there were some patients who had a significant clinical improvement, who had minimal lactobacilli in their stools. Obviously this hypothesis is a dud.

"Non-absorbable disaccharides" in general

Everybody knows and recognises lactulose. But many authors (Gluud et al, 2016) refer to this therapeutic approach by calling the entire class name. Are there any other non-absorbable disaccharides? Turns out, of the many disaccharides (sucrose, maltose, lactose) many are in fact absorbable. When one mentions non-absorbability, one ends up finding that lactulose is the only actual disaccharide used for hepatic encephalopathy. The other chemical which keeps coming up is lactitol, which is in fact not a dissacharide but a sugar alcohol like mannitol, sorbitol, xylitol etc. What good is it, one might ask? Particularly as the hospital formulary does not stock it, and often to get hold of it one might need to contact a wholesale supplier of reduced calorie baked goods? Well. It appears that interest in its use has waned since about the 1980s, which is unfortunate because Morgan et al (1987) ran it against lactulose in a clinical trial with encouraging results. Both drugs had about the same effect on hepatic encephalopathy but the lactitol group became sane more quickly.

Speaking of clinical trials. What evidence is there to support the use of any of these hideous substances on humans? Gluud et al (2016) identified 34 RCTs which fit their stringent criteria for meta-analysis, with a total n=1828. Mainly lactulose was studied. In short, lactulose seems to have a beneficial effect on mortality, and hepatic encephalopathy as well as on the risk of serious adverse events of hepatic failure such as hepatorenal syndrome and variceal bleeding.

Rifaximin and other antibiotics

Antibiotics for the management of hepatic encephalopathy have been trialled ever since it became apparent that bacterial activity had something to do with the neurological deterioration. Dawson et al (1957) were probably the first to try it; they gave 4-10g of neomycin to patients with cirrhosis. rifaximinNeomycin is an aminoglycoside which has been available since 1949. It hit gram negatives and a few gram positives. Unfortunately, people were discouraged by its nightmarish toxicity

Apart from neomycin, people have used paromomycin, vancomycin and metronidazole until finally settling on rifaxamin. Rifaxamin is a semisynthetic analogue of rifampicin; Norgine chemists added a benzimidazole ring making its absorption impossible. It is therefore only ~ 0.4% absorbed, and this proprotion remains stable even in inflammatory gastrointestinal disease states. Its concentration in the stool is so high that it is inevitably well above MIC for most organisms, and because of this high-caliber killing power it does not appear prone to drug resistance. Weirdly, it does not seem to harm many of the "benign" gastrointestinal flora like Bifidbacterium  and Lactobacillus (Adachi et al, 2006).

Evidence for rifaximin  is well-summarised by Kimer et al (2017) in a meta-analysis powered by an unrestricted grant from Norgine. A total of 21 RCTs involving 2258 patients were included. In summary, it appears to improve the quality of life but it does not seem to prevent episodes of hepatic encephalopathy. When used to treat acute exacerbations, its effect is greatest in the population of most encephalopathic patients (i.e. it's useless if you are only mildly confused), but it does seem to have a significant positive effect - for example, Wang et al (2019) found that the combination of rifaximin and lactulose (as opposed to lactulose alone)  reduced mortality (23.8% vs. 49.1%) and hospital stay (5.8 days vs. 8.2 days).


Strictly speaking if you're going to call lactulose "non-absorbable disaccharides", then you'd have to call probiotics "gut flora modulation", as the therapeutic approach is not limited to probiotics but also include prebiotics and synbiotics. We should be grateful to Shukla et al (2008) for unraveling the yarn of nomenclature. Generally speaking, gut flora modulation therapy consists of trying to outnumber enteric gram-negatives with an overwhelming exogenous influx of some microbes which you feel are beneficial (eg. Lactobacillus sp., Bifidobacterium sp., or yeasts such as Saccharomyces boulardii).russian kefir These immigrants then reproduce out of control in the presence of lactulose, and hopefully promote a gut environment where all ammonia is incorporated into bacteria and is cleared cathartically.

Realistically, this theory falls apart on a number of levels. Already mentioned is the lack of evidence that changing the gut microbiome has any influence on hepatic encephalopathy (again, a change in the metabolism among existing gut organisms is probably the beneficial effect seen with lactulose). The empirical evidence for any positive effect is also fairly ephemeral. A meta-analysis by Dalal et al (2017) identified 21 trials with 1420 participants, of which the majority "suffered from a high risk of systematic error (‘bias’) and a high risk of random error (‘play of chance’)". There was no effect on mortality when compared to placebo or even to no treatment, which is a serious allegation. Literally, doing nothing was as good as giving probiotics. The meta-analysis authors reluctantly concluded that probiotics probably improve recovery from hepatic encephalopathy as compared to placebo, but that they probably are not as good as lactulose.  One positive thing that can be said about them is that they are quite safe (they are, after all, just buttermilk and yoghurt). No authors so far have report any case of fatal Lactobacillus sepsis in a cirrhosis patient.

Other cathartics

If the cathartic effect of lactulose is the dominant mechanism behind its efficacy, one argues, then surely one should be able to achieve this by any number of other means. Facetiously, one might even make the argument for reclassifying Salmonella and rotavirus as probiotics. Less stupid suggestions might include such old favourites as polyethylene glycol.  

Small trials, eg. the HELP trial run by Rahimi et al in 2014 (with 25 patients in each arm) seem to favour PEG as a therapy when compared to lactulose. There was a small difference, but the PEG group seemed to recover their wits a little faster. The magnitude of difference is surprising, because the lactulose therapy (20-30g every 8 hours) was compared to a massive 4000ml of polyethylene glycol solution via NG. In other words, whole bowel lavage was performed. It is remarkable that there was so little comparative improvement in that group, considering that their effluent was probably clear at the end. In other words, their gut microbiome would have consisted of a tiny frightened population of residual colonic organisms, hiding deep in some diverticulae, trying not to call attention to themselves. If anything, these results suggest that catharsis is probably one of the minor effects of lactulose.


Flumazenil stashTo use of flumazenil in hepatic encephalopathy might seem like a weird thing to do, considering that the causes of decreased level of consciousness in HE are certainly multifactorial, and unlikely to be related to reversible GABA-antagonist activity of the hepatic toxins. However, some have suggested that the coma is at least to some extent mediated by unregulated out-of-control GABA activity, and that it might be laudable to somehow undermine this process with a competitive antagonist. Flumazenil binds the receptor in a manner which prevents benzodiazepine molecules from binding, but appears to have few intrinsic properties of its own. In high doses it appears to act as a weak partial agonist.  Neave et al (2000) were able to demonstrate a sedative effect in healthy volunteers with doses of around 5mg (for some reason the findings were published in the British Dental Journal).

Having said these negative things, it must be mentioned that the impressively detailed college answer to Question 10 from the second paper of 2017 regards flumazenil as a useful second line agent ("can result in clinical improvement but no mortality benefit", they said). That is certainly not an inaccurate statement. Grimm et al (1988) and Bansky et al (1989) found that hepatic encephalopathy patients responded nicely to flumazenil, and those who did not had objective features of cerebral oedema. Of the patients who were going to respond, most did so within minutes or even seconds of administration. The Grimm team actually gave fairly large doses, starting with 2mg and going up to a whopping 15mg as an infusion over several hours, which (doing a quick pharmacy survey) would have depleted the entire ward stock of a major tertiary ICU.  Fortyunately, Bansky's people found that even doses of 0.2mg produced some sort of clinically interesting effect (eg. whereas before you were too confused to do a number-matching test, following 0.2mg of flumazenil you had a good chance of getting full marks).

In summary, flumazenil seemed pretty promising in the 1980s. In 2002, Goulenok et al were able to perform a meta-analysis including 6 trials with a total of 641 patients, demonstrating a substantial superiority over placebo (all the trial patients were also receiving some mixture of lactulose and rifaximin). Goh et al (2017) repeated a very similar literature search and were able to scrape together only 842 patients, suggesting that this branch of research remained undersubscribed in the fifteen years that had passed. Of these trials, all demonstrated a short-term benefit in terms of an improvement in the level of consciousness but nothing in terms of long-term survival. 

Ammonia scavengers

Given how fixated everybody is on ammonia in hepatic encephalopathy, it is all the more surprising that ammonia-directed therapy is not employed more often. In other, more clearly ammonia-centric diseases (eg. genetic urea cycle diseases), ammonia scavengers are either a faithful mainstay of therapy or at least some sort of reliable rescue remedy. An excellent review of the state-of-the-art ammonia scavenger therapy by Las Heras et al (2017) reveals a vast and exciting range of such drugs. In order to simplify revision, a table is offered here. None of these are exactly an "ammonia scavenger" as such, in that the molecule of sodium benzoate does not bind ammonia and chelate it like one might expect a scavenger to do. Instead, these substances promote the incorporation of ammonia nitrogen into some sort of "readily excretable non-urea solute" which is then cleared by the kidneys instead. The caveat of course is that you need reasonably functional kidneys for this to work.

Ammonia Scavenger Therapies
Drug name Usual dose Expected side effects and limitations

Sodium benzoate

5g bd
  • Hypernatremia
  • Gastritis and reflux
  • Allergy-like reaction with rash

Sodium phenylacetate

5g/m2 q24 hrs
  • Anaemia
  • Vomiting
  • Hyperglycaemia
  • Hypokalemia
  • Neurotoxicity (confusion)

Sodium phenylbutyrate

5g bd
  • Hypernatremia
  • Amenorrhoea
  • Decreased appetite
  • Foul body odour
  • Rarely, aplastic anaemia

Sodium benzoate is a fairly ubiquitous pharmaceutic excipient, included as a preservative. It is present in the diet of most normal people at least to some small extent; doses in excess of 50g/day have been taken with minimal adverse effect, and the LD50 in rodents is almost 2g/kg of body weight. It is thought to activate a non-urea-cycle pathway for amino acid metabolism (Misel, 2013). To be more specific, benzoate is conjugated with coenzyme A to form benzoyl CoA, which then conjugates with glycine to form hippurate (hippiric acid is N-benzoylglycine) which is then excreted in the urine. As a means of getting rid of amino acid molecules, this is a damn good way. Las Heras et al report one study result where 60% of total nitrogen excretion was in the form of hippurate.

This substance was used by Sushma et al (1992) to impressive effect. The trial randomised patients to receive either lactulose or sodium benzoate, and found that they were equivalent in terms of controlling the psychomotor symptoms of encephalopathy. The more interesting finding was that sodium benzoate therapy was thirty times cheaper than a course of lactulose, amazing given the fact that 500mls of lactulose costs $9.95 at the local chemist. That bottle would last you about a week, however. Oral sodium benzoate is available for around $6.00 from Amazon, for 450g which would last 45 days. If cost were an issue, this substance is certainly a winner. 

Sodium phenylacetate and L-ornithine phenylacetate are an analogous alternatives which bring forth phenylacetate to bind coenzyme A and be conjugated to glutamine. The resulting "excretable non-urea metabolite" is PAGN, phenylacylglutamine. These substances are usually administered as IV preparations, and usually together with sodium benzoate (eg. as Ammonul, a commercially available 10% solution). Dosing recommendations are derived largely from literature regarding urea cycle disorders, and according to the Ammonul monograph people usually dose it to square metres of body surface area. 

Sodium phenylbutyrate and glycerol phenylbutyrate are metabolised in a similar manner, and produce byproducts such as phenylbutyrylglutamine (PBGN) which is also readily excreted. These are orally bioavailable; the FDA has approved 500mg Buphenyl tablets (as with sodium benzoate, the dose is 5g twice a day). 

Zinc supplements

Apparently, zinc is an important cofactor in the urea cycle enzyme pathway. Specifically, ornithine transcarbamylase and glutamine synthase apparently require zinc to function, and these liver and muscle enzymes are required for normal urea synthesis.  Zinc deficiency is usually seen in alcoholic cirrhosis patients, presumably because beer wine and spirits contain little zinc, and because their diet contains little else. It stands to reason that one should want to replace zinc in these patients. Because of its generally low toxicity (you'd need to eat approximately 80-400mg per kg to have serious adverse effects), it also stands to reason that one should want to replace it in non-alcoholic patients, because it will probably do no harm and maybe some good.

Chavez-Tapia et al (2013) were able to identify 4 trials which were worthy of their meta-analysis. Of these 233 patients, the majority received an oral supplement of about 600mg of zinc per day. Three of the trials found a significant improvement in performance on the number connection test, but little in the way of mortality benefits, quality of life, prevention of encephalopathy or relief of an acute episode. The only merit of this therapy appears to be its low cost and minimal toxicity.

Shunt reversal

If you have a portosystemic shunt, it is likely making things worse. If this is in the wake of a  TIPS procedure, it's clear that there is a shunt, and sensible advice would be to downsize or occlude the TIPS. A normal TIPS should have a flow rate in the order of 2-3L/min (according to ultrasonographic data by Pinter et al, 2015); it may be "downsized" to a lower flow rate by re-stenting with a smaller diameter stent.

But what if you haven't had a TIPS? The college mention that medically refractory hepatic encephalopathy should raise suspicion of a spontaneous splenorenal shunt. Apart from treatment failure, what would give you the impression that this has formed? Ohnishi et al (1986) review this topic. Apparently, the main clue is the size of their oesophageal or gastric varices: if they are small and few, then a shunt must be present (otherwise, you'd expect them to be huge). 

Nutrition for hepatic encephalopathy

So, the encephalopathic person has a high ammonia and is giving you every reason to form an impression of them as a poor metaboliser of protein. This is in fact true (they suck at protein). Previous strategies (eg. Schwartz et al, 1954) had made attempts to limit protein intake in such patients with the expectation that a decreased access to exogenous amino acids should give rise to an improvement in the ammonia load and therefore a clinical benefit. Not so! The encephalopathic patient, robbed of dietary protein or calorie restricted, will obstinately generate ammonia by dissolving structural body protein. The way to prevent this act of defiance is the administration of sufficient dietary protein and carbohydrate. Merli et al (2012) reviewed the guidelines and agreed on the following recommendations:

  • Larger than average caloric goals (CICM answer to Question 10 suggests 35-40 kcal/kg/day)
  • High protein intake (CICM suggest 1.2-1.5g/kg/day)
  • Vegetable-based protein
  • Oral branched-chain amino acid supplementation
  • Multiple small meals frequently, to shorten the between-meal fasting period

Prognosis of hepatic encephalopathy

Severe hepatic encephalopathy in ICU seems to actually have a slightly better outcome than other sorts of organ system failures.

  • The mortality at 1 year is about 54% according to one small study.
  • Those patients who require nothing other than mechanical ventilation (i.e. ones who got intubated for low GCS and airway protection) tend to have better outcomes.
  • The ones which have ascites, varices (which bleed) as well as sepsis - their 1-year mortalty tends to be as high as 60%.
  • In spite of these grim numbers, the admission of cirrhosis patients to ICU is no longer viewed as a completely futile exercise, because there has been a gradual expansion of the treatment options available to them, and because their outcomes have improved with time.


Oh's Intensive Care manual:

Chapter 49   (pp. 549) Disorders  of  consciousness  by Balasubramanian  Venkatesh

Chapter 44   (pp. 501) Liver  failure by Christopher  Willars  and  Julia  Wendon

Parsons-Smith, B. G., et al. "The electroencephalograph in liver disease." The Lancet 270.7001 (1957): 867-871.

Ramos, Juan Francisco Rivera, and Celina Rodríguez Leal. "Review of the final report of the 1998 Working Party on definition, nomenclature and diagnosis of hepatic encephalopathy." Ann Hepatol 10 (2011): S36-S39.

Blei, Andres T., and Juan Córdoba. "Hepatic encephalopathy." The American journal of gastroenterology 96.7 (2001): 1968-1976.

Ciećko-Michalska, Irena, et al. "Pathogenesis of hepatic encephalopathy."Gastroenterology research and practice 2012 (2012).

Bajaj, J. S., et al. "Review article: the design of clinical trials in hepatic encephalopathy–an International Society for Hepatic Encephalopathy and Nitrogen Metabolism (ISHEN) consensus statement.Alimentary pharmacology & therapeutics 33.7 (2011): 739-747.

Ferenci, Peter, et al. "Hepatic encephalopathy—definition, nomenclature, diagnosis, and quantification: final report of the working party at the 11th World Congresses of Gastroenterology, Vienna, 1998." Hepatology 35.3 (2002): 716-721.

Prakash, Ravi, and Kevin D. Mullen. "Mechanisms, diagnosis and management of hepatic encephalopathy." Nature Reviews Gastroenterology and Hepatology7.9 (2010): 515-525.

Norenberg, Michael D., and Antonio Martinez-Hernandez. "Fine structural localization of glutamine synthetase in astrocytes of rat brain." Brain research161.2 (1979): 303-310.

Skowrońska, Marta, and Jan Albrecht. "Alterations of blood brain barrier function in hyperammonemia: an overview." Neurotoxicity research 21.2 (2012): 236-244.

Zwingmann, Claudia, et al. "Selective increase of brain lactate synthesis in experimental acute liver failure: Results of a [1H‐13C] nuclear magnetic resonance study." Hepatology 37.2 (2003): 420-428.

Ott, Peter, Otto Clemmesen, and Fin Stolze Larsen. "Cerebral metabolic disturbances in the brain during acute liver failure: from hyperammonemia to energy failure and proteolysis."  Neurochemistry international 47.1 (2005): 13-18.

Nabi, Eiman, and Jasmohan S. Bajaj. "Useful Tests for Hepatic Encephalopathy in Clinical Practice." Current gastroenterology reports 16.1 (2014): 1-8.

Guillén, Juan C. Quero, and Juan M. Herrerías Gutiérrez. "Diagnostic methods in hepatic encephalopathy." Clinica chimica acta 365.1 (2006): 1-8.

Amodio, Piero, et al. "Prevalence and prognostic value of quantified electroencephalogram (EEG) alterations in cirrhotic patients." Journal of hepatology 35.1 (2001): 37-45.

Amodio, P., et al. "The EEG assessment of low-grade hepatic encephalopathy: comparison of an artificial neural network-expert system (ANNES) based evaluation with visual EEG readings and EEG spectral analysis." Clinical neurophysiology 117.10 (2006): 2243-2251.

Shawcross, Debbie L., and Julia A. Wendon. "The neurological manifestations of acute liver failure." Neurochemistry international 60.7 (2012): 662-671.

Ferenci, Peter, et al. "Hepatic encephalopathy—definition, nomenclature, diagnosis, and quantification: final report of the working party at the 11th World Congresses of Gastroenterology, Vienna, 1998." Hepatology 35.3 (2002): 716-721.

Ortiz, M., et al. "Development of a clinical hepatic encephalopathy staging scale." Alimentary pharmacology & therapeutics 26.6 (2007): 859-867.

Cadranel, Jean-François, et al. "Focal neurological signs in hepatic encephalopathy in cirrhotic patients: an underestimated entity?" The American journal of gastroenterology 96.2 (2001): 515-518.

Conn, Harold O., and Milton M. Lieberthal. The hepatic coma syndromes and lactulose. Williams & Wilkins, 1979. - this is a book, and not available for free online; it is reviewed by E.Elias, and is apparently "thoroughly entertaining and sometimes downright funny reading." I would love a copy. The reviewer recommends that this treatise on hepatic encephalopathy remain "within easy reach of the armchair".