The liver has a remarkable capacity to regenerate following even extensive necrosis. In some instances, however, progressive cirrhosis develops following injury induced by a variety of conditions, ranging from congenital biliary atresia to viral infections or alcohol abuse. In such patients, fatal complications of bleeding, infection, and progressive hepatic insufficiency can be reliably anticipated. Treatment with diet, various medications, immunosuppression, or even a few palliative operations is typically of short-term benefit at best. The only life-saving option for such patients, and also for some with acute fulminant hepatic failure, is replacement of the diseased liver with a healthy organ. Nevertheless, as recently as 1980, clinical liver transplantation still remained an investigational procedure which was being performed regularly in only two centres worldwide. Over the next 10 years, the practice of hepatology was completely transformed as hepatic replacement became accepted as not only a hypothetical possibility, but as the preferred therapeutic option that could be practically offered to a significant proportion of the thousands of patients who die annually from irreversible liver failure. As a result, decisions regarding the care of patients with liver disease should now be made with the perspective that future liver replacement may be required. Operative procedures in the portal area, such as portacaval shunts or complex biliary drainage manoeuvres, should generally be avoided since they greatly reduce the likelihood of a successful subsequent liver transplant. Similarly, holding the transplant in abeyance while awaiting the absolute terminal stages of hepatic failure significantly compromises the recipient's chances of survival and is no longer justified for suitable candidates with relentlessly progressive liver dysfunction.

This dramatic change in the recommended approach to the treatment of hepatic insufficiency is based upon the improvement in long-term survival and rehabilitation which has been achieved following transplantation. In marked contrast to the limited rehabilitation provided by other forms of intervention, over 80 per cent of the patients who survive transplantation return to full time employment, schooling, or homemaking. As described below, better definition of the most appropriate indications and timing for liver transplantation, improved methods of donor organ preservation, and refinements of surgical techniques and perioperative management, are among the factors which have contributed to the reduced morbidity and mortality. The availability of more selective, and therefore less toxic, immunosuppressive protocols, however, has been the most important development in the recent rapid growth of liver transplantation.

The almost simultaneous introduction into clinical trials of cyclosporin and monoclonal antibodies at the beginning of the last decade ushered in a new era of solid organ transplantation. Markedly improved results were immediately evident in patients following renal transplantation and even more dramatically, following liver transplantation where a doubling in 1-year survival was seen (Table 1) 238. Stimulated by this encouraging change in clinical outcome, a Consensus Development Conference on Liver Transplantation was convened in the United States at the National Institutes of Health in 1983. Review of the results then being reported led the participants to conclude that liver transplantation was no longer an experimental procedure but a practical therapeutic approach. This determination encouraged a much wider application of the procedure, and over the next 8 years, more than 150 centres were established worldwide to perform liver transplants for a continually broadening list of indications.

The first reference to hepatic replacement in the scientific literature was by Welch in 1955. Initial experimental efforts were directed to the transplantation of an extra liver into an ectopic site in the abdomen, typically with systemic venous inflow to the portal system. There are several theoretical advantages of such an auxiliary liver. The inherent risks related to the typically difficult host hepatectomy in the presence of severe portal hypertension and the instability that may develop during the anhepatic phase of the transplant procedure are avoided. In addition, any residual function of the retained native liver might provide a temporary supportive role during postoperative periods of compromised allograft function.

The early canine auxiliary allografts were rapidly rejected, making observations beyond a few days impossible. The more prolonged survival provided by the introduction of azathioprine immunosuppression revealed a rapid diminution of hepatic allograft mass beginning within 2 weeks after heterotopic implantation. The model thus led to the definition of the importance of hepatotrophic substances, including insulin, in the splanchnic venous blood. Techniques providing portal flow directly from the recipient's alimentary venous return into the transplanted liver proved essential for maintaining the long-term anatomical and functional integrity of the allograft. Perfection of these techniques led to the first clinical trial of auxiliary liver transplantation in November 1964. That attempt, and approximately 10 others over the next 4 years, all met with a fatal outcome, usually as a result of sepsis and hepatic failure. The first successful auxiliary liver transplant was performed in April 1969 at Memorial Hospital in New York. The procedure was performed in a 72-year-old patient with non-resectable cholangiocarcinoma. The allograft functioned normally for 9 months until the patient's death, secondary to infection in her own obstructed liver. The first heterotopic liver allograft to provide unquestionably prolonged survival was performed in 1972 by the same group for treatment of biliary atresia. That patient survived for more than 17 years. Nevertheless, overall clinical results following auxiliary liver transplantation have been poor: only two of 50 reported patients who had undergone the procedure by 1986 survived more than 1 year. Interest in the heterotopic procedure, however, has recently been renewed by reports of successful auxiliary partial liver transplantation in patients with complications of hepatic failure deemed to be so advanced that they were excluded from consideration for hepatectomy and orthotopic replacement.

The surgical techniques for orthotopic liver transplantation were first studied in canine experiments in 1956. Over the next 7 years, separate teams led by Moore, in Boston, and by Starzl, first at Chicago and then at Denver, independently identified the exacting technical requirements of the operation. Based upon these extensive studies, the first clinical attempt at liver replacement was undertaken at the University of Colorado in March 1963. Despite the experience of the surgical team, the procedure could not be completed due to massive haemorrhage. Over the next 4 years, isolated efforts to accomplish this formidable procedure at several institutions worldwide yielded no long-term successes. Hepatic failure leading to sepsis resulted from ischaemic damage and rejection in most of these recipients. It became clear that the successful clinical application of the technical skills already available would have to await the development of more effective immunosuppressive modalities. The addition of antilymphocyte serum to clinical protocols in 1966 helped to overcome this obstacle. With the use of triple-drug therapy including azathioprine, steroids, and antilymphocyte serum, improved results were first reported in renal allograft recipients. Soon afterwards the first human hepatic allograft recipient to achieve prolonged survival was given a new liver in Denver, in July 1967. This young patient, who underwent the procedure for treatment of an extensive hepatoma, lived with normal allograft function for more than a year before dying with recurrent cancer. Despite such occasional successes, however, frequent postoperative complications continued to plague the procedure and resulted in an early death rate of about 70 per cent. These results remained in stark contrast to the more satisfactory outcome which was being achieved following kidney transplantation. Most centres concluded that the technical complexity of the liver transplant operation would continue to limit its clinical relevance until more highly selective immunosuppressive agents, which would not render the host so vulnerable to infection, became available. Over the next 15 years, therefore, there was a virtual moratorium on the clinical application of liver transplantation except for continued efforts by the two teams in Colorado and Cambridge. As noted previously, more specific immunosuppresive agents, including cyclosporin and monoclonal antibodies, were eventually developed and after two decades of disappointments and tragic failures, these finally provided the effective immunosuppression that had been needed to demonstrate the clinical feasibility of hepatic replacement therapy. Further improvements in the management of rejection, such as the recent introduction of the experimental agent FK506, promise to continue to reduce the morbidity faced by liver allograft recipients and more widely extend the indications for the procedure.

Since there is no effective alternative treatment available, liver transplantation should be considered for virtually any patient with advanced hepatic failure and a predicted survival of less than 1 year without hepatic replacement. Typical candidates can be divided into five broad groups, as outlined in Table 2 239. In infants and younger children, biliary atresia has been the most common indication. Initial management of these patients usually includes an attempted porticoenterostomy (Kasai procedure) during the neonatal period. Successful biliary diversion and stabilization can be accomplished in 50 to 75 per cent of these attempts, at least until some growth has occurred. This greatly increases the likelihood of a size-compatible liver donor becoming available. If the initial Kasai procedure fails, most centres now recommend immediate consideration for transplantation, since multiple attempts at revision of biliary drainage greatly compromise the likelihood of successful hepatectomy and liver treatment.

In later childhood, most candidates for transplantation have inborn errors of metabolism, postnecrotic cirrhosis, and hepatic neoplasms which are otherwise unresectable. In children with benign conditions, a fall-off from the established growth curve is often the major indication to proceed with transplantation. If permanent growth retardation is to be avoided, the optimal time for transplantation must often be at a much earlier stage in the evolution of deteriorating hepatic function than would be considered reasonable in adults.

In adults, the leading indication has been postnecrotic cirrhosis. Most of these patients have non-A, non-B viral hepatitis. Interestingly, there initially seemed to be little risk of significant disease recurrence following transplantation. With the recent development of a serological marker for hepatitis C, the clinical relevance of this disease in the allograft will be more accurately assessed. Patients who are chronically positive for hepatitis B surface antigen are highly likely to retain or redevelop their original serological markers of infection. Various perioperative manipulations, including treatment with immune globulin, interferon, or hepatitis B vaccine, have not clearly affected the incidence of recurrence. Some of these recipients have died with fulminant hepatitis in the allograft. However, the recurrent hepatitis B manifests with varying degrees of severity. Many patients enjoy prolonged symptom free-survival despite early post-transplant return of detectable serum levels of surface antigen. Thus, persistent pretransplant antigenaemia has not proved to be an absolute contraindication to transplantation. Potential candidates with hepatitis B virus e antigenaemia, which indicates a greater viral concentration, are at higher risk of developing clinically significant recurrence and are, therefore, more frequently excluded from consideration for transplantation. Hepatic replacement for other conditions leading to end-stage cirrhosis, including primary biliary or cryptogenic cirrhosis, sclerosing cholangitis, and autoimmune hepatitis, has been highly successful and is generally accepted as the appropriate therapeutic approach.

Despite the high prevalence of cirrhosis secondary to alcohol abuse, many of these patients are unsuitable for transplantation, both because they typically have multisystem disease, including severe cerebral atrophy, and because they are prone to precipitous clinical deterioration. In addition, non-compliance with essential medical therapy after surgery has been found to limit long-term success. As a result, application of hepatic transplantation for this indication has remained somewhat controversial. Nevertheless, the results of transplantation, combined with a multidisciplinary treatment programme for substance abuse in individuals with progressive liver dysfunction after discontinuing alcohol intake, are comparable to those for other accepted conditions. Thus, liver transplantation for selected patients with Laennec's cirrhosis is again being more widely evaluated.

The critical decision for all of these patients is the timing of the transplant. Until recently, the procedure was often considered so drastic that it was recommended only as a last resort when all other palliative measures had failed. Clearly, allowing these patients to deteriorate to the point of repeated gastrointestinal bleeding, malnutrition, sepsis, and coma, so compromises the chances of successful transplantation that it is no longer acceptable. The role of the hepatologist in choosing the optimum time for referral of the patient for transplantation is perhaps the most important determinant of survival following the procedure. Unfortunately, the natural history of these diseases proceeds at various rates in individual patients; the decision to proceed to transplantation cannot therefore be simply based on some predetermined clinical or biochemical profile. A combination of factors, including the nature and stage of the disease, the patient's age, quality of life, and history of any previous complications must all be considered. Appropriate candidates with primarily parenchymal conditions often present with hypoalbuminaemia, coagulopathy, variceal bleeding, or hepatic encephalopathy, while serum bilirubin levels are not markedly abnormal. Portacaval shunting should generally be avoided for management of variceal haemorrhage in potential transplant candidates from this group of patients. If bleeding cannot be controlled by conservative measures, including endoscopic sclerotherapy or bonding, it is reasonable to recommend splenorenal or mesocaval anastomosis for patients with Child's Class A cirrhosis. The long-term survival rate after portosystemic shunting in patients with more advanced cirrhosis is so inferior to that achieved by transplantation that appropriate candidates with class B or C cirrhosis should generally be considered for immediate liver replacement.

Severe hyperbilirubinaemia with intractable itching and the disabling complications of hepatic osteodystrophy may be the major indications for liver replacement for patients with cholestatic conditions. In contrast to the patients with primarily parenchymal disease, synthetic function is relatively preserved and there is often only limited evidence of portal hypertension when transplantation is, nevertheless, clearly appropriate.

Fulminant hepatic failure secondary to viral, toxin, or drug-induced sudden massive necrosis of a previously healthy liver represents a particularly difficult indication for transplantation. Although some of these patients recover, in those who progress to stage III–IV encephalopathy, a mortality rate of over 80 per cent can be anticipated. This unpredictable prognosis often leads to a delayed decision to proceed to transplantation. Hepatic replacement in these patients, in order to be successful, should be performed before stage IV coma and the sequelae of advanced cerebral oedema are established. Transplantation in patients at this point is associated with a mortality rate of greater than 50 per cent. Ominous earlier signs include persistent coagulopathy, hypoglycaemia, renal failure, rapid shrinkage of liver mass, and progressive encephalopathy despite maximal medical and nutritional support. At this stage, these patients should be referred for urgent liver replacement, hopefully before metabolic acidosis and sepsis develop. Transplantation prior to the development of these more grave signs could result in replacement of an occasional liver with reversible damage. Nevertheless, a survival rate of 60 to 75 per cent can now be achieved by transplantation, whereas at least 80 per cent of these patients will die with medical management alone.

A wide variety of liver-based congenital errors of metabolism can be successfully treated with liver replacement. Patients with an inborn metabolic defect such as haemophilia who require liver transplantation for other reasons (such as hepatitis), naturally, therefore, enjoy the additional benefit of complete correction of the underlying defect.

Primary malignancy of the liver involving both lobes or occurring in the presence of pre-existing cirrhosis which precludes even partial hepatic resection would appear to be an ideal indication for total hepatectomy and transplantation. Absence of the multisystem ravages of chronic liver failure and freedom from severe portal hypertension usually make these patients excellent surgical candidates. Although early survival following transplantation is good, cancer recurs within 1 year following liver replacement in over 50 per cent of the survivors. The results have been particularly discouraging in patients with cholangiocarcinoma in whom extrahepatic recurrences, typically as peritoneal or diaphragmatic implants or even intrahepatic metastases develop within months. The prognosis is more favourable following liver replacement for relatively small hepatomas presenting in a cirrhotic liver or for the slow growing fibrolamellar variant of hepatoma. Long-term disease-free survival has been observed in over 50 per cent of these recipients. Pretransplant assessment of patients with malignancy requires extensive diagnostic studies, often including exploratory laparotomy, in the effort to rule out occult metastatic disease. Clearly, however, more effective adjuvant therapy will have to be identified before liver replacement can be considered an appropriate approach, except for a few highly selected individuals, to hepatic cancer.

As the number of patients undergoing hepatic replacement has increased, the need for hepatic retransplantation has become more common. The reported frequency of retransplantation has varied between 5 per cent and 25 per cent among different centres. Overall survival rates following retransplantation are approximately 20 per cent lower than those following the primary procedure. Repeat liver transplantation is performed primarily in recipients whose initial graft is failing due to technical complications, primary non-function, or irreversible rejection. Retransplantation for uncontrolled acute rejection is uncommon: more typically, one observes relentlessly progressive chronic rejection which is manifested by worsening cholestasis and a histopathologic picture of ‘vanishing bile ducts’. Retransplantation has generally been accepted as the only treatment available for these patients, although it has recently been suggested that the new immunosuppressive agent, FK506, may reverse these changes. Patients with primary graft non-function, secondary to unrecognized donor organ damage suffered during the agonal prerecovery period or to preservation injury, decompensate rapidly in the post-transplant period. As a result, they may be extremely unstable by the time a second donor is identified. The prognosis following retransplantation for this indication, therefore, is least favourable.

The selection from such a large patient population of the candidates most likely to benefit from liver transplantation relies primarily upon exclusion of coexisting conditions which have been found to increase the risks of early postoperative death unacceptably. Absolute contraindications (Table 3) 240 include malignancy or uncontrolled infection outside the liver, secondary malignancies, life-limiting heart disease, or significant pulmonary disease. Currently active drug or alcohol addiction is also considered an absolute contraindication. As noted earlier, experience in patients with active addictions has emphasized that non-compliance with the post-transplant medical regimen almost inevitably leads to fatal complications.

As surgeons have gained increasing experience with liver replacement, a number of factors which were previously felt to preclude successful transplantation have now been relegated to relative contraindications. These include, for example, age over 65 years (the oldest reported recipient of a liver was 76 years old at the time of transplantation), and portal vein thrombosis. Extensive clotting of the portal or even mesenteric veins, which previously made revascularization of the allograft impossible, can now be successfully managed through the use of vein grafts (see below). In patients with acute renal failure associated with hepatic decompensation (hepatorenal syndrome), renal function can be anticipated to improve rapidly following successful liver replacement. Patients with chronic renal dysfunction can be managed with simultaneous liver and kidney transplantation with only moderately increased risks.

Most suitable kidney donors can also be liver donors. Ideally, one prefers a donor less than 65 years of age, though older donors have been used. There should be no previous history of alcoholism, neoplasia, except for brain tumours or skin cancer, or any current systemic viral or bacterial infection. The ABO blood group should be identical or compatible with the recipient. The chances of success with ABO incompatibility are, nevertheless, sufficient to occasionally justify the use of an incompatible liver in a critically ill individual who has no other available donor. Reasonable matching for body size between donor and recipient is essential in order to limit the technical difficulty of the reimplantation procedure. It is best to have a donor organ from a person whose weight is between 25 and 125 per cent of the recipient's: size constraints are particularly exacting for small individuals. Because of the rarity of potential donors in the paediatric population, one must often consider using older donors for transplantation into children. In some cases, the large size of the allograft has made successful reimplantation impossible. This obstacle is now being overcome by using adult livers, reduced in size to a single lobe or even segment (see below). Recently, this approach has been further extended to division of a single cadaver donor liver in such a way as to obtain two viable grafts for implantation into different recipients, or to procurement of a single lobe or segment from a living related donor for transplantation into a small child.

Histocompatibility testing currently plays little role in selecting recipients for liver transplantation. Typically, the urgency of the recipient operation precludes donor selection based upon tissue matching. Furthermore, retrospective analysis has not revealed a significant correlation between matching and results. Successful liver transplantation, in fact, has not infrequently been performed even in recipients with measurable serum levels of lymphocytotoxic antibodies reactive with donor histocompatibility antigens. Such a ‘positive crossmatch’ would be anticipated to result in hyperacute rejection of kidney allografts. This phenomenon has seldom been documented following liver transplantation, and only a modestly compromised long-term survival rate has been observed in hepatic recipients of such donor organs.

The success of the multiple organ recovery involved during donor hepatectomy requires careful co-ordination among the teams to assure that there is no compromise in viability of any of the transplanted organs. In addition, it is critical to have anaesthesia support to monitor and maintain cardiovascular integrity of the donor during the meticulous dissection, which may take 3 to 4 h.

Although the details will differ, depending upon the combination of organs to be removed, certain common principles prevail. These include wide exposure, dissection of each organ to its vascular connection while the heart is still beating, placement of cannulas for in-situ cooling, and removal of the organs while perfusion continues, usually in the order of heart and/or lungs, liver, kidneys, pancreas.

The organs are exposed through a midline incision extending from suprasternal notch to the pubis (Fig. 1(a)) 705. Rapid inspection excludes unsuspected sepsis, neoplasia, or other significant pathology, and confirms the gross suitability of the organs to be procured. If the heart is to be used, it is usually mobilized as the first manoeuvre so that it can be quickly removed at any later stage should uncorrectable vascular instability occur during the dissection of other organs. The liver and often the pancreas are mobilized next. The coeliac axis is dissected to the aorta. If the pancreas is not to be used, the splenic and superior mesenteric arteries may be ligated and divided. The entire length of the coeliac axis, including a Carrel patch of donor aorta will be subsequently taken for anastomosis in the recipient. The common bile duct is transected and the gallbladder incised and flushed to prevent autolysis of the biliary epithelium. The portal vein is dissected to the confluence of the splenic and superior mesenteric veins where a cannula can be placed into the splenic vein for rapid portal perfusion (Fig. 1(b)) 705. Skeletonization of the liver is completed by isolating the vena cava posteriorly, ligating and dividing the right adrenal vein, and freeing the suprahepatic vena cava.

If the pancreas is to be transplanted, the spleen is mobilized, the short gastric vessels are divided, and the spleen and pancreas retracted to the right. The body and tail of the pancreas are then carefully dissected free.

The kidneys and major abdominal vessels are next exposed by retracting the right colon and small bowel to the left. The kidneys are mobilized from the retroperitoneum, and the distal aorta and vena cava are completely freed. The donor is heparinized, after which a perfusion cannula is placed in the aorta and a venous drainage cannula into the vena cava.

Liver precooling, usually with 1 to 2 litres of crystalloid solution, is initially accomplished via the portal vein cannula. When the body core temperature falls to about 30°C, or earlier if haemodynamic instability occurs, the aorta is cross clamped at the diaphragm and in-situ aortic and portal vein flushing is begun using University of Wisconsis (UW) solution (Table 4) 241. The cardioplegic infusion is next begun into the ascending aorta and cardiectomy is performed. The liver is removed next. Finally, the remaining mobilization of the kidneys is undertaken so that the entire block consisting of both kidneys, ureters, aorta, and inferior vena cava can be lifted out of the abdomen and placed in cooled perfusion solution. Segments of donor iliac artery and vein are also removed for possible use as vessel grafts during hepatic reimplantation. In donors from whom whole pancreaticoduodenal procurement is planned, we advise removing this organ block last in order to avoid possible contamination from the transected duodenum.

After completion of the donor hepatectomy, the allograft is further flushed through the portal vein and hepatic artery and then immersed in a sterile plastic bag containing UW preservation solution for storage on ice until transplantation. The anion, lactobionate, and the trisaccharide, raffinose, contained in this solution appear to be particularly effective impermeants for suppression of hypothermia induced cellular swelling. The introduction of this preservation solution into clinical practice has extended the safe period of cold storage to as long as 24 h. This relatively prolonged preservation time increases the capability for more distant procurement and also relaxes the logistical constraints of the procedure. The recipient operation can now be scheduled on a semi-emergency basis, thus allowing more flexible use of operative rooms and personnel. Perhaps most importantly, the enhanced margin of safety has made possible the development of new techniques, such as reduced size liver transplantation, and has further minimized organ wastage by allowing donor hepatectomy to proceed even in circumstances where a recipient team is not immediately available.

One of the most important determinants of the perioperative mortality rate is the metabolic and resuscitative management by the anaesthesia team. The complexity of the procedure requires a number of extra access lines and pieces of equipment to provide adequate intraoperative monitoring and support. In addition to the usual anaesthesia machine, ventilator, and cardiac monitor with strip chart recorder, the operating room must usually have available a blood salvage device, separate pump systems for blood warming and rapid infusion, venovenous bypass, and a patient warming blanket, and often other special devices, such as the argon beam coagulator. Some attention, therefore, must be given to providing a reasonably compact and functional layout of the operative area. In adult recipients, access for infusion and monitoring is typically accomplished via a right radial arterial line, a large bore intravenous site in the right arm, and introducer catheters in each internal jugular vein, one for placement of the pulmonary artery catheter and one for connection to the rapid infusion pump. With this system, infusion rates as high as 2 l/ min can be provided, if needed, during periods of rapid blood loss. Access sites in the legs are avoided since infusions would be unreliable during the period of vena cava occlusion. The left arm is avoided since the venovenous bypass return is usually via the left axillary vein. In small children, the use of pulmonary artery catheters is usually neither required nor practical. Adequate monitoring and infusion is provided by central venous pressure measurements and 14- or 18-gauge intravenous catheters.

Patients with hepatic failure are predisposed to developing hypoxia from atelectasis, due to ascites, and from intrapulmonary arteriovenous shunts. Thus, careful preoxygenation and rapid sequence induction anaesthesia are commonly used. An appropriately sized, low pressure cuff endotracheal tube is placed in anticipation of a possibly prolonged period of intubation. Adequate padding of the heels, sacrum, elbows, and head is provided to prevent pressure ulceration during the operation which not infrequently extends beyond 12 h. Hypothermia, resulting from the long duration of the procedure, extracorporeal blood circulation in the venovenous bypass, and the implantation of the iced allograft, must be anticipated. Measures to provide recipient rewarming are essential. These include warmed intravenous infusions, a heated ventilator circuit, a warming blanket on the operating table, and plastic drapes on the head and exposed extremities.

The selection of anaesthetic agents is governed by the high cardiac output, low peripheral vascular resistance state typically associated with end-stage liver disease. Isoflurane inhalational anaesthesia will reduce systemic resistance and may be used as tolerated. The preferred maintenance agent for haemodynamic stability is a narcotic anaesthetic employing fentanyl and/or morphine. Because some patients may not tolerate administration of routine anaesthesia, they are often given a benzodiazepine as well, to block memory. Lorazepam is frequently chosen, since this agent requires only glucuronidation for excretion.

Some of the intraoperative consequences of the transplant procedure are detailed in Table 5 242. The patients often develop metabolic acidosis, which should not be over-corrected with bicarbonate since they will also receive a large citrate load with administered blood products This will eventually be metabolized to bicarbonate and may lead to severe postoperative metabolic alkalosis. Although the determinants of need for packed red blood cells and fresh frozen plasma are independent, the patients typically require approximately equal volumes of each. Platelets are administered based upon pre- and intraoperative platelet counts and the number of blood volumes transfused.

Many of the occasionally catastrophic consequences of the anhepatic phase of the operation can be limited by use of veno-venous bypass. Up to 40 to 50 per cent of the cardiac output can be returned to the heart from the lower body and viscera via the bypass. Continuous monitoring of central venous, pulmonary arterial, and systemic arterial pressures, and frequent determinations of cardiac output are required during this critical period. Hypotension may be caused by hypovolaemia, inadequate bypass return, air or thromboemboli from the pump, or myocardial depression secondary to hypothermia or to the hypocalcaemia resulting from citrate intoxication during the rapid transfusion of blood products. Although there is a theoretical possibility of hypoglycaemia developing during the anhepatic stage, typically the blood sugar remains normal to elevated, presumably from the anticoagulant in banked blood and from numerous glucose containing carrier infusions.

Another critical period occurs at the time of liver reperfusion. Despite prior flushing of the allograft, the sudden bolus of cold, hyperkalaemic, acidotic blood from the liver and lower body may cause severe pulmonary artery vasoconstriction with resultant hypotension. It is thus essential to stabilize the patient by correction of acid–base abnormalities and hypovolaemia. Administration of calcium chloride just prior to vena caval unclamping may be required to antagonize the imminent hyperkalaemia. As the allograft function improves later, following rewarming and arterial reconstruction, a major fall in serum potassium often occurs, as a result of both uptake into the revascularized hepatocytes and increased availability of ionized calcium as citrate metabolism begins in the liver. Potassium supplementation may be required. A significant bleeding diathesis may also develop at this stage, secondary to the accelerated fibrinolysis which appears in association with increased blood levels of fibrin degradation products. This consequence of hepatic ischaemia may lead to substantial blood loss. The importance of aggressive replacement of coagulation factors during the early reperfusion period, sometimes including antifibrinolytic agents, cannot be overemphasized.

Removal of the diseased liver is a tedious process because of the coagulopathy inherent in patients with end-stage liver disease, the multiple venous collaterals and portal hypertension encountered, and the often dense adhesions from previous operations. Abdominal exploration in adults is carried out through bilateral subcostal incisions with a midline extension, usually including excision of the xiphoid. In children, this midline extension is seldom required. Meticulous haemostasis is obtained using a combination of electrocautery and suture ligation for essentially all dissection.

After entering the peritoneal cavity, the falciform and left triangular ligaments are divided, often requiring stepwise, silk ligatures for haemostasis. The gastrohepatic ligament (lesser omentum) is then exposed and similarly divided. If possible, the liver is left vascularized until the last stages, in order to retain as normal as possible levels of clotting factors and glucose control. In some patients with extensive bloody adhesions, it may be necessary to devascularize the liver earlier and then continue with the rest of the mobilization. The use of venovenous bypass at this point can help to stabilize the patients. Most teams, however, do not find the option of early bypass and rapid hepatectomy attractive.

The portal structures are generally isolated first. To maximize length for the subsequent biliary anastomosis, the recipient common bile duct is divided above the entrance of the cystic duct. If the patient has had a previous biliary enteric anastomosis, the anastomosis is taken down, protecting the portal area and the Roux loop which will be subsequently used for drainage of the donor liver. The proper hepatic artery is ligated and divided distal to the take-off of the gastroduodenal artery. The more proximal common hepatic artery is further exposed for subsequent anastomosis leaving the gastroduodenal artery intact (Fig. 2) 706. Finally the portal vein is skeletonized, ligating and dividing tributaries including the coronary vein and any pancreatic veins which may be present. If the portal vein is found to be thrombosed, more proximal dissection beneath the pancreas to the area of confluence between the splenic and superior mesenteric veins will usually identify a suitable anastomotic site. Occasionally it is necessary to expose the superior mesenteric vein below the pancreas at the level of the middle colic vein to obtain a patent vessel with adequate flow for portal revascularization. In both of these instances, vessel grafts taken from the donor may be required to span the distance from the more proximal recipient vessel to the donor portal vein (Fig. 3) 707.

Once the portal dissection has been completed, the infra-hepatic vena cava is exposed. Because of extensive retroperitoneal collateral vessels, the tissues must often be divided between ligatures, as is practised in this dissection during portasystemic shunting. The inferior vena cava is carefully encircled at a level just above the right renal vein. The right adrenal vein is ligated and divided. The right lobe of the liver is elevated as the peritoneal and diaphragmatic attachments are coagulated or suture ligated. The bare area of the liver is separated from the diaphragm and any additional venous branches entering the vena cava from posteriorly are ligated and divided. Mobilization of a lengthy segment of suprahepatic vena cava is accomplished by encircling the cava close to its exit from the liver. By working at this level, the recipient phrenic veins are usually not encountered—being left attached to the cuff above the site where the vena cava will be divided. After this dissection, the mobilized liver is held in place only by its venous attachments.

Although venous bypass during the anhepatic phase of the procedure is not an absolutely essential component of the operation, severe splanchnic venous congestion can occur in its absence (Fig. 4) 708, and many centres now use bypass for most adult recipients. Some of the potential benefits of venovenous bypass are summarized in Table 6 243. The haemodynamic stability provided by the high volume return of blood to the heart from otherwise obstructed venous beds allows revascularization of the donor liver in a less urgent fashion. A not insignificant advantage provided by such effective stabilization is the opportunity for training of inexperienced surgeons in this phase of the operation.

Heparin bonded catheters are inserted via a separate groin incision into the recipient's left femoral vein for decompression of the lower systemic venous bed. A second limb of the circuit is placed into the left axillary vein for blood return to the heart. An in-line centrifugal pump provides flow rates of 1 to 3 l/min (Fig. 5) 709. Final resection of the recipient liver is then completed by clamping the portal vein and transecting it high in the hilum. Vascular clamps are placed on the infrahepatic and suprahepatic cavae. A long suprahepatic caval cuff is developed by incising the liver parenchyma and transecting the hepatic veins within the substance of the liver (Fig. 6) 710. The infrahepatic cava is severed close to its junction with the caudate lobe. The liver is removed from the field and any major retroperitoneal bleeding sites are suture ligated. The third cannula of the venovenous bypass is then introduced into the lumen of the recipient portal vein to decompress the splanchnic venous system. This manoeuvre typically results in greatly diminished bleeding from the hepatectomy site. Final haemostasis can then be achieved by oversewing or cauterizing the exposed retroperitoneal tissues in the hepatic fossa (Fig. 7) 711.

The suprahepatic vena caval cuff is tailored by opening the recipient hepatic veins into the caval lumen. Sutures are inserted at the corners of donor and recipient veins and the allograft is gently lowered towards the diaphragm. The posterior anastomosis is accomplished using everting mattress sutures placed from within the lumen. The anterior rim is completed in simple over and over fashion. The infrahepatic cava is reconstructed by usual end-to-end venous anastomotic techniques. Prior to completion of the anterior wall of this anastomosis, the allograft is flushed via the donor portal vein with cold Ringer's lactate solution. This important manoeuvre removes air and any remaining preservation fluid, which contains high concentrations of potassium and acidic radicals that have accumulated during the ischaemic interval, from the intrahepatic venous spaces.

Preparation for the portal vein anastomosis requires removal of the outflow limb of the bypass circuit from the recipient vessel. A running continuous suture approximates the ends of the widely spatulated donor and recipient portal veins. As noted earlier, this anastomosis may have to be placed at the confluence of recipient splenic and superior mesenteric veins or, in some instances, on to a vein graft. The suture for all of the venous anastomoses is tied leaving 1 cm or more of ‘growth factor’ between the knot and the vessell wall. This allows expansion of any purse-stringing effect and limits the likelihood of anastomotic stricture. At this point, most units favour release of the venous clamps to restore portal and vena caval flow through the liver. If the patient is stable, venovenous bypass is discontinued and the remaining catheters are removed to prevent intraluminal clotting.

The hepatic artery anastomosis is usually constructed by suturing the donor coeliac axis with attached aortic Carrell patch to the recipient hepatic artery at the level of the gastroduodenal take-off. Various arterial anomalies may require modifications in the method of reconstruction, even including insertion of donor iliac artery as a vascular graft to provide inflow directly from recipient infrarenal aorta.

Biliary reconstruction is begun after the donor gallbladder has been removed and adequate haemostasis of the operative area has been accomplished. End-to-end choledochocholedochostomy using interrupted absorbable sutures is performed in recipients whose native biliary tree is not diseased. The anastomosis is stented with an appropriately sized T-tube brought out through the recipient duct (Fig. 8) 712.

Reconstruction is by choledochoenterostomy into a Roux-en-Y loop of jejunum for patients whose primary hepatic disease involved the biliary tree (such as biliary atresia, sclerosing cholangitis, cholangiocarcinoma). We favour stenting this anastomosis as well, using a small feeding tube brought out through the side of the Roux loop and the abdominal wall. This approach has the advantage of allowing postoperative radiographic studies of the anastomosis if any question of obstruction or leak develops.

Because of the limited availability of suitably sized organ donors for the relatively large numbers of infants and small children with end-stage liver disease, a significant proportion of these candidates die before transplantation can be accomplished. This problem has been addressed by developing reduced size orthotopic liver transplantation. With this approach, livers from cadavers weighing as much as 10 times that of the recipient can be successfully reduced in size and placed into paediatric recipients. Donor liver procurement is as described above. For partial liver transplantation, anatomic dissection of the graft is performed, ex vivo, to provide either a right or left lobe for reimplantation while discarding the resected portion. Growing experience with this technique has more recently progressed to partition of the vascular and biliary structures to obtain two viable grafts for use in two different recipients, the so-called split liver (Fig. 9) 713. In patients receiving the right lobe graft, recipient hepatectomy and all vascular anastomoses are performed as described above. Biliary reconstruction is by choledochoenterostomy. Recipient hepatectomy in patients receiving the left lobe graft involves mobilization of the liver from the retroperitoneum as described but preservation of recipient inferior vena cava, suture closure of the right and accessory hepatic vein orifices, and use of the left hepatic vein for anastomosis to the graft. The short lengths of donor left portal vein and hepatic artery have generally necessitated the use of vessel grafts for revascularization in the recipient.

Rarely, potential hepatic allograft recipients may be so unstable that they would be predicted not to be able to tolerate hepatectomy and orthotopic replacement. Transplantation of an auxiliary liver into a heterotopic position while leaving the diseased organ in situ could be a suitable alternative for such high-risk individuals. As noted above, long-term success following clinical attempts at heterotopic liver transplantation has been poor, primarily because of size constraints leading to technical complications and infection. Recent modification of the procedure, with reduction of the size of the graft by partial hepatectomy, may provide a satisfactory solution. For this approach, the donor liver is prepared ex vivo by removing the left lobe and reducing the vena cava to a short segment draining the right and middle hepatic veins. In the recipient, dissection is primarily limited to the portal area, as would be required for a portacaval shunt procedure. The infrarenal aorta is also freed over a length of 3 to 4 cm. The reduced size liver is oriented in the right subhepatic region with the donor vena cava overlying the recipient cava and the graft hilum facing the aorta (Fig. 10) 714. Vascular anastomoses include donor vena cava to side of recipient suprarenal vena cava; donor portal vein to side of recipient portal vein; and Carrel patch of donor aorta to side of recipient aorta. Biliary drainage is into a Roux-en-Y jejunal loop. Early experience with this approach appears promising but remains limited.

The early postoperative management of liver allograft recipients requires much more intensive monitoring and support than is required following kidney or pancreas transplantation. Because of their precarious pretransplant condition the prolonged operation and its attendant massive fluid shifts, and the coagulopathy and haemodynamic instability that can result if onset of normal allograft function is delayed, these recipients are initially cared for in a critical care unit. Satisfactory immediate postoperative cardiac output, maintenance of adequate renal function, and correction of abnormal clotting parameters may require aggressive blood and fresh frozen plasma replacement and support with vasoactive drug infusions. Other important aspects of the recipient's care can be considered as they relate to individual organ systems.

Satisfactory liver allograft function is manifested by the appearance of deeply pigmented bile in the drainage catheter and early restoration of clotting function. These recipients will typically have good renal function and will recover rapidly from anaesthesia so that successful extubation is possible within 12 to 48 h. If, however, there is minimal or watery depigmented biliary output (‘white bile’), the early postoperative care will be more complicated. These recipients can be reliably predicted to require assisted ventilation, intensive infusion of fresh frozen plasma, and other complex resuscitative measures before adequate stabilization may be achieved. As noted above, retransplantation may be the only recourse for some patients with early graft dysfunction. Timely identification of those recipients with irreversible ischaemic injury is critical if retransplantation is to be successfully accomplished before fatal cerebral oedema develops. As in the pretransplant candidate with fulminant hepatic failure, uncorrectable coagulopathy (prothrombin time over 20 s), hypoglycaemia, acidosis, and new onset of renal failure are ominous. These findings, which usually prompt the urgent search for a second allograft, are typically encountered in less than 5 per cent of liver recipients, though this indication for retransplantation has been reported to be as high as 10 to 15 per cent by centres initially willing to accept more marginal donors.

Technical complications most commonly include intra-abdominal bleeding, vascular thrombosis, biliary duct problems, or infected fluid collections. The incidence of postoperative bleeding is directly related to the severity of intraoperative bleeding and the quality of immediate allograft function. Despite meticulous surgical technique and haemostasis in the operating room, diffuse capillary oozing may persist until hypothermia and inadequate plasma clotting factor levels have been corrected. If major blood loss persists despite reversal of the coagulopathy, emergency re-exploration is indicated to identify a probable surgically correctable cause. In patients whose early bleeding ceases, laparotomy is still often required, usually 36 to 48 h after the initial surgery, in order to evacuate perihepatic collections which may otherwise become secondarily infected.

Hepatic artery thrombosis, particularly in small children, should be suspected in all patients who have unexplained fever, a bile leak, or a positive blood culture for Gram-negative organisms. Doppler ultrasound or angiography will confirm the diagnosis. If the clinical presentation is one of fulminant liver failure, biliary leak, or relapsing bacteraemia, emergency retransplantation is the only viable therapy. The small proportion of patients who develop only mild liver dysfunction, without abscess formation, may be conservatively managed while arterial collaterals to the allograft develop.

The presenting symptoms of portal vein thrombosis are usually less devastating. Rarely, in the immediate postoperative period, this complication can produce severe biochemical abnormalities and even hepatic necrosis requiring retransplantation. More commonly, portal vein thrombosis is suggested by recurrent variceal bleeding, intractable ascites, or an unexplained elevation in prothrombin time. Successful management of this complication in patients without severe hepatic dysfunction is usually achieved by operative thrombectomy and correction of any technical abnormality encountered. In the late post-transplant period, a spleno-renal shunt may be effective.

Bile duct problems, which historically accounted for the majority of postoperative complications, have greatly decreased in frequency since current reconstruction techniques became standardized. The routine use of a feeding or T-tube stent allows ready investigation of any suspected biliary tract complication. Anastomotic leak, usually manifested by appearance of bile in the abdominal drainage or an unexplained rise in serum bilirubin, occurs during the first several weeks after transplantation. Minimal, well drained leaks, not associated with hepatic artery thrombosis, will typically close spontaneously. More extensive bile collections require open drainage to create a controlled fistula or even resection and reanastomosis. Leaks secondary to arterial thrombosis are best managed by retransplantation. Biliary duct obstruction occurs more commonly in the later postoperative period, since the anastomoses are usually stented for several months following surgery. Short strictures can be dilated via endoscopic or percutaneous approaches. Surgical resection of the anastomosis and conversion to choledochojejunostomy may be required for more complex lesions. Earlier observations of diffuse obstruction by ‘sludge’ within the biliary tree are now seldom encountered. Control of this complication is presumably due to careful biliary flushing during the donor procedure and improved methods of reconstruction. An interesting and unusual cause of obstruction has been the development of a tension mucocele in a donor cystic duct remnant which shared a common wall with the bile duct, a phenomenon analogous to the ‘Mirizzi’ syndrome.

Postoperative hepatitis, usually caused by cytomegalovirus, can also cause graft dysfunction. Following serological or hepatic biopsy confirmation of the diagnosis, ganciclovir therapy is commonly instituted. Other viral infections, including recurrent hepatitis B or hepatitis C, for which interferon therapy may be indicated, can usually be histologically or serologically confirmed. Occasionally, differentiating these causes of allograft dysfunction from rejection may be difficult.

Aggressive efforts to prevent pulmonary complications are essential in all patients. Intensive chest physiotherapy, postural drainage, and endotracheal suctioning are routine. Flexible fibreoptic bronchoscopy may also be required to remove retained secretions causing segmental atelectasis. Right-sided pleural effusions are common in the early postoperative period and usually respond to diuretic therapy. Thoracentesis is rarely required to provide satisfactory lung re-expansion but should not be withheld when necessary, since even limited episodes of pulmonary dysfunction are poorly tolerated in these nutritionally depleted, debilitated, immunosuppressed patients. Attempted weaning from assisted ventilation is begun as soon as the recipient recovers from anaesthesia and demonstrates cough reflexes. If initial blood gases and mechanics are satisfactory, successful extubation can be accomplished usually within 12 to 48 h. Several unusual factors unique to liver allograft recipients may delay weaning. These include right diaphragmatic paralysis which may result from the intraoperative placement of the suprahepatic vascular clamp and the metabolic alkalosis which these patients often develop. Because the latter may contribute to compensatory hypoventilation, correction with potassium chloride, or rarely hydrochloric acid, may be necessary to allow successful early extubation.

Significant hypoxaemia resulting from development of intrapulmonary arteriovenous malformations in patients with chronic liver disease may persist into the early postoperative period. Although the prognosis for such patients is somewhat less favourable, such shunting, even advanced to the stage of digital clubbing, has been observed to reverse following successful liver transplantation.

Initial haemodynamic stability is maintained by volume administration and the addition of inotropic, chronotropic, or vasoactive agents as guided by clinical findings and continuous invasive monitoring of usual cardiopulmonary variables. Not infrequently, significant hypertension develops in patients receiving intravenously administered cyclosporin. The increased risks of intracerebral haemorrhage and seizures require aggressive treatment of sustained hypertension. Nitroprusside is typically the agent of choice in the immediate postoperative period because of the rapid reversibility of its effect. Subsequent therapy may require a combination of diuretics, a vasodilator, such as hydralazine, and often a calcium-channel blocker, such as nifedipine, as well. Arrhythmias are unusual and are typically associated with hypoxaemia or severe electrolyte abnormalities.

Acute renal failure may develop immediately postoperatively due to pre-existing renal insufficiency, intraoperative vena caval occlusion, extensive bleeding and hypotension, hepatic allograft dysfunction, and administration of nephrotoxic drugs, such as aminoglycoside antibiotics. All of these factors are exacerbated by the administration of cyclosporin, particularly via the parenteral route. Because survival is poor for patients who require dialysis, we attempt to limit the period of dysfunction by using induction immunosuppression with OKT3 rather than cyclosporin in recipients with persistent oliguria. As illustrated in Fig. 11 715, oral cyclosporin can be instituted after diuresis has resumed, and the biliary drainage tube has been clamped, parenteral administration of the drug being completely avoided. Extensive experience has established that in the presence of satisfactory hepatic allograft function, renal function regularly returns even in patients with pre-existing hepatorenal syndrome. Nevertheless, some patients with early acute tubular necrosis are left with persistently compromised renal function. Renal failure has gradually progressed, during chronic cyclosporin therapy, even to the need for renal transplantation in a few hepatic allograft recipients.

In addition to uraemia, other major metabolic derangements may require correction in the early postoperative period. These include hypothermia, hyperkalaemia, especially in the presence of poor hepatic and renal function, and hypocalcaemia and metabolic alkalosis resulting from citrate infusion.

Leucopenia and thrombocytopenia secondary to hypersplenism typically persist into the postoperative period. The leucopenia is usually of no clinical consequence; prolonged thrombocytopenia, however, may require platelet transfusions and limit the dosages of azathioprine that can be safely administered. The hypersplenism can be anticipated to resolve gradually over the first few weeks after successful liver transplantation. Haemolysis, at times requiring transfusion support, may occur 1 to 2 weeks postoperatively, most commonly in recipients of an ABO compatible, mismatched allograft. The haemolytic anaemia is the result of antirecipient isohaemagglutinins produced within the graft. Successful treatment has included transfusion, often using donor ABO group red blood cells, high dose steroids, and plasmapheresis during the typically transient period of haemolysis. Splenectomy has occasionally been required.

Another unusual haematological complication has been the development of aplastic anaemia in some patients who received a liver transplant for treatment of acute non-A, non-B viral hepatitis. The onset of aplastic anaemia has usually been within 1 to 6 months following transplantation and subsequent fatal infection is not infrequent. Some patients have had recovery of marrow function, either spontaneously or following antilymphocyte serum therapy, but the most effective therapeutic approach has not been defined.

Postoperative ileus usually resolves by the third or fourth day, after which progressive enteral alimentation is provided. In patients who remain obtunded, nasoenteric tube feedings are preferred to avoid the risks of aspiration. Some units routinely institute total parenteral nutrition in the immediate post-transplant period. Others, because of concerns over infection from the central line, reserve this approach for the few patients whose gastrointestinal tract cannot be used.

Seizures and other neurological or psychiatric disorders, such as persistent obtundation, expressive aphasia, confusion, and transient psychoses, are common after liver transplantation. Some of the identified aetiological factors have included air embolism, cerebral oedema, intracranial bleeding, electrolyte disturbances, cyclosporin toxicity, and sleep deprivation. Air embolism has seldom been seen since the importance of flushing the liver prior to revascularization was recognized. Hypomagnesaemia and hypocholesterolaemia have been implicated as contributing factors in seizures associated with cyclosporin therapy and thus should be corrected when present. In patients with recurrent seizures, many centres prefer to replace cyclosporin with an antilymphocyte serum preparation until the patient stabilizes in the later postoperative period. Acute treatment for seizures is usually with diazepam or phenobarbital. Tegretol therapy is usually satisfactory for chronic prophylaxis. We try to avoid the repeated use of phenobarbitol or long-term dilantin administration because of the induction of cytochrome P-450 microsomal enzyme systems which rapidly decrease blood concentrations of immunosuppressive agents. An important aspect of these patients' care is the avoidance of excessively frequent monitoring of night-time vital signs and medication administration so the patient can have appropriate periods of sleep.

The perihepatic drains are removed, usually within 48 h following transplantation, except for the one at the portal area. This drain is left in place until cholangiography, 7 to 10 days postoperatively, confirms the integrity of the biliary anastomosis. The T-tube stent is usually tied off at this point and then left in place for 4 to 6 months. Earlier removal not infrequently leads to bile spillage and localized peritonitis due to the lack of a well-healed exit tract around the T-tube.

Conventional immunosuppressive regimens most commonly use a combination of cyclosporin, azathioprine, and steroids. Our protocol includes intravenous cyclosporin (4–5 mg/kg.day) started on the day of surgery with oral cyclosporin (10–12 mg/kg. day) being added as soon as it can be tolerated postoperatively. It is generally impossible to achieve therapeutic blood levels of the drug with oral therapy alone during the period of external biliary drainage. Methylprednisolone or prednisone is tapered by 40 mg decrements over 5 days from 200 mg/day to an initial maintenance level of 20 mg/day. Azathioprine is initiated at a loading dose of 5 mg/kg and then continued at 1 to 2 mg/kg. day. In patients with early renal dysfunction, we substitute OKT3 monoclonal antibody for cyclosporin during the period of oliguria (Fig. 11) 715. More recently, some centres have replaced cyclosporin with the new macrolide antibiotic, FK506, with initial encouraging results.

Despite induction immunosuppression, 40 to 60 per cent of liver allograft recipients suffer one or more rejection episodes, usually beginning 5 to 14 days post-transplantation. Typical biochemical and clinical criteria suggesting rejection are listed in Table 7 244. Differentiating rejection from other disorders, such as allograft ischaemia, technical complications, or infection, can be difficult, and biopsy confirmation is frequently required. Histopathological changes of acute rejection include mononuclear cell infiltrates of portal tracts, subendothelial cells invading portal or central veins, cholestasis, bile duct damage, and variable hepatocellular necrosis. With more chronic rejection, manifested clinically by cholestatic jaundice, the hepatic artery intima is infiltrated by foamy macrophages which produce considerable narrowing of the vessel lumen. Interlobular bile ducts are often reduced in number and excess fibrosis appears in portal tracts. Unfortunately, many of the biopsy findings can also result from other conditions including cholangitis and viral hepatitis. Thus, the ultimate diagnosis must take into consideration the entire constellation of clinical, biochemical, and histopathological findings lest overtreatment of multiple suspected rejection episodes exposes the patient to unacceptable risks of infection.

Rejection episodes are initially treated with increased steroid doses. If satisfactory reversal does not ensue, OKT3 monoclonal antibody or other antilymphocyte preparations are added to the regimen. FK506 may also prove to be effective in reversing rejection which has proved unresponsive to conventional therapy. As previously noted, retransplantation may be the only recourse for some patients with relentless chronic rejection.

The intensive immunosuppression required in these debilitated patients, who not infrequently also suffer technical complications, provides an ideal setting for invasive infection. The overall incidence of infection has been reported to range from 45 to 80 per cent, which greatly exceeds that observed in recipients of other solid organ allografts. These complications are the major source of morbidity and the most common cause of death in liver transplant recipients.

The major infections likely to be encountered are summarized in Table 8 245. Infections occurring in the early postoperative period may arise from agents transmitted with the allograft, pre-existing recipient conditions, or the typical bacterial complications of surgery. With currently employed donor screening for hepatitis B surface antigen and hepatitis C or HIV antibodies, transmission of these conditions has been essentially eliminated. More problematic are bacterial or Candida infections acquired during the preprocurement care of the donor. If terminal cultures become positive, intensive antimicrobial therapy is added to the routine perioperative antibiotic regimen. As noted above, patients with uncontrolled extrahepatic infection should be excluded from consideration for transplantation. In addition, some centres recommend selective bowel decontamination, particularly with respect to fungi, for potential liver recipients. Because of the uncertainty of the timing of the procedure, however, this approach is rarely practical, and post-transplant enteric administration of antifungal preparations is more commonly advised.

Postoperative bacterial complications have most commonly been pneumonitis, abdominal or wound abscesses, or diffuse sepsis. The vast majority of these have been associated with technical complications, such as bleeding, arterial thrombosis, or biliary anastomotic problems. During the later (after 1 month) post-transplant period, viral and other opportunistic infections become more prevalent. Herpes group viruses are the most common: cytomegalovirus has been observed in 70 to 100 per cent of recipients, herpes simplex in approximately one-half, and herpes zoster in 5 per cent. Fortunately, the introduction of ganciclovir therapy has greatly reduced the morbidity and mortality due to these agents. Infection with the Epstein-Barr virus can produce conditions ranging from an infectious mononucleosis syndrome to a life-threatening lymphoproliferative disease similar to, or progressing to, B-cell lymphoma. This syndrome may regress if immunosuppressive dosages are decreased. Some groups recommend additional acyclovir therapy. As discussed above, reappearance of hepatitis B surface antigen is common in patients receiving liver transplants following hepatitis B. No effective therapy has been identified but the clinical course of the disease is not rapidly progressive in all patients. Interferon therapy appears to be effective in some patients who develop hepatitis C postoperatively.

Opportunistic infections are usually manifested as a primary infection of the lungs, most commonly due to Cryptococcus, Aspergillus, or Nocardia. During this later time period, pneumonitis caused by the protozoan Pneumocystis carinii, may also appear. Since these infections may progress rapidly or spread to other sites, aggressive efforts to identify the aetiologic agent and treat the infection are essential. These include examination of induced sputum, bronchoscopic brushings or biopsy, or even open lung biopsy, if necessary, to isolate the causative agent.

As noted above, current overall success rates after liver transplantation in the current era are over 70 per cent at 1 year, with experienced centres reporting 5-year survival of approximately 60 per cent (Fig. 12) 716. The expected long-term survival for these previously untreatable patients now exceeds that provided by modern care for many other commonly encountered conditions, including myocardial infarction and most solid tumours. Most deaths occur in the first 3 months after liver transplantation, early graft dysfunction or infection being the most common causes of failure (Table 9) 246. Rehabilitation in the majority of surviving patients is almost always excellent. In children, this includes a return of physical health and resumption of normal growth and development. In addition, emotional well being is restored so that these young recipients typically return almost immediately to their normally expected school level. Adult recipients are similarly able to return to a productive place in society: over 80 per cent resume their previous occupations or life-style and usually report that their performance is even more effective because of improved stamina and mental clarity.

Rehabilitation may be significantly delayed in some patients by the complications of osteoporosis which are commonly associated with chronic liver disease. Despite successful transplantation, the severe osteopenia which had developed preoperatively, consequent to altered vitamin D metabolism, may result in compression fractures, back pain, and prolonged disability. This condition generally stabilizes within 3 to 6 months following transplantation but again emphasizes the importance of the timely consideration of liver replacement before such secondary complications become too advanced.

Other limitations to complete rehabilitation include the side-effects of long-term immunosuppression, particularly nephrotoxicity and hypertension, development of recurrent or de novo malignancies, as may occur in all immunosuppressed patients, and chronic rejection, which may require retransplantation.

Because of the relatively recent development of liver transplantation, precise survival expectations beyond 5 years have not yet been established. Nevertheless, some of the earlier patients have now survived for over 20 years. Death in patients who survived at least 5 years after transplantation has occurred only rarely in such recipients. This suggests the prognosis for the 60 per cent of recipients currently expected to survive 5 years is indeed excellent.

Liver transplantation has now come of age. The operation is technically difficult and many patients are in poor condition when they are referred for transplantation. Nevertheless, the procedure provides a chance for long-term survival and excellent rehabilitation for more than half of a group of patients whose lifespan would be measured in terms of months with any other therapy currently available. The identification of more selective and efficacious immunosuppression together with better means of preserving the donor allograft have been two of the most important factors that have made transplantation the preferred therapy for selected patients with end-stage cirrhosis or unresectable cancer limited to the liver.

Now that acceptable survival results have moved the procedure from the experimental arena to a truly therapeutic modality, other factors limiting its more widespread application must be addressed. The first of these is the question of cost effectiveness. The average cost of a liver transplant has been estimated to be between 100 000 and 200 000 US dollars. In fact, these costs probably do not greatly exceed those required to care otherwise for many chronically ill patients with advanced liver disease. These would include loss of income, costs of disability, medications, hospital admissions, palliative surgical procedures, and, predictably, terminal care. Nevertheless, new medical approaches such as organ transplantation, are constantly reviewed in terms of whether the funds being allocated, despite providing life-saving treatment, might not be more justifiably spent in other areas. In a dramatic example of the type of painful decision necessitated by limited resources, one United States state legislature recently voted to discontinue payment for all extrarenal transplantation. The not unexpected nationwide reaction to the death of a 7-year-old boy who was denied a bone marrow transplant subsequently led to review of the decision and the attempt to provide compromise solutions. Whether effective solutions can be identified remains unclear.

Interestingly, questions regarding cost effectiveness are generally being asked only of new medical therapies. In part, this is the fault of the medical community, which perhaps should be more active in identifying areas of established therapy that should be re-evaluated. It was rather incongruous, for example, to note that in the same journal issue in which the above state legislative decision was reviewed, a report appeared which indicated that the cost of treatment for moderate hypertension is estimated to be as high as 20 000 to 30 000 US dollars per year of life saved. While it is natural for new medical approaches to be the most likely targets for cost containment, scrutiny will probably soon extend to routinely accepted practices. As has been established in patients with end-stage renal failure, the new approach (transplantation) may be found to provide not only a better quality of life but also a more cost effective long-term treatment than the alternative (dialysis).

The second major issue limiting liver transplantation is the inadequate supply of donor organs. Currently, some patients die while waiting for a suitable organ, yet allografts are successfully transplanted from fewer than one-third of the estimated 12 000 to 18 000 people who die yearly in the United States in circumstances that could make organ donation feasible. Clearly, new approaches are needed to increase the number of organs available for transplantation and to improve upon the utilization of those organs procured.

The extent to which clinical liver transplantation will be applied, therefore, will be greatly influenced, at least in the short term, by economic and ethical issues. Ultimately, new approaches, including the use of non-human donor organs and more specific immunosuppressive regimens (for example, FK506 or more selective monoclonal antibodies), will undoubtedly make the procedure not only more available but also more effective and economically attractive.

Asher N, Stock PG, Bumgardner GL, Payne WD, Najarian JS. Infection and rejection of primary hepatic transplant in 93 consecutive patients treated with triple immunosuppressive therapy. Surg Gynecol Obstet 1988; 167: 474–84.

Bismuth H, et al. Hepatic transplantation in Europe. Lancet 1987; ii: 674–6.

Brems JJ, Hiatt JR, Ramming KP, Quinones-Baldrich WJ, Busuttil RW. Fulminant hepatic failure: the role of liver transplantation as primary therapy. Am J Surg 1987; 154: 137–41.

Broelsch CE, et al. Liver transplantation including the concept of reduced-size liver transplants in children. Ann Surg 1988; 208: 410–20.

Calne RY. Liver transplantation: the Cambridge/King's College Hospital experience. 2nd edn. Orlando, Florida: Grune and Stratton, 1987.

Cosimi AB, Cho SI, Delmonico FL, Kaplan MM, Rohrer RJ, Jenkins RL. A randomized clinical trial comparing OKT3 and steroids for treatment of hepatic allograft rejection. Transplantation 1987; 43: 91–5.

Cosimi AB, Jenkins RL, Rohrer RJ, Delmonico FL, Hoffman M, Monaco AP. Randomized clinical trial of prophylactic OKT3 monoclonal antibody in liver allograft recipients. Arch Surg 1990; 125: 781–5.

Delmonico FL, et al. Procurement of a whole pancreas and liver from the same cadaveric donor. Surgery 1989; 105: 718–23.

Evans RW. Cost-effectiveness analysis of transplantation. Surg Clin North Am 1986; 66: 603–16.

Iwatsuki S, Gordon RD, Shaw BW, Starzl TE. Role of liver transplantation in cancer therapy. Ann Surg 1985; 202: 401–7.

Iwatsuki S, et al. Liver transplantation in the treatment of bleeding esophageal varices. Surgery 1988; 104: 697–705.

Kalayoghi M, et al. Extended preservation of the liver for clinical transplantation. Lancet 1988; i: 617–19.

Maddrey WC. Transplantation of the liver. New York, Amsterdam, London: Elsevier Science Publishers, 1988.

National Institutes of Health Consensus Development Conference. Liver Transplantation. Hepatology 1983; 4 Suppl: 107S–110S.

Otte JB, de Ville de Goyet J, Alberti D, Balladur P, de Hemptinne B. The concept and technique of the split liver in clinical transplantation. Surgery 1990; 107: 605–12.

Ramsey G, Nusbacher J, Starzl TE, Lindsay GD. Isohemagglutinins of graft origin after ABO-unmatched liver transplantation. N Engl J Med 1984; 311: 1167–70.

Shaw BW, et al. Venous bypass in clinical liver transplantation. Ann Surg 1984; 200: 524–34.

Starzl TE. Surgery for metabolic liver disease. In: McDermott WV, ed. Surgery of the Liver. Oxford: Blackwell Scientific, 1989: 127–36.

Starzl TE, Demetris AJ, Van Thiel D. Liver transplantation. N Engl J Med 1989; 321: 1014–22, 1092–9.

Starzl TE, Iwatsuki S, Shaw BW, Gordon RD, Esquvel CO. Immunosuppression and other nonsurgical factors in the improved results of liver transplantation. Semin Liver Dis 1985; 5: 334–43.

Starzl TE, et al. Evolution of liver transplantation. Hepatology 1982; 2: 614–36.

Tarter RE, Erbs S, Biller PA, Switala JA, Van Thiel DH. The quality of life following liver transplantation. Gastroenterol Clin North Am 1988; 17: 207–17.

Terpstra OT, Reuvers CB, Schalm SW. Auxiliary heterotopic liver transplantation. Transplantation 1988; 45: 1003–7.

Tzakis A, Todo S, Steiber A, Starzl TE. Venous jump grafts for liver transplantation in patients with portal vein thrombosis. Transplantation 1989; 48: 530.

Wall WJ. Liver transplantation: current concepts. Can Med Assoc J 1988; 139: 21–8.

Williams JW, Peters TG, Vera SR, Britt LG, van Voorst SJ, Haggitt RC. Biopsy-directed immunosuppression following hepatic transplantation in man. Transplantation 1985; 391: 589–96.