Adapted from Schindl MJ, Redhead DN, Fearon KC et al. The value of residual liver volume as a predictor of hepatic dysfunction and infection after major liver resection. Gut 2005; 54:289–96. With permission from the BMJ Publishing Group Ltd.

Common causes of acute liver failure: hepatic insufficiency following liver resections

Liver resection is the only treatment with the potential to cure patients with cancers that have originated in the liver itself (primary liver cancer) or that have originated elsewhere and have subsequently spread to the liver (metastatic liver cancer). Equally, it is a preferred therapy in patients with tumours in the liver that are benign, but with the potential of malignant transformation (uncertain benign primary liver tumours). Liver resection of even major parts of the liver (up to 70%) is feasible, because the liver has a remarkable capacity to regenerate. Within 6–8 weeks following 60–70% hepatectomy, the liver has regained nearly its original size and weight.

The most common cause of metastatic liver cancers is primary colorectal cancer, and it is estimated that in the West there is a yearly incidence of 300 new cases of liver metastases from colorectal origin per million population. The current estimate is that this should lead to approximately 100–150 patients per million eligible for liver resection for this indication. To this should be added the patients with primary benign and malignant liver tumours, and hence about 150–200 liver resections should probably be performed per million population each year.

Ever since the first liver resection by Langenbuch in 1887, this procedure has remained a major undertaking and even in the recent past, liver resection was still a dangerous surgical procedure with a high mortality of 20–30% in the 1970s. This was mainly due to excessive intraoperative bleeding but, over the subsequent decades, the procedure has become increasingly safe due to improvements in surgical and anaesthetic techniques. At present, mortality rates are reported to be well below 5%. Currently, the single most important cause of lethal outcome following surgical removal of major parts of the liver is liver failure. For this reason, many researchers and clinicians have attempted to design methods to identify patients at risk of liver failure (and hence mortality) following liver resection. The development of such a method has been hampered by several factors, as outlined below.

The critical point determining lethal outcome following liver resection has been a failure of the residual liver to function properly. Therefore, focus in this research area has always been in identifying a single liver function test that identifies those patients that have a liver with limited function. This has proven exceedingly difficult, and hence such a test is not available for a number of reasons.

First, as outlined above, the liver has a remarkable capacity to regenerate very rapidly, which underlines that there is tremendous overcapacity of several liver functions. In this context, it is known that it is entirely safe in most instances to resect 50% of the liver, because the residual half liver will simply take over all vital liver functions such as clearing bacteria, urea synthesis and synthesis of crucial proteins. From this, it has been estimated that a crucial liver function such as urea synthesis has an overcapacity of 300%, which implies that a static preoperative liver function test will be unable to assess this particular function. An alternative and innovative strategy would be to give a challenge to the liver and measure the ability of the liver to respond or cope – a dynamic test.

The critical minimum residual liver volume has been estimated to be approximately 25% after resection.²

The second crucial problem has been that there is only a poor correlation between volume and function. However, it is still unclear why some patients with smaller hepatic remnants do not develop liver failure whilst some with greater residual volumes do. These observations suggest, however, that peri- and intraoperative events superimposed on the innate hepatic capacity to withstand injury play a role. Hepatic insufficiency in this situation may arise either if not enough liver volume is left after partial hepatectomy or if the residual volume does not function properly. A functional limitation may arise, for example, in patients that have received aggressive chemotherapy in order to reduce the number and size of metastases prior to surgical treatment by liver resection. One of the factors contributing to defective defence may be preoperative fasting,³ but equally prior chemotherapy and pre-existent steatosis may play a role.

A third important aspect is that during liver surgery deliberate hypotension and temporary hepatic blood inflow occlusion (the so-called Pringle manoeuvre) are used by many surgeons to reduce blood loss during liver surgery (15 minutes ischaemia, 5 minutes reperfusion (15/5 Pringle)). Other surgeons do not use this manoeuvre, assuming that it causes oxidative stress and ischaemia/reperfusion (I/R) injury.⁴^,⁵ There is little doubt that this procedure does cause oxidative stress and I/R injury; however, the consequence of this is variable. In a situation where defence mechanisms against oxidative stress are deficient it may adversely affect liver function. In this situation hepatic steatosis may constitute an additional predisposing factor to damage by ischaemia/reperfusion.

Ischaemia/reperfusion is, on the other hand, the basis of ischaemic preconditioning, a process in which temporary clamping and release of the liver blood flow has been shown to be beneficial in terms of increasing resistance to subsequent injury.⁶

In this situation it is assumed that defence mechanisms against oxidative stress are adequate and are indeed enhanced by short-term I/R injury.⁷

The above three factors explain why it has been exceedingly difficult hitherto to design a proper liver function test that reliably singles out those patients at risk of liver failure following liver resection. The term ‘liver function’ is a rather crude denominator for a range of functions that includes ammonia detoxification, urea synthesis, protein synthesis and breakdown, bile synthesis and secretion, gluconeogenesis and detoxification of drugs, bacteria and bacterial toxins.

Chronic liver failure

The clinical signs of chronic liver failure are often insidious and can also be related to the type of disease. Cirrhosis is associated with a failure of hepatic function and the consequences of increased hepatic vascular resistance. Metabolic impairment is manifest by jaundice, coagulopathy, impaired ammonia clearance and encephalopathy, hypoalbuminaemia and oedema. The presence of increased vascular resistance is associated with the development of splenomegaly, ascites and gastro-oesophageal or abdominal wall varices. The slow progression of many chronic liver diseases, over years, implies a gradual, almost incremental, loss of liver cell mass or function. There are many causes of liver failure, including hepatitis B and C virus, autoimmune diseases such as primary biliary cirrhosis, primary sclerosing cholangitis and autoimmune hepatitis, alcoholic liver disease, Wilson’s disease, α₁-antitrypsin deficiency and others. All are associated with chronic or repeated cell injury and attempts at repair. The fibrosis and scarring associated with this regeneration and repair lead to the clinical condition termed cirrhosis, with a typically small shrunken irregular liver and an increased risk of cancer.

The Child–Pugh score for chronic liver disease ⁸ has served as a useful means of categorising patients based on the severity of their liver disease. It employs five clinical measures of liver disease and each measure is scored 1–3, with 3 indicating the most severe derangement (Table 1.1).

Table 1.1

Child-Pugh score for chronic liver disease

Metabolic liver function

The liver plays a central role in fat, carbohydrate and protein metabolism, as well as in acid–base homeostasis. In the context of liver failure, disturbances of fat metabolism are probably not crucially important. With respect to carbohydrate metabolism, it is well known that the liver plays a central role in the conversion of lactate to glucose. Part of this lactate is formed due to anaerobic metabolism of, amongst others, glucose in skeletal muscle. This metabolic route of glucose to lactate (muscle) and then back to glucose (liver) is very important for glycaemic homeostasis and is called the Cori cycle. Failure of the liver will be witnessed by lactic acidosis and hypoglycaemia.

Next to its role in carbohydrate metabolism, the liver plays a central function in nitrogen homeostasis. Hepatic synthesis and breakdown of proteins and amino acids, and detoxification and clearance of nitrogenous waste products of amino acid and protein metabolism in other organs are of central importance. For example, the gut uses the amino acid glutamine as a fuel for enterocytes, which give rise to the production of waste end-products of intestinal metabolism, like ammonia. This ammonia is then transported by the portal vein to the liver, where it is detoxified by the formation of urea.

Liver failure gives rise to multiple abnormalities in nitrogen metabolism, some of which are thought to play a crucial role in the characteristic syndrome of hepatic encephalopathy that accompanies liver failure. Hepatic encephalopathy is a reversible neuropsychiatric syndrome, with a probably multifactorial cause.⁹ The current belief is that ammonia is one of the key components in the aetiology of hepatic encephalopathy¹⁰ because liver failure is usually associated with moderate to severe hyperammonaemia. Hyperammonaemia leads to increased brain uptake of ammonia, followed by detoxification of ammonia in the brain by coupling to glutamate to form glutamine. This process consumes glutamate (an important excitatory neurotransmitter) and leads to the formation of glutamine, which acts as an osmolite causing brain oedema.

One other well-known metabolic abnormality during liver failure is an imbalance in plasma amino acids, notably the ratio between the branched chain amino acids (BCAAs) and the aromatic amino acids (AAAs).

Some 30 years ago, Fischer and colleagues published their ‘unified hypothesis on the pathogenesis of hepatic encephalopathy’,¹¹ based on the observation that, during hepatic failure, plasma levels of BCAAs were decreased and the AAAs were increased.^11–13

These changes in plasma levels were thought to be caused by increased BCAA catabolism in muscle and decreased AAA breakdown in the failing liver.¹⁴ A reduction in the insulin–glucagon ratio in this situation may play a key role in disturbing the balance between anabolism and catabolism. Accumulation of AAAs in the circulation in combination with the increased breakdown of BCAAs, particularly in skeletal muscle, would, according to this hypothesis, give rise to a decrease in the BCAA to AAA ratio, the so-called Fischer ratio. The increase in plasma AAAs in combination with increased blood–brain barrier permeability for neutral amino acids has been suggested to contribute to an increased influx of AAAs in the brain, since they compete for the same amino acid transporter. This, in turn, would lead to imbalances in neurotransmitter synthesis and accumulation of false neurotransmitters such as octopamine in the brain, which may contribute to hepatic encephalopathy.¹⁵

Measuring liver volume

Advances in imaging have permitted the development of in vivo imaging of the liver. Three-dimensional models of the liver can be constructed from computed tomography (CT) or other cross-sectional imaging modalities, such as magnetic resonance imaging (MRI). The volume of the liver can then be calculated based on known separation of image slices combined with planar mapping of cross-sectional areas. In addition, such three-dimensional computer models can be simulated to map the effects of surgery by performing virtual hepatic resection, and studies have demonstrated that there is a good correlation between computer modelling and actual resection weight of surgical liver specimens (Fig. 1.1).²^,¹⁶

Figure 1.1 Three-dimensional reconstruction of the liver preoperatively (red) showing tumours (green). Computer prediction of residual liver volume based on virtual hepatectomy of 3-D model (yellow) and actual photograph of resection showing residual liver segments. Reproduced from Schindl MJ, Redhead DN, Fearon KC et al. The value of residual liver volume as a predictor of hepatic dysfunction and infection after major liver resection. Gut 2005; 54:289–96. With permission from the BMJ Publishing Group Ltd.

This technology is useful as a research tool because it allows liver function to be put into the direct context of the volume of functioning liver tissue. In addition, this technology is useful for predicting the need for reconstruction of venous territories of the liver in split liver transplantation. Usually, liver volumetry is performed on software directly linked to the hardware MRI or CT. In recent years, however, stand-alone software has become available, which makes it possible to perform hepatic volumetry remote from the radiological hardware. Examples of such software are the freely downloadable program ImageJ (for Windows-based PCs) and OsiriX (for Apple Macintosh). Our group has recently shown that the ImageJ software is very useful in measuring liver volumes in patients referred with a CT undertaken in the referring centre ¹⁷ (Figs 1.2 and 1.3).

Figure 1.2 Correlation between volume of resection calculated with ImageJ and actual measured weights of the resection specimens (n = 15, Pearson’s test).¹⁴ Reproduced with permission from World J Surg.

Figure 1.3 Mapping the territory of the right hepatic lobe drained by the middle hepatic vein. The numbers represent the volumes of the territories at risk if segment 5 and 8 tributaries of the middle hepatic vein were not reconstructed in a potential right lobe living-donor liver transplant. Image reproduced with permission of MeVis imaging technologies, Bremen, Germany Kindly provided by H. Lang and A. Radtke, Plainz, Germany.

Blood tests of liver function

As part of many blood chemistry analyses, it is possible to request liver function tests. These tests refer to the transaminases, alkaline phosphatase, γ-glutamyl transferase and bilirubin. These tests are not truly measures of function but do give an indication of processes going on within the liver. Aspartate aminotransferase and alanine aminotransferase are hepatocyte enzymes that are released in conditions in which hepatocytes are damaged or killed, such as ischaemic injury, hepatitis, severe sepsis and in response to cancer. Liver-specific alkaline phosphatase is expressed predominantly in the biliary epithelium and is elevated in conditions such as cholangitis or biliary obstruction. γ-Glutamyl transferase is expressed by both hepatocytes and biliary epithelium, and can also be induced by high alcohol consumption.

Biochemical markers of true liver function vary depending on whether acute or chronic liver failure or injury is being considered (Table 1.2).

Table 1.2

Blood tests useful to assess function in acute and chronic liver injury

	Acute	Chronic
Albumin	−	+++
Prothrombin time	+++	+++
Bilirubin	+	+++
Lactate	++	−
Glucose requirement	++	−
Ammonia	+	+

Tests of liver function measuring substance clearance

The ability to accurately predict postoperative outcome based on preoperative liver function would be a valuable addition to preoperative assessment. However, while various tests have been developed to assess liver function there is little evidence that these tests have sufficient sensitivity or specificity to predict postoperative outcome at an individual patient level. The tests currently in common use include the indocyanine green (ICG) clearance test, hepatobiliary scintigraphy with radioisotope clearance, lidocaine clearance test, the aminopyrine breath test and the galactose elimination test. These tests aim to provide an indicator of dynamic liver function, in that they can provide real-time assessment of liver function in response to a challenge. However, none of these tests challenge the liver to demonstrate its full functional capacity. Serum bilirubin and clotting factors provide a static indirect estimation of liver metabolism and synthetic function, but are influenced by a range of other factors that limit their relevance and suitability to predict postoperative outcome. The most commonly used test for liver function prior to liver resections is the ICG clearance test.

Urea synthesis

In the recent past, we have explored the feasibility of measuring urea synthesis using stable isotopes and relating this to liver volume in patients undergoing liver resection.²⁰ This study was conducted against the background of the notion that liver failure is almost always accompanied by hyperammonaemia, related to a presumed failure of hepatic urea synthesis. Using stable isotopically ¹³C-labelled urea, urea synthesis was measured before and after major hepatic resection, and liver volumes before and after resection with CT scans.

Major hepatic resection did not affect total body ureagenesis, because the synthesis of urea per gram of residual liver went up and increased 2.6-fold.²⁰ Therefore, it is unlikely that urea synthesis is a limiting factor in the initial aetiology of liver failure and this test is not likely to contribute to predicting liver failure following liver resection.

Glutathione synthesis

Unfortunately, most of the above tests focus on very specific functions or pathways. None of them assesses the main hepatic protection system against many diverse forms of stress and intoxications: the intracellular content and synthesis of glutathione (GSH). It is generally accepted that GSH plays a key role in the protection of the liver against many forms of stress, ischaemia and toxic compounds such as paracetamol. Unfortunately, there is currently no adequate test to assess hepatic GSH synthesis and metabolism in vivo in humans, even though such a test would be of great clinical importance. We have previously explored the feasibility of measuring GSH synthesis in vivo during liver surgery in humans using stable isotopically labelled ²H₂-glycine, a component of GSH (γ-glutamyl-cysteinyl-glycine), but this approach was not suitable, because part of the deuterium label of glycine was lost (unpublished data). Future research will have to focus on designing a test that is both dynamic and which focuses on the GSH system, making it possible to determine liver function correlated to liver volume, and assess an individual’s risk of developing liver failure following liver resection.

Measuring liver blood flow

Blood flow in the splanchnic area, particularly the gut and liver, can be measured in a number of ways. These can basically be either invasive (i.e. intraoperative) or non-invasive. During surgery, when the abdomen is opened, blood flow can be measured in the portal vein and in the main hepatic artery. Portal vein blood flow measurements provide predominantly information on the flow across the intestines. By summing up the blood flow in the hepatic artery and the portal vein, total hepatic blood flow can be calculated. Theoretically, this could also be achieved by measuring hepatic venous outflow, but this is impractical in humans because of the short common outflow tract of the three hepatic veins. Non-invasive MRI-based techniques are being developed that may offer improved accuracy of measurement of liver blood flow and the potential for repeat measurements.²¹ The ratio of portal vein to hepatic artery blood flow changes with increasing resistance of the liver and may indicate the development of fibrosis or cirrhosis. Methodology for assessing the importance of blood flow as a predictor of liver parenchymal condition has not been fully evaluated but may provide a means of assessing safety of surgery and regenerative capacity in some patients.

Such measurements of hepatic and portal arterial blood flow can be obtained using 6–8 mm and 12–14 mm handle ultrasonic flow probes, respectively (Transonic Systems, Kimal PLC, Uxbridge, UK). Essentially, the vessels have to be dissected free for this flow measurement and the three-quarters circular probe is applied to the vessel. These probes are believed to provide the most accurate technique for assessing flow in relatively small vessels. However, there is considerable variability in measurement related to Doppler ultrasound signal strength and coupling with the vessel wall. Also, there are likely to be changes in diameter of the artery, in particular related to its handling during surgery. The advantage is that repeated measurements can be obtained and that the surgeon can operate this application without help from a radiologist. Likewise, post-resection blood flow measurements can be taken before closure of the abdomen, typically 1–2 hours after the first measurement. This gives an impression of blood flow across the residual liver following major resections.

During liver surgery, organ blood flow can also be measured by means of colour Doppler ultrasound scanning (e.g. Aloka Prosound SSD 5000; Aloka Co. Ltd, Tokyo, Japan). A 5-MHz probe is used to trace the vessels and calculate the cross-sectional area. Then, time-averaged mean velocities of the bloodstream are measured at the point where the cross-sectional area of the portal vein and hepatic artery is taken. For accurate velocity measurements, care must be taken to keep the angle between the ultrasonic beam direction and blood flow direction below 60°. The cross-sectional area of the vessel is calculated by drawing an area ellipse at the same point as where the velocity was measured. Portal venous and hepatic arterial blood flows can then be measured proximal to their hilar bifurcations. In the case of an accessory hepatic artery, both arteries should obviously be measured.²²^,²³ In our experience, this method gives roughly the same values as the ultrasonic flow measurement described above. Theoretically, it is possible to perform such flow measurements preoperatively or postoperatively using a percutaneous approach, although the measurement in the hepatic artery requires a skilled ultrasonographer.

In recent years, technical improvements in hardware and software applications for MRI have made it possible to measure blood flow in the portal vein and hepatic artery in a non-invasive manner. By linking this method of flow measurement to hepatic volumetry using MRI, blood flow per volume unit of liver can be calculated.²⁴^,²⁵ It has been suggested that MRI may provide a more accurate and reliable assessment of portal vein and hepatic artery blood flow than ultrasonography, particularly given the wide interobserver variability seen with the latter technique.²¹ Although limited to the preoperative period, MRI flow studies may provide complementary information to intraoperative ultrasonography as required.

A further technique that is emerging is the use of near-infrared spectroscopy. This technique measures absorption of near-infrared wavelength light and from this can be calculated tissue oxygenation, since haemoglobin oxygenation status alters absorption of this wavelength light. This technique is more useful for estimating tissue oxygenation and perfusion at a sinusoidal level, but could potentially be combined with other measures to estimate liver blood flow.²⁶

Effect of major liver resection on hepatic blood flow

Direct measurement of hepatic artery and portal vein blood flow before and after liver resection reveals interesting results. When expressed as absolute values portal blood flow does not change whereas hepatic artery blood flow falls. Typically, portal vein flow is approximately 840 mL/min and post-resection 805 mL/min. Using this method, hepatic artery flow pre-resection is about 450 mL/min and post-resection 270 mL/min. When these flows are expressed in relation to the preoperative liver volume and residual postoperative liver volume, it can be seen that the blood flow per gram of liver tissue increases in portal flow from a mean 0.55 mL/min per g liver to 1.09 mL/min per g liver and the hepatic artery flow remains relatively constant (Fig. 1.5). In experimental research, pressure measurements can also be obtained using radial artery invasive monitoring to estimate hepatic artery pressure and direct portal vein pressure measurement using a small needle coupled to a pressure transducer similar to that used for measuring central venous pressure. The combination of flow and pressure measurement then allows calculation of hepatic sinusoidal resistance (Fig. 1.5).

Figure 1.5 Directly measured blood flow intraoperatively in six patients during major hepatic resection. Measurements were taken from the main portal vein and the main hepatic artery simultaneously using multichannel Transonics ultrasound flow probes. During the liver resection one branch of each of the portal vein and hepatic artery is ligated. The post-resection blood flow measurement has been taken just before closure of the abdomen, typically 1–2 hours after the first measurement. Results are expressed per gram of liver tissue.

Effect of major liver resection on innate immunity

The liver forms an important part of the innate immune system by producing acute-phase proteins and other opsonins, proteins that bind to bacteria facilitating their phagocytosis. In addition, 85% of the reticuloendothelial system is located in the liver (Kupffer cells) and clearly surgical resection will involve a reduction of this cell mass.

It is not unreasonable to expect that major liver resection might result in some impairment of innate immunity. Our group has previously demonstrated that major liver resection is associated with increased frequency of infection as well as increased likelihood of objective evidence of liver function impairment.²

In a separate study, our group has also shown that major liver resection is associated with a temporary defect in the ability of the reticuloendothelial system to clear albumin microspheres that were used as a surrogate for bacteria.

Loss of approximately 50% of liver volume, such as might occur during a right hepatectomy, is associated with impairment of reticuloendothelial cell clearance equivalent to that of non-surgical patients with Child C chronic liver disease.²⁷

The liver also synthesises and exports many acute-phase proteins involved in innate immunity or homeostasis. C-reactive protein, for example, binds to phosphoryl choline moieties of encapsulated bacteria and acts as an opsonin, promoting phagocytosis. Mannan-binding lectin, complement fragments and α₁-acid glycoprotein (orosomucoid) can also act as opsonins. Transferrin and caeruloplasmin are important in the binding and carriage of free metal ions and α₁-antitrypsin and α₁-antichymotrypsin act as antiproteases. Liver failure or liver surgery may be associated with a reduction in synthesis of some of these acute-phase proteins (mannan-binding lectin, haptoglobin, α-fetuin and fibronectin), whereas the concentrations of others may be increased despite a reduction in functional liver tissue (C-reactive protein, liver fatty acid-binding protein; unpublished data). The exact significance of these changes is unclear but may contribute to a global impairment in innate immunity in the injured liver.

Liver regeneration

The liver is unique in that it is the only organ in the adult that is capable of regenerating or renewing itself to restore the ratio between pre-injury liver volume and body weight. Knowledge of the capacity for the liver to regenerate is presumed to be ancient and is the basis for the punishment meted out by Zeus to Prometheus, who according to Greek mythology was chained to a rock and had his liver eaten daily by a vulture (only for it to regenerate overnight). This continued for several years until the vulture was finally killed by Heracles, who also released Prometheus. While the speed of liver regeneration is exaggerated in this myth, it is true that it is an extremely rapid process. In the context of surgery, liver regeneration happens very rapidly, with most of the cell division required for regeneration occurring within 72 hours of injury. Full liver function and volume are usually restored within 6–12 weeks. In chronic injury or in the presence of fibrosis, liver regeneration can be chaotic with repeated insults causing scarring, and nodular regeneration with disordered architecture leading to cirrhosis.

Molecular signals for hepatic regeneration

At a cellular level, liver regeneration depends on the coexistence of three key factors: changes in the microenvironment of the liver cell supporting growth, the ability of differentiated hepatocytes to proliferate and inhibition of processes, linking injury to programmed cell death.

Stimuli for liver regeneration stimulate transcription factors that turn on a variety of genes expressing growth factors. Although not direct growth factors, the hormones insulin and adrenaline potentiate the effects of growth factors on hepatocyte regeneration. All elements of the liver are required to regenerate; however, the coordination of these processes is complex. Removal of the stimulus for regeneration by growth to pre-injury capacity and transforming growth factor-β act as brakes that slow regeneration of liver elements (Fig. 1.6).

Figure 1.6 Schematic of some of the factors known to regulate liver regeneration.

Macrophage-derived Wnt signalling directs hepatic progenitor cells to become hepatocytes in chronic liver injury by maintaining progenitor cell expression of Numb (a cell fate determinant).²⁸ Targeting progenitor cells may offer the potential to enhance regeneration of the chronically injured liver.

Barriers to hepatic regeneration include cirrhosis and fibrosis and ongoing liver injury such as might occur with biliary obstruction or sepsis.

Cell populations involved in liver regeneration

Histology of normal liver regeneration following resection or acute injury shows the presence of high mitotic rates in mature hepatocytes. Normally, these cells are mitotically quiescent but can move into S phase extremely rapidly. For example, following 70% hepatectomy in rat approximately 30–40% of hepatocytes are seen to be undergoing mitosis within 48 hours of surgery and indeed the liver will regain its normal size within 10 days. The situation is more complex in chronically injured liver (e.g. cirrhotic liver); here, the hepatocytes are less able to undergo mitosis and are frequently in cell cycle arrest. Furthermore, the accumulation of excess scar tissue deposited in cirrhosis contributes to the inability of the liver to respond to injury and regenerate effectively. In this setting a second population of cells becomes activated and can contribute to parenchymal regeneration. These intrahepatic cells are located in the canal of Hering (the most distal branch of the biliary tree); termed hepatic progenitor cells (HPCs), they are bipotential and are capable of giving rise to both biliary and hepatocyte populations under the influence of macrophage-derived factors (see above). This response is seen in chronic or severe injury and sometimes appears as a ductular reaction. Although it is now recognised that the HPCs can regenerate the liver in chronic liver disease, whether these progenitor cells are capable of responding to the acute demands of major hepatic resection is as yet unknown. It is also worth noting that there is an increasing recognition that intrahepatic stem cells are a likely source of a significant proportion of liver cancers. The role of circulating extrahepatic cells in liver regeneration has received interest recently and the potential bone marrow origin of hepatocytes has been suggested. However, if this phenomenon occurs at all, it is extremely rare. The bone marrow does, however, supply macrophages and myofibroblasts that are involved in the liver’s scarring response to injury. The relationship between bone marrow-derived cells and the response to injury is complex, with different macrophage subtypes shown to either promote fibrosis or repair. However, administration of bone marrow-derived macrophages to the fibrotic liver via the portal vein has been shown to reduce fibrosis and improve markers of regeneration in preclinical models.²⁹ The use of bone marrow populations to stimulate liver regeneration in both animal models and clinical studies is likely to be an area of future development (see later).

Consequences of surgery

Unfortunately, at present it is unclear what the key mechanisms of liver failure are, and why the liver usually regenerates but sometimes progresses into liver failure. It is believed that ischaemia/reperfusion (I/R) injury plays an important role in the sequence of events leading to liver failure. Hepatic resections are major surgical procedures, often leading to significant blood loss. In order to reduce blood loss, central venous pressure is reduced during liver surgery and hepatobiliary surgeons frequently occlude hepatic blood inflow temporarily (Pringle manoeuvre). Obviously, all these factors may contribute to I/R injury in the liver. A key component of I/R injury is the generation of oxygen free radicals. The latter can induce ischaemic necrosis and caspase-dependent apoptosis, and may contribute to failure of vital metabolic synthetic pathways. However, it remains to be investigated which one of these plays a key role during liver failure. In this context, it has been proposed that the balance between hepatocyte regeneration and apoptosis can be tipped towards either side by hepatic defence mechanisms against oxygen free radical damage. Also, oxygen free radicals play a role in determining whether apoptosis or ischaemic necrosis occurs in the liver. Apparently, the equilibrium between oxygen free radicals and their scavengers plays a pivotal role in determining whether regeneration or decay occurs. Glutathione (GSH) is the principal oxygen free radical scavenger in the liver and the principal defence mechanism against I/R damage. Hepatic GSH levels decrease following I/R damage, inflammation and nutritional deprivation. It seems conceivable that a reduction in liver volume following surgery contributes to insufficient hepatic free radical scavenging capacity as a consequence of reduced GSH synthesis. I/R injury may aggravate this situation.

Small-for-size syndrome

The original descriptions of small-for-size syndrome described a condition arising in split liver transplantation characterised by the development of ascites, portal hypertension and liver dysfunction in an otherwise healthy transplanted portion of liver. The underlying cause for this syndrome is believed to relate to blood flow and the failure of a small liver volume to cope with often very high blood flows of patients with previous chronic liver disease undergoing transplantation. The validity of this hypothesis was supported by the observation that partial diversion of portal blood flow into the graft using a portocaval shunt could limit or prevent the development of small-for-size syndrome. Subsequently, other manoeuvres have also been effected, such as ligation or embolisation of the splenic artery, which works in the same way by reducing portal vein flow.

In patients undergoing even very major liver resection it is rare to develop small-for-size syndrome. Some patients do, however, develop ascites, jaundice and chronic liver dysfunction, and it is more likely that this syndrome is more dependent on a failure to regenerate than on excessive blood flow.

Hepatic steatosis

Fat infiltration of the liver is an increasing problem with increased prevalence of obesity and the metabolic syndrome (obesity and type 2 diabetes). Macroscopically the liver may appear enlarged, pale or yellow coloured with rounded edges. Microscopically the liver can have microsteatosis (small fat droplets within every hepatocyte) or macrosteatosis (regional infiltration of hepatocytes with large fat droplets) (see Fig. 1.7).

Figure 1.7 Macroscopic and microscopic images of steatotic liver.

Assessment of steatosis

Assessment of hepatic steatosis is notoriously difficult. Experienced surgeons can estimate liver fat by judging the size, rounded or sharp edges of the liver and its appearance. Even using colour as an estimate is prone to error, as can be seen in Fig. 1.8.

Figure 1.8 Physical appearance of livers with varying fat content confirmed by histology to demonstrate the poor correlation between colour and objective measurement of fat content.

The gold standard for hepatic fat assessment is histology. Trucut or wedge biopsies can be assessed by a pathologist and a reliable estimate of the percentage fat content produced. In addition, useful information including the distribution – macrosteatosis or microsteatosis – and the presence of fibrosis or inflammation can be provided. New MRI techniques are, however, challenging the accuracy of pathological assessment of steatosis and offer the potential advantage of being non-invasive.³⁰

Chemotherapy-induced liver changes

Increased usage of chemotherapy in the neoadjuvant context has revealed changes in the liver associated with chemotherapy, particularly oxaliplatin and irinotecan. These range from a soft, fragile pale liver to steatosis, steatohepatitis and sinusoidal dilatation. Surgery should be deferred until 6 weeks after chemotherapy and studies, although conflicting, suggest that tolerance of major liver resection may be reduced and complications more frequent in individuals who have received chemotherapy. A study by Mehta et al. showed that oxaliplatin-based chemotherapy was associated with increased blood loss and prolonged hospital stay.³¹

Portal vein embolisation

Morbidity and mortality after hepatectomy have constituted a limitation on the number of patients eligible for resection, and currently only 8% of patients with colorectal hepatic metastases are candidates for curative liver resection. Liver function is correlated with liver volume, and consequently hepatic insufficiency in this situation may arise because not enough functional liver volume is left after surgical removal of part of the liver. Interestingly, following removal of part of the liver, the residual liver usually regenerates to the point where the preoperative liver weight–body weight ratio is regained. This notion has led to the belief that if it were possible to increase preoperatively the volume of the future residual liver, it would be possible to perform more extensive liver resections and hence cure more patients. It has long been recognised that interruption of one part of the liver portal blood flow usually leads to hypertrophy of normally vascularised liver. This has been observed in patients with Klatskin tumours, which have a tendency to invade the portal vein, causing ipsilateral atrophy and contralateral hypertrophy. This concept has subsequently been harnessed by manoeuvres such as embolising the right portal vein prior to surgical resection. This leads to hypertrophy of the left liver lobe prior to surgery and facilitates the subsequent safe extensive resection of the right liver (extended right hepatectomy) 6 weeks later (Fig. 1.9). This phenomenon has been harnessed to maximise the residual functional liver volume of patients who are predicted to have a small remnant liver volume. This approach is fully based on the concept that, in the normal liver, volume is correlated to function and hence liver failure occurs when residual liver volume is too small. A completely different and novel approach would be to improve liver function per volume unit of liver. Recent evidence from studies using mebrofenin suggests that functional improvement of the future liver remnant following portal vein embolisation (PVE) may precede changes in liver volume.³² This important observation suggests that surgery earlier after PVE may be possible. Limitations to PVE-induced hypertrophy include pre-existing hepatic fibrosis or cirrhosis and technical or anatomical inability to completely obstruct a major portal vein branch.

Figure 1.9 Portal venograms showing the main left and right branches prior to embolisation (a), and after embolisation of the right portal vein (b).

Technique

The most common technique of PVE is to puncture a branch of the vein using a percutaneous approach. A venogram is obtained to demonstrate all of the relevant branches and then the branch to be embolised is cannulated and coils and embolic material delivered to obstruct portal flow. A check angiogram can be performed to demonstrate success of the technique. Usually either a left or right main branch is occluded. To obtain hypertrophy of segments 2 and 3 in large right-sided tumours, it is not sufficient to embolise just the right portal vein and it is recommended that the branches supplying segment 4 should also be embolised. Patients usually tolerate PVE remarkably well, presumably because of the dual blood supply of the liver, and complications are uncommon. Significant hypertrophy can be achieved, as can be seen in Fig. 1.10.

Figure 1.10 Calculated residual liver volumes before and after portal vein embolisation (PVE) in patients scheduled to undergo major liver resection.

Therapy for liver failure

N-Acetyl cysteine

Glutathione depletion is a major problem in patients with paracetamol (acetaminophen) toxicity. N-Acetyl cysteine has been used for many years as a treatment for early paracetamol poisoning. It is thought to act by replenishing glutathione stores and by providing alternative thiol groups to which damaging reactive oxygen species can bind. The realisation that reactive oxygen species can be generated by conditions other than paracetamol poisoning such as sepsis and ischaemia/reperfusion has led to N-acetyl cysteine being used in a more general way to support patients with early evidence of liver dysfunction or failure.

Nutritional support in liver failure

The role of nutritional support in acute liver failure is uncertain, largely because of a lack of evidence in the world literature. Enteral nutrition is known to preserve gut barrier function and thus might be considered to be beneficial in the context of liver failure. In addition, the provision of energy might be considered beneficial in the context of glycogen storage failure, and to fuel the regeneration of liver tissue and recover function. The limited ability of the failing liver to handle nitrogen and synthesise urea (potentially exacerbating encephalopathy) would argue against excessive provision of proteins unless these were in a form where they did not contribute to the circulating ammonia load.

Artificial extracorporeal liver support

For the vast majority of patients who take toxic doses of paractetamol, suffer alcohol-induced liver injury or develop liver dysfunction following liver resection, the regenerative capacity of the liver is sufficient to prevent irretrievable liver failure and death. However, when this regenerative capacity is overwhelmed treatment strategies to temporarily or permanently replace the failing liver are required. The ability to provide short-term extracorporeal liver support, either during the wait for transplantation or to facilitate liver regeneration and avoid transplantation, is an attractive option. A range of devices have been developed, either focusing on the detoxification functions of liver (artificial liver support) or also incorporating bioreactors intended to also perform synthetic liver functions (bioartificial liver support). Assessment of efficacy has been hampered by the limited number of randomised controlled trials and small sample size, but a recent meta-analysis does suggest overall survival benefit in acute liver failure.³³

Artificial liver support

Artificial systems include the MARS (Molecular Adsorbent Recirculating System) device, Prometheus and the BioLogic-DT (now called the Liver Dialysis Device, currently being redesigned). The greatest experience has been with the MARS device, which deploys an albumin dialysis circuit to remove both water-soluble and protein-bound toxins.³⁴ Thus, a low Fischer ratio can be corrected by recirculating albumin dialysis.³⁵ Because the system preferentially removes AAAs, compared with BCAAs, the Fischer ratio significantly increases, predominantly by the removal of AAAs in a small series of patients.^35–38 MARS has been shown to be useful in fulminant hepatic failure, by attenuating the increase in intracranial pressure, which plays a major role in this situation.³² There may also be an effect on survival and improvement of degree of hepatic encephalopathy in patients with acute or chronic liver failure.³⁷^,³⁹ Equally, the system has been tested on artificial neuronal networks showing a normalisation of abnormal signals if the medium (plasma derived from rats with liver failure) was pretreated with MARS. The role of MARS in a more chronic situation of mild hepatic encephalopathy, when correction of an abnormal Fischer ratio would likely be more important if this were a major pathogenetic factor, is still largely unknown and deserves further study.⁴⁰ It has been suggested that the role of MARS and bioartificial liver support systems should be limited to carefully designed clinical trials.⁴¹ It is currently uncertain how hepatic excretory assistance devices, such as MARS, compare with bioartificial liver assistance devices, which in addition to their excretory functions aim to provide biosynthetic capacity.³⁷

Bioartificial liver systems

Bioartificial systems incorporate a bioreactor containing either human hepatoblastoma cell lines (e.g. the HepatAssist device) or porcine hepatocytes (e.g. the ELAD – Extracorporeal Liver Assist Device), through which the patient’s blood is perfused. An additional filter component may be included to aid detoxification and improve bioreactor survival.

One of the major problems with these systems is what type of cells to use, and a variety of different approaches have been taken. Animal hepatocytes perform many of the same functions as human hepatocytes, although some of the proteins produced are obviously different. Human immortalised cell lines are an attractive proposition and some of the more differentiated cell lines can replicate many of the normal hepatocyte functions. Regardless, the true functionality of these cells in the clinical setting is uncertain. The design of bioartificial liver systems is challenging and the large surface area of hepatocytes needed to be effective is difficult. Engineering scaffolds of membranes or tubules has been the most popular approach. In normal liver, hepatocytes are polarised and have an epithelial surface. However, it is still to be determined how to recreate this polarity and its absolute importance has yet to be defined. Hepatocytes proliferate and function better in association with non-parenchymal cells; however, the creation of co-cultures in reactors produces its own problems. Cells must maintain viability or be able to be replenished to provide liver support over a prolonged period of time. In addition, very sick patients require a short time period to set up the support system, and the reactor must be easy to use by critical care nurses, safe from contamination and not overly expensive. For all of these reasons, bioartificial liver systems remain a tantalising prospect that has yet to break through into routine clinical practice.

Liver transplantation

Irreversible acute or chronic liver failure is amenable to treatment by liver transplantation. It is extremely uncommon for patients who have undergone liver resection to subsequently require or proceed to liver transplantation. The most obvious reason for this is that many patients who undergo liver resection do so for metastatic or primary liver cancer and transplantation would be contraindicated because of the risk of immunosuppression and aggressive recrudescence of the tumour. A number of patients with bile duct injury have progressed to transplantation, usually in a chronic setting following the development of biliary stricture, cholangitis and secondary biliary cirrhosis. Similarly, a number of patients who have undergone a ‘cancer resection’ for what turned out to be a benign biliary stricture, perhaps as part of primary sclerosing cholangitis, fail to regenerate their livers and may progress to transplantation.

Cell therapy for liver failure: general principles

A number of key principles have operated as key drivers for the development of cell therapies for clinical treatment of liver failure. Firstly, it is recognised that the injured liver usually provides a rich environment stimulating tissue regeneration and the liver can normally ‘heal’ itself. Secondly, in animal models there is evidence that stem cells or non-parenchymal cells can support regeneration of hepatocytes. Thirdly, it is recognised that the difference between liver failure and compensated liver function in terms of cellular functional equivalents is probably very small. Finally, it would be preferable to support the liver by techniques that were within the body rather than using extracorporeal devices. This desire has stimulated research into therapeutic application of cell or stem cell transplantation.

The dual goals of stem cell therapy in the context of acute liver failure or injury are to promote rapid recovery of hepatocyte function and to allow regeneration of liver tissue without excessive scarring. Direct administration of hepatocytes or stem cell-derived hepatocytes to the injured liver has been met with little success in preclinical studies. However, using bone marrow-derived cells to support endogenous processes may support the regenerating liver, enabling effective regeneration.⁴²

Haemopoetic stem cell therapy for liver disease in humans

There are several reports in the scientific literature of bone marrow (BM) stem cell therapy in patients with advanced liver disease. It was first reported that BM stem cells could increase the liver’s ability to regenerate in patients who were undergoing hepatic resection for various liver cancers sited in the right lobe. Here the patients underwent embolisation of the right branch of the portal vein prior to surgery to stimulate compensatory hypertrophy of the left lobe. Autologous CD133-positive BM stem cells were injected into the blood vessels that supply the left liver lobe shortly after the surgery and accelerated regeneration of the non-embolised section of the liver was seen compared with control patients.⁴³ However, it must be stated that this was a small non-randomised study. The second report used BM stem cells in patients with liver cirrhosis.⁴⁴ CD34-positive stem cells were isolated from the patients’ own blood following granulocyte colony-stimulating factor (GCSF)-induced haematopoietic stem cell mobilisation and were re-injected into the blood supply to the liver – preliminary evidence appeared to show that improvement in liver function in three out of five of the patients occurred during this therapy. In the third study, patients with liver cirrhosis had mononuclear cells isolated from their own BM during general anaesthesia.⁴⁵ These cells were re-injected into the patient’s bloodstream and again the patient’s liver function appeared to improve. Although these studies are very encouraging, they are preliminary, of small numbers and non-randomised. Furthermore, in none of these studies were the cells marked to enable identification either by radiological tracking or in biopsies of the liver tissue. Therefore, a number of important questions are unanswered. It is not certain that these cells definitely settled in the liver over a period of time, whether some of the cells engrafted other organs in the body and by what mechanisms the cells were having their positive effects within the recipients’ livers.