Biliary Atresia Splenic Malformation Syndrome

Although the association of BA with polysplenia had been recognized for some time, clarification of what constitutes BASM (Table 40.2) is only relatively recent (Davenport et al, 1993). The constellation of other anomalies is peculiar, and the reasons for this are still obscure. The common embryologic insult may simply be timing (30 to 35 days) rather than a specific genetic defect. Key genes are important in both bile duct development (JAG1, HNF-6 [ONECUT1]; Kohsaka et al, 2002) and visceral and somatic symmetry (INVS, CFC1; Bamford et al, 2000; Shimadera et al, 2007), although correlation with the human condition is patchy. A possible genetic link has recently been reported by a French group, who found an increased frequency of mutations in the CFC1 gene (on chromosome 2) compared with controls (Davit-Spraul et al, 2008).

Table 40.2 Spectrum of Anomalies in Biliary Atresia Splenic Malformation Syndrome

A maternal diabetic milieu seems to predispose to BASM in the same way that it can cause transposition of the great vessels and sacral agenesis, although these are not part of the usual spectrum of defects. The macroscopic appearance of the biliary tree and liver is also somewhat different from that of isolated BA, with frequent absence of the common bile duct and a miniscule gallbladder and small proximal remnant. The liver is also usually symmetric, whatever the abdominal situs.

Epidemiology

A clear difference in incidence is seen across the globe. The highest is reported in Taiwan, at one in 5000 live births (Hsiao et al, 2008). In Japan, the incidence is one in 10,000 (Nio et al, 2003), with national studies from the United Kingdom and France reporting an incidence of one in 17,000 and 19,000, respectively (Chardot et al, 1999; Livesey et al, 2009; Serinet et al, 2006). Although there are no national studies in North America, smaller regional studies suggest an incidence closer to that of Europe (Yoon et al, 1997) but with some interracial variation (The et al, 2007). Significant regional differences within the United Kingdom have also been observed, again ascribed to variations in racial origin and the nature of the BA. For example, developmental BA is more common in white infants than in those of Asian origin (Livesey et al, 2009).

A distinct female predominance in seen in those with developmental BA that is not seen in the isolated BA group for unknown reasons (Davenport et al, 1993, 2006; Livesey et al, 2009). A seasonal variation in incidence has also been observed in some regionally based studies (Yoon et al, 1997; The et al, 2007), although this has not been confirmed in the larger, national studies (Chardot et al, 1999; Livesey et al, 2009).

Pathophysiology

The macroscopic appearance of the extrahepatic biliary tree ranges from being inflamed and hypertrophic yet intact to being an atrophic negligible remnant with absent parts. The histologic appearance within the liver appears more consistent with early portal tract inflammation, a mononuclear cell infiltrate, bile duct plugging, and proliferation and fibrosis giving way to overt biliary cirrhosis from about 3 months on.

The nature of the mononuclear infiltrate has assumed a degree of importance, and many authors suggest this is the specific primary destructive element of BA (Davenport et al, 2001; Mack et al, 2007). Although not necessarily uniformly applied across the spectrum of BA, the infiltrate is largely composed of CD4+ T lymphocytes, specifically Th1 (Mack et al, 2004) and CD56+ natural killer (NK) cells (Davenport et al, 2001; Shivakumar et al, 2009), which exhibit markers for proliferation (CD71+) and activation (particularly LFA-1+ but also CD25+). Infiltrating CD8+ cells appear to lack markers of activation—such as perforin, granzyme B, and Fas ligand (Ahmed et al, 2001)—in contrast to other diseases such as primary biliary cirrhosis (see Chapter 9).

Expression of cell adhesion molecules, proteins involved in cell-cell binding, is an important precursor to an immune-active cell infiltrate. Both interstitial cell adhesion molecule (ICAM)-1 and vascular cell adhesion molecule (VCAM)-1 are important in binding circulating leukocytes by interaction with cell-surface integrins, and they are expressed on sinusoidal and biliary epithelium of infants with BA (Dillon et al, 1997; Davenport et al, 2001).

Expression of soluble(s) ICAM-1 and sVCAM-1 in the circulation is also increased (Minnick et al, 1998; Davenport et al, 2005), as is seen in other immunologically mediated liver diseases such as primary biliary cirrhosis and sclerosing cholangitis; levels of sVCAM-1 at the time of presentation have been shown to be prognostic (Davenport et al, 2005). Increasing levels of various other cytokines (interleukin [IL]-2, tumor necrosis factor [TNF]-α, IL-18) can also be shown postoperatively (Fig. 40.2), but most discriminate poorly between patients who would clear their jaundice and those who would not; conversely, some cytokines (IL-2, interferon [IFN]-γ, IL-4, IL-10, TNF-α, and sICAM-1) were better predictors of subsequent need for early transplantation (Narayanaswamy et al, 2007).

FIGURE 40.2 Postoperative plasma tumor necrosis factor (TNF)-α levels in biliary atresia (n = 21), showing progressive increase whether clearance of jaundice was achieved (good, n = 11) or not (poor, n = 10).

(Modified from Narayanaswamy B, et al, 2007: Serial circulating markers of inflammation in biliary atresia: evolution of the post-operative inflammatory process. Hepatology 46:180-187.)

Resident (Kupffer cells) or systemic/recruited macrophages and monocytes seem to play a dual role in BA, as both the presenters of antigenic material in the first place and as the initiating driver for fibrosis in the development of chronic liver disease. Tracy and colleagues (1996) first showed increases in resident macrophages (CD68+) with marked expression of the lipopolysaccharide receptor CD14+. Increased levels of both CD68+ cells and circulating markers of CD68+, TNF-α, and IL-18, have been shown to predict outcome post-Kasai (Kobayashi et al, 1997; Narayanaswamy et al, 2007).

Viral Hypothesis of Biliary Atresia

One suggested cause of BA is infection with hepatotropic viruses, such as rotavirus, reovirus, and cytomegalovirus, either as an indirect trigger of an abnormal immune-mediated reaction or as an actual pathologic agent. The evidence is largely based on animal models in which intraperitoneal innoculation on the first day of life is able to mimic postnatal biliary inflammation (Riepenhoff-Talty et al, 1993; Petersen et al, 1998). Subsequent translation into humans has been more problematic. Serologic evidence of viremia is contradictory (Morecki et al, 1982; Tarr et al, 1996), and isolation of both viral RNA and DNA have (Riepenhoff-Talty et al, 1996; Tyler et al, 1998; Fischler et al, 1999) and have not (Brown et al, 1988; Steele et al, 1995; Bobo et al, 1997; Jevon & Dimmick, 1999) been identified by PCR techniques.

The most recent study is one from Hannover, Germany, in which liver biopsies from 74 infants obtained at the time of portoenterostomy were tested for a panel of DNA and RNA hepatotropic viruses (Rauschenfels et al, 2009). One or more viruses was detected in about one third of infants, and the rate of detection increased with infant age, which suggested that observed viral infection was more likely to be a secondary finding rather than a specific cause.

One problem with the viral hypothesis is the lack of an explanation as to why such ubiquitous organisms only damage the biliary tree in such a small fraction of infants and children. It may be that such infants are in some way susceptible to cholangiocyte damage and require a (viral) trigger to set off an immune-mediated destructive process (Mack, 2007).

RNA gene expression studies in rotavirus-induced murine models have shown upregulation of many proinflammatory genes, among them the IFN-γ inducers IRF7 and IRF9 (Carvalho et al, 2005). Similar gene-expression profiling in human BA samples has also shown overexpression of immune regulatory genes and differences between isolated and syndromic forms (Bezerra et al, 2002; Zhang et al, 2004).

A number of studies have shown different but specific elements of genetic variation between patients with isolated BA and unaffected control subjects. Thus, susceptibility may be imparted by polymorphisms in a number of relevant genes. Examples of this include vascular endothelial growth factor (VEGFA; Lee et al, 2010), macrophage migration inhibitory factor (MIF, Arikan et al, 2006), and intercellular adhesion molecules (ICAMs; Arikan et al, 2008).

Most recently, maternal microchimerism, a novel mechanism of immune damage, has been suggested based on the observation that male infants with BA have a threefold increase in cells of maternal origin in their livers (Hayashida et al, 2007). These were later shown to be maternal-origin chimeric CD8+ T cells and CD45 NK cells capable of initiating immune cholangiolar damage (Muraji et al, 2008).

Diagnostic Workup

Liver biochemistry will show a conjugated hyperbilirubinemia that is variable but rarely above 250 µmol/L (more than 15 mg/dL), modestly raised transaminases (aspartate aminotransferase, alanine aminotransferase), significantly raised γ-glutamyl transpeptidase (GGT), and usually normal protein and albumin levels. Hemoglobin and white cell counts are normal, although the platelet count may be raised.

The differential diagnosis of medical conditions that cause conjugated jaundice is long, and all can mimic the occluded bile duct of BA. Exclusion of TORCH infections—Toxoplasma, cytomegalovirus, hepatitis, and so on—α1-antitrypsin deficiency, Alagille syndrome, and cystic fibrosis should run in parallel with efforts to establish a diagnosis of BA.

US examination is a key part of the workup in order to exclude other possible surgical diagnoses (choledochal malformation, inspissated bile syndrome), and evidence of at least intrahepatic bile duct dilation should be visible (Davenport et al, 2003). Although suggestive and supportive, US is uncommonly diagnostic of BA and typically shows a shrunken, atrophic gallbladder with no evidence of filling between feeds. Some US signs do appear more reliable, such as the “triangular cord sign” (Park et al, 1997). This suggests a sonographic appearance of the proximal solid biliary remnant in front of the bifurcation of the portal vein, and advocates quote accuracy rates of greater than 80% (Humphrey & Stringer, 2007).

Radioisotope hepatobiliary imaging with technetium-labeled iminodiacetic acid derivatives has a role in some centers in showing the need for laparotomy. In BA there will be severe cholestasis and no bile excretion, but this can be seen in many cases of neonatal hepatitis and other medical conditions, depending on the severity of cholestasis. Placement of a nasoduodenal tube and aspiration over 24 hours has been used in many centers, chiefly Asian, as a simple, low-cost diagnostic measure, although this has rarely found favor in Western centers. Percutaneous liver biopsy to assess for histologic features of large-duct obstruction versus those of neonatal hepatitis is safe and well-tolerated but needs an experienced pathologist to interpret.

Direct cholangiography is becoming more prevalent in larger centers with direct access to endoscopic retrograde cholangiopancreatography (ERCP) (Shanmugam et al, 2009; Petersen et al, 2009). It requires a sufficiently small endoscope and a level of familiarity with the procedure, as failure to cannulate the bile duct are features of both BA and an inexperienced operator. Laparoscopy with or without liver biopsy offers an alternative to ERCP, and it may be the prelude to the definitive procedure. Direct puncture of the gallbladder is relatively straightforward to show presence or absence of bile and as an access point for a cholangiogram. This can be coupled with liver biopsy.

Cholangiogram

The diagnosis is always confirmed initially through a limited right-upper quadrant muscle-cutting incision, allowing access to the gallbladder (does it contain bile?) and a cholangiogram if needed. It should be obvious that a cholangiogram is not always possible, simply because the gallbladder may not have a lumen. Nonetheless, this is in itself presumptive of the diagnosis. To exclude BA, the cholangiogram should demonstrate proximal intrahepatic ducts. This can be difficult, as contrast preferentially fills the distal duct and duodenum. A small, distal vascular or “bulldog” clamp should aid reflux into the more proximal biliary tree if patent.

Mobilization

The liver should be fully mobilized by dividing the falciform, coronary, and triangular ligaments such that the organ can then be everted outside of the abdominal cavity (Fig. 40.3). This allows full exposure of the porta hepatis and facilitates dissection; it also reduces venous return, and an increase in intravenous volume support is always needed, together with close communication between surgeon and anesthesiologist.

FIGURE 40.3 Operative appearance of liver in biliary atresia. The liver is relatively soft, and the gallbladder is collapsed and contains no bile, which is consistent with a type 3 biliary atresia.

Portal Dissection

Initially the gallbladder is mobilized and the distal CBD divided (Fig. 40.4A), with the dissection consisting of elevation of the proximal pyramidal remnant from the right hepatic artery and bifurcation of portal vein (Fig. 40.4B). Small veins to the porta hepatis need division and facilitate downward traction of the portal vein and exposure of the posterior part of the portal plate and caudate lobe. On the left side, the recessus of Rex, where the umbilical vein joins the left portal vein, should be exposed; an isthmus of liver parenchyma from segment III to IV may need division by coagulation diathermy to achieve this. On the right side, remnant biliary tissue can be identified almost circumferentially around the right vascular pedicle. The division into right anterior and posterior pedicles should be visualized, and the posterior pedicle’s biliary elements should also be incorporated into the subsequent Roux loop. Excision of remnants flush with the liver capsule is accomplished by developing a plane between solid white biliary remnant and the underlying liver (Fig. 40.4C); excising liver parenchyma itself does not seem to improve bile drainage. All the denuded area of the porta hepatis needs to be incorporated into the Roux loop as the portoenterostomy. Some authors (Gittes & Koybayashi, 2007) have suggested a role for frozen section of the portal plate to determine that the observed ductules are large enough. My view is that there should be nothing left to resect from the initial dissection. Burrowing into liver parenchyma itself seems pointless, possibly because subsequent scarring obliterates any exposed ductules.

FIGURE 40.4 Schematic illustration of Kasai portoenterostomy. A, Mobilization of gallbladder and division of distal common bile duct. B, Elevation of biliary remnants and separation from vascular structures at level of porta hepatis (portal vein confluence, right hepatic arterties, and, to a lesser extent, left hepatic arteries). C, Transection of portal plate from umbilical “point” to around the division of right-sided (anterior and posterior) portal pedicles. D, Portoenterostomy with 40- to 45-cm retrocolic Roux loop.

Roux Loop and Portoenterostomy

A standard retrocolic Roux loop measuring 40 to 45 cm should be constructed (Fig. 40.4D). The jejunojejunostomy should lie about 10 cm from the ligament of Treitz and can be stapled or sutured. The proximal anastomosis must be wide (approximately 2 cm), and as such, an end-to-side tension-free arrangement is preferred. Fine, precise sutures (e.g., 6-0 PDS) at the edge of the portal plate are satisfactory. Remnant ductules are concentrated in the right and left recesses, and it is important that all parts of the transected plate are incorporated.

Aftercare

Nasogastric drainage and intravenous fluids are required for 2 to 4 days postoperatively. Antibiotics, intravenous at first and then oral, are given in our institution for 1 month. All these infants, even those clearing their jaundice, have subnormal quantities of intestinal bile and to ameliorate the effects of fat malabsorption, they should have a formula rich in medium-chain triglycerides. Vitamin supplementation is standard, both enteral and parenteral, and choleretic agents, such as ursodeoxycholic acid, may have some value but only in those patients who have reestablished bile flow. The value of steroids is discussed later.

Cholangitis

Cholangitis is presumed to occur by bacterial ascent from the Roux loop via patent biliary ductules (see Chapter 43). It is therefore uncommon in those who have no effect from Kasai surgery. Nonetheless, it may occur in 30% to 50% and is most frequent within a year of surgery (Ecoffey et al, 1987). Clinically, cholangitis is characterized by worsening jaundice, fever, and acholic stools. The diagnosis may be confirmed by blood culture or by percutaneous liver biopsy, but it is important to treat suspected cases early with broad-spectrum antibiotics effective against gram-negative organisms (e.g., ceftazidime, amoxycillin, piperacillin). Sometimes recurrent cholangitis may be problematic, which can be associated with intrahepatic cyst formation (Bu et al, 2002), presumably acting as a bacterial nidus. Most patients respond to prolonged antibiotic prophylaxis, either with intravenous broad-spectrum antibiotics via an indwelling vascular access device or with oral nonabsorbable antibiotics (e.g., neomycin). If cholangitis still recurs despite these measures, liver transplantation should be considered.

Sometimes cholangitis occurs some years after the portoenterostomy in children with otherwise good liver function. It is important in these cases to exclude a mechanical Roux loop obstruction, typically at the level of the mesocolon, because corrective surgery for obstruction is relatively simple (Houben et al, 2006). Percutaneous transhepatic cholangiography and radionuclide hepatic imaging may be useful for diagnosis.

Portal Hypertension

Elevated portal venous pressure can be shown in virtually all infants at the time of the Kasai operation (Duché et al, 2006). However, whether it persists or regresses depends on the degree of established fibrosis and, most importantly, the response to surgery (Kang et al, 1993). About 60% of children who survive to 2 years will have endoscopic evidence of esophageal varices, although only approximately half of these will ever bleed. The severity and degree of the varices themselves are not necessarily related to the original degree of liver fibrosis or to the number of episodes of cholangitis (Kang et al, 1993). Endoscopic surveillance of the upper gastrointestinal tract in children with treated biliary atresia has been suggested from about the age of 2 years because of this unpredictability of variceal formation (see Chapter 71).

The age at first hemorrhage is typically approximately 2 to 3 years, and it manifests as sudden hematemesis and melena from rupture of an esophageal varix. Ectopic intestinal variceal bleeding may also occur, usually in an older child with relatively normal liver function. The initial treatment of bleeding varices is supportive, with restoration of blood volume and correction of platelet deficiency and any coagulopathy. Drugs such as octreotide, or more recently terlipressin, may be used be used to reduce portal venous pressure. Banding has become the most efficacious technique to control varices, although it may be limited by patient size (Hall et al, 1988). For infants and small children, endoscopic sclerotherapy is still effective (Stringer et al, 1989), using sclerosants such as ethanolamine or sodium tetradecyl sulfate. Even with various pharmacologic and endoscopic alternatives, sometimes the only means of variceal control is a Sengstaken or equivalent pattern tube. Transjugular intrahepatic portosystemic shunt (TIPS) as a method of acutely reducing portal hypertension is also possible in children with BA, but it should be regarded only as a “bridge” to transplantation (Johnson et al, 1996; Cao et al, 1997), and it is not widely available.

Hepatopulmonary Syndrome and Portopulmonary Hypertension

Hepatopulmonary syndrome (HPS) is characterized by cyanosis, hypoxia, and clubbing and is due to shunting across the pulmonary vascular bed, likely caused by failure to deactivate vasoactive substances, possibly endothelin-1, by the damaged liver (Iqbal et al, 2008). Patients with HPS have higher levels of exhaled nitric oxide, presumably the final link in this chain. It is more commonly seen in those with BASM, suggesting a congenital element (Davenport et al, 2004a), and it can occur even in those who are anicteric. The diagnosis is made by arterial blood gas estimation showing significant hypoxia with no improvements seen on increasing ambient oxygen. Radionuclide lung scans with macroaggregated tagged albumin can be used to quantify the degree of shunting and can range from 4% to 50% (Al-Hussaini et al, 2010).

Portopulmonary hypertension is much less common and may share the same underlying vascular mechanism as HPS. Although it is often silent and perhaps only detected because of cardiomegaly on a chest radiograph, it can be a cause of sudden death and is associated with systolic pulmonary arterial pressures greater than 40 mm Hg. Although no real therapy is available for the pulmonary vascular anomaly, it is a definite indication for liver transplantation and is usually reversible, albeit with an increased risk of complications (Iqbal et al, 2008).

Liver Malignancy

The cirrhotic liver of treated BA is potentially premalignant, and there are several reports (Kohno et al, 1995; Iida et al, 2009; Hol et al, 2008) and a single series of three cases (Tatekawa et al, 2001). This of course may be due simply to the relatively small numbers of long-term survivors, but perhaps it implies only a modest risk. Malignant change may include hepatocellular carcinoma (Hol et al, 2008; Iida et al, 2009; see Chapter 80) or cholangiocarcinoma (Kulkarni & Beatty, 1977; see Chapter 50A, Chapter 50B, Chapter 50C, Chapter 50D ), but more benign changes, such as focal nodular hyperplasia (Okugawa et al, 2008), may also occur. Although the youngest reported case was diagnosed in a patient 10 months old, but most are seen during later childhood or adolescence as a nodule or mass on routine US screening of the liver. Because of the risk for malignant change, serum α-fetoprotein levels should probably be part of the routine follow-up for such children, although clearly in some, these levels will still be normal.

Laparoscopy

The open technique described above requires precision, maximal exposure of the portal plate, and delicate suturing. This can be approximated by laparoscopic or robotic means, and small series have been reported (Esteves et al, 2002; Lee et al, 2004; Dutta et al, 2007; Ayuso et al, 2008). However, now that the first wave of enthusiasm has perhaps abated, it appears that the results are not comparable to the open technique (Wong et al, 2008). The only outcomes that matter are clearance of jaundice and preservation of the native liver; other outcomes, such as early feeding and better cosmesis, are irrelevant. In many centers, there is currently a self-imposed moratorium, until the reasons for failure become clearer or techniques improve.

Adjuvant Steroids

The inflammatory nature of BA has been emphasized above, and postoperative steroids seem a logical approach to this. In addition, most steroids have a chloretic action, although actual evidence of effect is scant. A number of small, uncontrolled studies from the United States and Japan suggesting benefit have been published (Meyers et al, 2004; Kobayashi et al, 2005). Our large, randomized, double-blind, placebo-controlled trial starting with 2 mg/kg/day showed a significant early effect at 1 month after operation with a lower bilirubin (Davenport et al, 2007), but this did not translate to a diminished need for transplantation or to a significant difference in the proportion with complete clearance of jaundice. Further studies from Germany using a higher starting dose (10 mg/kg/day) showed no difference, although their overall clearance rates were poor (27% were jaundice free at 2 years; Petersen et al, 2008).

Effect of Age at Surgery

It is axiomatic that once cirrhosis is established, the outcome will be worse. And because cirrhosis takes time to develop, it is logical to think that the earlier the surgery, the better the outcome. However, given that we do not know when biliary obliteration starts in isolated BA, trying to identify a relationship with postnatal age may be difficult. Large, multicenter series with multiple surgeons have been unable to show a statistically significant relationship (Davenport et al, 2004a).

Figure 40.5 illustrates this for isolated BA derived from a single center with only two surgeons, and it uses clearance of jaundice (<20 Umol/L) as a measure of outcome. This shows that there is no discernable difference up to 100 days, and certainly no evidence of any “cutoff.” By contrast, Figure 40.6 shows the same analysis for one of the developmental forms, cystic BA, and here a relationship is obvious, with an impaired outcome the older the infant is at surgery (Davenport et al, 2008).

FIGURE 40.5 Relationship of age at surgery to clearance of jaundice in infants with isolated biliary atresia (n = 177).

(Modified from Davenport et al, 2008: Surgical outcome in biliary atresia: etiology affects the influence of age at surgery. Ann Surg 247:694-698.)

FIGURE 40.6 Relationship of age at surgery to clearance of jaundice in infants with cystic biliary atresia (n = 23). (Modified from Davenport et al, 2008: Surgical outcome in biliary atresia: etiology affects the influence of age at surgery. Ann Surg 247:694-698.)

Is there any benefit of surgery in those infants who come to surgery after 100 days? By now, this should be relatively infrequent in developed countries, but it still occurs. Such infants are much more likely to have established cirrhosis, and the prognosis is, without doubt, worse. Nevertheless, a Kasai portoenterostomy as the first step may be reasonable and has been associated with long-term jaundice-free survival in our series (Davenport et al, 2004b). The alternative is a primary liver transplant.

Kasai Variations

Although most surgeons operating today follow the original principles set out by Kasai and Suzuki (1959), most likely use a much wider portal dissection to try and incorporate as much as possible of the transected portal plate. There have been published variations in the construction of the Roux loop. Typically these were developed to reduce the postoperative incidence of cholangitis, and it included bringing the Roux loop out as a stoma and creating an antireflux valve (Tanaka et al, 1991; Nakajo et al, 1990). Hindsight suggests these have no real value and serve to impair the later transplant procedure, should it become necessary (Ogasawara et al, 2003). Use of the appendix as a conduit to the duodenum has also been described (Crombleholme et al, 1989), but again, later reports suggest no value (Lee et al, 2007; Tsao et al, 2003).

Biliary Atresia Variations

Types 1 and 2 BA may be found in up to 5% to 10% of all cases and were formerly described as “correctable,” because some form of lesser procedure, such as a hepaticojejunostomy, could be done. Nonetheless, care must be taken, and a radical resection of remnant tissue back to the portal plate is now usually considered a better option with an encompassing portoenterostomy.