Clinical Features and Diagnosis

Two forms of BA are described: (1) a more common (around 85% to 90% of cases) perinatal or postnatal sporadic form (“acquired”), possibly virus related, and (2) a less common (around 10% to 15% of cases) fetal or embryonic form (“congenital”), with a high frequency of associated malformations. BA splenic malformation (BASM) syndrome is characterized by polysplenia or asplenia, various laterality defects such as abdominal or complete situs inversus, mediopositioned liver and intestinal malrotation, cardiac laterality defects, and positional abnormalities of the major abdominal blood vessels. Intriguingly, BASM syndrome is less commonly seen in the Far East and Asia, where BA is twice as common as in the rest of the world. An increased incidence of maternal diabetes mellitus and female gender has been observed in the BASM syndrome. Children with this syndrome appear to have an increased frequency of infections, possibly leading to their poorer long-term prognosis compared with classic BA, including after liver transplantation, although no formal defect in their humoral immunity has been identified. It has also been suggested that the precarious blood supply to the biliary tree may be further jeopardized by their vascular abnormalities. A recent study suggests that among syndromic infants with BA, there may be a subgroup—around 6% of all BA children—who may have other, nonlaterality defects, including cardiovascular and genitourinary anomalies.

Clinical features of BA are jaundice, pale stools, and dark urine presenting at or soon after birth. Because physiologic jaundice, characterized by unconjugated hyperbilirubinemia, is common in neonates and most infants with BA have no major symptoms in the first few weeks of life, diagnosis is often delayed. This is a particular problem for infants with the perinatal or “acquired” form of BA, whose stools may have some pigment in the first few weeks of life before bile flow is completely obstructed ( Figure 69-1 ). Delayed diagnosis and surgical treatment carry a severe prognosis. Age at surgical correction is inversely correlated with the medium-term (up to 20 years) survival with native liver. Therefore, it is of paramount importance for health care professionals attending young infants to check the color of the urine and stools of all jaundiced babies, irrespective of their general health or age, and refer those with dark urine and pale stools promptly to specialized hepatology centers. No satisfactory screening test is available for BA, although promising results using universal neonatal stool color cards have been reported from Taiwan.

Acholic stool (left) strongly suggestive of a surgical problem during neonatal period. Children with biliary atresia can, however, have initially pigmented stools (right). (See plate section for color.)

Reproduced from Francavilla R, Mieli-Vergani G. Liver and biliary disease in infancy. Med 2002; 30 :45–7, with permission from The Medicine Publishing Company, Abingdon.

Physical examination and laboratory tests give little clue to the diagnosis of BA. Most of the affected infants have a mild degree of hepatomegaly and splenomegaly. Ascites or cutaneous signs of chronic liver disease are rarely detected in the early stages of the disease, when correct diagnosis is most important for effective surgical intervention. Biochemical findings are nonspecific, with levels of transaminases, γ-glutamyltranspeptidase (GGT), and alkaline phosphatase similar to those found in other forms of neonatal cholestasis. Coagulopathy, if present, is responsive to intravenous vitamin K. An ultrasound scan revealing an absent or abnormal gallbladder with an irregular wall or, in older infants, the “triangular cord” sign, is suggestive of BA. However, a normal gallbladder or absence of the triangular cord sign does not exclude BA.

Histologic examination of the liver by an experienced histopathologist leads to the correct diagnosis of BA in up to 90% of cases. Typical histologic findings are edematous portal tracts with inflammatory changes, bile duct proliferation, and intraportal bile plugs ( Figure 69-2, A ), but in very young babies these features can be much less obvious. Biliary radionuclide scans are only useful if isotope is demonstrated in the gut, thereby excluding BA and avoiding laparotomy. If the liver biopsy is ambiguous, but the stools remain acholic, endoscopic retrograde cholangiopancreatography (ERCP) is indicated to assess the patency of the biliary system. Currently, magnetic resonance cholangiopancreatography is a sensitive, but nonspecific, diagnostic tool for BA diagnostics. It is hoped that improvement in the magnetic resonance technique will offer a noninvasive way of assessing the biliary tree in suspected BA. If the cholangiography is not informative, an explorative laparotomy with intraoperative cholangiography is required but should be undertaken by an experienced surgeon, because hypoplastic extrahepatic ducts caused by a severe intrahepatic cholestasis may be interpreted as atretic, leading to an unnecessary and possibly damaging operation. The diagnosis of BA should always be confirmed by histologic examination of the excised biliary remnants ( Figure 69-2, B ).

Histologic appearance of biliary atresia. (A) Liver biopsy showing typical changes—edematous portal tract with increased fibrosis, duplicating bile ducts, and cholestatic plugs (hematoxylin-eosin stain, ×250). (B) Bile duct remnant obtained at Kasai portoenterostomy showing fibrosis and occlusion of extrahep­atic bile ducts (hematoxylin-eosin stain, ×125).

Reproduced from Francavilla R, Mieli-Vergani G. Liver and biliary disease in infancy. Med 2002; 30 :45–7, with permission from The Medicine Publishing Company, Abingdon.

Etiopathogenesis

Studies of bile duct remnants removed at surgery and from serial sectioning and reconstruction of surgical and postmortem liver specimens, indicate that BA is in most cases a sclerosing inflammatory process affecting previously formed bile ducts. The cause of such inflammatory process remains unknown. It is conceivable that BA represents a common final phenotypic pathway of neonatal liver injury caused by diverse causes, including developmental, vascular, or infectious factors, which may act antenatally or within the first 3 months of life in a genetically predisposed individual. Although BA is not an inherited disorder, because identical twins are usually discordant for the disease and occurrence of BA within the same family is exceedingly rare, it is possible that a genetic predisposition to an aberrant immune response against an exogenous agent and/or somatic mutations of genes regulating bile duct morphogenesis in fetal life are involved. Whatever the initiating event, as bile flow dramatically increases perinatally, bile leakage from the abnormal ducts is likely to trigger an intense inflammatory reaction, with consequent obliteration of the biliary tree. The detergent effect of the extravasated bile, however, cannot be the only explanation for the liver damage, because the disease can also progress in those patients in whom the Kasai portoenterostomy has achieved adequate bile flow. Proposed etiologic factors in BA include defective morphogenesis/genetic factors, vascular abnormalities, viral infection, exposure to toxins, and aberrant immune mechanisms.

Genetic Factors/Defective Morphogenesis

A separate clinical and etiologic subgroup, named BA splenic malformation (BASM) syndrome, is believed to be caused by defective morphogenesis of the biliary tree. A recessive insertional mutation in the proximal region of mouse chromosome 4 or complete deletion of the inversion (inv) gene in a murine model leads to anomalous development of the hepatobiliary system. However, no consistent mutations in the human homologue INV gene were identified in patients with BA, including those with BASM syndrome, suggesting that the INV gene is unlikely to be involved in the fetal cases of BA. CFC1 , coding for the CRYPTIC protein, is another gene investigated in BA. Although the precise function of the CRYPTIC protein is unknown, it is believed to act as a cofactor in the Nodal pathway that determines left–right axis development, disturbed in the BASM syndrome. Recently, a genetic mutation in exon 5 of the CFC1 gene, leading to the amino acid substitution Ala145Thr, was found in 5 of 10 infants with BASM syndrome. It is conceivable that CFC1 heterozygous mutations predispose to BASM, but then a second genetic or environmental factor is necessary to produce the disease phenotype.

Histologic features similar to the inherited group of disorders termed ductal plate malformation syndrome, which include congenital hepatic fibrosis and Caroli syndrome, have been reported in fetal type BA, suggesting that abnormalities in hepatocyte growth factor signaling during a critical period for mesenchymal/epithelial differentiation or other defects in intracellular adhesion molecule systems might be involved in the pathogenesis of this form of the disease. Of note, however, among nine children with BA diagnosed in utero, only one had BASM syndrome and none had histologic appearances of “ductal plate malformation.”

Cui et al. have recently used genome-wide associated studies to identify a potential susceptibility gene for BA at 2q37.3, encoding glypican 1 (GPC1), one member of proteoglycan family involved in intercellular signaling, including biliary development. They have used its analog gpc1 in a zebrafish model to demonstrate that gpc1 knockdown variant develops impaired bile production and intrahepatic biliary and gallbladder defects.

Tan et al. postulated that BA may derive from failure of the ductal plate structure remodeling between 11 and 13 weeks of gestation leading to the formation of an inadequate mesenchymal cuff around the developing hilar bile ducts, which could potentially be prone to rupture at the initiation of bile flow at 12 to 13 weeks of gestation.

Two studies report the presence of maternal microchimerism in children with BA. Hayashida et al. demonstrated three times more XX chromosome cells in males with BA compared with age-matched controls. Kobayashi et al. confirmed maternal microchimerism in male patients with BA and showed maternal microchimerism in females with BA by demonstrating presence of maternal human leukocyte antigens (HLAs). They suggested that maternal microchimerism could be a potential causative factor in BA, as maternal cells could elicit an immune response similar to graft-versus-host disease.

Vascular Abnormalities

Intrahepatic and extrahepatic bile ducts receive their blood supply exclusively from the hepatic artery. An arteriopathy affecting branches of the hepatic artery has been reported in patients with BA. These observations have led to the proposal that a vasculopathy may be the cause of BA, although whether vascular problems are primary or secondary to bile duct damage remains to be clarified.

Viral Infection

The reported seasonal clustering of human BA cases and experimental evidence of virus-induced BA have suggested a link between this disorder and exposure to viral agents. An initial bile duct epithelial injury caused by viral infection would lead to a progressive immune-mediated inflammatory and sclerosing process resulting in damage and eventually obstruction of the bile ducts. Several viruses have in turn been implicated in the pathogenesis of human BA, although published reports have been largely anecdotal and have given controversial results.

Epidemiologic studies on reovirus and rotavirus, currently believed to be the most likely infectious agents involved in the pathogenesis of BA on the basis of experimental models, are also conflicting. Interest in reovirus stems from the observation that infection in weanling mice causes bile duct and liver damage similar to that observed in BA, with the lesions persisting after the virus or viral antigens are no longer detected. Search for reovirus antibodies, however, in infants with BA has been inconclusive, possibly because of a high incidence of passively transferred maternal anti-reovirus immunoglobulin G. Studies in the liver tissue have also given discrepant results.

The demonstration that in mice BA can be induced by rotavirus has elicited a strong interest in the possible role of this pathogen in the causation of the human disease. Rotavirus infection of newborn mice in the first 24 hours of life leads to jaundice, acholic stools, and hyperbilirubinemia by the end of the first week of life. Progressive inflammation and obstruction of the extrahepatic bile duct is observed by 2 weeks of age, mimicking human BA. In this animal model, which has been replicated in various laboratories, bile duct injury is associated with an initial CD4 T helper 1 (Th1) immune response that through the release of interferon γ (IFNγ) induces macrophages to produce tumor necrosis factor α (TNF-α) and nitric oxide and immune activation persisting after viral clearance. Mack et al. have been able to provoke bile duct–specific inflammatory changes into naive syngeneic severe combined immunodeficiency (SCID) mice by adoptively transferring T cells obtained from mice in which rotavirus had induced BA. Emergence of the bile duct lesions in the absence of virus in the recipients suggests that biliary epithelial cell–specific autoreactive T cells are generated in the course of rotavirus infection and cause biliary damage. Although this animal model suggests that BA could be the result of rotavirus infection, its role in the human disease remains uncertain. It will be interesting to see whether the recent introduction of rotavirus vaccine into the routine immunization calendar of some countries, like Mexico and the United States, will affect the prevalence of BA.

Toxins

Outbreaks of BA in lambs and calves in Australia, possibly related to a fungal or other environmental toxin exposure, as well as the reported time and space clustering of BA cases in humans, have led to the proposal that an environmental toxin could be involved in its pathogenesis. However, to date no environmental agent, apart from viruses, has been clearly associated with BA in humans.

Immune Mechanisms

The presence of a portal tract mononuclear cell infiltrate in the liver biopsies of infants with BA has suggested a primary inflammatory process leading to bile duct obstruction. The mechanism by which immune cells induce bile duct damage, however, is unclear. The currently favored immune pathogenic scenario postulates that a viral or toxic insult to the biliary epithelium leads to the expression of new or altered antigens. In predisposed individuals, a peptide derived from these neoantigens would be presented to naive T lymphocytes by a professional antigen-presenting cell (APC), or directly by biliary epithelial cells that themselves express antigen-presenting HLA class I and II molecules. Of note, patients with BA have an increased number of resident liver macrophages—Kupffer cells, which can act as APCs. Primed Th1 lymphocytes would then orchestrate a damaging immune response, unfolding through the release of proinflammatory cytokines and recruitment of cytotoxic T cells, leading to progressive bile duct epithelial injury, fibrosis, and eventually occlusion of the extrahep­atic bile ducts. A recent study has demonstrated defective numbers and function of T regulatory cells in a murine model of BA with improved survival following adoptive transfer of T regulatory cells. No information is available on the role of the recently described proinflammatory Th17 cells.

Immune-mediated diseases are frequently associated with specific HLA molecules, which present peptide antigens to unprimed T lymphocytes. Several groups, therefore, have investigated a possible HLA predisposition to BA. Although an early report in a relatively small number of BA patients suggested that the possession of the allotype HLA-B12 (49% of BA patients versus 23% of control subjects) and haplotypes A9-B5 and A28-B35 could predispose to non-BASM BA in a European population, other groups could not replicate these findings, and a subsequent study from the same group using more sensitive molecular techniques in a larger number of patients (101 European children) failed to detect any HLA association. An Egyptian study showed an increased frequency of HLA-B8 and DR3 in 18 children with BA, 10 having the B8/DR3 haplotype that is associated with autoimmune hepatitis, primary sclerosing cholangitis, and inflammatory bowel disease. Although the available data do not support a strong link between HLA and BA, larger studies in patients of different ethnicities are needed. While exploring the involvement in BA predisposition of immunologically relevant non-HLA molecules that have been associated with other immune-mediated liver diseases, Donaldson et al. found no difference in polymorphism prevalence in genes encoding the proinflammatory TNF-α and interleukin (IL)-1 and the anti-inflammatory IL-10 cytokines in 101 children with BA when compared to 96 geographically and ethnically matched adult controls.

In keeping with a Th orchestrated immune process, an abundance of CD4+ T lymphocytes has been described in the liver and extrahepatic bile ducts of BA patients. These cells may have accumulated in the liver through the enhanced expression of adhesion molecules, an event that may also explain the high number of natural killer (CD56+) cells among the inflammatory infiltrate. DNA microarray techniques have shown upregulation of Th1 cytokine encoding genes, such as IFNγ, and downregulation of genes encoding Th2 cytokines. A possible Th2 involvement in the pathogenesis of BA, however, is suggested by the finding of circulating autoantibodies in affected children, including anti-neutrophil cytoplasmic antibodies and antibodies directed to alpha-enolase and vimentin. There is also evidence that in BA, biliary epithelial cells undergo an augmented rate of apoptosis promoted by de novo expression of Fas ligand. Bile drainage after portoenterostomy is reportedly better in patients with Fas ligand-negative biliary epithelial cells.

Recently, an attempt has been made to map the sequential behavior of an array of soluble mediators of inflammation reported or assumed to be involved in the pathogenesis of BA, including adhesion molecules and pro-inflammatory and anti-inflammatory cytokines, by prospectively studying 21 consecutive infants with BA at the time of Kasai portoenterostomy and serially thereafter for 6 months. As controls, other neonatal cholestatic diseases and infants with no liver disease were investigated. No significant differences in the baseline cytokine levels were found between BA and normal controls; however, IL-2 and IL-10 were significantly higher in comparison with other cholestatic disease controls. Soluble intercellular adhesion molecule-1 (sICAM-1) levels were substantially higher in BA at baseline compared to the two control groups. Within the first 6 months after portoenterostomy, all plasma cytokine and adhesion mol­ecule levels increased significantly, with the exception of IL-10, suggesting that the inflammatory process is progressive and involves both nonpolarized Th and macrophage immune responses that are not ameliorated by portoenterostomy. Among the circulating immune modulators investigated, sICAM-1 was the best candidate plasma biomarker for BA severity, with total serum bilirubin level being significantly positively correlated with the sICAM-1 levels at 1, 3, and 6 months after the surgery. Moreover, a cut-off level of serum sICAM-1 of 1779 ng/mL at 1 month after Kasai portoenterostomy predicted the need for transplantation in the first year of life, with 87% sensitivity and 92% specificity, suggesting that plasma sICAM-1 level could be used as a marker of disease progression and outcome in BA.

Treatment

The treatment of BA is surgical. In 5% to 10% of infants with BA, the surgeon can identify a patent common bile duct containing bile and in continuity with intrahepatic bile ducts. In these infants, a biliary-intestinal anastomosis via a long Roux-en-Y loop may allow bile to drain satisfactorily. In the majority of patients, however, the proximal common hepatic duct is absent or completely obliterated up to the point at which it enters the liver, and at the porta hepatis is replaced by fibrous tissue. This tissue needs to be transected flush within the liver, and then a Roux-en-Y loop of jejunum is anastomosed around the fibrous edges of the transected tissue, forming a portoenterostomy (Kasai procedure) ( Figure 69-3 ). For surgery to be effective, the intrahepatic bile ducts must be patent at the porta hepatitis. Later modifications of the Kasai procedure undertaken to reduce the risks of cholangitis are usually unsuccessful, unless there is a radiologically identifiable complication that is amenable to a simple surgical correction. Of note, these attempts could increase the operative risks during liver transplantation if this is required subsequently.

Schematic presentation of Kasai portoenterostomy with creation of jejunal Roux-en-Y loop.

There are three macroscopic types of BA:

• •
  

  Type I—affecting the distal part of the common duct

  
  
• •
  

  Type II—affecting the common hepatic duct, but sparing the gallbladder and common bile duct

  
  
• •
  

  Type III—affecting right and left intrahepatic ducts and the gallbladder

The most common form is type III (85% to 90% of cases), which is often referred to as “uncorrectable”; surgical reconstruction (portoenterostomy) is most challenging in this variant.

After surgery phenobarbital at a dose of 5 to 7 mg/kg/day is often used for long-term induction of the microsomal enzymes of the hepatocyte endoplasmic reticulum. If the jaundice reappears, the dose could be doubled, following exclusion of a mechanical problem with the Roux-en-Y loop. All children should be supplemented with fat-soluble vitamins, which should be continued medium or long term according to resolution of jaundice. Choleretic treatment with ursodeoxycholic acid (UDCA, 15 to 25 mg/kg/day) could also be considered. If the portoenterostomy is performed by an experienced surgeon, good bile flow with normal serum bilirubin values can be achieved in more than 80% of children operated on by 60 days of age, but in only 20% to 30% with later surgery. If bilirubin returns to normal, a 90% 15-year survival rate has been reported, with a good quality of life into the fourth decade. Up to 11% of children could be completely free of clinical and biochemical signs of liver disease after 10-year follow-up. If the bilirubin level is not reduced, the rate of progression of cirrhosis is not slowed and survival beyond the second birthday is unusual. If partial bile drainage is obtained, development of end-stage chronic liver disease may be delayed, but liver replacement usually becomes unavoidable by puberty. A Japanese report on long-term follow-up of BA patients, all of whom underwent a Kasai portoenterostomy in the 1970s, quotes a native liver 20-year survival of 44%, but with significant morbidity, including recurrent cholangitis and gastrointestinal bleeding in 37% and 17% of patients, respectively.

In view of the strong inflammatory component in the pathogenesis of BA, steroids have been in empirical use for many years. Besides their immunologic and anti-inflammatory effects, steroids can increase bile flow by inducing canalicular electrolyte transport. There is, however, no solid clinical evidence that steroids are of benefit. Most published studies are retrospective and uncontrolled, with only two being prospective and controlled. Davenport et al. performed a randomized double-blinded, placebo-controlled trial of low-dose oral prednisolone in 71 children with BA, and Petersen et al. used a high-dose steroid regimen to treat 20 consecutive patients after Kasai portoenterostomy and compared them with a historical control group. Neither study showed any difference in overall survival, liver transplantation requirements, survival with native liver, or jaundice-free survival with native liver. Since then three more studies, one open-label, one multicenter randomized, and one nonrandomized have investigated the use of high dose steroids after Kasai portoenterostomy (4 or 5 mg/kg/day starting at postoperative day 7) and concluded that this regimen improves serum bilirubin levels at 6 and 12 months after surgery, but not native liver survival at 2 years. Preliminary information from a North American multicenter, randomized, placebo-controlled Steroids in Biliary Atresia Randomized Trial (START) suggests that there is no difference in clearance of jaundice at 6 months or native liver survival at 2 years using a starting dose of 4 mg/kg per day of prednisolone tapered over 9 weeks post-Kasai.

An important postoperative complication of BA is cholangitis. This is seen in more than 50% of patients in the first 2 years after surgery, and a wide range of microorganisms could be implicated. Cholangitis is characterized by fever, recurrence or aggravation of jaundice, and, frequently, clinical features of septicemia. Blood culture, ascitic aspirate, or liver biopsy to identify the organism responsible should precede intravenous antibiotic therapy, which is continued for 14 days if a pathogen is identified. Often, however, the diagnosis of cholangitis is not obvious, and unexplained fever may be the only symp­tom. Intravenous antibiotics are then started empirically, after taking a blood culture and assessing liver function, C-reactive protein level, and full blood count. If the fever responds to the antibiotics, these are continued for 5 days. Should the fever recur after stopping them, a liver biopsy is performed for histologic examination and culture. Piptazobactam is currently the authors’ initial choice pending in vitro sensitivities. Long-term prophylaxis with rotating antibiotics may be indicated for recurrent cholangitis.

A degree of portal hypertension is present in almost all patients at the time of initial surgery. Approximately 50% of all survivors aged 5 years, even those with normal bilirubin levels, have esophageal varices, but only 10% to 15% have gastrointestinal bleeding. For these, elective endoscopy with variceal banding, if necessary, has been advocated as the treatment of choice.

Nontransplanted children with BA have a small long-term risk of developing hepatocellular carcinoma, which is not always associated with elevated serum α-fetoprotein. In the presence of unsuspected or small hepatocellular carcinoma detected at the time of the procedure, liver transplantation represents an effective method of treatment with no need for additional chemotherapy.

In approximately 10% of patients in whom the serum bilirubin level returns to normal, intrahepatic cholangiopathy progresses and complications of biliary cirrhosis ultimately develop. For these patients, and those with persistent jaundice, liver transplantation should be considered. With 1-year survival rates approaching 90%, and 5-year survival rates of more than 80%, liver transplantation is now a standard therapeutic option, although it remains a formidable medicosurgical procedure. The recipient is likely to have one or more life-threatening complications in the perioperative or postoperative period. Lifelong immunosuppressive therapy is required, frequently resulting in chronic nephrotoxicity and increased risk of opportunistic and community-acquired infections and malignancies, all necessitating a close medical and surgical supervision. Most of the survivors have a good quality of life and attend school, although the long-term medical and psychologic effects of liver transplantation in childhood are yet largely understudied. The supply of donors of suitable size and blood group, even with an increased use of split grafts where one donor liver is used for two recipients (usually one child and one adult), remains a major limiting factor in liver transplantation. Segmental graft transplant from living relatives has given survival rates of 90% in infants in whom Kasai portoenterostomy was unsuccessful. The results are better in children transplanted when heavier than 10 kg (or after the age of 1 year) and when the procedure is done electively.

Liver transplantation in patients with BA should be complementary to portoenterostomy, except for infants in whom decompensated cirrhosis has developed because of delayed diagnosis. The combination of Kasai portoenterostomy followed by transplantation in case of failure has considerably improved the survival of children with BA. The reported 4-year actuarial survival rate with native liver is 51%, and an overall (with native liver and post liver transplantation) 4-year actuarial survival rate is 89%. The precise indications, timing, and optimal management of some of the intraoperative and postoperative problems, including the control of rejection, remain the subject of ongoing research.