Definitions and Epidemiology
Biliary atresia (BA) is a progressive idiopathic, necroinflammatory process involving the extrahepatic biliary tree, which can be either segmentally or entirely affected. As the disease progresses, there is obliteration of the extrahepatic bile duct lumen and obstruction to bile flow (Figure 23–1). The result is cholestasis and chronic liver damage. With time, the intrahepatic biliary system becomes increasingly involved. BA is the most common cause of neonatal jaundice for which surgery is indicated and the most common indication for liver transplantation in children.
Many different classifications of BA have been proposed over the years. One classification system focuses on the anatomy of the biliary tree. The most comprehensive is that used in the Japanese BA Registry (Figure 23–2).1 In this classification system there are three major types of atresia: type I, atresia of the common bile duct (10% of patients); type II, atresia of the hepatic ducts (2% of patients); and type III, atresia at the porta hepatitis (88% of patients). Within each type there are several subtypes, which will not be discussed here. The first two types of BA are sometimes labeled “correctable atresia,” whereas type III corresponds to the so-called non-correctable type of atresia, and accounts for the majority of patients. However, with current hepatoportoenterostomy (HPE) techniques, most patients achieve at least some bile drainage, and thus the correctable versus non-correctable distinction is rarely used.
Three major types of BA, based on anatomic findings.1 This image was adapted from an image available at www.mayoclinicproceedings.com. The original image is found in reference #1. Type I (about 10%): atresia of the common bile duct. Type II (about 2%): atresia of the hepatic duct. Type III (over 88%): atresia at the porta hepatis.
Another classification system is based on the associated malformations and onset of jaundice. Many textbooks refer to embryonic and perinatal BA. In the former, there is typically onset of jaundice at birth and often multiple major malformations, the most common of which is heterotaxy. In the latter, jaundice appears several weeks after birth and there are usually no associated malformations. Perinatal BA accounts for 70–80% of all cases. Because there are so many exceptions to these patterns, new nomenclature systems have been proposed. Biliary atresia splenic malformation (BASM) is a term that specifically captures those previously termed “embryonic” who have heterotaxy. BASM applies to all infants with BA and either asplenia or polysplenia. The typical associated malformations include midline liver, malrotation, preduodenal portal vein, interrupted inferior vena cava, and cardiac malformations. Ten to 15% of all cases of BA have this categorization. Another 10–15% have major malformations in association with BA but do not fall into any stereotypical pattern. As our understanding of BA improves, it is likely that there will be an improved classification system.
BA occurs with an estimated frequency of 1 in 8000 to 1 in 20,000 live births, depending on the country. In the United States, there are 0.65–0.85 cases per 10,000 live births,2 resulting in 250–400 new cases per year. The highest incidence occurs in Asians. In Taiwan, there are 1.46 cases per 10,000 live births. The female-to-male ratio is close to 1, and sexual predominance varies based on the study. BA appears to occur more frequently in certain racial groups in the United States. The most recent epidemiologic study of potential risk factors for BA identified 112 cases of BA using data from the National Birth Defects Prevention Study. This study found that patients with BA were more likely to be born to non-Hispanic black mothers than to non-Hispanic white mothers, OR = 2.29 (95th CI = 1.07 – 4.93).3 Another study used a population-based birth defects surveillance system in Atlanta and identified 112 patients born with BA from 1968 to 1993 and also found higher rates of BA among non-white infants compared to white infants (0.96 versus 0.44 per 10,000 live births).4 Caton et al. identified 369 BA cases from 1983 to 1998 in New York State and found that infants with BA were born to black mothers at an increased rate when compared to white mothers.5 These reported racial differences in incidence may reflect genetic, socioeconomic, and/or environmental factors.
Studies looking at seasonal variations of BA have shown inconsistent results. The Atlanta study by Yoon et al. found seasonal clustering with a greater number of cases occurring from December to March.4 In the New York Study, seasonal patterns varied by region, with infants in New York City more likely to be born in the spring, and infants from outside of the city more likely to be born in the period from September to November.5
Several epidemiologic studies have looked at gestational age, birth weight, and maternal age in BA patients. The findings from one study to the next are conflicting and therefore no definitive risk factors have been identified. For example, the Atlanta study found an increased incidence in term infants with low birth weights (<2500 g). On the other hand, The et al. found that patients with BA were more likely to have normal birth weight.3 This study also found that infants with BA were more likely to be preterm (≤36 weeks). The largest and most comprehensive epidemiologic study used the Swedish national database, which included 99% of all children born in a 10-year period and identified 85 cases of BA.6 They found that high maternal age (over 34 years of age), parity of at least 4, prematurity (22–32 weeks), and small for gestational age (SGA) were all associated with an increased incidence of BA. Although these findings are somewhat inconsistent between studies, the individual findings related to racial clustering, seasonal patterns, and birth weight have led researchers to investigate several possible environmental and genetic factors that may be involved in BA.
The cause of BA is unknown, but is most likely multifactorial (Figure 23–3).7,8 Some of the proposed mechanisms include: (1) defects resulting from a viral infection; (2) injury caused by toxin exposure; (3) immune or autoimmune dysregulation; (4) genetic predisposition. All four of these proposed etiologies may be involved, and there is some speculation that BA actually represents several different diseases all with the same common phenotype. This theory is supported by the fact that there is great variability in the timing of presentation, rate of disease progression, and phenotype, with the majority of patients having only BA and a minority having multiple other congenital anomalies.
There are several hypotheses regarding the pathogenesis of BA. The cause of BA is most likely multifactorial. Adapted from Sokol et al.23
Early reports of seasonal variation, with a predominance of cases in the winter months, suggested a viral etiology. However, as mentioned in the previous section, further studies looking at seasonality patterns are inconsistent. Nonetheless, various different viruses have been implicated, including reovirus, rotavirus, cytomegalovirus, and human papillomavirus. Of all these proposed viral etiologies, the evidence for reovirus is strongest. Identification of a specific viral agent is complicated by the fact that although a virus may trigger the first stage of inflammation in BA, by the time the patient has been diagnosed with BA, the virus may have cleared.
Reports of increased incidence of BA in specific areas, mainly urban locations, suggest a possible toxin-mediated defect. The strongest evidence for a toxin as a possible etiology for BA comes from three reported outbreaks of BA in lambs in Australia in 1964, 1988, and 2007. During gestation, the ewes that gave birth to affected lambs had grazed on lands that had recently been flooded. A significant number of offspring were thin, jaundiced, had acholic stools, and eventually died. The lambs had enlarged, firm, dark livers with shrunken gallbladders. These findings are consistent with BA. The proposed theory is that the pregnant ewes ingested a toxin when grazing on lands previously submerged. The toxin may have triggered BA. To date, a specific toxin has not been identified.
Mack et al. and Bezerra et al. suggest that immune or autoimmune dysregulation in response to an initial insult, such as a virus, leads to BA. Immunohistochemical staining of livers from infants with BA revealed increased CD4+ and CD8+ T lymphocytes as well as CD68+ Kuppfer cells, compared to normal controls.9 In addition, increased production of pro-inflammatory cytokines, such as IL-2, IL-12, interferon-γ, and tumor necrosis factor-α, was found. Bezerra et al. demonstrated over-expression of interferon-γ RNA in the livers of infants with BA. In addition, his group used a rhesus rotavirus (RRV) model of BA and found that when interferon-γ knockout mice are inoculated with RRV, periductal inflammation develops, but the lumen of the bile ducts remains patent.10 They therefore propose that interferon-γ may play a regulatory role in biliary obstruction.
The role of genetics in the development of BA is currently of great interest. It is clear that BA does not follow Mendelian inheritance. For the most part, twin studies to date show discordance. However, there are multiple reports of recurrence of BA in the same families.11–13 We know that the incidence of BA is higher in certain Asian populations, raising the possibility that there are unique polymorphisms among Asians resulting in susceptibility to BA. One study by Zhang et al. looked at gene expression profiles in liver tissue of perinatal BA compared to embryonic BA.14 The two forms of BA had unique expression patterns, with the embryonic form showing over-expression of genes involved in regulatory functions and the perinatal form having over-expression of genes involved in metabolic functions.14 The CFC1 gene has been implicated in BASM. In one study the gene was analyzed in 10 patients with BASM.15 Five of the 10 patients had a heterozygous transition in exon 5 of this gene. The frequency of this mutation was two times higher than in controls. The inversin gene has also been implicated in mice with BASM; however, this gene does not seem to be causative in humans. These examples suggest that genetics may play a role in BA. How much it actually contributes to the etiology of BA and how is yet to be determined. Schreiber and Kleinman proposed that perhaps BA is the result of a multi-hit pathologic process, in which a viral or toxic insult in a genetically predisposed individual leads to BA.16
Jaundice is the first sign of BA. All patients with BA have jaundice. Other signs and symptoms may occur, but vary from patient to patient (Table 23–1). Jaundice typically develops in the first weeks of life but it can also occur at birth. Some infants have acholic stools (Figure 23–4). The color of the stool may vary from day to day. Most infants have dark urine because of bilirubin excretion into the urine. In the majority of cases, infants with BA are born full term, have a normal birth weight, and initially thrive and seem healthy.
|Time of Onset
|Secondary to conjugated hyperbilirubinemia; may be difficult to assess in a dark-skinned infant
|Color may vary from day to day
|Occurs as a result of bilirubin excretion in the urine
|May make stool in diaper appear darker than it really is
|Patients may or may not have hepatomegaly
|Rarely present in first few months of life
|Rarely present in first few months of life