Postoperative liver dysfunction is a common problem. Although the incidence after elective abdominal surgery is less than 1 per cent, much higher rates occur after major surgery, multiple trauma, and prolonged intervention. The majority of cases are mild, transient, and resolve spontaneously, but occasionally the liver injury may be severe and result in fulminant liver failure and/or chronic liver disease. There are many aetiological factors, and in any one patient the pathogenesis is often multifactorial (Fig. 1) 76.

In this section the clinical presentation, investigation, diagnosis, treatment, and prevention of hepatic problems in the surgical patient are discussed. Patients with normal preoperative liver function are considered separately from those with pre-existing liver disease. Primary liver diseases, details of hepatobiliary surgery, liver transplantation, and surgical treatment of portal hypertension and ascites are dealt with elsewhere.

Postoperative hepatic dysfunction in surgical patients with normal liver function can be classified into three groups: those due to (1) overproduction of bilirubin; (2) hepatocellular dysfunction; and (3) extrahepatic biliary obstruction (see Table 1 103).

In a healthy individual the liver conjugates up to 500 &mgr;mol of bilirubin per day as a result of the breakdown of red blood cells. The liver is capable of handling several times this quantity without the occurrence of hyperbilirubinaemia and only if haemolysis is severe or occurs in conjunction with hepatocellular insufficiency does jaundice develop. Unconjugated bilirubin comprising 90 per cent of the total is suggestive of haemolysis. When the level of unconjugated bilirubin is excessively high there appears to be a concomitant rise in the conjugated fraction. The cause of significant haemolysis may be haemolytic anaemia, blood transfusion, resorption of haematomata, sepsis, or open-heart surgery.

In surgical patients with sickle-cell disease there are increased risks as acute haemolysis and severe pain can be precipitated by infection, dehydration, acidosis, and hypoxia. In the postoperative period the earliest signs of infection must be treated promptly, especially as these patients have splenic hypofunction and are susceptible to bacterial infection. Patients from the African continent, parts of Asia, the Arabian peninsula, and southern Europe should be screened for sickle-cell disease.

Patients with hereditary spherocytosis may also experience a haemolytic crisis following infection, and in the postoperative period this can cause an unconjugated hyperbilirubinaemia. The diagnosis is suggested by the family history or the presence of a raised mean corpuscular haemoglobin concentration, with more than 1 to 2 per cent of spherocytes on the blood film.

Surgery, infection, acidosis, and many drugs, including antibiotics and analgesics (see Table 3 105), may precipitate haemolysis in patients with glucose 6-phosphate dehydrogenase deficiency. At least 10 million people worldwide have this red cell enzyme deficiency, and thus patients from the Mediterranean, South East Asia, the Middle East, and West Africa should be screened preoperatively. The cresyl blue decoloration test or the methaemoglobin reduction test can be used in screening, and the diagnosis made by enzyme assay.

Pyruvate kinase deficiency is another red cell enzymopathy in which infection can precipitate haemolysis. Patients should be aware of their diagnosis but macrocytosis and an abnormal enzyme assay will confirm the diagnosis.

Causes of non-immune acquired haemolytic anaemia include disseminated intravascular coagulation, vasculitis, pneumococcal, meningococcal, and Gram-negative sepsis, Clostridium perfringens (was C. welchii) infection, burns, drowning, and some drugs. These are covered in other parts of this section.

Immediate and delayed haemolytic reactions may occur following blood transfusion. Within 24 h of the transfusion of one unit of stored blood at least 10 per cent undergoes haemolysis. Transfusion of two units of blood should not result in an increase in the serum bilirubin. If transfusion is rapid, massive, or occurs in a patient with impaired liver function, the capacity of the liver to conjugate bilirubin may be exceeded. Jaundice in this situation occurs 10 to 12 h after transfusion. Incompatibility of transfused blood may result in a severe immediate haemolytic reaction, which may occur if there are antibodies to the donated blood in the recipient's plasma. Jaundice appears at 12 h after commencing transfusion, peaks at between 24 and 36 h, and lasts for a total of 4 or 5 days.

Delayed haemolytic transfusion reactions are seen between 3 days and 3 weeks post-transfusion, with the peak reaction being at around 7 to 10 days. They are due to a secondary immune response, and in the majority of cases there has been sensitization to red cell antigens through past transfusion or pregnancy. This response is often to Rhesus and Kidd antigens, and is seen clinically as extravascular haemolysis with fever, jaundice, and anaemia. A serum sample should be screened for antibodies and future transfusions preceded by careful compatibility testing.

Large haematomata, crush injury, and bleeding from major vessels result in large pools of extravascular blood which, when resorbed, can result in an unconjugated hyperbilirubinaemia. As these patients often have hepatocellular dysfunction due to hypotension, hypoxia, and major surgery, as well as renal impairment, the severity and duration of jaundice may be marked. In a similar way, massive pulmonary infarction can cause hyperbilirubinaemia.

A massive haemolysis can occur in association with Clostridium perfringens (was C. welchii) infection 24 to 72 h after gastric, biliary tract, and colonic surgery. The typical clinical picture is of a restless, hypotensive patient, an acute rise in serum bilirubin, and crepitus around the wound site. Several causes of liver dysfunction probably occur simultaneously in these patients as the conjugated bilirubin level can be greater than the unconjugated. As these cases can be fatal, prompt treatment with massive doses of penicillin and hyperbaric oxygen are imperative. Meningococcal, pneumococcal, and Gram-negative sepsis can cause haemolysis through disseminated intravascular coagulation and secondary microangiopathic haemolysis. As sepsis can also cause intrahepatic cholestasis, a combination of the two factors may cause marked jaundice.

Early and late rises in bilirubin are seen after open-heart surgery. Early onset jaundice may be seen in up to 23 per cent of such patients and the main contributing factors are hypoxia, severity of right-heart failure preoperatively, and number of units of blood transfused. Although it has been suggested that cardiopulmonary bypass and prosthetic valves cause haemolysis, these are probably not significant contributors to the increased bilirubin. Late jaundice due to an autoimmune haemolytic anaemia has been reported where anaemia and jaundice, exacerbated by repeat transfusion, occur a few weeks after surgery. The presence of antiglobulin antibodies confirms the diagnosis; steroids are the treatment of choice.

Gilbert syndrome is low-grade chronic hyperbilirubinaemia found in 3 to 7 per cent of the population, with a male to female ratio of 2–7:1. It presents in the second or third decade as a raised level of unconjugated bilirubin. The serum bilirubin does not exceed 100 &mgr;mol/l and is usually less than 50 &mgr;mol/l. Liver function tests, routine haematology tests, and liver biopsy are normal. Haemolysis and acquired liver disease should be excluded and the diagnostic tests are: a two- to threefold increase in bilirubin induced by a 48-h fast and resolved by resumption of a normal diet; normal postprandial serum bile acids and reduced bilirubin UDP-glucuronyl transferase activity in a liver biopsy. The syndrome is probably inherited in an autosomal dominant fashion with variable penetrance.

The reduced bilirubin UDP-glucuronyl transferase activity is associated with reduced hepatic clearance of bilirubin and this, in combination with a possible mild compensated haemolytic state, is thought to cause the hyperbilirubinaemia. The main precipitating factor seems to be fasting, and it is primarily lipid withdrawal that is to blame. In the postoperative period, a raised bilirubin level may be noted particularly during episodes of vomiting, especially in pregnancy and febrile illnesses. Gilbert syndrome must be considered in the differential diagnosis of hepatic dysfunction in such patients.

Circulatory failure contributes to hepatic dysfunction in many surgical situations, although it is rarely the sole cause of the liver abnormality. Major trauma, burns, sepsis, massive blood loss, and surgery can be precipitants of ‘shock’, and these factors often occur together. In particular, gastrointestinal blood loss and septicaemia increase the risk of liver dysfunction when associated with hypotension. Cholestasis is the most common pattern of injury following hypotension, and this is a benign complication with a good prognosis. Prolonged hypotension, which is often associated with increased right atrial pressure, results in an ischaemic hepatitis, for example in open-heart surgery. There is an initial striking elevation of serum transaminases, up to 200 times the normal level, a marked decrease in prothrombin time, and a typically delayed bilirubin rise. These dramatic changes are seen within hours of surgery, and where no severe liver damage has occurred they revert rapidly to normal with restoration of liver blood flow and oxygenation. However, massive centrilobular hepatic necrosis can occur, and the ischaemic hepatitis can progress to fulminant hepatic failure, which has a high mortality rate. The clinical manifestation of hypoxic liver cell necrosis inevitably postdates the hypoxic event, and other causes, especially a viral hepatitis, must be considered.

Massive haemorrhage in combination with massive transfusion (for example, more than 20 units of blood) puts the liver particularly at risk of liver damage, should the patient survive. Major trauma patients are particularly at risk of this form of liver damage as well as that due to direct liver injury. In one study, 2 per cent of patients with major trauma and shock developed significant jaundice.

Patients with major burns form another group in which circulatory failure is an important factor in the aetiology of the associated hepatic dysfunction. Haemolysis often adds to the bilirubin load on the liver.

The normal liver usually tolerates hepatic artery ligation without significant sequelae unless the flow of portal-vein blood is inadequate because of vascular stricture and sepsis. Minimal derangement of bilirubin and alkaline phosphatase levels occur, and moderate increases in the transaminase levels in the first week may be the only consequence. Hepatic arterial collateral vessels develop very rapidly and this, in combination with the portal circulation, reduces the ischaemic insult. Extensive mobilization of the liver can involve division of the ligamentum and triangular ligament and if this precedes hepatic artery ligation, massive liver necrosis may result. If infarction occurs, the amounts of bilirubin and transaminases rise rapidly to high levels.

The first case of post-transfusion hepatitis was reported in 1885 after a patient received human serum. In the early 1980s the incidence of post-transfusion hepatitis in Europe and the United States was between 2 and 20 per cent. In 1991 it was suggested that the incidence in Italy may have fallen to around 5 per cent. The change in blood donor selection, surrogate testing for non-A, non-B post-transfusion hepatitis (with alanine aminotransferase and anti-HBc (antibody to hepatitis B core antigen) testing), and the modification of transfusion practice (less homologous and more autologous transfusions) have reduced the incidence. With an increase in the sensitivity and specificity of diagnosing hepatitis C virus (HCV) carriage in the future, the low risk of developing post-transfusion viral hepatitis will become extremely low. The viruses known to cause post-transfusion hepatitis are listed in Table 4 106.

The estimated number of carriers of HBV worldwide is 120 million, and in the United States of America there are at least 900 000 carriers. Transmission of HBV by blood transfusion has been reduced dramatically by the screening of donor blood, although today parenterally acquired HBV infection can still occur in transfused patients because genetic mutations in the virus result in falsely negative screening tests.

The clinical picture of acute HBV infection is very variable, from asymptomatic to fulminant hepatic failure. A fulminant fatal post-transfusion illness is seen in the elderly. Recovery without sequelae is the rule, but 5 to 10 per cent of infants and children with acute HBV infection develop chronic infection, as do 90 per cent of neonates. The mortality of hepatitis B infection is 1 to 3 per cent, and treatment with &agr;-interferon is available for selected patients with chronic HBV infection. However, post-exposure prophylaxis by administration of hepatitis B immunoglobulin and HBV vaccine is also available.

Prevention of post-transfusion hepatitis B is mainly through the screening of donated blood. Donor education, voluntary and not commercial donations, and attempts to discourage donors at risk of carrying transmissible diseases are other important measures. A combination of alanine aminotransferase and antihepatitis B core (anti-HBc) antibody assays have been used since 1986 to exclude many donors at risk of hepatitis C and human immunodeficiency virus infection as well as HBV infection. More recently, more specific tests have become available.

Hepatitis B surface antigen (HBsAg) was used to screen those infected with HBV, but more recently it has been shown that HBV variants exist due to mutations in the genome. Such mutations can result in the HBsAg test being negative and post-transfusion hepatitis B infection, including fulminant disease, has occurred. Screening for anti-HBc positivity should prevent this.

Hepatitis C virus is an RNA virus with homology to the flaviviruses and is the most common cause of post-transfusion hepatitis, accounting for up to 95 per cent of cases. Both plasma and cell products are infective. The infection rate of transfused patients is 7 to 10 per cent. The estimated prevalence of HCV infection among volunteer blood donors around the world varies from 0.2 to 1.7 per cent. For example, in New York it is 0.9 to 1.4 per cent; in western Europe, 0.7 per cent; in Japan, 1.5 per cent; and in Hungary, 1.7 per cent.

The incubation period is 5 to 12 weeks, although after transfusions of a particularly large volume of infected blood this may be shortened to between 1 and 4 weeks. Clinically, the disease is mild with only a moderate elevation of transaminase levels. Rarely, a fulminant hepatitis may occur which has a high mortality. Levels of bilirubin and the transaminases may rise for a second, or even third, time in the first few months of the infection. Diagnosis is mainly by exclusion as seroconversion may take up to 6 months or more. Tests for antibodies to HCV are widely available but the false positive rate is high. Repeat assays using different varieties of antibody, or the immunoblot technique as opposed to the enzyme-linked immunosorbent assay (ELISA), improve the specificity. The polymerase chain reaction (PCR) method will identify viral RNA in the serum and is the most accurate technique, but at the present time it is expensive and not widely available. Histologically, there are lymphoid follicles in the portal tracts, showing mild chronic inflammation bordering on chronic active hepatitis. There may be parenchymal inflammation and focal necrosis, as well as some fatty change. Chronic infection is also clinically mild and occurs in 85 per cent of patients or more. At least 41 per cent of patients develop chronic active hepatitis and 20 per cent develop cirrhosis. HCV also greatly increases the risk of development of hepatocellular carcinoma, four times more so than HBV. The time interval between infection and end-stage liver disease may be in the order of 25 years. Treatment of HCV-infected patients with &agr;-interferon has been proved to be effective, but the optimum timing of treatment in such a protracted disease process is still unclear.

Prevention was initially through donor education, exclusion of high-risk individuals as donors, and surrogate testing using serum aspartate transaminase levels and anti-HBc tests. Since September 1991 blood donors in the United Kingdom have been screened for HCV infection by methods involving the use of antibodies to viral components. Any positive blood is rejected despite the high false positive rate.

Hepatitis D virus only infects individuals already infected with HBV. The incubation period is 30 to 50 days and chronic infection is the most common outcome. Diagnosis is made by finding serum IgM anti-delta. Acceleration towards cirrhosis may occur when HBV-infected patients then acquire HDV infection. Prevention of post-transfusion HDV infection is the same as for HBV.

Human immunodeficiency virus, when acquired by transfusion, can cause a hepatic illness very similar to hepatitis B or C. The antibody appears between 1 and 2 months after exposure. The risk became apparent in 1982 to 1983 and was highest with unheated, non-pasteurized pooled plasma products (e.g. factor VIII). Appropriate treatment and screening of all donors now exists and should prevent infection in this way. Through aggressive donor education the incidence of HIV positivity among donors in the United Kingdom is 1 in 50 000.

Cytomegalovirus transmission commonly occurs after transfusion but infection is usually subclinical and benign. A glandular fever-like illness is typical at 2 weeks to 3 months after exposure. Clinically, fever, splenomegaly, jaundice, raised aminotransferase levels, and atypical lymphocytes may occur. However, massive hepatic necrosis and granulomatous hepatitis have been reported. Immunocompromised patients are at most risk and may also develop a fatal pneumonitis or disseminated infection. Virus can be cultured from saliva or urine and IgM antibody to CMV can be detected in the serum. In the United Kingdom, 50 to 60 per cent of the population is anti-CMV antibody positive. Only a few of these are infective but there is no readily available test for infectivity.

The EBV causes infectious mononucleosis or glandular fever and can be transmitted parenterally. Clinically there is fever, right upper quadrant pain, with or without pharyngitis, and lymphadenopathy. Hyperbilirubinaemia occurs in about 50 per cent, transaminases are raised to 20 times the normal level in up to 80 per cent, and one-third of patients have a raised alkaline phosphatase. The Paul–Bunnell or monospot test is usually positive and the diagnosis is made by finding a raised IgM anti-EBV capsid antibody. The sinusoids and portal tracts are infiltrated with large mononuclear cells. The histology may be similar to that for hepatitis A, B, or C. Fatal acute hepatic necrosis is rare and chronic hepatitis and cirrhosis do not occur. As infection with this virus is common, and infectivity cannot be specifically tested for, screening is not performed.

Patients incubating a hepatitic virus may come to surgery and anaesthesia. Postoperative deterioration in liver function frequently occurs and a mortality rate of 31 per cent in 36 such patients has been reported. Those patients who died had viral or alcoholic hepatitis. Thus viral serology should be part of the investigation of a patient with hepatitic liver dysfunction occurring early after surgery.

Many drugs used in the peri- and postoperative period have been associated with liver dysfunction. Almost every naturally occurring liver disease affecting man can be mimicked by the toxic effect of drugs on the liver. Drugs can affect bilirubin metabolism at any stage, causing hyperbilirubinaemia. The drug or its metabolite can be hepatotoxic or can precipitate a hypersensitivity reaction. Hepatocellular dysfunction may be due to cellular necrosis or intrahepatic cholestasis. Factors that increase the risk of drug-induced hepatic injury include liver disease, increasing age, female sex, and genetic polymorphism.

The list of potentially hepatotoxic agents is large and ever increasing. Some of the drugs used in the perioperative period are listed in Table 5 107. The general anaesthetic drugs are discussed separately. Alternative causes of liver dysfunction should be sought but there are often several potential candidates. A hepatitic picture must lead to exclusion of a viral aetiology, and the differentiation of intrahepatic and extrahepatic cholestasis is important. Liver biopsy will only rarely give a diagnosis. Diagnostic challenge is not recommended, as a severe reaction can occur and the mortality from drug hepatitis with jaundice is approximately 10 per cent.

Halothane, a haloalkane, was first introduced in 1956. Within 4 years there had been several reports of postoperative liver necrosis ascribed to halothane usage. The National Halothane Study gave the incidence of fatal hepatic necrosis as 1 in 35 000. Two subsequent, again retrospective, studies gave an incidence of between 1/6000 and 1/20 000.

Halothane hepatitis is associated with a 75 per cent incidence of multiple exposures, particularly where subsequent exposure is within 3 months. Female and obese patients are more at risk, and enzyme induction by other drugs further increases this risk. Liver injury occurs 1 to 15 days after exposure, and jaundice appears on approximately the seventh day, but may be later if liver injury follows the first exposure to halothane. There is fever, eosinophilia (in 8–32 per cent), arthralgia, and a non-specific skin rash. The transaminase levels are grossly elevated, for example 500–2000 IU/l, whereas the alkaline phosphatase level is often less than twice the normal level.

Histologically, the main feature is centrilobular necrosis, varying from multifocal spotty necrosis to massive necrosis. Ballooning degeneration of hepatocytes, inflammatory infiltrate, stromal fibrosis, fatty change, and occasionally granulomatous aggregates are also seen. Distinction from a viral hepatitis may be difficult.

Two types of halothane-induced hepatotoxicity are thought to exist: a mild subclinical form occurring in up to 20 per cent of patients exposed (abnormal liver function tests are the only manifestation) and a second, rare, fulminant form, due to severe necrosis which may be fatal. Five factors have been postulated in the pathogenesis of halothane hepatotoxicity: toxic products of metabolism, hypersensitivity, regional hepatic hypoxia, genetic predisposition, and altered hepatocellular calcium homeostasis.

Enflurane hepatitis has been proposed in 30 to 50 reports and, although these remain a subject of contention, the liver injury reported is similar to that induced by halothane.

Isoflurane has also been reported to cause liver damage. Whereas halothane undergoes 30 per cent biotransformation, with enflurane it is only 2 per cent and with isoflurane 1 per cent. Metabolic transformation is probably the key to haloalkane liver injury and the difference in the degree of biotransformation may explain the much lower incidence in the last two agents. Desflurane is the newest and most promising agent but is still being studied.

Since its advent in the 1960s, parenteral nutritional support has become safer, more reliable, and progressively more efficient. However, complications still occur and hepatobiliary abnormalities are the second most common problem, after catheter sepsis, which result in cessation of parenteral feeding. There are a number of different patterns of liver dysfunction that have been attributed to total parenteral nutrition, but the patients in whom these problems occur have several other concomitant risk factors for hepatic disease and it is often impossible to blame any one of these. The clinical picture, hepatobiliary dysfunction, liver histology, pathogenesis, and management of hepatic problems associated with total parenteral nutrition can usefully be discussed by comparing adults and infants.

Whereas the main hepatic problem in infants on total parenteral nutrition is cholestasis, in adults it is hepatocellular damage. In adults the presentation of liver injury is less severe and can be divided into abnormalities occurring during short-term therapy and those occurring during longer-term total parenteral nutrition. Fatty liver, intrahepatic cholestasis, and non-specific triaditis are features of short-term treatment, and steatonecrosis and chronic liver disease are seen with long-term feeding. When dextrose was the primary source of non-protein calories, quite dramatic rises in transaminase levels occurred. As the formulation of total parenteral nutrition has changed, so have the abnormalities seen. Since the introduction of lipid emulsions as an additional source of calories, abnormalities in liver function tests have become less frequent, a late and slow rise in alkaline phosphatase and bilirubin is the most frequent finding. These changes are most notable when excessive amounts of fat (e.g. more than 3 g/kg body weight) are given.

Fatty change is the most benign hepatic lesion seen within the first 14 days of total parenteral nutrition, and is often paralleled by a rise in the transaminase levels. Periportal fat infiltration is the first change, but this may progress to pan- or centrilobular infiltration. The rise in transaminase levels that usually accompanies these changes most commonly resolves spontaneously, even when total parenteral nutrition is continued. The lipid is mainly triglyceride, and the likely cause of its accumulation is increased hepatic synthesis combined with impaired export of triglyceride. Dextrose- and glucose-based feeds are clearly associated with fatty liver. Most total parenteral nutrition formulations should now contain lipid emulsions, and this has been shown to decrease the incidence of steatosis. Excess lipid will result in fat accumulation, but this is seen within the Kupffer cells and hepatic lysosomes. Essential fatty acid deficiency can result in a fatty liver and tryptophan degradation products have also been implicated. Primary systemic deficiency of carnitine also results in a fatty liver. Low systemic and hepatic carnitine levels have been found in patients on long-term total parenteral nutrition, with sepsis, and in stress states. Whether the inclusion of carnitine, a non-essential amino acid, in total parenteral nutrition would prevent or reverse the fatty change is not known. Malnutrition itself can produce a fatty liver, as can starvation and sepsis. Increased tissue release and increased hepatic uptake of free fatty acids, combined with increased synthesis of triglyceride, result in the fatty change.

Intrahepatic cholestasis occurs after more than 2 weeks of treatment, at a time when the transaminase levels are returning to normal and the bilirubin and alkaline phosphatase levels are starting to rise. The histological changes that accompany this cholestatic picture are bile-duct proliferation, canalicular bile plugging, centrilobular cholestasis, bile pigment within hepatocytes, and a periportal infiltration with granulocytes and lymphocytes. In the majority of cases the cholestasis resolves on discontinuation of feeding. Reduction in the dose of lipid infused will reduce the levels of bilirubin and alkaline phosphatase. In one study, cholestasis was reported in 14 of 27 patients on total parenteral nutrition, and it was suggested that cholestasis in these patients predisposes them to cholelithiasis. Patients on total parenteral nutrition for more than 6 weeks were also shown to develop biliary sludging, as shown by ultrasound scanning. Normal feeding for 4 weeks returned the bile to normal. Progressive liver disease has not been seen in adults receiving total parenteral nutrition for up to 6 months.

The pathogenesis of the cholestasis remains unclear, but several factors have been implicated. The underlying condition and nutritional status of the patient, as well as the infection and treatment (including transfusion, surgery, and drugs), may also play a role. Various amino acids and toxins have been blamed. Deficiency of taurine may be a factor in as much as taurine is involved in the metabolism of bile acids and may prevent the accumulation of lithocholic acid. This secondary bile acid has been found in excess in patients with both inflammatory bowel disease and cholestasis associated with total parenteral nutrition, and has been implicated as a hepatotoxin. Metabolic products of tryptophan have been suggested as toxins. Lipid emulsion in doses above 3 g/kg.day have been shown to cause intrahepatic cholestasis; however, a further study giving 3.5 g/kg.day of fat was not accompanied by cholestasis. Decreased hepatic biliary flow due to reduced neural and hormonal stimulation from a rested bowel may contribute to this pattern of cholestasis. Bacterial overgrowth secondary to stasis or intestinal surgery and portal endotoxaemia are possible additional factors in these patients.

Non-specific triaditis occurs in patients with inflammatory bowel disease receiving total parenteral nutrition; this relatively minor change is accompanied by very marked derangement of liver function tests. This group of patients may have liver abnormalities that predate the total parenteral nutrition, and these may have been exacerbated by the parenteral nutrition.

Chronic progressive liver disease in adults on total parenteral nutrition is rare but has been demonstrated in patients receiving treatment for more than 6 or even 12 months. Such long-term total parenteral nutrition is given to patients who often require multiple transfusions and hepatotoxic drugs, and in whom the underlying disease is chronic, complicated, and may have involved extensive and repeated surgery. Again, it is impossible to isolate total parenteral nutrition as the only factor causing liver problems. Three studies give an incidence for progressive liver disease of between 5 and 15 per cent. The changes seen in short-term total parenteral nutrition were also seen in many of these patients, and beyond 6 months' therapy a cholestatic picture of liver function was the common finding. Histologically the common changes seen in these studies were cholestasis, hepatocyte necrosis, an alcoholic hepatitis-like picture, steatonecrosis, and early cirrhosis. The pathogenesis could involve any or all of the factors mentioned under shorter-term total parenteral nutrition, as the limited number of cases makes studies difficult. Total parenteral nutrition is the only source of nutrition in these patients, and juggling with the constituents of the feed formula and cyclical feeding are often the only therapeutic options. The prognosis may be poor in patients with short-bowel syndrome.

Nutritional support of premature infants is now the most common use of total parenteral nutrition in paediatric medicine, and liver dysfunction has been noted in this group of patients throughout the development of this therapy. Children with chronic and/or extensive gastrointestinal disease make up the majority of the remainder. Infants receiving total parenteral nutrition are at risk of five types of hepatobiliary disease: fatty change, intrahepatic cholestasis, biliary sludging, gallstones, and acute acalculous cholecystitis.

Early in the course of total parenteral nutrition a liver biopsy may demonstrate mild fatty change and hydropic swelling of hepatocytes. These changes are not precursors of cholestatic disease, they occur infrequently and are reversible on stopping the total parenteral nutrition. In animal and human studies the data suggest that excess carbohydrate is to blame. However, deficiency of essential fatty acids may be another cause.

The first reported case of cholestasis related to total parenteral nutrition was in a premature infant in 1971. Most premature infants are at risk of heart failure, sepsis, necrotizing enterocolitis, and treatment with blood transfusion and multiple drugs. All of these may cause hepatobiliary problems and thus it is difficult to ascribe hepatic complications to total parenteral nutrition alone. The incidence of cholestasis increases with duration of total parenteral nutrition and also with decreasing gestational age and birth weight. In 1979 it was found that cholestasis occurred in 23 per cent of infants with respiratory problems receiving total parenteral nutrition. If treatment was for more than 60 days, the incidence rose to 60 per cent and to 90 per cent after 80 days. Duration of therapy is often a function of the gestational age, as is birth weight, and these are therefore not independent variables. Immature hepatic function may be the primary factor in the liver disease seen in infants.

Monitoring of premature infants on total parenteral nutrition is important and the commonly used ‘liver function tests’ are not useful. A rise in unconjugated bilirubin occurs late in the course of cholestatic disease in these patients and total bilirubin is commonly raised in premature infants with no liver disease. Alkaline phosphatase is unhelpful as there is a preponderance of the bone isoenzyme in infants, and this isoenzyme is also affected by the child's nutritional status. Transaminase levels can also be unreliable as the levels do not correlate with the degree of cholestasis associated with total parenteral nutrition. &ggr;-Glutamyl transpeptidase is the most sensitive test, but lacks specificity. Serum bile salt levels have also been found to be good indicators of cholestasis, but there is a normal developmental delay in bile salt metabolism, giving rise to elevated levels. A combination of &ggr;-glutamyl transpeptidase and serum bile salts may be a more specific test. Ultrasound scanning and endoscopic retrograde cholangiopancreatography (ERCP) will help to exclude extrahepatic cholestasis and it may also be necessary to perform a liver biopsy. Although the liver histology may be non-specific, a biopsy can identify other causes of cholestasis, such as cytomegalovirus hepatitis or extrahepatic obstruction. It can also indicate the urgency of cessation of the total parenteral nutrition and the prognosis of the liver pathology. Cholestasis early in treatment with total parenteral nutrition is seen as bile pigment in the hepatocytes, bile plugs in canaliculi, pseudorosette formation, and a varying degree of triaditis, with or without eosinophils. Cholestasis with longer-term total parenteral nutrition may show portal and lobular fibrosis, expansion of the portal tracts with portal ductular proliferation, and bile plugs in the interlobular ducts. Patients with intrahepatic cholestasis in whom total parenteral nutrition is continued may develop micronodular cirrhosis. The clinical cholestasis that reverses on stopping the total parenteral nutrition is accompanied by reversal of most of the histological findings. Some portal fibrosis may persist and where cirrhosis occurs the prognosis is poor, liver failure being the most common cause of death in these patients.

The pathogenesis of this cholestasis is still unclear, but there are many candidates. As has been stressed previously, it is often difficult to separate several risk factors for hepatic injury in the population of patients on total parenteral nutrition. Sepsis, transfusion, drug therapy (including anaesthesia), and the underlying disorder may all contribute. In premature infants there is, in addition, immature hepatic function: there is decreased hepatic uptake of bile salts, bile salt synthesis is reduced, ileal uptake of bile salts is inefficient, and the ability to detoxify lithocholic acid is impaired. These factors may make the premature infant particularly susceptible to hepatic injury. Hepatotoxins of various types have been suggested and most of the contents of total parenteral nutrition have been implicated.

Biliary sludging is seen in children and adults on long-term total parenteral nutrition and in 1983 it was shown that up to 100 per cent of adults receiving more than 6 weeks' total parenteral nutrition develop ‘sludge’. This thick bile may be responsible for varying degrees of extrahepatic cholestasis, and liver histology may show bile plugging of interlobular ducts. Two infants have been described with refractory cholestasis that improved after surgical disimpaction of biliary sludge. Thus, an infant with cholestasis during total parenteral nutrition should also be investigated for an extrahepatic cause, with ultrasound scanning, computerized tomographic scanning, ERCP, possibly liver biopsy, and biliary decompression if necessary.

Gallstones can occur in infants and adults on total parenteral nutrition. In 1990 a study reported that, over a 3-year period, five of seven infants with choledocholithiasis were premature and had sepsis and/or were receiving total parenteral nutrition. Even in infants, ERCP is an effective therapeutic approach. Biliary sludging, cholelithiasis, and acalculous cholecystitis associated with total parenteral nutrition are more thoroughly discussed in relation to adults (see above).

The majority of abnormalities are minimal and self-limiting and may require no therapy. The only effective means of reversing cholestasis in these patients is to replace total parenteral nutrition with oral feeding. Infants, in particular, may be dependent on parenteral feeding, and in such patients there are several important factors to consider. Liver disease due to drugs or infection should be excluded as well as any extrahepatic obstruction. The latter may need urgent correction. If hepatic dysfunction persists and no other cause is found, the carbohydrate content of the feed should be adjusted. The Harris and Benedict equation is used to predict the patient's carbohydrate need, but this is not accurate and does not account for the ‘stress state’ (for example nutritional state, tissue damage, infection) of the patient. Glucose utilization has been shown to be maximal at around 5 to 6 mg of dextrose/kg.min, and this is probably the maximum dose above which lipogenesis occurs. Overfeeding should also be excluded and corrected. Any calorie input lost in reducing the dextrose delivered is replaced by protein or fat. Fat should supply approximately 30 per cent of the required calories but the total quantity should not exceed 3 g of fat/kg.day. In feed formulations that do not contain fat, an essential fatty acid deficiency may be the cause of the liver dysfunction, and a lipid emulsion should be introduced. Four per cent of the daily non-protein calories supplied by linoleic acid will correct this deficit. As discussed above, there are several amino acids that, when added to the feed formulation, may help to prevent the hepatic problems. If the quantity and source of calories is appropriate, then ‘cyclic’ nutrition can be tried. This involves giving all total parenteral nutrition over a 10- to 12-h period, with no calories given during the remainder of the 24 hours. This has been shown to prevent or reverse the liver disease. In patients with increasing cholestasis, in whom total parenteral nutrition cannot be stopped, copper, manganese, and aluminium should be withheld as they are normally excreted in the bile and may accumulate in the liver and basal ganglia, causing permanent damage.

Mild hyperbilirubinaemia can be precipitated by fasting and is due mainly to an unconjugated bilirubin rise. The majority of patients showing this effect are probably those with Gilbert syndrome. Fatty change is also seen, particularly in acute weight loss or starvation. This is related to the increase in serum fatty acids and increased fatty acid turnover precipitated by decreased availability of glucose, a rise in glucagon levels, and increased sympathetic nervous activity. Obese subjects who lose weight rapidly may show a transient elevation of serum liver enzymes.

Fatty change in the liver is seen in up to 50 per cent of obese subjects, with occasional periportal inflammation and fibrosis. Steatonecrosis and cirrhosis have been reported but this may be due to coexistent diabetes mellitus or alcoholic liver disease. Fifty per cent of obese patients can be shown to be glucose intolerant, and this and excess dietary fat and carbohydrate in relation to protein intake may be involved in the aetiology of steatosis. The fatty infiltration is perivenular and diffuse. Liver function tests may be abnormal and reflect more severe histological change. The changes are, in general, benign and non-progressive, and can be reversed by weight loss.

Diabetic patients also show fatty change in the liver; the majority are non-insulin-dependent diabetics who are also overweight. Steatosis is very rare in juvenile-onset insulin-dependent patients. Symptoms are rare, an enlarged, slightly tender liver may be found on examination, and liver function tests may be slightly deranged in about 20 per cent of diabetics, but do not correlate with histology. The fatty change is centrilobular and diffuse. Weight loss and good diabetic control will resolve these abnormalities.

Steatonecrosis may also occur and this is seen in the non-insulin-dependent group. It has been suggested that the incidence of cirrhosis among diabetics is twice that of the general population. This suggestion is unproven and may originate in the number of cirrhotic patients with glucose intolerance that have wrongly been classified as primary diabetics.

Emergency biliary surgery in diabetic patients has a higher than expected mortality. This is due in part to the disruption of glucose control caused by surgery, the increased risk of infection due to leucocyte dysfunction, and poorer wound healing.

Sepsis can produce a deep jaundice, which may be cholestatic and occurs 2 to 4 days after the onset of bacteraemia. In 1969 a study showed an incidence of jaundice of 0.6 per cent in 1140 patients with bacteraemia; however, hepatic dysfunction was shown in 1979 to have little effect on survival when compared with the primary infection. Pneumonia, Gram-negative bacteraemia, intra-abdominal abscess, and pyelonephritis can all cause a raised bilirubin. Gram-negative infection in infants frequently causes cholestasis. Endotoxins may be the main culprit as they have been shown to inhibit bile secretion, although cytokines have been implicated more recently. As in most cases of hepatic dysfunction discussed here, sepsis is rarely the only factor that can be implicated and thus the aetiology may be multifactorial. Biochemically and histologically, the changes are very similar to those observed with circulatory failure, with a moderate rise in conjugated bilirubin, aminotransferases, and alkaline phosphatase levels. However, an increase in the unconjugated bilirubin level also occurs, giving a rise in total bilirubin out of keeping with the increase in liver enzymes. Hepatic histological changes include biliary stasis, fatty change, and periportal inflammation. Extrahepatic biliary obstruction must be excluded. Pneumococcal, meningococcal, and Gram-negative sepsis may cause haemolysis by disseminated intravascular coagulation or a secondary microangiopathic haemolysis, and in these conditions the rise in unconjugated bilirubin will be prominent.

‘Benign postoperative intrahepatic cholestasis’ is unlikely to be a specific entity. It occurs in situations where blood loss is a prominent problem and is probably due to a combination of hypotension and multiple blood transfusions. Caroli in 1950 was the first to describe the occurrence of postoperative cholestatic jaundice. Benign postoperative intrahepatic cholestasis has been included in all lists of causes of postoperative jaundice since about this time. The aetiology of postoperative cholestasis is discussed within this section, and the majority of cases given this label in the past now have a definable cause.

Bile-duct injury can follow cholecystectomy, common bile duct exploration, or any upper abdominal operation. If unrecognized at operation, jaundice, biliary fistula, or biliary peritonitis will occur in the early postoperative period. Prompt surgical repair helps to prevent permanent liver damage.

Retained common bile-duct stones after cholecystectomy and/or exploration of the common bile duct are uncommon. In the majority of cases, ERCP will both diagnose and treat this problem by sphincterotomy. Reoperation is required if ERCP fails or is not available. Some practitioners advocate visualization and, if necessary, clearance of the bile duct at ERCP prior to cholecystectomy. Occasionally, blood may collect in the common bile duct and cause obstruction.

Acute postoperative pancreatitis is uncommon and the cause is unknown. Thirty per cent of patients may be jaundiced, and oedema of the head of the pancreas is thought to result in some degree of obstruction and a low-grade hyperbilirubinaemia.

Acute non-calculous cholecystitis can occur after major trauma, burns, surgery that does not involve the upper abdomen, and in patients receiving long-term total parenteral nutrition, especially infants. It accounts for about 1 per cent of all cases of cholecystitis. A Japanese series of acalculous cholecystitis after gastrectomy demonstrated an incidence of 0.64 per cent. The aetiology is unknown, but it has been suggested that biliary stasis is important.

Postoperative cholecystitis occurs most commonly in the fifth to seventh decade, but in patients with trauma or burns this form of cholecystitis is seen most frequently in the second to fourth decade. The sex ratio is also different in these two groups; females predominate in the former, males in the latter. The postoperative form tends to follow a major surgical procedure. This form of cholecystitis can occur up to 1 month after the operation. Right upper quadrant pain and tenderness is usually accompanied by nausea, vomiting, and fever. The observed bilirubin rise is variable but may be up to 85 &mgr;mol/l; levels of transaminases and alkaline phosphatase are only mildly raised. Ultrasound may show enlargement of the gallbladder and, by definition, no gallstones are seen. ERCP is often necessary to exclude other causes of obstruction, although surgery should not be delayed in these already seriously ill patients. Histologically, the gallbladder shows vascular dilatation, congestion, and oedema in all layers, without fibrosis. Abscesses of varying size may be seen in the gallbladder wall and the mucosal surface is necrotic and ulcerated. Perforation is frequent.

Acalculous cholecystitis occurs in patients receiving total parenteral nutrition for more than 3 months with an incidence of 4 per cent.

The level of mortality has been given as between 33 and 75 per cent. However, this may pertain only to major trauma patients and reflects the already much increased mortality in this group.

Patients with liver disease may require surgery as a consequence of the disease or for unrelated problems. Any surgery involving the liver itself, the biliary system, or the blood vessels associated with the liver has the potential to cause hepatocellular dysfunction, bleeding problems, nutritional difficulties, and infection. Where there is impaired liver function in such patients the problems discussed in this section become doubly important.

Anaesthesia and surgery in a patient with pre-existing liver disease have the potential to cause deterioration in liver function and even acute liver failure. Peri- and postoperative haemorrhage and postoperative infection are the other major complications. Factors that should be considered in these patients are the nature of the liver disease, the preoperative assessment of the liver disease, fluid balance, renal function, drug metabolism and adverse effects, nutrition, maintenance of liver blood flow, and prevention of complications.

The nature and severity of the liver disease should be confirmed before surgery whenever possible, as there are specific risk factors related to the underlying liver pathology.

In patients with known liver disease, assessment of hepatic function is important preoperatively. Standard liver tests (aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, &ggr;-glutamyl transferase) do not give a good guide to hepatic function. Serum bilirubin in cirrhotic patients can be used as a predictor of dysfunction. Serum albumin and liver-dependent clotting factor measurement will assess the synthetic capacity of the liver. The indocyanine green clearance test can also be used to assess function and hepatic blood flow. As haemorrhage and infection are important complications in these patients, platelet count, haemoglobin level, white blood-cell count and differential, blood grouping, and cross-matching of sufficient fresh blood is essential. Serum urea estimation is often unreliable in assessing renal function in liver disease, as hepatic dysfunction causes a low urea. Electrocardiogram, chest radiograph, and screening for any infection are also mandatory. Further specific tests may be required and are mentioned below as appropriate.

(1)correction of coagulopathy to normal by administering vitamin K and fresh frozen plasma;

(2)improving the nutritional status;

(3)treatment of renal impairment;

(4)treatment of infection;

(5)control of ascites.

The nutritional status of patients with liver disease is discussed separately below.

The HBV- or HCV-positive patient that comes to surgery is at risk of postoperative deterioration in liver function and is also a risk to all staff involved in his or her care. Hepatitis B surface antigen assay is a good screening test but anti-HBc positivity is more reliable in atypical cases (see above). Anti-HCV antibody can be detected and if specific testing for HCV infection by PCR is not freely available, positivity should be confirmed by alternative antibody assay techniques. Both HBV and HCV infection are associated with periods of mild, or even subclinical, liver disease, and both can result in chronic liver disease and cirrhosis. Liver function and associated complications of chronic liver disease must therefore be identified prior to surgery.

Surgery should be avoided if possible in acute hepatitis as acute liver failure may be precipitated, liver function is unpredictable, and the specific cause may not yet be identified (for example, viral serological tests may not become positive until some weeks after the acute illness). One study reported a 61 per cent morbidity and 31 per cent mortality following surgery in patients with undiagnosed acute liver disease. The deaths from hepatic failure occurred in those patients with viral or alcoholic hepatitis.

Patients with obstructive jaundice are at risk of postoperative renal failure, haemorrhage, and deterioration in liver function. It has been shown that patients with liver disease and hyperbilirubinaemia have abnormal renal structure and function, abnormal circulatory haemostasis, and deterioration in the gastrointestinal barrier to infection, and all these contribute to the postoperative risk of renal failure. Increased levels of unconjugated and conjugated bilirubin and bile salts damage the middle segment of the proximal convoluted tubule, producing changes very similar to those caused by anoxia. Decreased creatinine clearance occurs preoperatively in 30 per cent of patients with obstructive jaundice, and this is seen particularly in patients with coincident sepsis. This is due to decreased renal blood flow, which is more marked in the cortex and results in impaired ability of the kidney to concentrate urine, as well as susceptibility to sodium and water depletion. The cause may be decreased sensitivity to catecholamines. Patients with obstructive jaundice have lowered peripheral vascular resistance, renal salt wasting, some loss of left ventricular function, and pooling of blood in the splanchnic bed. Thus, a small volume of blood loss can result in a marked fall in arterial blood pressure, which will obviously exacerbate any renal failure. Endotoxins have been found in the circulation of 50 to 70 per cent of patients with obstructive jaundice, and these toxins also cause defects in renal structure and function. In particular, they increase renal vascular resistance, cause endothelial swelling, fibrin deposition, and low-grade disseminated intravascular coagulation. Bile salts disrupt endotoxins in the gut lumen but in obstructive jaundice this is prevented and endotoxins readily reach the portal circulation. Hyperbilirubinaemia, bile salt retention in the liver, and the endotoxins impair the reticuloendothelial phagocytotic function of the Kupffer cell system and endotoxins enter the general circulation. Table 6 108 lists the parameters that help identify the patient at risk of postoperative renal failure.

The principles of preoperative treatment of these patients include the treatment of sepsis, avoidance of nephrotoxic drugs, treatment of renal impairment, and the correction of hypovolaemia, hypoalbuminaemia, hyponatraemia, and anaemia. Preoperative renal failure will require correction of fluid balance, administration of antibiotics, and dialysis if necessary.

Chronic liver disease can be associated with adequate liver function where there is no hyperbilirubinaemia, hypoalbuminaemia, or coagulopathy. If infection and renal impairment are also excluded, specific preoperative treatment may not be necessary. Perioperatively, particular consideration of fluid balance, ventilation, liver blood flow, and drug metabolism is important. There is still a risk of liver decompensation, haemorrhage, and infection postoperatively. Peri- and postoperative management are discussed below.

Cirrhosis may be associated with well-compensated liver function, assessed as above. However, surgery in cirrhotic patients has a high mortality, reported to be 80 per cent in patients with advanced cirrhosis undergoing cholecystectomy. Child's classification is a useful guide to the preoperative assessment of risk in these patients, which is made more accurate by the addition of the prothrombin time (see Table 7 109). Prothrombin time, serum albumin, and the presence of infection (white blood-cell count >10 000/mm³) are regarded as the most useful indicators of risk. The risk is further increased when these patients undergo gastrointestinal surgery and is highest following hepatobiliary surgery.

Cirrhosis is also associated with renal abnormality. The latter is accompanied by proteinuria and haematuria. Renal failure may occur in 50 to 75 per cent of cirrhotics. It may be prerenal, due to acute tubular necrosis, or it may have no obvious cause, when it may be labelled the hepatorenal syndrome. Preoperatively these patients should be assessed daily for weight change, ascites, oedema, and pyrexia. The sodium balance must be estimated; urine volumes, urine and serum osmolarity, serum creatinine, urea, sodium, potassium, and albumin must be checked; and haemoglobin, haematocrit, and liver function must be tested. Deranged coagulation will require daily vitamin K injections, but where hepatocellular function is poor these may not correct the abnormalities and fresh frozen plasma will be needed to cover any invasive tests or bleeding episodes as well as perioperatively.

Infection, hypoalbuminaemia, and ascites should be treated. The last should be treated by nutritional support (see below) along with paracentesis and albumin infusion. Urinary catheterization and a central venous catheter are essential but, again, infection must be avoided.

Perioperative malnutrition increases the morbidity and mortality of any operation. In the patient with liver disease, malnutrition will compound the already significant risk of complications. Obstructive jaundice results in malabsorption of fat and steatorrhoea. If obstruction is prolonged, malnutrition will develop. Inevitably, chronic parenchymal liver disease is associated with protein–calorie malnutrition. Glucose intolerance is seen in 50 to 80 per cent of cirrhotics, deficiencies of vitamins A, C, and E, folic acid, and zinc result, and there is a deficiency of branched-chain amino acids. These, and the nature of the liver disease, result in compromised immune defence, and postoperative infection and wound dehiscence are common problems.

Malnutrition can be identified from a history of weight loss, the patient's height to weight index, body composition, protein turnover, serum albumin, urinary urea, and the immune status, as well as a number of other detailed tests.

Nutritional support can be given as oral supplements, enteral or parenteral feeding. Oral feeding is obviously the most efficient, cheap, and safe, when it is appropriate. Enteral feeding is well tolerated in obstructive jaundice. Special formula feeds are available for patients with chronic liver disease. They contain a high proportion of branched-chain amino acids which may be deficient and which also help reduce protein catabolism, another factor in the malnutrition in these patients. When such patients require surgery, parenteral feeding is often the only method available. Extensive investigation and preparation often dictates periods of starvation preoperatively and may also be a factor delaying surgery. Parenteral feeding regimens need to be tailored to the patient. If there is no glucose intolerance, glucose solutions can be the major source of calories; but if sugar intolerance exists, fat and protein content must be increased. Lipid emulsions seem to be well tolerated by patients with liver disease. Insulin resistance may be a problem, particularly in patients with infection. In patients with well-preserved liver function, and no past or presenting encephalopathy, standard formulations are acceptable. If there is a risk of encephalopathy, or if it already exists, a mix containing less phenylalanine, tyrosine, tryptophan, and methionine, and more arginine and branched-chain amino acids is recommended. Vitamins A, D, E, K, and C, folic acid, zinc, and copper must be regularly supplemented.

The maximum benefit to be gained from preoperative parenteral feeding comes from 10 to 14 days of therapy. The aim should be to correct the amino acid profile, reach a positive nitrogen balance, and scrupulously avoid infection. Renal failure is often a further problem, volume and electrolytes must be carefully monitored, and the formula may need to be adjusted further. If the patient is requiring haemodialysis or haemoperfusion, this will assist the control of fluid balance. In obstructive jaundice, if there is no hepatocellular dysfunction, standard supplemented oral, enteral, and parenteral feeding regimens can be used. Obstructive jaundice should not be left untreated any longer than necessary (see above).

Monitoring is vital and should consist of at least 8-hourly tests of blood sugar; daily weight, urea, creatinine, and electrolyte measurements; biweekly determinations of calcium, phosphate, liver function, and albumin; and weekly measurement of zinc and magnesium levels and culture of blood, urine, and any other drained fluid. The duration of feeding should be reviewed constantly.

The principal intraoperative concern is anaesthetic technique. Basic principles pertain to patients with liver disease, but protein binding, metabolism, and liver blood flow are particularly important considerations. An experienced anaesthetist should be employed where liver disease and hepatobiliary surgery are combined. In patients with well-compensated liver disease, standard drugs may be appropriate. The premedication may be affected by albumin levels, and the response to opiate drugs at this time may be helpful in assessing the dose requirements postoperatively. Narcotic analgesics and the benzodiazepines may have very prolonged action in patients with hepatocellular dysfunction.

During surgery there are several specific problems. Periods of hypotension or hypoxia will compromise liver function and precipitate renal failure more readily than in other patients. There may be derangement of blood coagulation. Rapid or large-volume transfusions of blood products may be required. Hepatobiliary disease and renal impairment will modify most drug pharmacokinetics and dynamics.

Of the haloalkanes, isoflurane is the best agent to use in these patients (see above); however, inhalation agents should be avoided. It should also be noted that prolonged use of nitrous oxide may cause additional liver damage. Liver blood flow, hepatocellular dysfunction, and plasma protein concentration are the factors affecting anaesthetic drug kinetics and dynamics intraoperatively. All anaesthetic techniques cause a reduction in hepatic blood flow, as do hyperventilation, increased sympathetic tone, and surgical manipulation of the liver and abdominal viscera. Good liver perfusion is also important for the clearance of endotoxins and lactate, and metabolic acidosis is another potential problem. A fall in liver blood flow will reduce the clearance of drugs and enzyme activity intraoperatively and the potential for a postoperative deterioration in liver function will be increased.

Control of ventilation must also be precise, as hypoxia, acidosis, and hyperventilation can all affect liver and renal function adversely, both intra- and postoperatively. Central venous catheterization, a urinary bladder catheter, and an arterial line are mandatory. In patients with cirrhosis and renal impairment, sodium-containing fluids should be used judiciously. Fresh blood and fresh frozen plasma must be readily available, and in long operations electrolytes, blood sugar, and coagulation will need to be monitored regularly.

The importance of avoiding infection during all anaesthetic and surgical techniques cannot be stressed too often.

The immediate postoperative concern is whether to reverse the anaesthetic or whether ventilatory support should be continued in an intensive care unit. Reasons to continue ventilation in these patients include:

(1)the persistent effect of anaesthetic and/or neuromuscular blocking agents—in obstructive jaundice, neuromuscular blocking drugs will have prolonged action;

(2)massive intraoperative blood loss and transfusion;

(3)anticipation of further bleeding;

(4)any episode of cardiac arrest during the operation;

(5)significant preoperative lung disease;

(6)failure to maintain adequate arterial oxygenation;

(7)Post-sternotomy, cardiopulmonary, or left-heart bypass.

Regular assessment of liver function in the postoperative period will allow the early identification of deterioration. All the factors discussed in the first part of the section pertain to these patients also.

Central venous catheters should be left in place to monitor fluid balance, especially in relation to risk of bleeding and renal failure. Haematological monitoring must include daily coagulation testing and the prompt correction of abnormalities. Intravenous vitamin K should be given daily.

Constant vigilance against infection is mandatory in postoperative patients with liver disease. All catheters and drainage systems must be watched and handled by experienced staff. Blood, urine, and drained fluids must be monitored regularly for infection. Cirrhotic patients are particularly prone to chest infection, and chest physiotherapy should be rigorous in these patients. Consideration of peritoneal infection is important, as this can be an occult cause of deteriorating liver function. Biochemical monitoring should be routine as renal function is at risk. When haemorrhage, hypotension, and infection coexist, small volumes of concentrated urine, with a urine to plasma osmolarity ratio greater than 1.05, indicate potential renal failure. This is reversible if recognized and treated promptly. Established renal failure can be oliguric or high output.

Upper abdominal, subcostal, and sternotomy incisions are associated with severe postoperative pain, which is made worse by prolonged procedures. Patients with liver disease often require these operative procedures, and thus pain relief is particularly important. Septicaemia and coagulation abnormalities may limit the administration of analgesia by epidural methods. Doses of other intravenous or inhalational analgesics should be low initially, and they must be monitored carefully. The duration of action of benzodiazepines and narcotic analgesics can be very prolonged in patients with hepatocellular disease.

The surgical patient with normal preoperative liver function

Barbara JA, Contreras M. Non-A, non-B hepatitis and the anti-HCV assay. Vox Sang 1991; 60(1): 1–7.

Clark RJS, Doggart JR, Lavery T. Changes in liver function after different types of surgery. Br J Anaesth 1976; 48: 119–22.

Du Priest R, Khaneja S, Cowley A. Acute cholecystitis complicating trauma. Ann Surg 1979; 189: 84–6.

Flint LM Jr, Polk HC Jr. Selective hepatic artery ligation: limitations and failures. J Trauma 1979; 19: 319–21.

McIntyre N, Benhamou J-P, Bircher J, Rizzetto M, Rodes J, eds. Oxford textbook of clinical hepatology. Oxford: University Press, 1991.

Nunes G, Blaisdell FW, Margaretten W. Mechanism of hepatic dysfunction following shock and trauma. Arch Surg 1970; 100: 546–56.

Powell-Jackson P, Greenway B, Williams R. Adverse effect of exploratory laparotomy in patients with unsuspected liver disease. Br J Surg 1982; 69: 449–51.

Ray DC, Drummond GB. Halothane hepatitis, Br J Anaesth 1991; 67: 84–99.

Saidi F, Donaldson GA. Acute pancreatitis following distal gastrectomy for benign ulcer. Am J Surg 1963; 105: 87–96.

Sirchia G, et al. Prospective evaluation of post-transfusion hepatitis. Transfusion 1991; 31: 299–302.

Takahushi T, Yamamura T, Utsunomiya J. Pathogenesis of acute cholecystitis after gastrectomy. Br J Surg 1990; 77: 536–9.

Utili R, Abernathy CO, Zimmerman HJ. Endotoxin effect on the liver. Life Sci 1977; 20: 553–68.

Zimmermann HJ. Even isoflurane (editorial). Hepatology 1991; 13(6): 1251–3.

Zimmerman HJ, Fang M, Utili R, Seeff LB, Hoofnagle J. Jaundice due to bacterial infection. Gastroenterology 1979; 77: 362.

Total parenteral nutrition

Allardyce DB. Cholestasis caused by lipid emulsion. Surg Gynecol Obstet 1982; 154: 641–3.

Beale EF, Nelson RM, Bucciarelli RL, Donnelly WH, Eitzman DV. Intrahepatic cholestasis associated with parenteral nutrition in premature infants. Paediatrics 1979; 64: 342–7.

Bowyer BA, Fleming CR, Ludwig J, Petz J, McGill DB. Does long-term parenteral nutrition in adult patients cause chronic liver disease? J Parenteral Enterol Nutr 1985; 9: 11–17.

Craig RM, Neumann T, Jeejeebhoy KN, Yokoo H. Severe hepatocellular reaction resembling alcoholic hepatitis and cirrhosis after massive small bowel resection and prolonged total parenteral nutrition. Gastroenterology 1980; 79: 131–3.

Fisher RL. Hepatobiliary abnormalities associated with total parenteral nutrition. Gastro Clin N Am 1989; 18(3): 645–66.

Fouin-Fortunet H, Le Quernec L, Erlinger S, Lerebours E, Colin R. Hepatic alterations during total parenteral nutrition in patients with inflammatory bowel disease: a possible consequence of lithocholate toxicity. Gastroenterology 1982; 82: 932–7.

Grant JP. Effects of cyclic vs. continuous TPN in hospitalized patients. American Gastroenterology Association Postgraduate Course. Nutrition in Gastroenterology. 1987.

Grant JP, et al. Serum hepatic enzyme and bilirubin elevation during parenteral nutrition. Surg Gynecol Obstet 1977; 145: 573–80.

Jeejeebhoy KN, Anderson GH, Nakhooda AF, Greenberg GR, Sanderson I, Marliss EB. Metabolic studies in total parenteral nutrition with lipid in man. Comparison with glucose. J Clin Invest 1976; 57: 125–36.

Jonas A, Yahav J, Fradkin A, Kessler A. Choledocholithiasis in infants: Diagnostic and therapeutic problems. J Pediatr Gastroenterol Nutr 1990; 11(4): 513–17.

Messing B, De Oliveira FJ, Galian A, Bennier JJ. Cholestase au coeurs de la nutrition parenterale totale; mise an evidence de facteurs favorisants; association à une lithiase vesiculare. Gastroenterol Clin Biol 1982; 6: 740–7.

Rabeneck L, Freeman H, Owen D. Death due to TPN-related liver failure. Gastroenterology 1984; 86: 1215–19.

The surgical patient with pre-existing liver disease

Aranha GB, Sontag SJ, Greenle HB. Cholecystectomy in cirrhotic patients; a formidable operation. Am J Surg 1982; 143: 55–60.

Garrison RN, Cryer HM, Howard DA, Polk HC. Clarification of risk factors for abdominal operations in patients with hepatic cirrhosis. Ann Surg 1984; 199: 648–54.

Gelman SI. Disturbances in hepatic blood flow during anaesthesia and surgery. Arch Surg 1976; 111: 881–3.

Greenway B, Williams R. Adverse effect of exploratory laparotomy in patients with unsuspected liver disease. Br J Surg 1982; 69: 449–51.

O'Keefe SJ, El-Zayadi AR, Carraher T, Davis M, Williams R. Malnutrition and immune competence in patients with liver disease. Lancet 1980; ii: 615–17.

Silk DBA. Parenteral nutrition in patients with liver disease. J Hepatol 1988; 7: 269–77.