Figure 6.1 Facsimile reproduction from Edinburgh Medical Journal of John Thompson’s original case of ‘congenital biliary obstruction’.

6.3    PHENOTYPIC CLASSIFICATION

The cause of BA is not clear in most cases, and there is much speculation as to the importance of certain factors without much in the way of observable evidence. We have tried to develop a classification based more on observable traits and evidence rather than assumption – as the older simplistic division into ‘embryonic’ versus ‘perinatal’ was.

We can therefore recognise four variants, within which there is a high degree of clinical homogeneity, and thus a consistent aetiological mechanism arises as a result.

6.3.1  Biliary atresia splenic malformation syndrome (see Chapter 4)

This syndrome encompasses BA, splenic anomalies (usually polysplenia), vascular anomalies (usually preduodenal portal vein and absence of vena cava), visceral asymmetry (usually situs inversus) and cardiac anomalies (Table 6.1 and Figure 6.2).

For these peculiar anomalies to occur together, there must have been some insult operating at a critical point during embryonic development [11–13]. Our clinical series strongly suggested a much higher incidence of early trimester problems in those where the offspring had BASM (vs. otherwise isolated BA), including a marked association with maternal diabetes and possibly in vitro fertilisation [11,12]. The formation of the extrahepatic duct occurs from 20 to 38 days, in tandem with key events in the formation of the heart, spleen, determination of situs and so forth. At operation, these infants typically have no common bile duct (CBD); a tiny, atrophic gallbladder; and no evidence of inflammatory response. The liver is also usually symmetrical, whatever the nature of the abdominal situs. Furthermore, despite the protracted timeline, the liver parenchyma appears entirely normal at birth [14].

Table 6.1 Recognised anomalies in the BASM syndrome

Figure 6.2 BASM syndrome. Typically, infants will have polysplenia (a), situs inversus (b) and a preduodenal portal vein (c).

BASM is the variant that seems most likely to have an underlying genetic defect, although it is not a straightforward Mendelian^* one. Mutations in the CFC-1 gene (Ch2q 21.1 loci) were found in 50% of infants with BASM in one French study, but the same pattern was observed in a high proportion of normal controls as well [15]. The gene itself encodes for CRYPTIC protein, which is related to heterotaxy^† and cardiac anomalies (e.g. transposition of the great vessels and double-outlet right ventricle), and may also have a role in mesoderm and neural patterning during gastrulation.

A smaller number of infants with BASM have immotile cilia syndrome (Kartagener’s^‡ syndrome), which provides an interesting speculation about mechanism. Dysfunctional cilia could be incriminated in determination of visceral situs, but how cilial dysfunction interacts with the developing biliary tree is not known. Normally, only rats and squirrel monkeys have ciliated intrahepatic bile ducts, although there may be chemosensory cilia on cholangiocytes in humans.

There are other syndromic associations which are not within the BASM spectrum. Thus, a relationship with the cat eye syndrome has been reported characterised by coloboma, anorectal atresia and chromosome 22 aneuploidy [16].

Finally, other more common congenital abnormalities, such as oesophageal atresia, jejunal atresia and anorectal malformations, arise more commonly than would be expected by chance in infants with otherwise isolated BA.

6.3.2  Cystic biliary atresia

This subtype accounts for about 10% of all BA cases and is caused by extrahepatic cystic formation in an otherwise obliterated biliary tract. The cyst may be filled with bile or mucus depending on the degree of preservation of the connection with the intrahepatic ducts. The largest can be detected antenatally. In our series [17], 50% were detected between 18 and 20 weeks’ gestation at maternal antenatal ultrasound (US) screening. It is important that this type of BA is not confused postnatally with an early-obstructing choledochal malformation, and precise diagnosis can only be confirmed at surgery with a cholangiogram. Various patterns of intrahepatic ducts can be recognised, including that of a grossly abnormal, irregular, pruned tree, or it can have a cloud-like appearance caused by multiple interconnections of the filamentous intrahepatic biliary ductules (Figure 6.3). There does not appear to be any ethnic or genetic predilection, but there is undoubtedly a better outcome following KPE [17,18].

6.3.3  Cytomegalovirus-associated biliary atresia

The proposal that viruses may be responsible for numerous infantile cholestatic conditions is attributed to an American pathologist, Benjamin Landing,^* working in the 1960s. Since then, a number of candidate viruses have been suggested that may trigger bile duct injury in some way. The earliest focus was on reovirus Type 3, and serological studies seemed to show an increased frequency of positivity in affected infants compared with those with neonatal hepatitis [19]. Other viruses suggested included human papillomavirus, Epstein–Barr virus, cytomegalovirus (CMV) and rotavirus [20].

Figure 6.3 Cystic BA. Left: Operative image with cannulated cyst. Note the solid, atrophic gallbladder and string-like distal CBD. Right: Cholangiogram showing cyst and ‘cloud-like’ appearance of the primitive intrahepatic ducts.

More recent work has been based on identification of viral nucleic acid in the principal target organ of the liver and biliary tree. For instance, the Hannover group looked at 74 infants with BA and screened for a panel of 11 viruses using RNA and DNA profiling [21]. A minority of BA infants (42%) showed some evidence of intrahepatic viral RNA or DNA (33% reovirus, 11% CMV). This group, however, concluded that viruses were not ‘main players’ in the aetiology of BA, but were merely a secondary phenomenon and did not influence prognosis in any way [21,22].

Fischler et al. from Sweden [23] first showed a higher prevalence of anti-CMV antibodies in the mothers of BA infants and higher serum CMV immunoglobulin M (IgM) levels in infants with BA. They also described greater amounts of immunoglobulin deposits on the canalicular membrane of the hepatocytes in infants with BA with ongoing CMV infection. Our focus has been on CMV as a possible aetiological candidate. This is a double-stranded DNA virus of the Herpesviridae family that can infect biliary epithelial cells and hepatocytes and occasionally, but certainly not commonly, where CMV inclusion bodies can be seen.

In our recently published clinical study of 210 infants with BA, 20 (9.5%) of the patients were CMV IgM +ve at the time of presentation [24]. CMV +ve infants were characterised by late presentation, more deranged biochemical derangement and a non-Caucasian origin. They also scored higher for a range of semiquantitive inflammatory and fibrotic scores on liver histology, even when corrected for age. Interestingly, using anti-CMV immunohistochemistry staining, we could find no evidence of the actual virus at the time of surgery in any of those identified. We also looked at the various T cell subsets in a further cross-sectional study based on putative underlying aetiology in 37 infants and showed that those who were CMV IgM +ve were of the predominantly Th-1 (Tbet) subtype compared with both BASM and isolated BA groups [25]. Brindley et al. have shown that there appears to be a specific T cell profile suggestive of prior exposure to CMV in about half of their American infants with BA, together with defects in peripheral blood Treg cells, which might have allowed early exposure [26].

What was clearly different in our CMV IgM +ve infants was their outcome (Figure 6.4). They had much less effect from KPE than those with isolated BA, and this translated into markedly reduced survival [24]. There is a potential for antiviral therapy in addition to surgery for these infants, and we have started using adjuvant ganciclovir, although there are no convincing studies which have been published to date.

Figure 6.4 Influence of CMV – King’s College Hospital experience (2004–2011). Native liver survival (a) and survival (b) for infants with BA – CMV IgM +ve (n = 20) vs. CMV IgM–ve (n = 111).

The mechanism by which CMV may work is still very much open to debate in the clinical situation, but it seems more rational in the animal models (Figure 6.5). Thus, it is possible that the virus acts as a trigger, activating an immune-mediated destructive process during the immediate perinatal period. In mice, this is facilitated by a reduction in regulatory T cells (Tregs) and the inflammatory cascade [27–29].

Figure 6.5 Animal models. The usual model involves perinatal administration of rhesus Rotavirus to a susceptible strain of BALB/c mice. The affected offspring develop jaundice (a) due to intrahepatic inflammation (b) and extrahepatic biliary obliteration (c). (d) Scanning electron microscopy shows a small gallbladder and a vanishing CBD. There is only a narrow window perinatally where this can happen, and it can be prevented by maternal vaccination [28]. (Reproduced with permission from Petersen C, Davenport M, Orphanet Journal of Rare Disease 2013, 8: 128. doi: 10.1186/1750-1172-8-128.)

6.3.4  Isolated biliary atresia

This is the largest group for which there is actually no clear aetiological mechanism at work.

It may still be a developmental problem, but it must have later onset in fetal life than those with BASM, as no other system or organ is affected. One hypothesis is that as the intraand extrahepatic bile ducts develop from different sources with differing timescales, linkage must occur to provide for luminal bile duct patency from the canalicular membrane to the duodenum. This is achieved at 10–12 weeks of gestation, and if disturbed, perhaps BA is one consequence (‘interface’ BA). For example, in a histological study by Tan et al. [30], sections of the porta of normal fetuses at 12 weeks’ gestation were compared with porta hepatis resections from infants with BA. These proved to be remarkably similar, suggesting developmental arrest. Alternatively, bile ducts may form completely normally in the first trimester, but obliteration occurs at some later stage, possibly even postnatally. This may be associated with a possible ‘neonatal hepatitis’–type picture, or postnatal destruction of the extrahepatic biliary tree could occur secondary to an ischaemic insult affecting the bile ducts or following a bile duct perforation. However, in these cases the intrahepatic ductules dilate, which does not occur in ‘true’ BA cases.

There are other more circumstantial pieces of evidence suggesting a prenatal origin of even isolated BA. Thus, studies have found low levels of the hepatic-specific enzyme γ-glutamyl transpeptidase (γ-GT) in amniotic fluid in the second trimester levels in infants who later turned out to have BA [31]. Mushtaq et al. [32] measured bile acids in the Guthrie blood spot (taken in the first days of life) in an attempt to find a screening test for BA. Just over three-fourths (77%) of 61 infants who later proved to have BA had elevated total bile acids (>97th percentile, 33 μmol/L). More recently, Harpreet et al. from Texas [33] retrospectively identified split bilirubin values in 31 infants with BA obtained when they were <48 h old (about half their overall cohort with BA). All were abnormal and significantly higher than controls. These disparate studies strongly imply that the cause of the cholestasis is fully established at birth in more than half of infants with what we have been labelling as isolated BA – making even those ‘developmental’ BA.

There are other less well-developed hypotheses on aetiology, including maternal microchimerism (Box 6.1) [34,35] and a recently developed one based on an environmental toxin, biliatresone, identified in sheep and cattle [13,36] (Box 6.2).

We currently think of BA as a final common pathway or phenotype with a number of different causative factors (aetiological heterogeneity) (Figure 6.6). Some of these may be developmental (i.e. cystic BA and BASM) and include some in the isolated group. Others may be perinatal in origin and related to viral damage, although our CMV work would suggest a figure of only around 10%. Another factor which might be important, although difficult to prove, is a possible genetic susceptibility to BA – ADD3 has been suggested in some Chinese infants.

6.4    PATHOGENESIS

BA is an occlusive panductular cholangiopathy, which in most infants is probably present at birth (contentious). Progression of the condition results in cholestasis, hepatic fibrosis and ultimately cirrhosis. The histological appearance of the liver is characterised by portal tract oedema, bile duct plugging and proliferation, a small cell infiltrate and occasionally giant cell formation. The immunhistochemical appearance is characterised by abnormal expression of Class II antigens in about one-third and cytokines such as intercellular adhesion molecule (ICAM) (predominantly on the biliary epithelium) and vascular cell adhesion molecule (VCAM) (predominantly on the sinusoidal endothelium) [37–39]. There is also an infiltration of activated CD4 +ve lymphocytes and CD56 +ve natural killer (NK) cells. Most studies suggest polarisation with a predominantly Th1 and Th17 effector profile [25]. Infiltrating portal macrophages and Kupffer^* cell activation may also be prominent in some cases, both acting as antigen-presenting cells and ensuring the continuation of the chronic inflammatory process.

There is a systemic response to hepatobiliary inflammation which can be detected as increased levels of cellular adhesion molecules (sICAM and sVCAM) and pro-inflammatory cytokines such as interleukin-2 (IL-2), tumour necrosis factor-α (TNF-α) and IL-18 [40]. Those quoted remain high postsurgery and continue to rise up until at least 4 months after surgery.

The Japanese classification is used in most centres throughout the world (Figure 6.7) to describe the macroscopic appearance of the extrahepatic ducts. It is based on the level of the most proximal obstruction:

• Type 1 (5%): CBD, often associated with a cyst which therefore should contain bile. The gallbladder should also contain bile.

• Type (2%): Common hepatic duct. Transection of the most proximal porta hepatis should show both right and left ducts to contain bile.

• Type 3 (>90%): Transection of the porta hepatis should not show any remnant bile ducts – these being microscopic (Figure 6.8). Subdivided on the basis of the distal remnants into (1) patent CBD and usually mucocele of gallbladder, (2) intact but obliterated distal ducts and (3) absent CBD and usually atrophic gallbladder.

Figure 6.6 Aetiology of BA – aetiological heterogeneity.

Figure 6.7 Classification of BA. Obstruction at the level of the CBD (Type 1), at the level of the common hepatic duct (Type 2) and involving the whole of the extrahepatic biliary tree (Type 3). Type 3 can be subdivided according to the state of the distal duct (a, b or c).

6.5    CLINICAL FEATURES

Infants with BA invariably have a conjugated jaundice, dark urine and pale, acholic stools. When asked, these signs are usually discernible from the time of birth. Birth weight and gestational age are usually normal for both those with BASM and those with isolated BA, and initially infants feed apparently normally, although their weight gain is subnormal [4]. The liver is usually enlarged and firm, and in older infants (>10 weeks), there may be ascites or splenomegaly, indicating portal hypertension.

Low intestinal bile salt levels interrupt vitamin K absorption, leading to a coagulopathy and raised international normalised ratio (INR) levels. Bleeding may therefore be a presenting feature, before the underlying biliary pathology is recognised. Sometimes this is relatively trivial, such as persistent bleeding of the umbilical stump, but it can be serious and catastrophic with an intracerebral haemorrhage. Vitamin D levels are also low due to the same mechanism and can even be very low in the United Kingdom and especially in those of Asian ancestry.

Figure 6.8 Histological appearance of transected porta hepatis in Type 3 BA showing preservation of multiple biliary ductules.

Figure 6.9 Diagnostic approach to infants with conjugated jaundice.

The algorithm and investigation strategy at King’s College Hospital is given in Figure 6.9. The blood biochemistry shows a raised total and conjugated bilirubin (also known as the direct fraction, referring back to the original van den Bergh^* test reaction), with usually high levels of γ-GT (<200 IU/L) and raised levels of aminotransferases such as aspartate aminotransferase (AST)/alinine aminotransferase (ALT).

*  Hijmans van den Bergh (1869–1843), Dutch clinician. He discovered (in 1918) that the reaction of a diazo reagent with bilirubin produces a measurably coloured product alone (direct, conjugated) or with the addition of alcohol (indirect, unconjugated).

An accurate diagnosis of BA may be established preoperatively in about 80%–90% of cases. In our institution, this is achieved by exclusion of various medical causes of conjugated jaundice (e.g. α-1-antitrypsin deficiency), and a combination of ultrasonography and percutaneous liver biopsy. The former is relatively nonspecific, but if it does show intrahepatic bile duct dilatation then this will not be BA. Dilatation more suggestive of the other less common surgical causes such as choledochal malformation, inspissated bile syndrome and the rare spontaneous perforation of the bile duct (Table 6.2). Percutaneous biopsy is accurate in about 85% of cases, and for those not quite meeting histological criteria, then cholangiography is required. In an increasing number of specialist centres, including our own, endoscopic retrograde cholangiopancreatography (ERCP) would be the next step [41,42]. Other centres have used duodenal aspiration, looking for bile and radioisotope scans documenting failure of bile excretion.

Table 6.2 ‘Surgical’ causes of conjugated jaundice in infancy

6.6    SURGERY: KASAI PORTOENTEROSTOMY (FIGURES 6.10 AND 6.11)

For most infants with BA, an attempt should be made to restore bile flow with a KPE. There is a dearth of preoperative predictors of success, including the age at surgery, and none with absolute discrimination. The groups which do have a worse prognosis are those coming to surgery at >100 days of age, those with other malformations [12] those with CMV IgM +ve BA [23], while infants with cystic BA [17] and low AST to Platelet Ratio Index (APRI) values [43,44] have a better outlook.

Because of the lack of absolute predictive factors, primary liver transplantation should probably only be considered for those with overt clinical features of cirrhosis (e.g. obvious ascites), US evidence of cirrhosis (parenchymal heterogeneity) and liver failure (increasing INR and decreasing albumin). This group should only form <5% of all those treated [45].

The technical features of the operation are illustrated in Box 6.3. Complete resection of all biliary remnants should be the object in all morphological types, including those with cystic change, even when there may be an obvious patent proximal duct. In my practice, frozen section analysis, formerly used to confirm patent ductules, is not necessary, as it should not be possible to resect any further biliary tissue anyhow. We have also found that a straightforward Roux loop reconstruction is all that is required. There is no advantage to the creation of mucosal valves or stomas in the loop, which in any case do not reduce postoperative cholangitis and may make a subsequent transplant dissection more hazardous.

6.6.1  Operative problems

Is it BA? The diagnosis is uncommonly a problem at laparotomy. Biliary hypoplasia and neonatal hepatitis should be considered if there is bile in a normal-appearing gallbladder. Here, the cholangiogram can be difficult to interpret as the ducts are small and the key feature is to document communication to intrahepatic ducts. But if there is bile, then it cannot be Type 2 or 3 BA, and in most cases of Type 1 BA, there should be an obvious cyst.

BASM syndrome? The operator should probably stand on the left side of infants with situs inversus. Extra care is needed to preserve the exposed preduodenal portal vein and avoid kinking and so forth. The biliary appearance is usually different and resembles an upside-down octopus, with an absent CBD and little inflammatory change. Formation of a Roux loop with malrotated bowel needs extra care to avoid large mesenteric gaps, predisposing to internal hernia.

Is there ascites? A moderate degree of ascites is common, and in most KPE, no drain need be used. Consideration should be given to a drain if this is excessive. Fluid from the drain will need to be replaced with albumin solution until it diminishes. In any case, it should be removed at 7–10 days; otherwise, it will form an ascitic fistula.

Is there bleeding? Significant portal hypertension may lead to venous collateralisation around the porta, rendering dissection difficult. Bipolar diathermy is the principal method of haemostasis. The transected portal plate should not, however, be diathermied at all to avoid bile ductule damage. Bleeding from this source invariably stops with pressure alone.

6.6.2  Surgical controversies

6.6.2.1  SURGICAL EXPERTISE

This is not a simple operation and is rarely performed in general paediatric surgical practice. Studies looking at experience of the centre (by implication, the surgeon carrying out the procedure) have shown significant differences in outcome between low-flow and high-flow centres. For instance, 5-year native liver survival was 61% in UK high-volume centres during the 1990s compared with 14% in low-volume centres [45]. The obvious solution to this problem is to maximise centre experience and concentrate patients – which is what has been achieved in England and Wales since 1999 with the designation of three distinctly funded units. Results have suggested an improvement in national outcome compared with both historical experience [44,46,47] and national results from the European mainland [48–51], Canada [52] and Japan [53]. It is noteworthy that there are no national outcome statistics from the United States [17,54]. This centralisation approach is probably only possible in densely populated countries where surgical care is provided largely from a monopolistic institution such as the British National Health Service. A similar system has now been adopted in Switzerland [49], Finland [6] and Denmark [50].

Stay updated, free articles. Join our Telegram channel

Join