Biliary Atresia (including Syndromic Forms)

Progressive obliterative cholangiopathy of the infant biliary tree, of multifactorial and genetically complex aetiology, which is the most frequent identifiable cause of neonatal cholestasis and the main indication for paediatric liver transplantation; it includes the isolated form, the biliary atresia splenic malformation syndrome, cystic biliary atresia, the cytomegalovirus-associated form and the biliary atresia seen in chromosomal and monogenic syndromes.

Literature and regulatory information update: 28 October 2026.

0. Initial Reference Table

1. Introduction

1.1. Definition and place in neonatal cholestasis

Biliary atresia (BA) is a progressive obliterative and fibrosclerosing cholangiopathy that affects the extrahepatic bile ducts and, to a variable degree, the intrahepatic ducts, with onset in the embryonic or perinatal period, and which manifests as severe, persistent neonatal jaundice with acholic stools [1, 2]. Pathological cholestasis affects about 1 in every 2500 term newborns, and BA accounts for 25% to 40% of these cases, ahead of a growing group of monogenic diseases, which explain about 25%, and of multifactorial causes, such as parenteral nutrition-associated cholestasis [36, 42]. Among more than a hundred specific causes of neonatal cholestasis, BA is the most common identifiable cause and the one that requires the most rapid diagnosis, because early surgical intervention is associated with better outcomes [50]. Cholestatic jaundice, that is, conjugated hyperbilirubinaemia, is always pathological, and its early detection and treatment can prevent death or the need for liver transplantation [40]. Even with timely portoenterostomy, bile flow is restored in only 50% to 75% of patients, and 60% to 75% eventually need liver transplantation by the age of 18 years, which makes BA the leading indication for liver transplantation in childhood [2, 40].

1.2. Historical overview

The association between polysplenia and BA was first noted in 1929, and isolated cases of heterotaxy with BA were described in 1892 and 1941 [3, 7]. Kasai and Suzuki published in 1959 the Sendai experience, in Japan, with 12 infants operated on over the preceding 8 years, proposing transection of the fibrous remnant at the liver hilum (porta hepatis) and its anastomosis to a jejunal loop; Kasai portoenterostomy (KPE, Kasai portoenterostomy) became widespread in the West during the 1980s [2, 37]. Ohi later extended the hilar dissection to the liver capsule [2]. The first liver transplant in a child with BA was attempted by Starzl in Denver in 1963, without success, and the technique only became effective in the 1980s with ciclosporin and, later, tacrolimus [47]. In 1991, the Denver series of Lilly and Karrer described 16 of 131 infants (12%) with a syndromic form and suggested for the first time that KPE could also succeed in these cases [3]. In 1993, Davenport and colleagues, at King's College Hospital in London, analysed 308 infants with BA, identified 23 with polysplenia and a further 4 with other splenic malformations (2 with double spleen and 2 with asplenia), found that situs inversus and portal vein anomalies cut across all these categories and proposed the term biliary atresia splenic malformation syndrome (BASM) as an aetiological and prognostic subgroup [51]. In the following decades, cystic BA was recognised, as were the form associated with anti-cytomegalovirus (CMV) immunoglobulin M (IgM), described in 1998 by Fischler and colleagues in Sweden, and the occurrence of BA in cat eye syndrome, whose first series, by Allotey and colleagues, brought together five infants from London and Paris [3, 37].

1.3. Milestones in genetic research

Molecular research on BA has followed two paths. In the isolated form, genome-wide association studies (GWAS, genome-wide association studies) identified common susceptibility loci, ADD3 being the only one replicated in Chinese, Thai and European populations; these were followed by EFEMP1 at 2p16.1 in a European-American cohort (2018), ARF6, MAN1A2 and, in 2023, AFAP1 and TUSC3 in a GWAS of 811 transplanted patients that pointed to polygenic susceptibility in the ciliogenesis and planar polarity pathway (CPLANE complex) [2, 52, 53]. Trio exome sequencing showed, in Hong Kong, an excess of rare deleterious variants in ciliary genes (2021) and, in the United States, the absence of recurrent de novo variants (2023) [8, 54]. In the syndromic form, the first variants were described in CFC1 and NODAL, and in 2019 the North American ChiLDReN consortium identified biallelic PKD1L1 variants in 5 of 67 patients with BASM, the most consistent genetic finding to date [7, 37]. In 2025 it was shown in human cholangiocyte organoids and in mice that ADD3 loss of function increases biliary epithelial permeability and the severity of experimental BA, and in 2026 a GWAS in Taiwan proposed three new loci not yet replicated [55, 56].

1.4. Current nomenclature and classifications

BA is classified according to two complementary logics. The anatomical Kasai classification, adopted by the Japanese society, defines the level of the proximal obstruction: type I, obstruction of the common bile duct; type II, obstruction of the common hepatic duct with patent right and left hepatic ducts; type III, obstruction at the level of the porta hepatis, which is the most frequent form [2, 5]. In the full version of the Japanese Biliary Atresia Society (JBAS, 1975), used by the 2nd edition of the Japanese guideline, the main type (I, I-cyst with a cystic structure communicating with the intrahepatic ducts, II and III) is defined by the site of extrahepatic atresia; the subtypes (a1, a2, b1, b2, c1, c2 and d) describe the macroscopic appearance of the common bile duct and the presence of the common hepatic duct; and the subgroups (α, dilated hepatic radicles with an internal diameter of 1 mm or more and communicating; β, hypoplastic radicles under 1 mm; γ, bile lake; μ, fibrous radicles; ν, fibrous mass; ο, aplasia of the radicles) describe the pattern at the hilum [57]. The Japanese Biliary Atresia Registry (JBAR), established in 1989, included more than 3950 cases in 2025 [57]. The aetiological and clinical-phenotypic classification, proposed by Davenport and adopted by the 2025 international guideline, distinguishes syndromic BA (BASM and cat eye syndrome, among others), cystic BA, CMV-associated BA and isolated BA [4, 5]. Isolated BA is the majority and has the least understood pathogenesis; the cystic and syndromic forms relate to failures of biliary morphogenesis and ciliary pathways, and the CMV-associated form to perinatal immune dysregulation [4]. Davenport considers BA a final common pathway, with a stereotyped presentation but without aetiological homogeneity, at whose core there always seems to be a failure of biliary tree development [3]. Orphanet organises these entities in the group «biliary atresia and related diseases» (ORPHA:498345), which includes isolated BA (ORPHA:30391) and the syndromic BA group (ORPHA:498350), where BASM is found (ORPHA:244283) [1, 18, 32, 33]. In France, the surgical classification of Chardot is mainly used, which is based on cholangiography and examination of the remnant and distinguishes four types: type 1 (3% of cases), atresia limited to the common bile duct; type 2 (6%), hilar cyst communicating with dystrophic intrahepatic ducts; type 3 (19%), patent gallbladder, cystic duct and common bile duct, with acholic fluid in the gallbladder; and type 4 (72%), atresia of the entire extrahepatic biliary tree. Except in type 1, the intrahepatic ducts are always affected [38]. The French and Japanese numbering rest on different criteria (patency of the distal segments in the French, proximal level of obstruction in the Japanese) and cannot be converted directly into one another.

1.5. Definition of the entities within scope

Isolated, or non-syndromic, BA occurs in an infant without other malformations and with normal initial growth, and accounts for about 80% to 90% of cases [3, 52]. BASM, in the original definition that the author himself recommends keeping, combines BA with four core components: situs inversus, preduodenal portal vein, absent intrahepatic inferior vena cava (IVC) and splenic malformation (polysplenia, asplenia or double spleen); Davenport rejects extending the definition to inguinal or umbilical hernias and to renal anomalies, which are probably coincidental, and considers too inclusive the designation «biliary atresia structural malformation syndrome» proposed in Canada [3]. For Orphanet, BASM represents the embryonic or syndromic form of BA and may include heart disease, situs inversus and preduodenal portal vein [18]. Cystic BA is characterised by a cystic dilatation in an obliterated extrahepatic tree; it generally corresponds to type I, is detectable on prenatal ultrasound and is the only variant in which cholangiography opacifies intrahepatic ducts, with an appearance ranging from a smooth tree to irregular «cloud-like» aggregates [2, 37]. CMV-associated BA is defined by positive serum anti-CMV IgM and/or detection of CMV deoxyribonucleic acid (DNA) in urine or blood at presentation [5]. CABA designates BA with congenital heart disease, a subset that overlaps largely with BASM and cat eye syndrome [3]. Finally, BA may be an occasional finding in defined chromosomal and monogenic syndromes, from trisomy 18 to Kabuki syndrome, covered in section 3 [3, 58, 59].

2. Molecular Genetics and Pathophysiology

2.1. Development of the biliary tree and of laterality

The extrahepatic biliary tree forms from the 4th and 5th weeks of gestation by ventral evagination of the foregut; the hepatic bud differentiates into a hepatic portion and a cystic portion under the control of the NOTCH, TGF-β (transforming growth factor beta), BMP (bone morphogenetic protein) and FGF (fibroblast growth factor) pathways, with participation of the transcription factors Hes-1, Pdx-1, Prox-1, HNF-6, HNF-1β and Foxf1 [3, 4]. The factors SOX17, which suppresses pancreatic fate, and PDX1, which represses biliary differentiation, separate the extrahepatic biliary tree from the ventral pancreas [4]. The intrahepatic tree arises at around the 8th week from the ductal plate, which organises centrifugally from the hilum around the branches of the portal vein, guided by a TGF-β gradient and by the non-canonical Wnt, Notch and Hippo pathways, with the factors HNF6, HNF1β and Sox9 [3, 4]. Determination of the left–right axis begins at gastrulation: the cilia of the embryonic node rotate at about 10 Hz and generate a leftward flow that is detected by the PKD1L1-PKD2 complex in immotile cilia and restricts NODAL, LEFTY1/2 and PITX2 signalling to the left lateral plate mesoderm; when the cilia are immotile, situs becomes random [3, 9, 60]. The temporal coincidence between the definition of situs and the budding of the biliary primordium explains why, in BASM, the abnormalities begin in the laterality phase, whereas in cystic BA the presence of residual bile in the cyst indicates that the process sets in after the onset of bile secretion and the connection between the intra- and extrahepatic trees [4]. Absence of the IVC and the preduodenal portal vein reflect defects of vascular remodelling, and splenic malformations, situs inversus and heart disease reflect defects of laterality [4].

2.2. Pathogenic model: susceptibility, injury and response

The dominant model proposes that genetically susceptible children, with immature or abnormal ducts, suffer injury from bile acids, environmental toxins or viruses, followed by an exaggerated immune response, fibrosis and ductal obliteration [4]. Two North American studies, with about 400,000 newborns in total, showed that conjugated bilirubin was raised in the first days of life in 100% of the children who would go on to have BA, which places the initial injury in utero, followed by a postnatal phase of response to injury [2]. Neonatal extrahepatic ducts are vulnerable because they have low glutathione, immature tight junctions, a collagen-poor submucosa and an immature glycocalyx; oxidative stress and hypoxia, with activation of HIF1α in cholangiocytes, contribute to this susceptibility, and the foetal extracellular matrix of the bile duct is rich in hyaluronic acid and poor in collagen [2]. Among the proposed insults, CMV-associated BA has been reported in 10% to 30% of cases in Europe, 60% in China and 78% in South Africa, with variable definitions and methods of positivity, without causality being proven; group C rotavirus was detected in 50% of livers with BA in one study, without confirmation in others [2]. The French PNDS (French National Diagnosis and Care Protocol) adds to the list of viruses described in association with BA respiratory syncytial virus, Epstein-Barr virus and human papillomaviruses, accepts that the frequency of viral infections in the neonatal period may explain chance associations, considers prenatal ischaemia of the bile ducts a possible mechanism and concludes that BA appears to be the final phenotype of multiple, intertwined mechanisms [38]. Biliatresone, an α-methylene ketone from Australian plants eaten by pregnant livestock that binds to reduced glutathione, selectively injures the extrahepatic bile duct in zebrafish and in newborn mice, but probably has no direct human relevance; microcystin-RR from algal blooms and maternal aflatoxin B1, in an Egyptian study, remain unconfirmed hypotheses [2, 4]. The same Egyptian group proposed calling «Kotb disease» a variant of BA attributed to congenital aflatoxicosis, with a high hepatic load of aflatoxin B1, inflammation and accelerated cirrhosis, in infants described as GSTM1-null and with p53 and GSTP1 alterations, whose mothers were reported to be heterozygous for the absent allele; OMIM records this proposal from a narrative review by the authors themselves, without independent replication, and the accompanying recommendations for prenatal aflatoxicosis screening are not in the guidelines consulted [22]. Other exogenous explanations historically mentioned in OMIM include ascending cholangitis from intrauterine Listeria monocytogenes infection (which may recur in successive pregnancies and mimic familial recurrence), antibodies against reovirus 3 and clustering in time and space; in Taiwan, the incidence peak of 2002 (1.90 per 10,000 live births) coincided with a dengue epidemic, an ecological association that does not prove causality [22].

The response to injury includes a TH1-type inflammatory phase, with interferon gamma (IFN-γ) and tumour necrosis factor, followed by a TH2 phase with IL-33, group 2 innate lymphoid cells and IL-13; B lymphocytes with IgM anti-cholangiocyte autoantibodies associated with worse outcome, deficiency of CD4+FOXP3+ regulatory T cells, CD177+ neutrophils with extracellular traps and maternal microchimerism also contribute [2]. Hypomethylation of the IFNG gene and increased methylation of the FOXP3 promoter in regulatory T cells have been described [4]. Fibrosis progresses through ductular reaction and activation of stellate cells into myofibroblasts, with TGF-β1 and platelet-derived growth factor (PDGF); matrix metalloproteinase 7 (MMP-7) is highly expressed in the BA liver, in hepatocytes, cholangiocytes and Kupffer cells, and remains raised years after KPE even when jaundice clears [4, 37]. Overexpression of the secretin receptor and a raised serum fibroblast growth factor 19 (FGF19) at the time of KPE are associated with lower native liver survival [2, 37]. Inflammation also reduces the expression of bile acid transporters in the murine rhesus rotavirus infection model [2].

2.3. Genetic architecture of isolated biliary atresia

Isolated BA does not follow a Mendelian pattern, and the variants identified show no clear genotype–phenotype correlation [4]. The lack of familial aggregation and discordance in twins were the classic arguments against a genetic cause, and twin studies show low concordance in both monozygotic and dizygotic twins, which suggests a role for epigenetics and post-zygotic mutations [4, 7]. Nevertheless, BA behaves as a complex disease with a contribution from common variants of modest effect, with odds ratios (OR, odds ratio) of 1.4 to 2.7, and from rare variants of reduced penetrance, with locus heterogeneity between populations: with the exception of ADD3, the loci have been reported only in Caucasian or only in Asian patients [2].

Table sources: [2, 4, 8, 52, 53, 56]. The cytobands of AFAP1, MAN1A2 and ARF6 come from the gene identification fields in GeneGuidance [61].

The ADD3 locus is the most consistent. Adducin 3 is a cytoskeletal and cell adhesion protein, and both ADD3 and GPC1 mediate the Hedgehog pathway; in zebrafish, reduced expression of add3 or gpc1 causes bile duct defects, more severe when both are reduced [4]. In 2025, inactivation of ADD3 in cholangiocyte organoids derived from human pluripotent stem cells caused defective differentiation, failure to recruit βII-spectrin, abnormal primary cilia, fewer tight junction proteins and greater epithelial permeability; Add3+/− and −/− mice exposed to the rhesus rotavirus model had a higher incidence and greater severity of BA, a preclinical result that supports the link between loss of ADD3 and foetal or neonatal susceptibility to biliary injury [55]. The 2018 GWAS, with 343 European-American patients with isolated BA and 1716 controls, plus a replication cohort of 156 patients with other extrahepatic anomalies, identified three single-nucleotide polymorphisms (SNPs) in the 5th intron of EFEMP1; the protein, absent in normal cholangiocytes, was detected in cholangiocytes and vascular smooth muscle of livers with BA and with other cholestatic diseases [52]. The 2023 GWAS, with 811 European patients transplanted for BA and 4654 genetically matched controls, found intronic SNPs in AFAP1 and TUSC3, both expressed in foetal liver organoids and in the BA liver, and an aggregate signal in 102 genes of the CPLANE complex; genome sequencing of 100 of these patients revealed rare variants in CPLANE genes with hepatobiliary HPO (Human Phenotype Ontology) terms in 87%, against 40% in non-CPLANE genes, and, in most cases, the rare variants were in genes different from those carrying common variants [53]. The authors suggest that abnormal vasculogenesis would explain the portal vein anomalies, present in up to half of patients [53].

Exome sequencing in 89 trios from Hong Kong and Vietnam with non-syndromic BA found rare deleterious de novo or biallelic variants in liver-expressed ciliary genes in 31.5% (28 of 89) of patients; livers with variants in KIF3B and TTC17 lacked cilia on cholangiocytes, knockdown of PCNT, KIF3B and TTC17 reduced cilia in fibroblasts and cholangiocytes, zebrafish with inactivation of these genes had reduced bile flow and nasal nitric oxide was abnormally low in 8 of 10 patients with ciliary variants [8]. A 2025 review describes this result as a 23-fold increase in variant burden; the original article and the 2024 review give an OR of 2.58, which is the correct value [2, 8]. In the United States, the National Birth Defects Prevention Study sequenced the exomes of 54 trios and 1 duo (91% with isolated BA) and compared them with 1513 parents of children with other birth defects: it found 42 rare de novo variants in 27 children (50.0%), 5 of them loss-of-function, none recurrent, and an association of rare variants in IFRD2 (P=3.75×10⁻⁶; Bonferroni P 0.03), concluding that rare high-penetrance variants do not explain a relevant proportion of cases [54]. An earlier study of 101 children with isolated BA, cited by Sok and colleagues, described 66 de novo variants in 66 genes, including STIP1 and REV1 [54]. In 2026, the first Taiwanese GWAS, with 189 cases and 558 controls, proposed loci at 8q22.1, 4q32.1 (PDGFC) and 10q25.1, which await replication [56]. Polymorphisms of TGFB1, VEGFA, PDGFA and CTGF and ABCB11 variants have also been described in patients with BA, and the mouse with an Efemp1 mutation has fewer elastic fibres, with biliary barrier dysfunction [4].

Copy-number variations gave the first signal at 2q37.3. In a screen of 35 patients with BA, Leyva-Vega and colleagues (2010) found a single relevant CNV, a 1.76 Mb heterozygous deletion at 2q37.3 containing 30 genes, among them AGXT, present in two unrelated boys and in none of 2026 controls: the first had BA at 3 weeks, transplantation at 8 months and normal development at 17 years, and the deletion came from his unaffected father; the second, with a 47,XXY karyotype, polysplenia, hypothyroidism, developmental delay, seizures and dysmorphism, had a larger, 5.87 Mb deletion encompassing the first, and another at 2q33.1; neither had the phenotype of the 2q37 deletion syndrome [22]. This region contains GPC1, which links the finding to the rare deletions in the previous table; the OMIM text itself writes «2q27.3» for «2q37.3» at one point, a typographical error that reappears in a 2025 review; the gene entry places GPC1 at 2q37.3 [4, 22, 30]. OMIM also refers to an unconfirmed association with variation in XPNPEP1 (entry 602443) [22]. In a classic analysis of 251 cases of extrahepatic BA, 51 had associated anomalies: 15 (29.4%) with combinations from the laterality spectrum, 30 (58.8%) with 1 or 2 anomalies, mainly cardiac, digestive and urinary, without a recognisable pattern, and 6 with intestinal malrotation, some with preduodenal portal vein, which the authors interpreted as a limited expression of a lateralisation defect; they proposed that BA with laterality defects was the most likely candidate for a major gene, and that teratogenic, infectious and multifactorial causes weighed more in BA with non-syndromic anomalies [22].

2.4. Genetics of the biliary atresia splenic malformation syndrome

PKD1L1 (7p12.3; OMIM 609721; MANE transcript NM_138295.5) encodes a PKD1-family protein that interacts with PKD2 at the embryonic node; biallelic variants cause autosomal visceral heterotaxy 8 (HTX8, OMIM 617205), with a spectrum ranging from situs inversus totalis to heterotaxy, and the gene–disease relationship is classified as «definitive» in GenCC [3, 9]. In the ChiLDReN study, which included patients from the PROBE and BASIC studies recruited between May 2004 and August 2016 in 18 centres in the United States and Canada, exome sequencing of 67 patients with BASM (58 trios), filtered for 2016 genes of ciliogenesis, ciliary dysfunction and cholestasis, found rare, potentially deleterious biallelic PKD1L1 variants in 5 patients and heterozygous variants of interest in another 3; immunohistochemistry of one liver showed reduced PKD1L1 expression in the biliary epithelium [7]. The cohort had a slight female predominance, KPE in 61 patients at a mean age of 59 days (21 to 122), 61% already transplanted and 27% alive with native liver at the time of selection, and no clear correlation between genotype, phenotype and outcome was identified [7].

The published table does not state the transcript; NM_138295.5 is the MANE transcript used by Sok and colleagues, and the variants c.6473+2_6473+3del, c.6949C>T, c.2399T>C and c.7937C>G were validated on it in VariantValidator (GRCh38), while the others are reproduced as published [7, 54, 62]. The variant NM_138295.5(PKD1L1):c.6473+2_6473+3del is classified in ClinVar (VCV000235796) as pathogenic or likely pathogenic by several submitters, without conflicts, for visceral heterotaxy 8 and situs inversus; it had been described in the homozygous state in two siblings with heterotaxy and severe heart disease [7, 63]. The allele frequencies of these variants in gnomAD range from 0.0004% to 0.61% [7]. PKD1L1 also appears in isolated BA: the National Birth Defects Prevention Study found two children without associated malformations with compound heterozygosity, NM_138295.5:c.8485G>C p.(Glu2829Gln) with c.7552G>A p.(Ala2518Thr), and c.6473+2_6473+3del with c.731C>T p.(Pro244Leu) [54]. In the mouse, the Pkd1l1−/− genotype is lethal; the heterozygote has hypertrophic, fibrotic bile ducts with fewer primary cilia, but patent, and only develops cholestasis with ductular proliferation after exposure to the dietary agent DDC (3,5-diethoxycarbonyl-1,4-dihydrocollidine), and only 35% to 45% of mice homozygous for Pkd1l1 mutations have heterotaxy, which reflects incomplete penetrance and randomisation of laterality [3, 7]. Hepatic deletion of Pkd1l1 combined with bile duct ligation reproduces the reduction in cilia, cholangiocyte proliferation and peribiliary fibroinflammation [2]. Lethality depends on the allele: according to OMIM, the induced rikishi mutant (p.Asp441Gly) is lethal before 15.5 days of gestation, with right pulmonary isomerism and absence of the NODAL cascade, whereas some homozygotes for the targeted null allele (tm1) reach adulthood, with left pulmonary isomerism and bilateral expression of the NODAL cascade; in neither are the nodal cilia altered, which led to the hypothesis that PKD1L1 represses PKD2 at the node and that nodal flow lifts this repression only on the left side [23].

In the HTX8 entry, OMIM describes mainly heterotaxy with complex heart disease, without BA in the cases gathered: two boys of northern European ancestry with a homozygous splicing variant, who died in the neonatal period; a 46-year-old Iranian woman, born to consanguineous parents, with situs inversus totalis and congenitally corrected transposition; two sibling foetuses with compound heterozygosity (p.Arg1347Ter and c.4798_4799del), one of them with polysplenia, malrotation and duodenal atresia and the other with a concomitant 22q11 deletion; and two foetal sisters with p.Arg2669Ter and c.160+1G>A, variants that OMIM presents without a reference transcript [23]. A South Asian patient homozygous for p.Arg2669Ter also had congenital sideroblastic anaemia without an identified cause, and the authors suggested monitoring the blood count in HTX8, an association not yet confirmed [23]. PKD1L1 BA thus comes mainly from the BASM cohorts, and not from the cardiac heterotaxy families that defined the entry.

The remaining laterality genes have weak evidence in BASM. CFC1 (2q21.1; OMIM 605194) encodes the protein Cryptic, an EGF-CFC cofactor of NODAL; a French laboratory described variants in 5 of 10 patients with BASM, all with situs inversus or dextrocardia, and a later variant was also present in the proband's father, who had situs inversus without BA [3]. CFC1 variants have been described in BA with polysplenia and in BASM, and Orphanet records CFC1 as a candidate gene tested in BASM, but the attributable fraction is very small and disputed: the variants N21H, R47Q and R78W (historical numbering, = NM_032545.4:c.61A>C p.(Asn21His), c.140G>A p.(Arg47Gln) and c.232C>T p.(Arg78Trp), classified in ClinVar as benign, benign and likely benign) occur at appreciable frequency in African American controls and behave as susceptibility alleles, loss-of-function variants are probably not associated with heterotaxy, and the gene is poorly accessible to short-read variant calling because of a paralogue [10, 18]. OMIM takes another view: it classifies HTX2 as autosomal dominant, on the basis of two heterozygous loss-of-function variants described in 2000 (p.Arg112Cys, inherited from a parent without phenotype, and the frameshift deletion 522delC, present in a normal parent of one of the patients), both defective in a zebrafish rescue assay, which points to incomplete penetrance and modifiers; in 2002, 2 of 86 patients with transposition of the great arteries or double-outlet right ventricle without a laterality defect had heterozygous CFC1 variants [20, 27]. The two sources therefore diverge on the effect of loss-of-function variants, and a CFC1 result should be interpreted with both in mind. ZIC3 (Xq26.3; OMIM 300265) causes X-linked visceral heterotaxy 1 (OMIM 306955), explains about 1% of sporadic heterotaxy and less than 3% to 5% of heterotaxy in boys, and has been mentioned in some infants with BASM with lesser evidence; the reference books assign it divergent cytobands (Xq28 and Xq26.2), and the correct location is Xq26.3 [3, 11, 28, 41, 60, 64]. Extrahepatic BA already appears in the family of African ancestry that defined HTX1 in 1987, with 9 affected boys in two generations and 11 deaths before 2 years, and in one of two cousins from a family described in 1994 with atrial septal defect, anomalous IVC, polysplenia and intestinal non-rotation, who died at 7.5 months of heart and liver failure [19]. In the screen of 194 patients, ZIC3 variants included a girl with heterotaxy limited to the heart and a patient with only atrial septal defect and pulmonary stenosis, and in a Lebanese family a carrier maternal uncle was healthy, the first case of incomplete penetrance in a male carrier [19]. A heterozygous deletion of FOXA2, a gene that activates Pkd1l1 in the mouse, was described in a child with BA, malrotation and interrupted IVC whose father had situs inversus and polysplenia without BA [7]. In the ChiLDReN study, no deleterious variants were found in CFC1, FOXA2, INVS, NODAL or ZIC3 [7]. In heterotaxy in general, only about 20% of cases have a variant in a known gene, and an exome study of 323 individuals with laterality defects found known variants in only 7%; the list includes ZIC3, LEFTY2, CFC1, GDF1, NODAL, ACVR2B, MMP21, FOXH1, CCDC11, MEGF8, NKX2-5, NPHP4, PKD2 and PKD1L1 [41, 60]. The OMIM visceral heterotaxy phenotypic series numbers 14 types: HTX1 (ZIC3, Xq26), HTX2 (CFC1, 2q21), HTX3 (locus at 6q21), HTX4 (ACVR2B, 3p22), HTX5 (NODAL, 10q22), HTX6 (CCDC11, 18q21), HTX7 (MMP21, 10q26), HTX8 (PKD1L1, 7p12), HTX9 (MNS1, 15q21), HTX10 (CFAP52, 17p13), HTX11 (CFAP45, 1q23), HTX12 (CIROP, 14q11), HTX13 (DAND5, 19p13) and HTX14 (CIROZ, 1p36), with associations not yet confirmed for SMAD2, ANKS3, MGRN1 and others [19]. In BASM, the hypothesis of a cholangiociliopathy gains strength with PKD1L1 and with KIF3B, which takes part in nodal flow; variants in INVS, TTC17 and MAN1A2 and enrichment of ciliary pathways have also been described [4, 7].

2.5. Maternal and environmental factors

Maternal diabetes is particularly associated with BASM: 4 of 23 mothers (17%) in the 1993 series and 7 of 56 (12.5%, against 1.2% in controls) in a later series, with overlap with other anomalies of diabetic embryopathy, such as caudal regression, transposition of the great arteries and double-outlet right ventricle; the probable mechanism involves hyperglycaemia, free radicals and epigenetic changes demonstrated in murine models [3, 51]. Half of the cases of left isomerism occurred in children of diabetic mothers in a 1999 series [64]. In non-syndromic BA, a Taiwanese national study of 285 patients found an OR of 2.2 (95% CI 1–4.5) for maternal type 2 diabetes and of 3 (95% CI 1.3–6.8) for maternal drug use, and the Texas registry, with 305 live births with BA, an OR of 2.3 for isolated BA, of 6.7 in the 27 with laterality defects and of 26 with pregestational diabetes [3]. Maternal diabetes is 10 times more strongly associated with BASM than with isolated BA, and pregestational diabetes more than doubles the risk of BA in two population studies [2]. In Sweden, maternal age over 35 years and parity of 4 or more were associated with a 2 to 3 times higher risk, and maternal intestinal and genitourinary infections in the 3rd trimester with 6 and 1.6 times higher risks; use of anti-inflammatory drugs for asthma in the 1st trimester and drug use triple the risk, and in one cohort 26% of 39 children with BA had been conceived by in vitro fertilisation, a finding not confirmed in population studies [2]. Prematurity doubles (under 37 weeks) or quintuples (under 32 weeks) the incidence, and being small for gestational age multiplies the risk by about 5 in Sweden [2].

2.6. Allelic disorders and the significance of the genes in other contexts

Several BA susceptibility genes have clinical curation for diseases unrelated to it, which precludes reading a panel result by analogy. ADD3 is associated, when homozygous for p.Gly367Asp, with autosomal recessive spastic quadriplegic cerebral palsy 3 (OMIM 617008); EFEMP1 with Doyne drusen; KIF3B with autosomal dominant retinitis pigmentosa 89, with one proband with hepatic fibrosis, retinitis and polydactyly; PKHD1L1 with DFNB124 deafness; none of these curations supports, by itself, an association with BA [61]. OMIM confirms this mismatch: BA does not appear among the phenotypes in the entries for the ADD3, EFEMP1 and GPC1 genes. ADD3 spastic quadriplegic cerebral palsy 3 is autosomal recessive and was described in four Jordanian siblings homozygous for p.Gly367Asp, in a Syrian patient and in other patients with p.Asn332Ser in the homozygous or compound heterozygous state, with loss of the actin filament capping function; EFEMP1 is associated with Doyne honeycomb retinal dystrophy and open-angle glaucoma 1H (both autosomal dominant) and with autosomal recessive cutis laxa type ID; and the GPC1 entry describes the function of glypican 1 in cancer and angiogenesis, without a Mendelian phenotype [29–31]. RFX6 causes, in the biallelic state, Mitchell-Riley syndrome, whereas truncating heterozygosity can cause MODY-type diabetes (maturity-onset diabetes of the young) of reduced penetrance and without malformations, so these heterozygotes should not be described as mere carriers [12]. Besides heterotaxy, ZIC3 causes X-linked non-syndromic congenital heart disease and X-linked VACTERL association, sometimes with hydrocephalus, and some heterozygous women have situs inversus totalis [11]. CFC1 is also associated with conotruncal heart disease (transposition of the great arteries, double-outlet right ventricle, atrioventricular septal defect, anomalous venous return) [10]. KDM6A, on the X chromosome, and KMT2D, at 12q13, cause Kabuki syndrome, and MED12 Hardikar syndrome, besides other disorders [3]. According to OMIM, Kabuki 1 (KMT2D) is autosomal dominant and almost always de novo, with KMT2D variants in 32% to 74% of patients with a clinical diagnosis, depending on series selection (12 of 40; 50 of 81, or 61.7%; 35 of 53 families, or 66%; 81 of 110, or 74%), about 70% of them truncating, and Kabuki 2 (KDM6A, Xp11.3) is X-linked dominant [13, 14]. Hardikar syndrome results from heterozygous de novo nonsense or frameshift variants of MED12, distributed along the gene and predicted to cause messenger RNA decay and haploinsufficiency, with skewed X inactivation; complete loss would be lethal in males [15]. Zimmermann-Laband shows genetic heterogeneity: KCNH1 (ZLS1, de novo missense variants), ATP6V1B2 (ZLS2) and KCNN3 (ZLS3) [16].

3. Clinical Spectrum and System-Specific Manifestations

3.1. Presentation of isolated biliary atresia

Cholestasis is present from birth, with raised conjugated bilirubin, although the liver appears normal and without fibrosis in the first weeks [3]. Jaundice develops gradually and often cannot be distinguished in the first 2 to 3 weeks, because unconjugated jaundice is almost universal in that period: in a Birmingham series, 15% of normal infants were clinically jaundiced at 12 to 14 days, 2.6% at 4 weeks and 0.2% at 6 weeks [37, 40]. The infant with isolated BA generally has normal initial weight and growth; pale or white stools and dark urine must be actively sought, hepatosplenomegaly appears later and poor weight gain is subtle [3, 37]. The French PNDS (French National Diagnosis and Care Protocol) describes a triad present from the first 15 days, in term newborns with a normal examination at birth: jaundice persisting beyond the 10th day, persistent discoloured stools (putty-grey, beige or pale yellow) with dark urine, and progressively developing hepatomegaly. Signs of cirrhosis (hard liver, splenomegaly, ascites) may appear only from 2 to 3 months, and a bulky, firm or hard hepatomegaly extending beyond the costal margin suggests a late-diagnosed BA; collateral thoracoabdominal venous circulation, ascites and hepatocellular failure are late signs. Liver function is as a rule normal at diagnosis, with normal factor V, although the prothrombin time may be low because of vitamin K malabsorption [38]. Fewer than 5% of infants present with coagulopathy and haemorrhage due to vitamin K deficiency, mainly those who did not receive prophylaxis at birth [37]. Biochemistry shows moderately raised aminotransferases, raised alkaline phosphatase and GGT and rising conjugated bilirubin [37]. Cholestasis impairs the absorption of the fat-soluble vitamins A, D, E and K, with poor weight gain, bruising and haemorrhage, and its natural history is progression to fibrosis, cirrhosis, portal hypertension, synthetic dysfunction and risk of hepatocellular carcinoma [36]. The feature that distinguishes BA from other neonatal cholestatic diseases, such as Alagille syndrome or alpha-1 antitrypsin deficiency, is early progression of fibrosis to cirrhosis [37]. Prognosis depends mainly on age at KPE: in isolated BA operated on before 30 days, jaundice clears in more than 70%, and after 100 days in less than 40% [2].

3.2. Treated disease: system-specific manifestations after portoenterostomy

The manifestations of treated BA are separate from the natural history because they depend on the success of KPE. In the liver, 43% of protocol biopsies at school age after a successful KPE show cirrhosis, and 79% to 97% of adult survivors with native liver have signs of cirrhosis; benign liver nodules occur in about 6% and hepatocellular carcinoma in about 1% of survivors with native liver [2]. Portal hypertension was present, as splenomegaly, thrombocytopenia or both, in 62% of 348 paediatric survivors with a median age of 8.5 years; oesophageal varices were found in 60% of those who underwent endoscopy, ascites in less than 10%, and at a mean age of 19 years 69% had portal hypertension and 64% varices [2]. The 5-year cumulative incidence of variceal haemorrhage exceeds 10% in those with splenomegaly and thrombocytopenia, and a systematic review found gastrointestinal haemorrhage in 21% of adults [2]. Cholangitis is the most frequent complication of KPE, between 40% and 93% depending on the series, and never occurs if bile flow has not been restored; two thirds of 219 paediatric survivors aged 5 to 18 years had at least one episode, 17% in the previous year, as did 31% of adults aged 17 to 30 years [2, 5, 37]. Intrahepatic bile lakes appear in 10% to 15% and predispose to recurrent cholangitis [2]. In the lung, hepatopulmonary syndrome affects 5% to 7% of children with native liver and portopulmonary hypertension less than 1%, mainly female adolescents [2]. Growth and puberty are generally normal, but vitamin D deficiency is frequent even after a successful KPE and despite supplementation, and fractures occur in 12% to 15% [2]. Neurodevelopmental delay is more common after a failed KPE, and a multicentre study of patients aged 2 to 25 years found psychosocial functioning lower than that of controls, especially at school, and mild gross motor and language delays; at 9.7 years, the quality of life of 221 children was lower than that of controls, and adult women report poorer perceived health, although education and employment are similar to those of the general population [2, 5].

3.3. Biliary atresia splenic malformation syndrome

BASM is the most frequent syndromic form and combines BA with defects of laterality and vascular remodelling. The biliary tree is affected in all cases, in a non-inflammatory manner and almost always with absence of the common bile duct and an atrophic gallbladder; the liver has a normal, «mirror-image» or symmetrical appearance, and variations of the hepatic arteries and a superficial preduodenal portal vein may confound the dissection [3].

Table source: Davenport 2025 [3]. A clinical review by the same author gives situs inversus in 30% to 40%, preduodenal portal vein in about 40% and heart disease in about 50% [37].

In the original 1993 series, age at presentation and preoperative biochemistry in BASM did not differ from those in BA without anomalies, but actuarial transplant-free survival was worse [51]. The heart disease of BASM generally corresponds to left isomerism, also called polysplenia syndrome, which results from bilaterally left-sided signalling pathways, with two morphologically left atria, multiple small spleens, bilateral bilobed lungs, interruption of the IVC with azygos continuation and, because of absence of the sinoatrial node, possible complete heart block; BA appears in the Oxford Desk Reference as a typical abdominal anomaly of left isomerism [64]. Interruption of the IVC with azygos continuation occurs in 75% of left isomerism and 7% of right isomerism, atrioventricular septal defect in 68% and 93%, and double-outlet right ventricle in 7% and 47%, respectively [64]. In the laterality sequences, BA, malrotation and anomalous portal and hepatic vessels are among the occasional anomalies, and polysplenia and asplenia may coexist in different members of the same family as expressions of the same lateralisation defect [60]. From the opposite perspective, BA occurs in about 6% of children with heterotaxy and cardiovascular malformation, about 1000 times the risk of the general population [41]. Hepatopulmonary syndrome is associated with BA in general and possibly more with BASM, and pulse oximetry is insensitive compared with the alveolar–arterial oxygen gradient [3]. In a comparison of 32 patients with BASM and 55 BA controls, there were no differences in the incidence of cholangitis or in humoral immunity (lymphocytes, immunoglobulins, tetanus antibodies), but upper respiratory tract infections were more frequent in BASM [3]. Anatomical asplenia and the functional hyposplenism of heterotaxy are associated with an annual infection rate of 4.2%, and infections requiring hospitalisation occurred in 50% of patients with anatomical asplenia against 8% of those with polysplenia or a normal spleen [41]. Intestinal malrotation was found in 58% to 72% of screened children with heterotaxy [11, 41].

3.4. Cystic biliary atresia

Cystic BA accounts for 5% to 10% of series and generally corresponds to type I, with bile in the cyst [2, 37]. The cyst contains clear mucus in most cases, but bile in about 25%, which indicates persistent communication with the intrahepatic tree, and can be detected on prenatal ultrasound, which requires differentiation from choledochal cyst malformation, which is much more benign and which only cholangiography resolves [37, 47]. With radical resection, jaundice clears in more than 80%, although there is late failure in adolescence [2].

3.5. Cytomegalovirus-associated biliary atresia

In Sweden, in 1998, anti-CMV IgM was present in 38% of children with BA against 6% of controls; in London, infants with positive anti-CMV IgM were older at KPE and had more inflammatory injury and histological fibrosis, even after correcting for age, with a TH1 response profile, and in England they more often belonged to families of non-Caucasian origin [37]. In Denver, just over half of patients with BA had an increase in hepatic IFN-γ-producing T cells in response to CMV [37]. At King's College Hospital, between 2003 and 2017, 38 of 376 patients (10%) with histologically confirmed BA had positive anti-CMV IgM at presentation [65]. This form has later KPE, less jaundice clearance, lower native liver survival and higher mortality before transplantation, and higher anti-CMV IgM titres are associated with worse prognosis after KPE [2]. CMV-associated BA is more frequent in Asia and BASM in the West [2].

3.6. Biliary atresia with congenital heart disease

Congenital heart disease accompanies BA in 6% to 8% of series from the United Kingdom, the Nordic countries, Korea and Canada, in 3% in China and India and in 15% to 16% in Sweden and North America [2]. At King's College Hospital, between 1998 and 2018, 37 infants (7%) had heart disease; most had BASM (48%) or cat eye syndrome (11%), a total of 21 of 37 (57%) with one of these two forms; 62% had acyanotic heart disease, 27% cyanotic and 11% trivial, and 5 also had duodenal atresia [3]. Mortality was 40%, against 4% in contemporary controls, and 5-year native liver survival 54% against 65%; the authors recommend treating and prioritising the heart disease, because KPE rarely succeeds when cardiac function is poor [3]. Cardiac repair before KPE was associated with greater jaundice clearance [2].

3.7. Non-random associations and special populations

In the national series for England and Wales between 1999 and 2019, with 867 patients, there were 8 oesophageal atresias (expected frequency in the United Kingdom 1 in 3500), 13 duodenal atresias (expected 1 in 5000) and 4 jejunal atresias (expected 1 in 5000); duodenal atresia coexisted with BASM in 10 of 13 cases (77%) and oesophageal atresia in 1 of 8 [3]. In the same series there were 7 cases of cat eye syndrome (0.8%), 3 of Kabuki, 1 of Zimmermann-Laband, 2 of Hirschsprung disease and 2 of Kartagener syndrome, both with BASM [3]. Other series describe intestinal atresia in 0.5% to 5%, anorectal malformations, abdominal wall defects, hypospadias in less than 1%, oesophageal atresia in 1% to 3% and duodenal atresia in 1.5% in the United Kingdom, more frequent in BASM and absent in the series from Korea, India and China [2]. Preterm infants with BA have more associated malformations (19%, 30% and 40% in different series) and, in a British study, 48% were twins; KPE is later, but jaundice clearance and native liver survival are similar [2].

3.8. Syndromic forms and syndromes with biliary atresia: table of types

Table sources: [3, 6, 7, 9–16, 19–25, 36, 54, 58–60, 66–71].

3.9. Syndromes with biliary atresia: description

Cat eye syndrome was described by Schachenmann and colleagues in 1965 in four patients with iris coloboma and a small extra chromosome; it is due to a supernumerary bisatellited chromosome inv dup(22)(q11), with tetrasomy of the segment 22pter→q11, and overlaps with supernumerary der(22) syndrome and mosaic trisomy 22, chromosome 22 aneuploidy being the common denominator [3, 66]. Isolated cases with BA were reported in the 1970s; in the first true series, from London and Paris, four of five infants had BA and the fifth a choledochal cyst malformation, two with features of BASM, including one with polysplenia, and cases from Sweden and France were added later [3]. The occasional anomalies of the syndrome also include malrotation and colonic aganglionosis [66]. The association between iris coloboma and anal atresia was noted by Haab in 1879, and the first report with the small extra chromosome came from Schmid, in Zurich, and Fraccaro, in Pavia, who proposed the name by analogy with cri du chat syndrome; more than half of patients do not have coloboma, and the review of 105 published cases (2001) identified preauricular tags or pits as the most constant feature [21]. In decreasing order of frequency, the major malformations are anal atresia with fistula, coloboma of the iris, choroid or optic nerve (with almost always unilateral microphthalmia), cleft palate, heart disease (mainly total anomalous pulmonary venous return and tetralogy of Fallot), renal malformations, hernias and reduced pinnae with atresia of the auditory canal; intrafamilial variability ranges from almost asymptomatic carriers to lethal forms, most have normal or low-normal intellectual development, and, apart from early deaths due to multiple malformations, life expectancy is not greatly reduced [21]. BA has been reported since 1972, including a mother with cytogenetically confirmed syndrome whose son, with coloboma, preauricular pits and anal stenosis, had BA diagnosed in the first year, and a girl with a type II chromosome, microcephaly, bilateral anotia, coloboma, anal stenosis, ductus arteriosus and intra- and extrahepatic BA, who died at 1 month of heart failure [21]. The marker is always dicentric, formed so far from material of the two maternal chromosomes 22, and its breakpoints cluster in two intervals that coincide with those of the 22q11.2 deletion: type I chromosomes, smaller and symmetrical, have both breakpoints in the proximal interval, and type II, larger, have one breakpoint in each interval or both in the distal one; the size of the duplicated segment did not correlate with clinical severity [21]. Diagnosis rests on demonstrating, by FISH or genomic study, a chromosome 22-derived marker with two extra copies of the proximal 22q11 critical region; treatment is surgical for anal atresia and complex heart disease, and marked short stature warrants a search for hypothalamic growth hormone deficiency [21].

In Kabuki syndrome (Niikawa-Kuroki syndrome, recognised in 1981), caused by KMT2D variants in 55% to 80% of cases and by KDM6A variants in a minority, there are four reports with five cases of BA, two of them from Manchester in 2000; the diagnosis of BA often preceded that of Kabuki by years and KDM6A was implicated in at least two cases, including a cystic BA with a KDM6A variant and facial features reminiscent of Alagille syndrome, described in section 10 [3, 72]. The reports gathered in OMIM include the two girls from 2000 with extrahepatic BA in the first year of life; a female infant with atresia of the common bile duct and extrahepatic ducts operated on at 44 days, with complete clearance of jaundice at 5 months, later recurrence and successful liver transplantation at 20 years; a patient with central airway stenosis and extrahepatic BA; and, in 1998, an infant with hypoglycaemia, ileal perforation and hydronephrosis who developed cholangitis and liver abscess and was transplanted at 8 months, with histology of neonatal sclerosing cholangitis, a differential diagnosis to bear in mind [13]. Hardikar syndrome, described in 1992 and linked to pathogenic MED12 variants, combines cleft lip and palate, pigmentary retinopathy, coarctation of the aorta, hydronephrosis, malrotation and, sometimes, biliary disease in girls; in a series of 6 patients, 2 had a probable choledochal cyst, 2 BA and 1 absent gallbladder [3]. The 2025 review describes the variants as missense, but the 2021 study that identified them, as summarised by OMIM, found de novo nonsense or frameshift variants in 7 unrelated girls, aged from infancy to 21 years [3, 15]. In that series all had cardiovascular involvement (coarctation or dilatation of the aorta in 5, pulmonary stenosis, ductus arteriosus with partial anomalous pulmonary venous return, carotid aneurysm), 6 had genitourinary anomalies, 2 deafness, and one died at 21 years of intracranial haemorrhage; neurodevelopment was normal in all, which distinguishes Hardikar from Kabuki and Alagille syndromes, with which it is confused [15]. In the original patients, the biliary disease required surgery for recurrent obstructive jaundice at 13 years in one and liver transplantation at 15 years in another, and a third patient with cirrhosis was transplanted at 24 months; the authors stress the need to investigate and follow the cardiovascular manifestations to prevent serious events [15]. Zimmermann-Laband syndrome, due to KCNH1 variants, combines craniofacial anomalies, gingival fibromatosis, aplasia or dysplasia of the distal phalanges and variable delay, and has only rarely been reported with BA [3]. OMIM adds to the description dysplastic or absent nails, scoliosis, hepatosplenomegaly, hirsutism and anomalies of the nasal and ear cartilages, and records that a boy published in 2005 as a new syndrome of brachydactyly, extrahepatic BA, ductus arteriosus and seizures, who also had nephrolithiasis, non-autoimmune diabetes and bilateral cataracts, was reclassified as Zimmermann-Laband at 5 years [16].

Mitchell-Riley syndrome, caused by biallelic RFX6 variants, combines permanent neonatal diabetes in the first days of life, low birth weight, hypoplastic or annular pancreas with exocrine insufficiency, intestinal atresia or malrotation and gallbladder hypoplasia or agenesis; RFX6 explains about 5% of permanent neonatal diabetes, and Orphanet records 16 cases or families, with a prevalence below 1 per million [12, 35, 71]. The Oxford Desk Reference describes extra- and intrahepatic biliary atresia in this syndrome, but Davenport considers that there are biliary anomalies without obligatory true BA: some patients have bile in the obstructed duodenal segments, proof of some biliary patency, so surgery should restore continuity between duodenum, jejunum and bile duct; King's treated 3 cases, 2 of them siblings, of whom 2 are alive [3, 6]. The syndrome was described in 2004 in 5 infants (two sibling pairs, born to Pakistani first-cousin parents and to non-consanguineous Asian parents, and a girl conceived by in vitro fertilisation with a donated oocyte), with neonatal diabetes, hypoplastic or annular pancreas, duodenal and jejunal atresias and absent gallbladder, without dysmorphism; both sibling pairs died in the first year [24]. In 2010, Smith and colleagues found biallelic RFX6 variants in 5 of 6 probands: the 3 with complete loss of function died in the first months, and the 2 with residual function were alive at 4.5 and 9 years [17, 24]. Biliary involvement varies: a girl born to consanguineous Spanish Roma parents had acholia and absent extrahepatic ducts confirmed by scintigraphy and died at 60 days; in a Moroccan boy with congenital haemochromatosis, initial cholangiography suggested BA, but reoperation showed cystic dilatation of the extrahepatic ducts, with patent biliary and pancreatic ducts; and in two Turkish cousins compound heterozygous for two nonsense variants diabetes appeared only at 3 and 6 years, one without a gallbladder and the other without hepatobiliary anomaly, which extends the phenotype to attenuated forms [24]. Heterozygous parents with diabetes, gestational or of another type, have been described in several families [17]. Martínez-Frías syndrome, autosomal recessive, combines low birth weight, tracheo-oesophageal fistula, duodenal atresia, intra- and extrahepatic biliary atresia and hypospadias, and overlaps with Mitchell-Riley, from which it is distinguished by oesophageal atresia and the absence of neonatal diabetes; reports of BA in this syndrome are poorly defined [3, 6]. OMIM, which assigns it the % symbol because its molecular basis is unknown, gathers three reports: the original from 1992, in two siblings born to consanguineous Spanish Roma parents (the proband, with hypoplastic pancreas, duodenal atresia and atresia of the extrahepatic ducts, died on day 2; her brother, with tracheo-oesophageal fistula and hypospadias, on day 5); two siblings born to consanguineous parents with duodenal and oesophageal atresias, malrotation, intra- and extrahepatic BA and hypoplastic pancreas, who died at 25 and 81 days (1998); and a boy with tracheo-oesophageal fistula, duodenal atresia, annular pancreas, agenesis of the gallbladder and extrahepatic ducts and anorectal atresia, who died at 10 months (1999) [25]. No RFX6 variants were found in the parents of the latter, and in 2008 it was proposed that Martínez-Frías and Mitchell-Riley were a single syndrome with six core features (neonatal diabetes, intestinal atresia, malrotation, BA, gallbladder hypoplasia and absent or abnormal pancreas), a proposal not adopted by the separation of the two OMIM entries [24, 25]. Other diseases with diabetes and gallbladder anomalies, such as those associated with GATA6 (heart disease, gallbladder agenesis, diaphragmatic hernia) and CNOT1 (holoprosencephaly, pancreatic and gallbladder agenesis), enter the differential diagnosis of these forms [71].

Extrahepatic BA is among the abdominal anomalies of trisomy 18, alongside pyloric stenosis, hypoplastic gallbladder, gallstones and imperforate anus [58]. In Emanuel syndrome, due to der(22)t(11;22), BA, Hirschsprung disease, abnormal hepatic lobulation, abnormal extrahepatic ducts, absent gallbladder and polysplenia are occasionally seen [67]. In Simpson-Golabi-Behmel syndrome type 1, due to GPC3 variants, polysplenia, choledochal cysts and BA have been described [68]. In fumarate hydratase deficiency, hepatic involvement occurs in 10% of patients and includes cirrhosis, neonatal liver failure and one report of non-inflammatory BA [69]. In the 17q12 deletion with loss of HNF1B, several infants had neonatal cholestasis with paucity of interlobular ducts and variable periportal fibrosis, one of them requiring portoenterostomy [70]. There is also an isolated case of achondroplasia with BA [3].

4. Clinical Recognition and Warning Signs

4.1. When to suspect

Suspicion rests on a simple rule: every infant who is jaundiced after 2 weeks of life should be assessed with total and direct bilirubin, and a direct bilirubin above 1.0 mg/dL (17 µmol/L) requires timely evaluation and referral to paediatric gastroenterology or hepatology [42]. The guideline itself (grade 1A for formula-fed infants) and the 2024 review accept, depending on local practice, clinical follow-up until 3 weeks of a breastfed infant who looks well, with a normal examination, pale urine and pigmented stools, measuring bilirubin fractions then if jaundice persists; in European countries where the first visit is at 4 weeks, any jaundiced infant should be investigated immediately [2, 42]. The 2025 FISPGHAN position statement does not make this distinction and asks for fractionated bilirubin in all infants with jaundice beyond 2 weeks, whatever their feeding, and in all those with pale or acholic stools or hepatosplenomegaly at any age [73]. Bilirubin must always be fractionated (grade 1A), because visual estimation does not distinguish the direct fraction [42]. The «Alerta Amarelo» (Yellow Alert) campaign of the Brazilian Society of Paediatrics and the Ministry of Health applies the same principle: jaundice persisting beyond 14 days, acholic or hypocholic stools or dark urine, at any age, require clinical assessment, measurement of bilirubin with the direct fraction and referral to a specialised service when stools are suspicious or direct bilirubin is raised [74]. Stool and urine colour must be observed by the doctor personally, because the infant with BA usually looks well and has normal initial weight gain [37, 40]. The 2017 guideline considers direct observation of stools an essential part of the assessment (grade 1A), recalling that only 63% of health professionals correctly identified acholic stools, and the 2nd edition of the Japanese guideline recommends examining every newborn or infant with pale yellow stools, because before 1 month of age many patients do not yet have acholic or clay-coloured stools (strong recommendation, evidence C, agreement of 73%), as well as those with prolonged jaundice with hepatomegaly (strong, C, 85%) [42, 57]. The targeted examination looks for dysmorphism (Alagille, metabolic diseases), neurological deficits and a large fontanelle (Zellweger, Niemann-Pick), cataract (galactosaemia), hepatosplenomegaly, heart murmur (Alagille), bruising (vitamin K deficiency) and contractures (arthrogryposis, renal dysfunction and cholestasis syndrome, ARC) [40]. The 2024 French PNDS (French National Diagnosis and Care Protocol) adopts an earlier threshold than the 2017 guideline: all neonatal jaundice lasting beyond 10 days of life should be investigated, and the cholestatic nature of jaundice that persists or appears after 10 to 15 days should always be suspected; the same document uses 14 days in the section on the first-line doctor's role and 15 days in the summary for the attending doctor. Complete, prolonged stool discolouration for more than 7 days should lead to BA being assumed until proven otherwise, and complete discolouration with firm hepatomegaly, or the presence of elements of polysplenia syndrome or of abdominal or thoracic situs inversus, make BA very likely on clinical grounds alone, by an argument of frequency. Physiological jaundice, of unconjugated bilirubin, affects 60% of term newborns and 80% of preterm infants, and breast milk jaundice 2% to 3% of breastfed infants, lasting as long as breastfeeding continues; in both the stools are pigmented and the examination is normal apart from jaundice, which becomes visible above 40 to 50 µmol/L of total bilirubin [38].

4.2. Contexts of increased risk

Some situations warrant active surveillance. In children with heterotaxy and cardiovascular malformation, BA occurs in about 6%; since routine abdominal ultrasound has poor sensitivity as a screening tool, the 2022 multidisciplinary consensus recommends teaching families to recognise, in the first 8 weeks, jaundice, acholic stools and dark urine, with urgent assessment by paediatric gastroenterology or surgery if they appear [41]. In the United States there is no systematic screening for BA in these children [41]. The PKD1L1 genetic information recommends prompt attention to signs of BA, which is more common in heterotaxy [9]. Other risk contexts are children of mothers with pregestational diabetes, preterm infants and newborns small for gestational age, and children with cat eye syndrome, Kabuki syndrome or another syndrome in the table in section 3.8 [2, 3]. A cystic lesion under the liver on prenatal ultrasound should be reassessed soon after birth, because of the risk of cystic BA [5, 37].

4.3. Warning and emergency signs

Warning signs are jaundice that persists or reappears after 2 weeks, pale, creamy or white stools, dark urine that stains the nappy, firm hepatomegaly and poor weight gain [1, 37]. The NASPGHAN/ESPGHAN guideline defines as signs requiring investigation for BA acholic stools, cholestasis with raised GGT without another cause and an abnormal or absent gallbladder on ultrasound, and FISPGHAN adds, for BA and for genetic and metabolic diseases, abnormal coagulation, poor growth, vomiting or lethargy and neurological signs [42, 73]. Severe coagulopathy that does not respond to parenteral vitamin K and is disproportionate to the liver injury suggests gestational alloimmune liver disease, metabolic disease or sepsis, and not BA [42]. Haemorrhage (cutaneous, umbilical, gastrointestinal or intracranial) due to vitamin K deficiency is an emergency: it occurs in less than 5% of infants with BA, mainly when there was no neonatal prophylaxis, and the international guideline recommends intravenous vitamin K before any surgery [5, 37]. In BASM, early bilious vomiting suggests duodenal atresia or volvulus on malrotation, which requires urgent surgery, and fever above 38.5 °C in a child with asplenia or hyposplenism requires immediate assessment for sepsis and empirical antibiotic therapy [3, 41]. After KPE, fever without a focus, stools losing their bile colour and worsening jaundice suggest cholangitis, which should be treated as an emergency [5, 37]. The French PNDS lists the errors to avoid: failing to notice stool discolouration in the maternity unit, whose earliness strongly favours BA; considering jaundice beyond the 10th day physiological; attributing jaundice too readily to breast milk; being reassured by a normal growth curve, by a normal transcutaneous bilirubin or by the presence of a gallbladder on ultrasound; and being misled by stool colouring added by food or drug pigments or by dark urine. Transcutaneous bilirubinometry estimates only total bilirubin and should not be continued beyond 2 weeks without plasma measurement of total and conjugated bilirubin. After diagnosis, parents should seek care quickly in the event of fever above 38 °C, sudden stool discolouration, significant vomiting, haematemesis or melaena, rapid increase in abdominal volume or weight, or worsening or recurrence of jaundice [38].

4.4. Newborn screening

BA meets most of the criteria for newborn screening, and the meta-analysis cited by the international guideline recommends screening within the first 30 days [2, 5]. The stool colour card (SCC, stool colour card) is used in Japan, Taiwan and parts of Canada, Germany and Switzerland; in Taiwan it brought KPE forward, reduced late cases and improved native liver survival, but long-term universal benefit is not established, and the test is subjective [2, 37]. The sensitivity of the SCC is 79.6% and the specificity 99.9%, and the guideline recommends it as a simple, inexpensive and non-invasive screening tool that can be attached to maternal and child health records (recommendation 1, grade 1B) [5]. Mobile applications are under study (PoopMD, PopòApp), still to be validated because of chromatic aberration in screens and cameras [5, 37]. Brazil included a stool colour scale in the Child Health Record (Caderneta de Saúde da Criança), and in Portugal the inclusion of the card in the digital version of the Child and Youth Health Record (Boletim de Saúde Infantil e Juvenil) has been described [74, 75]. The 2017 guideline cited a sensitivity of 95.2% of the card for pale stools in Taiwan, cost-effective home screening in Canada and a pilot project in Switzerland, without systematic implementation in the United States or Europe [42]. FISPGHAN adds South Korea and Argentina to the countries with validated cards distributed to parents and first-line professionals, proposes handing out a neonatal cholestasis card already during pregnancy, as part of the maternal and child protection card, and considers the commitment of health authorities indispensable for universal screening [73]. The 2nd edition of the Japanese guideline recommends screening all newborns and infants at around 1 month of age (strong recommendation, evidence B, agreement of 95%), as the first feasible moment for universal screening in Japan, and insists that this screening does not authorise waiting: in the presence of symptoms before 1 month, assessment should be immediate, and KPE should ideally be performed before 30 days [57]. In France, the 2024 PNDS considers newborn screening justified and the colour card the simplest method: the stool colour scale appears in the child health record (carnet de santé), and the «Alerte jaune» leaflet reproduces it, with colours 1 to 3 indicating biliary obstruction. Despite repeated calls for observation of stool colour during the first month, the median age at KPE in France has hardly changed since 1986 (59 days), and the PNDS insists that observation of the stools by the professional personally in the maternity unit and at clinic visits and investigation of all jaundice beyond 2 weeks remain the best means of avoiding delay; systematic screening by conjugated bilirubin, under study, does not yet meet the HAS criteria, with efficacy and cost–benefit still to be assessed [38].

Biochemical screening measures direct or conjugated bilirubin in the first days. In Texas, a two-phase programme in 124,385 newborns from 14 hospitals (January 2015 to June 2018) measured bilirubin in the first 60 hours, considering positive a value above the reference 95th percentile, and repeated it up to the 2-week visit, considering positive a value higher than in the first phase or above 1 mg/dL; it identified the 7 cases of BA, with a sensitivity of 100% (95% CI 56.1–100), specificity of 99.9%, positive predictive value of 5.9% (95% CI 2.6–12.2) and negative predictive value of 100%, and was associated with younger age at KPE [76]. In the before–after component of the same study, at the largest paediatric hepatology centre in the region, mean age at KPE went from 56 days (SD 19; 24 infants, 2008–2011) to 36 days (SD 22; 19 infants, 2015–2018), a difference of 19 days (95% CI 7–32; P=0.004); of the 19 operated on after implementation, 6 came from the screening study, 7 from screening in other hospitals and 6 from referral for symptoms [76]. The authors stress that the confidence interval for sensitivity is wide and that the design did not guarantee complete detection of false negatives [76]. The international guideline considers direct bilirubin in heel-prick blood promising (recommendation 3, grade 1B), with a positive result above 3.4 µmol/L up to 60 hours and above 17.1 µmol/L in a second serum measurement up to 2 weeks, but less cost-effective and more invasive than the SCC; the conversions to mg/dL printed in the guideline are wrong (3.4 µmol/L corresponds to about 0.2 mg/dL and 17.1 µmol/L to 1.0 mg/dL) [5]. Dried blood spot screening faces technical difficulties of scale [2]. The 2025 position statement of FISPGHAN (Federation of International Societies of Pediatric Gastroenterology, Hepatology, and Nutrition) advocates universal newborn screening by SCC or by direct bilirubin, to identify patients before symptoms, with local flexibility according to resources [73]. There is no proven way to prevent BA [2].

5. Diagnosis: Criteria, Tests and Algorithms

5.1. Criterion for cholestasis and first-line laboratory tests

Diagnosis begins by demonstrating cholestasis: direct or conjugated bilirubin above 1.0 mg/dL (17 µmol/L), according to the joint 2017 NASPGHAN/ESPGHAN (North American and European Societies for Pediatric Gastroenterology, Hepatology and Nutrition) guideline [42]. The diazo method (van den Bergh) measures direct bilirubin, which includes conjugated bilirubin and albumin-bound delta bilirubin, and the direct value rises with total bilirubin even when this is almost all unconjugated; the guideline therefore prefers conjugated bilirubin when available, considers abnormal any direct or conjugated value above 1.0 mg/dL, regardless of the total, and rejects the criterion of a direct fraction above 20% of the total, as physiologically and clinically complex [42]. FISPGHAN accepts the ratio above 20% only when the direct value is doubtful, warns of false positives of the diazo method with high indirect hyperbilirubinaemia and recalls that, rarely, cholestasis manifests without raised direct bilirubin, through changes in aminotransferases, GGT or serum bile acids, the latter with a normal value below 10 µmol/L [73]. Other authors propose a threshold of 0.3 mg/dL (5 µmol/L), and direct or conjugated bilirubin may not be an exact marker of cholestasis [36]. The 2025 international guideline considers direct bilirubin equal to or above 17.1 µmol/L and above 20% of total bilirubin suspicious for BA; a GGT above 100 U/L suggests BA and above 300 U/L makes it very likely [5]. For FISPGHAN, a GGT above about 200 to 300 U/L suggests BA, although lower values do not exclude it [73]. GGT is low or normal in most progressive familial intrahepatic cholestases (PFIC), in ARC syndrome and in bile acid synthesis defects, and raised in diseases with abnormal ductal morphology; alkaline phosphatase is of little use because it also reflects growing bone [36, 40]. A GGT below 300 IU/L before KPE was associated with worse prognosis in a Chinese series [2]. The initial investigation of neonatal cholestasis, which the ERN RARE-LIVER/BARD (Biliary Atresia and Related Diseases) position statement refers to the 2017 guideline, includes review of newborn screening results (galactosaemia, hypothyroidism, cystic fibrosis), full blood count, alpha-1 antitrypsin level and phenotype, T4 and TSH, bile acids, cortisol, glucose, lactate and metabolic parameters, urinary reducing substances, bacterial and viral infections including polymerase chain reaction (PCR) for CMV in urine, liver ultrasound and spine radiograph looking for butterfly vertebrae, and an echocardiogram if there is a murmur [47]. The Oxford Desk Reference adds a search for haemolysis, congenital serologies, thyroid and pituitary function and, in a second phase, very-long-chain fatty acids, 7-dehydrocholesterol, mass spectrometry of urinary bile acids, organic acids, ophthalmological examination and genomic analysis when there are syndromic features [40]. FISPGHAN organises the assessment in two levels: level 1, which it considers mandatory and available almost everywhere, includes total and direct bilirubin, AST, ALT, GGT, albumin, glucose, full blood count, coagulation, blood cultures, urine examination and culture, urinary non-glucose reducing substances, TSH, thyroxine and cortisol, and fasting abdominal ultrasound; level 2, directed by the clinical picture or by the paediatric hepatologist, includes congenital serologies, sweat chloride (or immunoreactive trypsinogen or CFTR testing), alpha-1 antitrypsin phenotype or genotype, total serum bile acids and mass spectrometry of urinary bile acids, metabolic screening, ferritin, alpha-fetoprotein, MR cholangiography, scintigraphy, endoscopic retrograde cholangiopancreatography (ERCP) or percutaneous cholecystocholangiography, chest and spine radiographs, MMP-7, elastography and consultations with ophthalmology, interventional radiology or hepatology and genetics [73]. Mass spectrometry of urinary bile acids should be performed before starting ursodeoxycholic acid, so as not to miss a treatable synthesis defect [42]. The Italian (SIGENP, 2022, with GRADE methodology) and Spanish (SEGHNP, AEPap and SEPEAP, 2025, with 26 recommendations for primary and hospital care) national consensus statements follow the same logic, considering cholestasis always pathological and BA the most common hepatic cause in the first months of life [77, 78]. The French PNDS (French National Diagnosis and Care Protocol) defines cholestasis of the newborn as conjugated bilirubin above 17 µmol/L (10 mg/L) or a conjugated fraction above 10% of the total, in the section on positive diagnosis, and above 20% of the total, in the section on differential diagnosis, an internal inconsistency of the document that adds to the rejection of the proportion criterion by the 2017 guideline; it adds that cholestasis is always accompanied by fasting total bile acids above 10 µmol/L, except in bile acid synthesis defects, and that alkaline phosphatase does not assess cholestasis in children, because of the predominance of the bone isoform. The initial PNDS panel includes full blood count with platelets and reticulocytes; prothrombin time before and after vitamin K and factor V; ALT, AST, GGT, total and conjugated bilirubin and total bile acids; total cholesterol; blood glucose; vitamins A, D and E; quantitative alpha-1 antitrypsin; checking of the newborn screening immunoreactive trypsinogen or sweat test; and cultures (blood cultures, urine) in the presence of signs of infection or deterioration of general condition. The profile suggestive of BA combines raised total and conjugated bilirubin, aminotransferases usually moderately increased and raised GGT, which may nevertheless be normal for age without excluding the diagnosis. Correction of the prothrombin time after vitamin K, with normal factor V, distinguishes cholestasis from hepatocellular failure, in which factor V is low and the prothrombin time does not correct [38].

5.2. Tests and diagnostic accuracy

Table source: international guideline 2025 (Table 1) [5]. AUC: area under the ROC curve (receiver operating characteristic).

Postnatal ultrasound is the imaging test of choice (recommendation 6, grade 1A) and looks for the triangular cord sign, hepatoportal fibrous plaques, absence of the common bile duct, a hypoplastic or atrophic gallbladder with abnormal wall and morphology, a hepatic artery of larger diameter and cystic components; in a multicentre study, ultrasound supported by machine learning outperformed conventional interpretation [2, 5]. The presence of bile in the gallbladder almost always excludes BA, except possibly in type I [2]. Shear-wave elastography is useful (recommendation 6, grade 2B), with stiffness values above 7.7 kPa up to 90 days and above 8.8 kPa after 90 days associated with good accuracy, but multicentre studies are lacking [2, 5]. Serum MMP-7 has an area under the curve between 0.90 and 0.99 and is higher in BA than in normal neonates, alpha-1 antitrypsin deficiency and Alagille syndrome, but has no consistent threshold across manufacturers, so it is a potential indicator (recommendation 5, grade 2C) and not a routine test [2, 5, 37]. Plasma β-amyloid combined with biochemistry is experimental [2]. Hepatobiliary scintigraphy, which demonstrates absent biliary excretion, is poorly discriminating and is no longer recommended by the European position statement [2, 47]. The 2017 guideline summarises the figures: in a meta-analysis, absence of drainage had a sensitivity of 98.7% and a specificity of 70.4% for BA, so excretion of the tracer into the intestine almost excludes it, but ductal paucity, idiopathic neonatal hepatitis, low birth weight and parenteral nutrition also give studies without excretion; premedication with phenobarbital for 5 days often delays diagnosis and KPE without demonstrated utility; the limited specificity precludes its use alone (grade 1B) and definitive demonstration of bile flow can serve to exclude BA (grade 1B) [42]. In the same guideline, ERCP had a sensitivity of 86% to 100% and a specificity of 87% to 94%, but requires an experienced endoscopist, neonatal equipment and general anaesthesia; three-dimensional MR cholangiography had a sensitivity of 99% and a specificity of 36%; and percutaneous cholecystocholangiography may miss a proximal obstruction [42]. Abdominal ultrasound (grade 1A) excludes choledochal cyst and gallstones, and findings of abdominal heterotaxy, midline liver, polysplenia, asplenia or preduodenal portal vein increase suspicion of BA with malformations; a normal ultrasound does not exclude non-syndromic BA [42]. According to FISPGHAN, the gallbladder should be assessed after 3 to 4 hours of fasting, with intravenous fluids if fasting is longer, and suggestive findings include a non-visualised or small gallbladder (less than 1.5 to 2 cm), with abnormal wall and shape, without contraction after a feed, and the triangular cord sign, findings that are highly specific but only moderately sensitive [73]. Data from the Japanese registry show that ultrasound is the most used test in Japan, followed by scintigraphy, serum bile acids and duodenal fluid analysis; the Japanese guideline considers the triangular cord sign the most important ultrasound finding, but recalls that many newborns do not reach the 4 mm criterion and that some subtypes do not form a fibrous mass at the hilum, so that, if BA cannot be excluded, one should proceed without delay to direct cholangiography [57]. The French PNDS asks for 4 to 6 hours of fasting, with an infusion if necessary, and highlights as highly suggestive a microcyst within the hyperechoic fibrous cone of the hilum; a larger hilar cyst, in BA, is never accompanied by dilatation of the bile ducts, which separates it from a choledochal cyst. If the gallbladder is visible, ultrasound can be repeated after a bottle feed to assess emptying, and liver elastography, when available, shows increased stiffness. Visualisation of the gallbladder or a normal ultrasound do not exclude BA limited to the hepatic ducts. Unlike the 2017 guideline and FISPGHAN, the PNDS considers MR cholangiography and hepatobiliary scintigraphy useless for this diagnosis, because they neither confirm nor exclude it with certainty and delay KPE [38].

Percutaneous liver biopsy shows portal inflammation, bile plugs, ductular proliferation and reaction and fibrosis, with an accuracy of about 90% (95% CI 85–95), although none of these findings is exclusive and some patients lack early ductular proliferation; it is the final step before cholangiography in most European centres, and the international guideline considers that it may help, with agreement from only 58% of the panel (recommendation 7, grade 2B) [2, 5, 47]. The NASPGHAN/ESPGHAN guideline sees in a timely biopsy, read by an experienced paediatric pathologist, the test that best supports the diagnosis (grade 1B), with a correct diagnosis in 90% to 95% of cases; in the North American consortium, pathologists recognised obstruction in 79% to 98% of proven BA, with a positive predictive value of 90.7%, but also classified as obstructive 14 of 15 cases of parenteral nutrition-associated liver disease and the 3 cases of alpha-1 antitrypsin deficiency [42]. Ultrasound-guided percutaneous biopsy has complications in 1.7% of children, but in a small series of infants 3 of 65 (4.6%) had haemorrhage requiring intervention [42]. The 2nd edition of the Japanese guideline partially reversed the position of the 1st, which advised against preoperative biopsy, and now recommends it weakly and in a limited way for differential diagnosis (strength 2, evidence D, agreement of 79%), given advances in genetics and the rarity of complications [57]. The 1st edition recognised the high sensitivity and specificity of biopsy, but suggested not performing it because of the risk of serious complications, including death, and the delay to surgery; in the abridged English version this suggestion has strength 2 (weak), evidence C and agreement of 70%, whereas the 2nd edition describes it as strength 3 [57, 79]. FISPGHAN adds that Alagille syndrome, alpha-1 antitrypsin deficiency, cystic fibrosis, parenteral nutrition-associated cholestasis and some PFIC can mimic BA histology when the sample is small [73]. Biopsy distinguishes biliary obstruction from PFIC: in severe ATP8B1 deficiency there is bland intracanalicular cholestasis, preserved lobular architecture and small ducts without paucity or the ductular reaction typical of obstruction [80]. In Alagille syndrome, by contrast, ductal paucity may be absent before 6 months and biopsy may show ductular proliferation, with a risk of misdiagnosis as BA [50]. For the French PNDS, signs of obstruction on liver biopsy (bile plugs, ductular proliferation, portal oedema or fibrosis) are not specific, also occur in neonatal sclerosing cholangitis and cystic fibrosis and may be subtle in the first weeks, so the diagnosis of BA never rests on liver histology alone; biopsy quantifies fibrosis and, when the diagnosis is another, immunohistochemistry with antibodies against proteins involved in cholestatic diseases points to a genetic cause [38].

5.3. Confirmation: surgical exploration and cholangiography

A definitive diagnosis is only possible by intraoperative cholangiography (recommendation 8, grade 1A), which diagnoses BA and excludes choledochal cyst [2, 5, 36]. In the European survey, 80% of surgeons confirm the diagnosis through a small right upper quadrant incision and 60% consider cholangiography mandatory; an atrophic gallbladder without a lumen, which is the most common situation, precludes cannulation and practically confirms BA, whereas about 20% have a structurally normal gallbladder filled with clear mucus, with a cholangiogram that opacifies a patent common bile duct without a proximal tree [47]. According to the international guideline, an atrophic gallbladder without bile indicates BA, and a gallbladder with bile requires cholangiography to exclude choledochal cyst; the subcapsular «spider» telangiectases seen at exploration are associated with BA [5]. Cholangiography is not always conclusive: passage of contrast from the gallbladder to the intestine does not exclude proximal obstruction, and a hypoplastic but patent tree may correspond to Alagille syndrome [2]. Endoscopic retrograde cholangiopancreatography (ERCP) is an alternative where expertise exists, and some Asian centres prefer diagnostic laparoscopy followed by immediate KPE [2, 5]. Once BA is confirmed, 95% of European surgeons actively look for features of syndromic BA, such as polysplenia [47]. The NASPGHAN/ESPGHAN guideline considers intraoperative cholangiography with histological examination of the ductal remnant the reference standard (grade 1A), but warns that up to 20% of cholangiograms suggest a wrong diagnosis when the tree is hypoplastic, in Alagille syndrome and cystic fibrosis, and therefore recommends investigating these two diseases before surgery; the decision to perform KPE or to proceed directly to transplant assessment rests with the team, as there is no diagnostic tool to guide it [42]. FISPGHAN recommends simultaneous liver biopsy, to assess severity and diagnose other diseases, requires the cholangiogram to assess patency proximal and distal to the gallbladder and mentions laparoscopic cholangiography with biopsy followed by KPE in the same procedure [73]. The Japanese guideline recommends classifying the type of BA by intraoperative cholangiography, because of its prognostic value (strong recommendation, evidence C, agreement of 88%), and pathological examination including hilar tissue, because the diameter of the hilar ducts is associated with native liver survival (strong, C, 82%) [57]. The French PNDS describes three routes for cholangiography, chosen according to availability and, whenever possible, under the same anaesthetic as KPE: surgical, when suspicion is strong, especially after 45 days; radiological, by transhepatic cholecystography, if the gallbladder is patent, of sufficient volume and accessible; and endoscopic retrograde, more rarely, when suspicion is weak (normal GGT, pale yellow stools, ultrasound without suggestive signs). Inability to opacify the whole biliary tree establishes BA, and complete visualisation of the tree with passage into the gut excludes it. The operation begins with a right transverse minilaparotomy over the gallbladder, at the level of the 10th rib, with transvesicular opacification whenever the gallbladder content exceeds 1 mL: yellow content, or opacification of the confluence and the common bile duct, excludes BA and limits the procedure to a liver biopsy. Definitive diagnosis requires histological examination of the remnant, sent oriented by the surgeon (hilar and distal portion) with an explanatory diagram, which shows complete obstruction of the ducts at at least one level of section [38].

5.4. Investigation of syndromic forms

When BASM or heterotaxy is suspected, investigation includes echocardiography, ultrasound or tomography to define situs, spleen, IVC and portal vein, and a peripheral blood smear for Howell-Jolly bodies, which is unreliable before 2 years, so functional hyposplenism should be presumed until assessed [41, 60]. Heterotaxy occurs in 9% to 12% of patients with primary ciliary dyskinesia, and Kartagener syndrome was recorded in two patients with BASM in the England and Wales series [3, 81]. When there are respiratory signs, primary ciliary dyskinesia requires at least two of four clinical criteria (unexplained neonatal respiratory distress, year-round wet cough, year-round nasal congestion and laterality defect) and confirmation by nasal nitric oxide repeatedly below 77 nL/min, genetic testing or electron microscopy [41]. The heterotaxy consensus recommends assessment by a geneticist in all patients and the study of copy-number variations (CNV) and sequence variants with heterotaxy, laterality, heart disease, ciliary dyskinesia or ciliopathy panels, ZIC3 testing being important for recurrence risk; there are no formal guidelines for genetic testing [41]. Dysmorphism, coloboma, cleft palate, neonatal diabetes or other malformations point to karyotype, chromosomal microarray or exome, depending on the suspected syndrome (table in section 3.8) [3, 40]. In CMV-associated BA, the definition requires serum anti-CMV IgM and/or CMV DNA in urine or blood at presentation, and the European position statement supports testing for CMV at the time of KPE [5, 47].

5.5. Role and limits of genetic testing

BA is an anatomical and surgical diagnosis, and no genetic test confirms or excludes it. Cholestasis panels serve the differential diagnosis with monogenic causes, which explain about a quarter of neonatal cholestasis [42, 43]. The largest published panel study (10,894 samples between 2016 and 2022, 0 to 88 years, 51.1% under 1 year, with 57 to 77 genes) expressly excluded BA, choledochal cyst and large-duct sclerosing cholangitis; the overall yield was 9.0% (6.8% definitive and 2.2% potential), mainly from JAG1, SERPINA1, ABCC2, ABCB11, CFTR, POLG and NOTCH2, and turnaround time fell from 3 months to 3 to 4 weeks [43]. This panel included PKD1L1, INVS and KMT2D, but PKD1L1 gave no definitive diagnosis, only one potential diagnosis and 10 monoallelic results [43]. The 2022 European position statement considers that genetic analysis is not yet a prerequisite before cholangiography and KPE because of turnaround time, which may change with faster results, especially in infants seen in the first weeks, before acholia [47]. The 2025 FISPGHAN position statement estimates that genetic causes explain 25% to 50% of neonatal cholestasis, depending on the population, and that, once BA is excluded, next-generation sequencing reaches a diagnosis in 10% to 30% of cholestatic infants, depending on the rigour of selection; it accepts that confirmation of biliary patency and genetic testing may be performed simultaneously or sequentially, states that Mendelian factors do not appear to play a role in isolated BA and that, among the susceptibility genes studied, only PKD1L1 has been implicated in a subset of BASM, which it includes in its table of genes with autosomal recessive inheritance [73]. In low-resource regions, FISPGHAN prioritises genetic testing in infants with low GGT, hypoglycaemia or an abnormal metabolic profile, consanguinity or several cases in the family, in progressive liver disease of unknown cause and in transplant candidates without BA; it recommends reanalysis of the data in patients with chronic disease and an initially negative result, because variant classification changes, and caution with heterozygosity in ATP8B1, ABCB11 and ABCB4 associated with transient neonatal cholestasis, whose relationship with severity is not established [73]. The 2017 guideline recalls that the significance of a variant depends on the clinical context [42]. The reverse risk also exists: in the National Birth Defects Prevention Study, a child registered with BA had the de novo variant NOTCH2 NM_024408.4:c.5194C>T p.(Gln1732Ter), which the authors interpret as possible misdiagnosed Alagille syndrome [54, 62].

A genetic test report in the context of BA should state the gene, the versioned transcript (for example, PKD1L1 NM_138295.5, NOTCH2 NM_024408.4, KDM6A NM_001291415.2), the variant in HGVS (Human Genome Variation Society) nomenclature, zygosity and phase when there are two variants, and the ACMG/AMP (American College of Medical Genetics and Genomics/Association for Molecular Pathology) class [62, 63]. A single heterozygous PKD1L1 variant does not explain BASM, and CFC1 variants must be read as susceptibility alleles according to the GeneGuidance curation or as a dominant cause with incomplete penetrance according to OMIM, in a region that short reads cover poorly because of a paralogue [10, 20, 43]. In heterotaxy, next-generation sequencing identifies a causal variant in only 20% to 30% of cases, so a negative result does not exclude a genetic cause [9]. Variants of uncertain significance require segregation analysis in the parents, and a variant inherited from a parent with situs inversus but without BA, as seen in CFC1 and FOXA2, is not in itself evidence against its relevance, given incomplete penetrance [3, 7]. Exome sequencing does not reliably detect structural variations, balanced rearrangements and low-level mosaicism, and the supernumerary chromosome of cat eye syndrome requires a karyotype or cytogenomic study [66].

5.6. Practical algorithm

The minimum assessment by any professional faced with an infant jaundiced after 14 days includes a complete, family and gestational history, physical examination, observation of the stools and fractionated bilirubin; in the presence of cholestasis, targeted and rapid investigations follow and, in liaison with the paediatric gastroenterologist or hepatologist, a staged investigation [42]. FISPGHAN proposes a conceptual flowchart with levels 1 and 2 of section 5.1 and encourages each reference centre to build its own algorithm according to resources and the local spectrum of causes, to minimise the time to diagnose or exclude BA [73]. The pathway recommended by the sources consulted can be summarised in six steps. First, total and direct bilirubin in every infant jaundiced after 2 weeks, or earlier if there are pale stools or dark urine [42, 74]. Second, in the presence of cholestasis, immediate referral to a centre with paediatric hepatology and hepatobiliary surgery, parenteral vitamin K and assessment of coagulation [5, 42]. Third, abdominal ultrasound, GGT, targeted metabolic, infectious and endocrine investigation and, where available, MMP-7 and elastography [5, 47]. Fourth, percutaneous liver biopsy in the centres that use it, when time allows [5, 47]. Fifth, surgical exploration with intraoperative cholangiography and, if BA is confirmed without contraindication, KPE in the same operation [2, 5]. Sixth, investigation of syndromic and CMV-associated forms, which modifies treatment and counselling, with phenotype-directed genetic testing and a cholestasis panel when cholangiography excludes BA [41, 43].

The French PNDS, based on Gonzales and Jacquemin, proposes its own algorithm. In the presence of frank and persistent stool discolouration, ultrasound determines whether there is dilatation of the bile ducts, which points to gallstones, choledochal cyst or another extrahepatic cause; without dilatation, Alagille syndrome (spine radiograph, ophthalmological examination, echocardiogram, facies), alpha-1 antitrypsin deficiency (serum level) and cystic fibrosis (immunoreactive trypsinogen, sweat test) are sought; if these are negative, cholangiography follows, preceded by liver biopsy when this shows signs of obstruction, and cholangiography separates BA from neonatal sclerosing cholangitis. In the newborn, causes requiring urgent treatment (BA, urinary tract infection or sepsis, cortisol deficiency, obstruction of the main bile duct) and easily detectable causes (Alagille, choledochal cyst, cystic fibrosis, alpha-1 antitrypsin deficiency) are investigated first, with the aim of diagnosing BA before 1 month, and suspicion of an infectious foetopathy should not delay that diagnosis. With strong suspicion, urgent surgery with intraoperative cholangiography and KPE in the same operation is scheduled; with weak suspicion, especially with partially discoloured stools, endoscopic retrograde or percutaneous cholangiography and liver biopsy are performed first. If the initial investigation does not find the cause, the child remains under hospital observation: normalisation of stool colour within about 15 days points to transient neonatal cholestasis, a diagnosis of exclusion of multifactorial cause (hepatic immaturity, ischaemia, sepsis, interruption of the enterohepatic cycle, genetic factors) that accounts for about 10% of neonatal cholestasis, with early jaundice with raised GGT and hepatomegaly and spontaneous resolution at around 1 year, which must be confirmed by prolonged clinical and biochemical follow-up until complete normalisation; absence of frank recolouration requires cholangiography and liver biopsy [38].

5.7. Prenatal diagnosis

Cystic BA and the heart defects of BASM can be detected on prenatal ultrasound, and microcysts at the liver hilum, a dilated right hepatic artery and peritoneal effusion have been described as prenatal signs, with scarce evidence [2, 5]. Prenatal ultrasound alone is not recommended for diagnosis (recommendation 2, grade 2C), but foetuses with abnormal findings should be followed after birth (grade 1B) [5]. For laterality without a specific diagnosis, the Oxford Desk Reference proposes foetal echocardiography at 14, 19 and 22 weeks in subsequent pregnancies [64]. Molecular prenatal or preimplantation diagnosis is only possible when the monogenic or chromosomal cause has been identified in the family, as with biallelic PKD1L1 or RFX6 variants [12, 40]. For the French PNDS, the main prenatal signs are apparent absence of the gallbladder and heterotaxy. A non-visualised gallbladder may correspond to type 4 BA, isolated gallbladder agenesis, cystic fibrosis (microgallbladder) or simply a failure to visualise a normal gallbladder, but always requires assessment from birth, with monitoring of stool colour, bilirubin, ultrasound and specialist opinion. A cystic structure at the hilum suggests cystic BA or congenital dilatation of the common bile duct. GGT in amniotic fluid collected before closure of the anal sphincter (at about 18 weeks) may suggest BA when it is low, but never confirms it. In the presence of prenatal suspicion, the PNDS recommends consultation with a paediatric hepatologist or hepatobiliary surgeon during pregnancy and specialist postnatal assessment by 15 days of life, recalling that stool discolouration may appear after the 15th day, especially in preterm infants [38].

6. Differential Diagnosis

6.1. Comparative table

Table sources: [2, 5, 36–38, 40, 50, 80, 82–88].

6.2. Main differential diagnoses

Alagille syndrome is the most important genetic differential, because cholangiography may show a hypoplastic but patent tree and early biopsy may show ductular proliferation instead of paucity; involvement of the extrahepatic bile ducts occurs in about 20% of patients with Alagille [2, 50, 89]. Alpha-1 antitrypsin deficiency, present in 14% to 29% of infants with neonatal hepatitis syndrome, may present with acholic stools and mimic extrahepatic BA, but jaundice usually resolves by 7 months; in Sveger's Swedish cohort, 11% of 127 ZZ homozygous newborns had prolonged neonatal jaundice [40, 82]. After BA, alpha-1 antitrypsin deficiency is the most frequent cause of liver transplantation in childhood [40]. In cystic fibrosis, presentation with isolated biliary obstruction in infancy is rare, and immunoreactive trypsinogen and faecal elastase are normal in BA [83]. Citrin deficiency is distinguished by lower direct bilirubin and ALT and higher total bile acids and alkaline phosphatase [84]. Niemann-Pick disease type C was found in 27% of infants initially classified as having idiopathic neonatal cholestasis and in 8% of all infants with cholestasis in a Colorado study [85].

Among surgical causes, choledochal cyst malformation is the one most often confused with cystic BA: both may be detected before birth, but choledochal malformation has a much better prognosis and only cholangiography, by opacifying a dilated and communicating intrahepatic tree, reliably separates them [37, 47]. The international guideline recommends excluding, before treating as BA, PFIC, citrin deficiency, inborn errors of bile acid synthesis, alpha-1 antitrypsin deficiency, Alagille syndrome, choledochal cyst and idiopathic infantile cholestasis [5]. Non-genetic causes to remember also include hypothyroidism, parenteral nutrition-associated cholestasis and gestational alloimmune liver disease [86, 87]. In syndromic forms, the differential diagnosis goes through the syndromes in the table in section 3.8: the combination of BA with neonatal diabetes points to RFX6, that of BA with oesophageal atresia to Martínez-Frías syndrome, and gallbladder agenesis with diabetes to GATA6 or CNOT1 [6, 71, 90].

In the French PNDS (French National Diagnosis and Care Protocol), BA accounts for 25% to 40% of neonatal cholestasis in the section on differential diagnosis and 45% in the section addressed to the first-line doctor; Alagille syndrome 10% to 15%, alpha-1 antitrypsin deficiency 5% to 10% and transient neonatal cholestasis about 10%. In alpha-1 antitrypsin deficiency, the PiZZ phenotype is the one that most mimics BA, with severe cholestasis and complete stool discolouration, and about 10% to 15% of these children progress to cirrhosis with an indication for transplantation; the SZ and MZ phenotypes may accompany neonatal cholestasis, but are not enough to stop the search for BA, and isoelectric focusing or genotyping should not delay diagnosis. In Alagille syndrome, JAG1 and NOTCH2 sequencing establishes the molecular diagnosis in more than 95% of cases, an atrophic gallbladder may mislead towards BA and, when the neonatal picture is incomplete, the diagnostic and therapeutic decision should be made in a reference centre. Neonatal-onset sclerosing cholangitis mimics BA, with discoloured stools, but jaundice usually regresses spontaneously after a few months, with persistent biochemical cholestasis and cirrhosis in some patients; diagnosis requires direct cholangiography (thin or irregular ducts), because MR cholangiography lacks sufficient resolution in the newborn, and histology shows signs of obstruction, lymphocytic cholangitis and sometimes ductal paucity. The PNDS adds UNC45A deficiency and familial hypercholanaemia to the genetic intrahepatocellular cholestases and, among those with raised GGT, highlights PFIC type 3 [38].

History and examination guide the differential diagnosis: other affected relatives suggest dominant transmission and consanguinity recessive transmission; pruritus, cholestasis of pregnancy or gallstones in the family point to PFIC or benign recurrent intrahepatic cholestasis; passage of meconium after 48 hours to cystic fibrosis; diarrhoea to an infectious cause or to PFIC1; prematurity with parenteral nutrition to parenteral nutrition-associated cholestasis; vomiting and irritability to metabolic disease, sepsis or foetopathy; splenomegaly with hard hepatomegaly to BA; and micropenis to panhypopituitarism. Fractionation of bilirubin immediately excludes the much more frequent unconjugated jaundices, with pale urine, pigmented stools and no hepatomegaly: constitutional haemolysis (membrane, haemoglobin, G6PD or pyruvate kinase deficiencies) and acquired haemolysis (maternofoetal alloimmunisation, immune, mechanical and infectious causes) and conjugation defects (physiological and breast milk jaundice, Gilbert syndrome, Crigler-Najjar disease), as well as hypothyroidism and trisomy 21 [38].

Primary bile acid synthesis defects are the treatable differential diagnosis most easily missed among cholestases with normal GGT. The French PNDS dedicated to them, from 2019, attributes 1% to 2% of childhood cholestasis to them; in an international survey, the combined minimum prevalence of 3β-hydroxy-Δ5-C27-steroid oxidoreductase (3β-HSD, gene HSD3B7) and Δ4-3-oxosteroid 5β-reductase (Δ4-3-oxo-R, gene AKR1D1) deficiencies was 1.13 per 10 million, with predominance of the former, and reached 4.02 per 10 million in France, partly because of referral of patients from other countries; the true prevalence is probably higher, because these diseases are underdiagnosed. The decisive feature is the contrast between signs of cholestasis (conjugated bilirubin jaundice, discoloured stools, hepatomegaly) and a normal GGT with normal or low total serum bile acids, without pruritus and without dilatation of the bile ducts, whereas in the other cholestatic diseases fasting total bile acids rise above 10 µmol/L and, in BA, so does GGT. In the PNDS algorithm for the child, if ultrasound shows no dilatation of the bile ducts, GGT separates diseases of the hepatocyte (normal GGT, in which the absence of pruritus points to a synthesis defect, especially 3β-HSD deficiency) from those that injure the biliary epithelium (raised GGT: Alagille syndrome, cystic fibrosis, alpha-1 antitrypsin deficiency, cholestasis of exclusive parenteral nutrition, sclerosing cholangitis), and a high serum bile acid concentration rules out synthesis defects. PFIC and related hepatocellular cholestases, which together account for 10% to 15% of childhood cholestasis, share a normal GGT (except PFIC3), but have raised serum bile acids and intense pruritus from the first months. In synthesis defects, ultrasound serves mainly to exclude bile duct disease, and biopsy, which is not indispensable, shows hepatocellular or canalicular cholestasis without signs of obstruction, sometimes giant cell hepatitis and portal and lobular fibrosis of variable severity [88].

Confirmation is made in a highly specialised laboratory, by analysis of urinary and serum bile acids by chromatography coupled with mass spectrometry, which shows absence or traces of primary bile acids and an excess of atypical metabolites specific to each deficiency, and then by study of HSD3B7 and AKR1D1. The urinary profile of Δ4-3-oxo-R deficiency is not pathognomonic, because the enzyme's activity falls in any advanced liver disease (for example, in neonatal haemochromatosis), and only study of AKR1D1 confirms the primary deficiency. Δ4-3-oxo-R deficiency is much rarer than 3β-HSD deficiency, manifests earlier, at around 3 months, with a more intense clinical and biochemical picture and faster progression to hepatocellular failure; 3β-HSD deficiency, with more than 70 published cases and about fifteen pathogenic variants in the homozygous or compound heterozygous state without known genotype–phenotype correlation, also has late and varied forms: liver disease that regresses spontaneously or progresses in flares, a digestive form with steatorrhoea and rickets, hepatomegaly with poor weight gain and almost normal liver tests, asymptomatic forms detected in sibling studies and diagnoses in adolescence or adulthood. Among the other synthesis defects, those of the steroid nucleus (for example, oxysterol 7α-hydroxylase deficiency, CYP7B1) cause isolated severe liver disease; those of the side chain (sterol 27-hydroxylase, alpha-methylacyl-CoA racemase, D-bifunctional protein) manifest mainly as neurological disease and malabsorption, with secondary hepatic involvement, because residual synthesis of primary bile acids persists; and those of conjugation (BAAT, SLC27A5) cause neonatal cholestasis with malabsorption of fat-soluble vitamins [88].

Without treatment, 3β-HSD and Δ4-3-oxo-R deficiencies progress to cirrhosis and hepatocellular failure; with oral cholic acid, jaundice and liver tests normalise within months, histological lesions improve and transplantation is avoided in the vast majority of patients, with the effect maintained in series with a mean follow-up of 12.5 years. For infants, children, adolescents and adults, the PNDS sets 5 to 15 mg/kg/day orally, with a minimum of 50 mg between 1 month and 18 years and adjustments in 50 mg steps guided by urinary and serum bile acids, divided into two doses when there is more than one capsule, and a maximum of 500 mg/day in adults, under the terms of the European marketing authorisation of Orphacol®, granted in 2013 for these two deficiencies; the US label of Cholbam® (approved in 2015) gives, for comparison, 10 to 15 mg/kg/day in one or two doses, in children and adults. Ursodeoxycholic acid may serve only as a bridge while awaiting confirmation, because it does not restore negative feedback on cholesterol 7α-hydroxylase or reduce toxic metabolites, and is not indicated as monotherapy [88, 91, 92]. The PNDS requires urgent referral of any newborn with cholestasis and discoloured stools to a centre of the French network for BA and genetic cholestasis with experience of KPE, where exclusion of BA and the search for a synthesis defect proceed in parallel, and requires assessment to begin in any newborn still jaundiced at 14 days [88].

The same 2019 PNDS divides cholestasis of the newborn into about 5% of causes limited to the extrahepatic bile ducts (gallstones, choledochal cyst, congenital stenosis, spontaneous perforation, tumour), 47.5% of causes affecting both the extra- and intrahepatic ducts (mainly BA, with polysplenia syndrome in about 10% of cases, and neonatal sclerosing cholangitis) and 47.5% of intrahepatic and hepatocellular causes; it does not state the source series for these proportions. In Alagille syndrome, it reports a molecular diagnosis by JAG1 and NOTCH2 in more than 90% of cases, and in alpha-1 antitrypsin deficiency it associates liver disease with the PiZ phenotype, with neonatal cholestasis in 10% to 15% of affected newborns and the diagnosis often suggested by absence of the alpha-1 globulin peak on protein electrophoresis. Congenital CMV infection affects 1% to 2% of newborns and is symptomatic in 5% to 10%, but attributing cholestasis to it requires caution, because BA can coexist with an infectious foetopathy; urinary tract infection, which improves within a few days with antibiotic therapy, is also only accepted as the cause after BA has been excluded, and persistence of jaundice requires the diagnosis to be revisited. Stones in the main bile duct generally appear after a free interval of a few weeks, with partial or transient stool discolouration and a triggering factor (dehydration, infection, parenteral nutrition, haemolysis) in about half of cases; a cystic formation of the bile duct accompanied by neonatal cholestasis should raise the possibility of cystic BA and prompt referral to a reference centre [88].

7. Treatments and Risk Management Strategies

7.1. Principles and organisation of care

Treatment of BA is sequential: timely KPE to restore bile flow, adjuvant and nutritional therapy, prevention and treatment of complications and liver transplantation when KPE fails or the disease progresses [2, 5]. Currently, 85% to 95% of patients undergo KPE, at around 2 months of age, and 5% to 15% undergo primary transplantation because of late diagnosis [2]. The 2025 international guideline, prepared by the Hepatobiliary Surgery Section of the Paediatric Surgery Branch of the Chinese Medical Association and the Paediatrics Section of the Chinese Chapter of the International Hepato-Pancreato-Biliary Association, with 50 experts from the United States, Europe and Asia, GRADE grading (Grading of Recommendations, Assessment, Development and Evaluation) and 23 recommendations, advocates treatment in reference centres linked in national and international networks, with patient registries [5]. Mandatory centralisation in England and Wales, in three centres since 1999, later extended to the rest of the United Kingdom and the Republic of Ireland, lowered the median age at KPE to under 50 days and increased jaundice clearance, and was adopted in Norway, the Netherlands, Sweden and Finland; in Germany, without centralisation, the national jaundice clearance rate is around 24% to 25% [2, 37]. France chose to keep care decentralised and, from 1997, to strengthen collaboration between centres, with a national observatory and support from experienced centres to those with small caseloads; in those born in 1997–2002, overall 4-year survival was similar to that in England and Wales in 1999–2002 (87.1% against 89%), but native liver survival 4 years after KPE was lower (42.7% against 51%, a non-significant difference) and the proportion of transplanted patients among those alive at 4 years was higher (51% against 43%), a difference the authors attributed to centres performing 2 or fewer KPEs per year [39]. KPE performed in a centre without a transplant programme is associated with a longer and more haemorrhagic hepatectomy at subsequent transplantation and with worse transplant prognosis [47]. In a survey by the ESPGHAN quality of care group, the mean age at referral was 55 days and at KPE 61 days, and most regions of the world do not publish age at KPE [73]. In France, the 2024 PNDS (French National Diagnosis and Care Protocol) requires urgent referral of every newborn with cholestasis, especially with discoloured stools, to a reference or competence centre of the biliary atresia and genetic cholestasis network with experience in KPE; the case is discussed at a multidisciplinary medical–radiological–surgical meeting, KPE is performed in an experienced paediatric surgery department, ideally in a paediatric liver transplant centre, and the team brings together hepatologists, surgeons, paediatric anaesthetists and intensivists, neonatologists, radiologists, the attending doctor, psychologists, psychomotor therapists, speech therapists, dietitians and social workers. The role of the first-line doctor is to diagnose cholestasis without delay (clinical assessment attentive to stool and urine colour, conjugated bilirubin and ultrasound in every child jaundiced at or after 14 days), give parenteral vitamin K and refer; the aetiological diagnosis is made in the specialised centre [38].

7.2. Preoperative preparation

The risk of haemorrhage due to vitamin K deficiency justifies intravenous vitamin K at a dose of 0.2 to 0.3 mg/kg per dose, 1 to 2 hours before surgery (recommendation 9, grade 1D, agreement of 100%), especially where neonatal prophylaxis is not universal [5]. The French PNDS recommends an intravenous or intramuscular injection of 10 mg of vitamin K, whatever the weight, as soon as cholestasis is diagnosed, to prevent intracranial haemorrhage, and repeats it as the first action of the first-line doctor [38]. The dose differs from the international guideline, whose dosage of 0.2 to 0.3 mg/kg corresponds to 0.8 to 1.2 mg in a 4 kg infant; the two regimens serve different purposes (correction of deficiency at the time of diagnosis and immediate surgical preparation) and have not been compared [5, 38]. The guideline considers that KPE is not indicated in the presence of severe fibrosis on two-dimensional shear-wave elastography (means of 12.2 kPa, range 10.8–13.5, for F3–F4, and of 18.7 kPa, range 15.7–24.4, for F4 or cirrhosis, on the Aixplorer device), of severe associated malformations that greatly increase surgical risk, such as severe heart disease, and of liver failure with severe coagulopathy or haemorrhage [5]. The French PNDS sets the timing of KPE after umbilical healing, because of the infectious risk, generally from the 15th day and ideally between the 15th and 30th day in the term newborn. When cholestasis is identified late, assessment in a specialised centre, preferably with paediatric transplantation, determines the degree of portal hypertension and the feasibility of KPE, which may be decided during the operation itself; KPE performed after 3 months can still give prolonged native liver survival, so the decision not to perform it because of disease that is too advanced should be discussed at a multidisciplinary meeting with a paediatric transplant centre, which then proposes primary transplantation; the PNDS summary accepts this course when diagnosis is made after 3 to 4 months and there are signs of portal hypertension [38]. The PNDS does not use elastography as a criterion for excluding KPE, unlike the international guideline [5, 38]. In BA with heart disease, the King's authors recommend treating and prioritising the heart disease, because KPE rarely succeeds with poor cardiac function, and prior cardiac repair was associated with greater jaundice clearance [2, 3]. In Mitchell-Riley syndrome, early glycaemic control and measurement of faecal elastase are part of the preparation [3].

7.3. Kasai portoenterostomy

KPE consists of transection of the fibrous remnant at the porta hepatis and reconstruction with a retrocolic Roux-en-Y jejunal loop of 30 to 40 cm [2]. In the 2021 European survey (43 respondents, 22 surgeons, 18 countries), liver mobilisation was total in 33%, partial in 43% and absent in 24%; 90% used loupes, 78% scalpel dissection, 52% exposed the Rex recess, 67% performed extended dissection, 90% used a retrocolic loop of 25 to 50 cm (mostly about 40 cm), 57% performed an end-to-side anastomosis with 6/0 absorbable suture and surgery lasted 120 to 240 minutes in 76% of cases; only one respondent performed laparoscopic KPE [47]. The technical appendix of the French PNDS, which attributes the greater morbidity and mortality of transplantation after a KPE performed outside a transplant centre partly to haemorrhage during hepatectomy and to disregard of these rules, describes: exploration of the cavity (ascites, appearance of the liver, elements of polysplenia and other malformations, Meckel diverticulum, which is resected because of the risk of haemorrhage in portal hypertension) and measurement of portal pressure via the recanalised umbilical vein, which reflects fibrosis; exteriorisation of the liver by gentle upward tilting without division of the triangular and falciform ligaments, to avoid adhesions; dissection of the remnant between the left portal branch and the hepatic artery branches on the right, with transection flush with the parenchyma without entering it and without dissecting the lower part of the pedicle or the arteries; a 40 to 50 cm Roux-en-Y jejunal loop, transmesocolic and preduodenal, without torsion and fixed at the mesocolic passage; end-to-side anastomosis to the hilar plate with fine absorbable suture (for example 6/0); biopsy of the free edge of segment III; no drains, which create adhesions and collateral circulation pathways; and anti-adhesion membranes on the inferior surface of the liver and under the abdominal wall closure, as haemorrhage prevention for a future transplant. The technique is adapted to the French anatomy: in type 1, cholecysto-enterostomy or hepato-enterostomy; in type 2, cysto-jejunostomy if the cyst contains yellow bile and communicates widely with the intrahepatic tree, and resection with hepatoportoenterostomy otherwise; in type 3, hepatoportocholecystostomy or hepatoportoenterostomy, depending on the centre. Hepatoportocholecystostomy preserves the sphincter of Oddi and may reduce ascending cholangitis, but exposes to bile leak (10% to 20%), diagnosed by a fluid bilirubin at least twice the serum level or by scintigraphy and treated by conversion to hepatoportoenterostomy, and to late progression of lesions in the preserved extrahepatic duct, so the PNDS asks for ultrasound on the 10th day and before discharge (between the 21st and 30th day). Postoperatively, gastric aspiration is maintained until return of bowel function (at least 48 to 72 hours), with progressive refeeding over 2 to 3 days, and intraoperative antibiotic prophylaxis, dependent on local practice, generally combines piperacillin–tazobactam and an aminoglycoside. The PNDS does not recommend laparoscopic KPE, because there is no certainty of a success rate equivalent to that of laparotomy [38]. Laparoscopic KPE, widespread mainly in China, has a more limited dissection and jaundice clearance rates ranging from 41% to more than 80%; the only long-term study found 10-year native liver survival of 45% (5 of 11) with the laparoscopic approach against 85% (17 of 20) with the open approach (P=0.03) [2]. The international guideline considers that the two approaches have comparable results at 2 years and that the choice should follow the centre's experience (recommendation 10, grade 1B), whereas the 2nd edition of the Japanese guideline, of 2026, made no recommendation on laparoscopic surgery because of very weak evidence (agreement of 73% for the absence of a recommendation): no differences in outcomes or complications were observed, and the laparoscopic approach accounted for 13.8% (73 of 529) of KPEs registered in Japan since 2018 [5, 57]. The same guideline recommends performing KPE in the first 30 days of life (strong recommendation, evidence B, agreement of 77%, strengthened relative to the 1st edition), because of its association with greater jaundice clearance and greater native liver survival for at least 15 years [57]. Robotic KPE has been introduced, with a hospital stay of about 10 days, against 6 to 7 days for open surgery in the West [2, 5].

Several technical modifications have not shown consistent benefit. In a Chinese trial with 166 infants, the short loop (13 to 20 cm) and the standard loop had cholangitis in 47% and 43% and jaundice clearance in 50% and 45%, without differences [47]. Intussusception anti-reflux valves showed no benefit in a Tokyo trial with 20 infants and are used in half of Chinese centres; the 2024 review considers that they add no benefit, whereas the international guideline reports a reduction in cholangitis, to be interpreted with caution [2, 5, 47]. Portocholecystostomy, used mainly in France when the gallbladder and distal common bile duct are patent, almost eliminates cholangitis, but requires more revisions for leaks or obstruction [47]. Cystic BA is treated by radical resection of the cyst, with jaundice clearance in more than 80% [2]. In BASM, the position of the liver, the remnants, the preduodenal portal vein and arterial variations alter the dissection, malrotation may require modification of the Roux-en-Y loop, and in Mitchell-Riley syndrome surgery should restore continuity between duodenum, jejunum and bile duct [3]. A systematic review of 7 series with 195 patients with BASM found a response to KPE similar to that in isolated BA, with earlier surgery and mortality mainly cardiac, but heterogeneity precluded statistical conclusions [3].

A second KPE is only considered in patients who had good drainage and sudden cessation of flow, and only once; reoperation is also justified by uncontrollable haemorrhage, anastomotic fistula or intestinal obstruction [5]. The Japanese guideline proposes a repeat KPE in patients with temporary clearance of jaundice followed by its reappearance, or with good initial bile excretion that suddenly stops, provided indications are restricted and transplantation is not delayed (weak recommendation, evidence C, agreement of 91%) [57]. To reduce adhesions that hinder a future transplant, the European position statement advises avoiding unnecessary exteriorisation of the liver and drains; hyaluronic acid barriers and hydrogels have no evidence [47].

7.4. Liver transplantation

The international guideline indicates transplant assessment when total bilirubin remains above 100 µmol/L (about 5.8 mg/dL) 3 months after KPE (recommendation 16, grade 1B, agreement of 95%); with bilirubin between 34 and 100 µmol/L or normal, transplantation is assessed if cirrhosis or portal hypertension do not respond to conservative treatment [5]. The other indications are insufficient drainage with persistent jaundice, advanced cirrhosis despite good flow, late diagnosis with severe injury (primary transplantation), intractable nutritional deficiencies with growth failure and bone disease with fractures, recurrent, multidrug-resistant or septic cholangitis, complications of portal hypertension (refractory variceal haemorrhage, ascites, spontaneous bacterial peritonitis), symptomatic thrombocytopenia, severe pruritus, pulmonary vascular disease, hepatopulmonary and hepatorenal syndromes, and hepatocellular carcinoma or cholangiocarcinoma [5]. Up to 12 years, the optimal timing lies at a PELD score (Pediatric End-stage Liver Disease) between 15 and 25; a score of about 10 warrants counselling, between 10 and 15 the benefit is limited and above 25 the urgency is greater but so are perioperative complications; allocation uses PELD up to 12 years and MELD (Model for End-stage Liver Disease) from 12 years (recommendation 17, grade 1C) [5]. Adult indices are not appropriate for deciding on transplantation in survivors with BA [2]. The 2026 Japanese guideline abandoned the PELD score, which the 1st edition had considered inconclusive, and bases the decision on the presence of oesophageal varices and portal hypertension: it recommends transplantation in refractory cholangitis and in treatment-resistant portal hypertension (strong recommendation, evidence C, agreement of 76%), proposes it to correct growth failure in native liver survivors (weak, C, 85%) and proposes primary transplantation, in a limited way, when the disease is already advanced at diagnosis and KPE itself carries a risk of death (weak, D, 79%) [57]. The most severe form of portal hypertension is portopulmonary hypertension, which may progress without symptoms until sudden syncope and be detected too late for transplantation, which is why the guideline asks for a follow-up programme targeting this risk [57]. Pre-transplant assessment includes liver function, nutrition, development, cardiopulmonary and renal function, oral hygiene, anaesthetic risk, vaccination and viral screening (recommendation 18, grade 1C), and weight of 10 kg or less, portal vein thrombosis, mechanical ventilation and high bilirubin increase preoperative mortality [5].

In a cohort of 3438 patients on the transplant waiting list, 15% were transplanted without prior KPE, 17% for early KPE failure (before 1 year) and 67% for late failure (after 1 year) [2]. In the United Kingdom, fewer than 5% undergo primary transplantation, typically because of presentation after 100 days with cirrhosis, ascites or portal hypertension [2, 37]. In a Brazilian centre, between 2013 and 2023, 42 children were transplanted for BA (60% girls, 83% with a living donor): the 19 with prior KPE were older (12 against 8 months) and had a lower PELD (13.2 against 21.4) than the 23 without KPE, portal vein thrombosis was 39% against 11% without statistical significance, and post-transplant survival did not differ [93]. Transplantation does not correct the extrahepatic manifestations of the syndromic forms, and in BASM the small numbers suggest possible differences in post-transplant sepsis [3, 36]. In France, BA accounts for about one third of paediatric liver transplant indications, the leading cause. With immediate KPE failure, transplantation usually takes place before 3 years, sometimes from 6 months in rapidly progressive cirrhosis; after initial success it may be needed later in childhood or adulthood, because of recurrence of jaundice or complications of cirrhosis and portal hypertension. Survival 5, 10 and up to 20 years after transplantation for BA exceeds 80%, and growth, development and fertility of transplant recipients approach normal, although the frequent hospital admissions of the first years may leave anxiety or attention disorders in the child and family. The French list of indications for assessment at a transplant centre essentially coincides with that of the international guideline (primary or secondary KPE failure, late diagnosis, recurrent cholangitis, with multidrug-resistant organisms or severe sepsis, complications of cirrhosis and portal hypertension, hepatopulmonary and hepatorenal syndromes, refractory pruritus) and explicitly adds symptomatic thrombocytopenia, chronic hepatic encephalopathy, ascites with fluid infection, a nodule suspicious for dysplasia or carcinoma and malnutrition requiring invasive nutritional support or fractures [38].

7.5. Adjuvant therapy after portoenterostomy

The evidence for the drugs used after KPE is limited. The international guideline summarises it as follows: corticosteroids and ursodeoxycholic acid improve jaundice clearance without clear benefit in native liver survival; antibiotics have no demonstrated benefit in preventing cholangitis or in survival; and antivirals in CMV-associated BA improve both jaundice clearance and native liver survival [5]. In European practice in 2021, 82% of centres gave prophylactic antibiotics for more than 1 week, 94% ursodeoxycholic acid and 50% corticosteroids, and none used intravenous immunoglobulin (IVIG), farnesoid X receptor agonists or IBAT inhibitors; worldwide, corticosteroid use ranges from 0% to 100% of centres [47]. The 2024 French PNDS states that no adjuvant treatment currently improves surgical outcomes: it considers corticosteroids controversial, because of undemonstrated long-term benefit and a possible increase in severe postoperative cholangitis; IVIG ineffective; oral antibiotic prophylaxis after KPE supported by few studies, with contradictory results; probiotics yet to be assessed; anti-CMV antivirals, used by some teams, without demonstrated long-term benefit; and obeticholic acid, Chinese herbs, phenobarbital and colestyramine without proven efficacy. At the time the PNDS was written, EMBARK had already failed to meet its primary endpoint and BOLD was under way [38].

Corticosteroids

Corticosteroids inhibit inflammation, promote enterohepatic recycling and reduce bile acid synthesis, but do not reduce cholangitis and carry a risk of infection, gastrointestinal injury and delayed growth and development, which with high doses (methylprednisolone above 4 mg/kg/day) may persist up to 6 years [5]. The START trial (Steroids in Biliary Atresia Randomized Trial), multicentre, double-blind and placebo-controlled, randomised 140 North American infants with a mean age of 2.3 months, between 2005 and 2011, and started treatment within 72 hours after KPE [44]. The active regimen lasted 13 weeks: intravenous methylprednisolone 4 mg/kg/day in 2 doses on days 1 to 3; oral prednisolone 4 mg/kg/day in 2 doses on days 4 to 7 and in week 2; 2 mg/kg/day in weeks 3 and 4; 1 mg/kg/day in weeks 5 and 6; 0.8; 0.6; 0.4; 0.2 and 0.1 mg/kg/day in weeks 7 to 11; 0.1 mg/kg on alternate days in weeks 12 and 13; discontinuation in week 14; the intravenous route was maintained for at least 2 days or until enteral feeding resumed, and the dose was adjusted monthly to weight [44]. Total bilirubin fell below 1.5 mg/dL with native liver at 6 months in 58.6% (41/70) of treated infants and in 48.6% (34/70) of the placebo group (adjusted relative risk 1.14; 95% CI 0.83–1.57; P=0.43), and transplant-free survival at 24 months was 58.7% against 59.4% (adjusted HR 1.0; 95% CI 0.6–1.8) [44]. Serious adverse events were similar overall (81.4% against 80.0%), but appeared earlier with corticosteroids (37.2% against 19.0% in the first 30 days, P=0.008), and a later analysis found impaired growth for at least 6 months [2, 44]. The trial was powered for an absolute difference of 25%, and in the subgroup operated on before 70 days the difference was 71.8% against 56.8%, without statistical significance, so a small benefit was not excluded [2, 44].

Other studies used different doses and populations. A British trial with 71 patients, using oral prednisolone 2 mg/kg/day for 4 weeks, showed a fall in bilirubin at 3 months (P=0.06), greater in the youngest [2]. A prospective King's study of 153 infants with isolated BA, negative anti-CMV IgM and KPE before 70 days compared low-dose prednisolone (2 mg/kg/day, n=18), high-dose (5 mg/kg/day, n=44) and no corticosteroid (n=91): jaundice clearance (bilirubin below 20 µmol/L) at 6 months was 67% and 66% against 52% (P=0.037), with no difference in 4-year survival (96%, 94% and 95%) or in 4-year native liver survival (50%, 57% and 46%) [94]. A Chinese study with 200 patients and prednisolone 4 mg/kg/day for 10 to 12 weeks found jaundice clearance at 6 months of 54% against 31% and native liver survival at 12 months of 66% against 50% [2]. Many centres use prednisolone above 4 mg/kg/day tapered over 4 to 6 weeks, and a systematic analysis suggests a gain of 10% to 15% in jaundice clearance and in native liver survival at 2 years; in Japan it is common to repeat the course in relapses [2, 37].

The sources diverge. The international guideline recommends postoperative corticosteroids because they improve jaundice clearance, with an unknown effect on native liver survival (recommendation 11, grade 2B), and acknowledges that the European meta-analyses and several trials show no benefit and that there is no consensus on duration [5]. The 1st edition of the Japanese guideline made no recommendation on corticosteroids because of scarce evidence, and the START authors concluded that the high dose cannot be recommended [44, 79]. The 2nd Japanese edition (2026) now recommends them after KPE (strong recommendation, evidence B, agreement of 73%), considering the adverse effects lesser and because improvements in jaundice clearance and native liver survival have been described; the English publication, which carries only the clinical questions, does not state the drug, dose or duration [57]. FISPGHAN, by contrast, concludes that attempts to curb the inflammatory response with corticosteroids, intravenous immunoglobulin or rituximab have not succeeded, and that individualised use of corticosteroids and antibiotics after KPE still requires validation in larger cohorts [73]. The decision therefore rests with the centre, weighing the modest benefit in jaundice clearance against the early adverse events and the effect on growth, with attention to the rules for vaccination under corticosteroids (section 7.8).

Ursodeoxycholic acid

Ursodeoxycholic acid (UDCA) is a hydrophilic bile acid that reduces ileal reabsorption of the more hydrophobic endogenous bile salts and has a hepatoprotective effect in vitro [95]. Despite almost universal use, the evidence is scarce: only one study, of withdrawal and reintroduction in 16 stable children, showed biochemical improvement, and the international guideline considers that it improves biochemical indices (recommendation 12, grade 2D) without setting a dose, noting that doses and discontinuation criteria vary between centres [2, 5]. The 2024 French PNDS is the only guideline consulted that sets a dose for BA: 600 mg/m² body surface area per day, ideally in 2 doses, with regular adjustment to weight and no maximum dose stated, using 250 or 500 mg tablets, 200 mg capsules or, in infants, preferably the 250 mg/5 mL suspension (Ursofalk®), available in France through compassionate access (formerly a cohort temporary authorisation for use); it acknowledges that the drug has not proven efficacy in BA, but records a consensus to use it when bile flow is restored, postoperatively and in the long term even with partial success, because of its choleretic, immunomodulatory and hepatoprotective action, and considers its use more controversial when KPE fails [38]. This is the reference dose. For comparison, the British summary of product characteristics (SmPC) of Ursofalk® 250 mg/5 mL oral suspension, whose only paediatric indication is hepatobiliary disease of cystic fibrosis, from 1 month to 18 years, gives 20 mg/kg/day in 2 to 3 doses, up to 30 mg/kg/day if necessary (up to 10 kg, 20 mg/kg/day; above 10 kg, 20 to 25 mg/kg/day) [95]. Converted to weight using the Mosteller formula, with usual lengths for each weight, the PNDS dose corresponds to about 32 to 37 mg/kg/day in a 4 to 6 kg infant, above the SmPC maximum, and to about 27 mg/kg/day at 10 kg, because body surface area per kilogram decreases with growth. Use in BA is off-label in all the labels consulted. The same SmPC contraindicates the drug in BA with unsuccessful portoenterostomy or without recovery of good bile flow, as well as in bile duct occlusion and acute inflammation of the gallbladder or bile ducts [95]. The suspension contains sodium benzoate (7.5 mg/5 mL), which may worsen neonatal jaundice, and propylene glycol (50 mg/5 mL) [95]. Frequent adverse effects are soft stools or diarrhoea; nausea, vomiting, urticaria and pruritus have been reported with unknown frequency, and decompensation of cirrhosis has been observed, very rarely, in advanced stages of primary biliary cholangitis, with partial regression after discontinuation; the SmPC recommends monitoring AST, ALT and GGT every 4 weeks for the first 3 months and then every 3 months [95]. According to the SmPC, diarrhoea requires dose reduction and, if it persists, discontinuation of treatment; the contraindication in BA also applies when bile flow is lost after an initial response [95]. The 2nd edition of the Japanese guideline proposes ursodeoxycholic acid after KPE (weak recommendation, evidence C, agreement of 88%), because improvement in native liver survival without major adverse effects has been suggested, also without a dose [57]. In the United States, ursodeoxycholic acid is FDA-approved, but its use in paediatric cholestasis is off-label [36]. In Portugal, Infomed lists Ursofalk® as 250 mg capsules and 500 mg film-coated tablets (SmPCs revised in November 2024); only the capsules have a paediatric indication, limited to hepatobiliary disease associated with cystic fibrosis between 6 and 18 years, with the same dose of 20 mg/kg/day in 2 to 3 doses, increasable to 30 mg/kg/day, and a table of capsules by weight from 20 kg; a search for marketed medicines in Infomed, on 29 October 2026, returned no oral liquid form [95, 96]. The Portuguese SmPC repeats the paediatric contraindication of unsuccessful portoenterostomy or without recovery of good bile flow in BA, bile duct occlusion and acute inflammation of the gallbladder or bile ducts, monitoring of AST, ALT and GGT every 4 weeks for the first 3 months and then every 3 months, and dose reduction or discontinuation in the event of diarrhoea [96]. The practical difference from the British SmPC is age: the latter covers cystic fibrosis from 1 month, with an oral suspension, and the Portuguese one only from 6 years [95, 96]. The text of the Brazilian package leaflet was not obtained.

Antivirals in cytomegalovirus-associated biliary atresia

At King's College Hospital, from 2011, intravenous ganciclovir 5 mg/kg twice daily and/or oral valganciclovir 520 mg/m² twice daily, started in the early postoperative period in selected cases and continued until CMV DNA became negative, were associated with jaundice clearance in 75% against 21% (P=0.009) and 2-year native liver survival of 75% against 25% (P=0.04), with no difference in overall survival, in a non-randomised comparison of 8 treated patients with 28 controls, with level of evidence III [65]. The international guideline considers that the antiviral improves native liver survival when anti-CMV IgM, urine PCR or blood PCR are positive (recommendation 14, grade 2C, agreement of 79%), and some European centres stop it when CMV DNA falls below 250 IU/mL, a criterion yet to be validated [5]. The 2024 review prefers oral valganciclovir, and the European position statement supports testing for CMV at the time of KPE and considering an antiviral [2, 47]. This use is off-label: the US label of valganciclovir (Valcyte®) authorises it in paediatrics only for prevention of CMV disease in kidney (4 months to 16 years) and heart (1 month to 16 years) transplantation, with a single daily dose, taken with food, calculated as 7 × body surface area × creatinine clearance by the modified Schwartz formula, with clearance capped at 150 mL/min/1.73 m²; the King's dose of 520 mg/m² twice daily does not follow this formula [65, 97]. The sources consulted do not describe the adverse effects observed in the King's series; the label carries a boxed warning about severe haematological toxicity (leucopenia, neutropenia, anaemia, thrombocytopenia, pancytopenia and bone marrow failure), impaired fertility, foetal toxicity and mutagenic and carcinogenic potential in animal data, which requires regular blood counts during treatment [97].

Intravenous immunoglobulin

The open-label phase I/IIa PRIME trial gave IVIG 1 g/kg per dose at 3 to 5, 30 and 60 days after KPE to 29 infants; it was feasible, but did not increase the proportion with bilirubin below 1.5 mg/dL at 90, 180 or 360 days compared with the START placebo arm, and native liver survival at 360 days was 58.6% against 70.5%; 90% had a serious adverse event, none directly attributed to IVIG [98]. IVIG has no place as an adjuvant to KPE [2, 98].

Prophylaxis and treatment of cholangitis

The meta-analysis of retrospective and cross-sectional studies does not show that antibiotic prophylaxis reduces cholangitis, and there is no evidence in favour of any regimen or duration; nevertheless, the international guideline recommends it in the absence of better evidence (recommendation 13, grade 1C, agreement of 84%), with the typical regimen of an intravenous third-generation cephalosporin, with or without a nitroimidazole, for at least 2 weeks, followed by an oral third-generation cephalosporin alternating with a sulfonamide every 2 weeks for 3 to 6 months [2, 5]. A randomised trial at Fudan University (NCT05925309, 356 patients) is studying oral prophylaxis [99]. The Japanese guideline addresses the question in two stages: to prevent early cholangitis, intravenous antibiotics for about 2 weeks after KPE followed by prolonged oral antibiotics (weak recommendation, evidence C, agreement of 94%), and longer-term antimicrobial prophylaxis (weak, C, 95%), without setting the duration or the drug of the oral phase [57]. Cholangitis is treated aggressively (recommendation 20, grade 1A): broad-spectrum intravenous antibiotics for 1 to 2 weeks, starting with a third-generation cephalosporin with or without metronidazole, switching to a carbapenem if fever persists for more than 24 hours and adjusting according to blood culture if there is no improvement within 48 to 72 hours; in refractory cholangitis, a carbapenem for 2 to 4 weeks; in frequent cholangitis, enterococcal cover with vancomycin or linezolid; with high fever and negative blood cultures, consider fungi [5, 37]. Cultures are rarely positive and isolate Enterococcus faecium, Escherichia coli, Enterobacter cloacae or Klebsiella pneumoniae, and in an open-label Chinese trial with 211 children meropenem was equivalent to cefoperazone in mild to moderate cholangitis, whereas in severe cholangitis adding IVIG to meropenem shortened the fever [2]. In late cholangitis, the Japanese guideline proposes corticosteroids or other choleretics according to symptoms and advises against fasting, which has no evidence (weak recommendation in favour of the former and weak against the latter, evidence D, agreement of 80%), and proposes biliary drainage in the presence of dilatation of the intrahepatic ducts or intrahepatic biliary cysts after the operation (weak, D, 88%) [57]. Recurrent cholangitis requires a search for bile lakes by ultrasound and magnetic resonance imaging, and new-onset cholangitis in an older child who has long been anicteric suggests partial obstruction of the Roux loop, with possible surgical revision; without a surgical solution, a long-term venous catheter and cyclical intravenous antibiotic therapy are used [5, 37].

The French PNDS defines cholangitis as the combination of signs of sepsis (fever, hypothermia, haemodynamic instability, deterioration of general condition or consciousness), worsening jaundice with discoloured stools and abdominal pain, especially on palpation of the liver, and confirms it by worsening cholestasis with an inflammatory syndrome, by positive blood cultures in 30% to 50% (taken, if possible, before antibiotics) and, on ultrasound, by increased signs of portal hypertension or ascites, which should be tapped for culture; cholangitis is suspected in almost three quarters of operated patients, but documented bacteriologically in only one third of suspected cases, and is more frequent in the first weeks or months after hepatoportoenterostomy or cysto-jejunostomy. Without an identified organism, puncture of a biliary cavity or a liver biopsy with culture may be discussed case by case, and next-generation microbiological sequencing is not yet routinely available. Urgent treatment combines resuscitation and broad-spectrum intravenous antibiotic therapy, first line piperacillin–tazobactam with or without amikacin, then adapted to cultures, to any previous cholangitis and to local ecology, ideally for 21 days in total, with amikacin stopped on day 3 and at least 10 to 15 days intravenously; an oral switch to ciprofloxacin may be discussed case by case after at least 7 to 10 days intravenously with the infection controlled. When in doubt, the antibiotic is started and stopped at 48 to 72 hours if clinical findings, biochemistry and blood cultures do not support it; relapsing forms or latent chronic cholangitis may exceptionally justify prolonged prophylaxis for months, and recurrent or late cholangitis requires a search for a surgical cause (short or antiperistaltic Roux-en-Y loop, loop torsion, band, intrahepatic biliary cavity) [38]. The PNDS sets a dose only for piperacillin–tazobactam, on the emergency card (80 to 100 mg/kg every 8 hours), which coincides with the US label: 80 mg/kg of piperacillin (90 mg/kg of the combination) every 8 hours from 2 to 9 months and 100 mg/kg (112.5 mg/kg of the combination) from 9 months, up to 40 kg, for the indications of appendicitis and peritonitis [38, 100]. For amikacin, the label gives 15 mg/kg/day in 2 or 3 doses in children and older infants and, in the newborn, a loading dose of 10 mg/kg followed by 7.5 mg/kg every 12 hours, not exceeding 15 mg/kg/day, with peaks above 35 µg/mL and troughs above 10 µg/mL to be avoided [101]. For oral ciprofloxacin, the labelled paediatric dose for complicated urinary tract infection, from 1 to 17 years, is 10 to 20 mg/kg every 12 hours, up to 750 mg per dose; use in cholangitis is an off-label extrapolation, and the PNDS recalls that in France ciprofloxacin has no paediatric authorisation in this situation [38, 102]. The PNDS diverges from the international guideline, which starts with a third-generation cephalosporin and escalates to a carbapenem, for 1 to 2 weeks [5, 38].

Nutrition and fat-soluble vitamins

Energy requirements are increased and fluid tolerance limited, so formulas with 80 to 100 kcal/100 mL are used (fortified breast milk, high-energy formulas, carbohydrates or medium-chain triglycerides, MCT); without breast milk, the postoperative formula is hydrolysed and rich in MCT, and parenteral nutrition is reserved for intolerance or failure of tube feeding (recommendation 15, grade 2B for MCT and 1B for vitamins) [5]. A Tongji trial (NCT06764082, 120 patients) is studying a high-MCT formula [99].

Table source: international guideline 2025, Table 3 [5]. TPGS: d-α-tocopheryl polyethylene glycol succinate; PIVKA-II: protein induced by vitamin K absence.

The French PNDS (French National Diagnosis and Care Protocol) follows a different logic for fat-soluble vitamins, with periodic intramuscular administration while jaundice persists. After the 10 mg of vitamin K given at diagnosis, if the child has persistent chronic cholestatic jaundice with total bilirubin above 60 µmol/L, it proposes vitamin D 200,000 IU intramuscularly every 3 months, vitamin A 50,000 IU intramuscularly monthly, vitamin K as a micellar solution (10 mg/mL) according to one of several regimens and oral vitamin E as tocofersolan (Vedrop®, 50 mg/mL, equivalent to 75 IU/mL) 0.35 mL/kg/day, without a stated maximum and adjusted by the ratio of serum vitamin E to the sum of cholesterol and triglycerides; with total bilirubin below 60 µmol/L, it proposes an oral combination of vitamins A, D, E and C (Uvestérol ADEC®, in France under a temporary use restriction) 1 to 2 mL/day and oral vitamin K 20 mg every 15 days. In France there are also an oral multivitamin supplement (DEKAs Essential®) and a vitamin E supplement (DEKAs AquaE®), classified as foods for special medical purposes and not as medicines, whose use in this indication is under evaluation. Whatever the regimen, vitamin A, D and E concentrations and the prothrombin time are checked regularly and supplementation is adapted to each child [38]. The European SmPC of Vedrop® indicates it for vitamin E deficiency due to digestive malabsorption in congenital or hereditary chronic cholestasis, from the term newborn to 18 years, an indication that covers BA, at a dose of 0.34 mL/kg/day (17 mg/kg of d-alpha-tocopherol), without a fixed maximum dose and adjusted by the plasma vitamin E level, measured monthly in the first months; it contraindicates it in preterm infants and asks for monitoring of the prothrombin time and INR, because high doses of vitamin E increase the bleeding tendency in vitamin K deficiency, and of renal function in dehydrated children or those with renal failure [103]. The PNDS dose coincides in practice with that of the label and, expressed in units (about 26 IU/kg/day), lies at the upper limit of the international guideline range (15 to 25 IU/kg/day) [5, 38].

Table sources: [5, 38]. IM: intramuscular; IV: intravenous.

The PNDS gives nutrition a central role, because cholestasis reduces the absorption of long-chain triglycerides, essential fatty acids and fat-soluble vitamins, cirrhosis creates hypercatabolism and anorexia due to gastric fullness, malnutrition accelerates hepatic deterioration and nutritional status conditions transplant morbidity and mortality. It recommends a specialised dietitian; the monthly mid-upper arm circumference/head circumference ratio until 3 to 4 years, because weight may be falsely increased by organomegaly, ascites or oedema; and a high-energy diet started as soon as necessary (KPE failure, slowing of growth, worsening cholestasis). Breastfeeding may be maintained for a few weeks, enriched with MCT and maltodextrin, but the priority is energy intake; in the other situations an extensively hydrolysed protein formula rich in MCT (Pregestimil®, Alfaré®) is used, reconstituted at 15%–17% instead of the usual 13.5%, enriched with maltodextrin (4%–6%) and, by tube, with MCT (2%–4%); whole-protein formulas very rich in MCT (Lipistart®, Monogen®) are second line orally, because they contain more sodium and are less well tolerated in enteral nutrition. The diet is introduced in steps of 24 to 48 hours (3 to 4 days if tolerance is doubtful). If intake or weight gain is insufficient, constant-rate overnight enteral nutrition by nasogastric tube is started early, for 8 to 12 hours, with 60 to 100 mL/kg (sometimes 50 mL/kg is enough), about half of daily energy and a maximum total volume of 200 mL/kg/24 h in infants, later switching to ready-to-use products rich in MCT (Infatrini Peptisorb® before 1 year; Peptamen Junior® or Nutrini Peptisorb® thereafter; 1 to 1.5 kcal/mL); if daytime intake falls, 1 or 2 daytime boluses are added. Weaning begins when the child accepts the spoon. Most teams avoid gastrostomy, because of the risk of peristomal varices, and parenteral nutrition is reserved for failure or intolerance of the enteral route (ascites, vomiting). In KPE failure, nutritional follow-up is closer and the nasogastric tube is placed early. Oral feeding disorders are frequent, especially with tube feeding, and require speech therapy and psychomotor therapy, while maintaining daytime pleasure feeding. Bone mineralisation is monitored through calcium–phosphate balance, with adequate vitamin D supplementation, because of osteopenia and the risk of fractures from the first months [38]. The PNDS algorithm, adapted from the 2019 joint NASPGHAN/ESPGHAN position paper, escalates in steps of 2 to 4 weeks, with growth checked every 2 to 4 weeks: standard feeding with a target of at least 130% of requirements; enrichment (at least 150% of requirements in infants, with a formula with about 30% MCT; larger portions, a night meal or a 1 to 1.5 kcal/mL supplement in children); admission for enteral nutrition by nasogastric or nasoduodenal tube, overnight for 10 to 12 hours or continuous over 24 hours if there are hypoglycaemic episodes; and, finally, parenteral nutrition [38].

In portal hypertension, protein-losing enteropathy (intestinal loss of albumin, lipids, lymphocytes and immunoglobulins, with diarrhoea, abdominal pain and chronic hypoalbuminaemia) is treated with a fat- and salt-controlled diet: foods low in long-chain triglycerides, fats replaced by MCT oil or by the Liquigen® emulsion in infants, removal of salt-rich foods and fat-free cooking. Chronic hepatic encephalopathy, due to portosystemic shunting with hepatocellular failure and hyperammonaemia, warrants a protein-controlled diet (1 to 1.5 g/kg/day, adjusted to ammonia and intake), divided equally between meals, possibly with ammonia-lowering treatment [38].

Pruritus and drugs acting on bile acids

For the pruritus of cholestasis, the GeneReviews overview lists colestyramine, rifampicin (with monitoring for drug-induced hepatitis), naloxone, hydroxyzine and sertraline, without specific doses for BA [36]. The IBAT inhibitors odevixibat and maralixibat are effective for pruritus in PFIC and Alagille syndrome, but failed their trials in BA (section 7.7) and are not authorised for this disease [37, 104, 105]. For pruritus, the French PNDS recommends rifampicin at a starting dose of 5 mg/kg/day, increased in steps up to a maximum of 20 mg/kg/day, and reserves for refractory pruritus, exceptional in BA, other medical or surgical treatments discussed case by case [38]. The US rifampicin label has indications only for tuberculosis and meningococcal carriage, with a paediatric dose for tuberculosis of 10 to 20 mg/kg/day, not exceeding 600 mg/day; use for pruritus is off-label, and the PNDS maximum coincides with the upper limit of the label, which adds the absolute daily ceiling [106]. In a non-randomised prospective Japanese study with 57 infants, eicosapentaenoic acid (n=25) did not alter jaundice clearance or 2-year native liver survival, although platelets and the APRI index (AST-to-platelet ratio) were lower [37].

7.6. Specific management of syndromic forms

In BASM and heterotaxy, asplenia or functional hyposplenism should be presumed until assessed, with daily antibiotic prophylaxis: phenoxymethylpenicillin (penicillin V) 125 mg twice daily before 3 years and 250 mg twice daily from 3 years, or amoxicillin 20 mg/kg/day in 2 doses (without a maximum dose defined in the source); in penicillin allergy, erythromycin 125 mg (before 3 years) or 250 mg (from 3 years) twice daily [41]. Prophylaxis reduces invasive pneumococcal infection by 50% to 80% and may be stopped at 5 years if vaccination is complete, if there is no history of serious infection and if there is rapid access to care, a criterion extrapolated from the recommendations for sickle cell disease [41]. The French PNDS considers prophylaxis with penicillin V (Oracilline® in France) indispensable in BA with polysplenia syndrome, together with early vaccination against pneumococci and meningococci, and accepts that in older children with complete pneumococcal vaccination it may be replaced by amoxicillin in the event of fever, with urgent medical consultation; it does not set doses or an age for stopping, which follow the 2022 consensus [38, 41]. For curative amoxicillin, the US label gives, from 3 months, 20 to 45 mg/kg/day in doses every 8 or every 12 hours, depending on the infection, and up to 3 months a maximum of 30 mg/kg/day every 12 hours [107]. The pulmonary arteriovenous shunts of hepatopulmonary syndrome are more frequent, and sometimes early, when there is polysplenia syndrome [38]. Vaccination includes the routine schedule, the 23-valent pneumococcal polysaccharide vaccine at 2 years with a booster 5 years later and the conjugate meningococcal vaccine at 11 to 12 years with a booster at 16, in the US schedule, and any fever above 38.5 °C requires immediate assessment for sepsis and empirical antibiotic therapy [41]. Radiological screening for malrotation in asymptomatic children with heterotaxy is not recommended; volvulus requires an urgent Ladd procedure, whose mortality in heterotaxy is 13% to 22%, and an elective Ladd procedure, when performed, should await the cavopulmonary connection or complete cardiac repair [41]. In a prospective cohort, 7 of 38 children underwent a Ladd procedure (4 prophylactic and 3 urgent) and none of those managed expectantly had volvulus over a median of 1.6 years [11]. In BASM, correction of malrotation may be performed at the same time as KPE, as in the case described in section 10 [108]. Suspected primary ciliary dyskinesia in the presence of chronic lung disease warrants the investigation described in section 5.4 [9, 41].

In Mitchell-Riley syndrome, neonatal diabetes is treated with insulin, with long-acting analogues or continuous subcutaneous infusion in small infants, without the possibility of transition to a sulfonylurea, and exocrine pancreatic insufficiency with pancreatic enzymes [12, 71]. In cat eye syndrome and CABA, the order of surgery follows the severity of the heart disease, as described in section 7.2 [3].

7.7. Clinical trials, regulatory status and developments

Table sources: [2, 44–46, 65, 94, 98, 109–111].

These studies cannot be compared with one another: they differ in age at KPE, in the proportion of syndromic and CMV-associated forms, in the criteria for jaundice clearance (bilirubin below 1.5 mg/dL, 20 µmol/L or 2 mg/dL) and in design, and only START, EMBARK and BOLD were randomised and placebo-controlled [44, 45, 109]. The relative gains in observational studies, such as the King's study with corticosteroids or with antivirals, must be read together with the absolute rates and with the risk of selection bias, and START itself did not exclude an absolute benefit of about 10% [44, 65, 94]. The negative trials of IBAT inhibitors and the closure of the obeticholic acid trials show that, to date, no pharmacological therapy has demonstrably altered native liver survival in isolated BA [45, 46, 111].

Table sources: [36, 38, 95–97, 103–105, 111–113].

Developments in the last 12 months have different levels of evidence. In terms of trials, BOLD, with 254 patients randomised in 19 countries, ended in June 2026, and the sponsor (Ipsen) announced in a press release of 24 July 2026 that odevixibat did not improve native liver survival at week 104, with a safety profile similar to that of the approved indications; subgroup analyses and the decision on the open-label extension (BOLD-EXT, NCT05426733) await full review of the data, which had not been published at the time of this update [46, 109, 110]. The phase 2/3 trial of obeticholic acid ended in October 2025 by sponsor decision, without hepatic events in either arm [111]. In terms of guidelines, the 2025 international guideline (Ge et al., 2026) and the 2nd edition of the Japanese guideline (2026) were published, the latter with 23 clinical questions, recommendations for 22 and none for laparoscopic surgery; the FISPGHAN position statement (2025) and the Spanish consensus (2025) address the diagnosis of infant cholestasis [5, 57, 73, 78]. The APSPGHAN (Asia Pacific Society of Pediatric Gastroenterology, Hepatology and Nutrition) statement on endoscopic haemostasis of varices, published in January 2026, addresses screening endoscopy in BA and the dose of octreotide; the French PNDS (French National Diagnosis and Care Protocol), of November 2024, falls outside the 12-month window, but is, among the guidelines consulted, the most recent European national guideline dedicated to BA [38, 114]. In terms of research, the Taiwanese GWAS (2026) proposed new loci, and the demonstration in organoids and mice of the role of ADD3 (2025) is preclinical [55, 56]. Trials are registered of enteral N-acetylcysteine (NCT06260566, phase 1, 12 patients, not yet recruiting in October 2026), umbilical cord mesenchymal stem cells (NCT07011199 and NCT07797673, phase 1), granulocyte colony-stimulating factor (G-CSF) after KPE (NCT06708572), indocyanine green for diagnosis (NCT07513038 and NCT07250854) and maralixibat for pruritus in rare cholestatic diseases (EXPAND, NCT06553768) [99]. A Hannover group proposed rectal budesonide after KPE in 2025 [115]. N-acetylcysteine reduced injury in the rhesus rotavirus model and, in a small pilot study, was associated with better bile flow; autologous bone marrow mononuclear cells (New Delhi, Hanoi) improved biochemical parameters without altering fibrosis, and G-CSF appeared safe and associated with less cholangitis [2, 37].

7.8. Vaccination

The international guideline recommends the routine vaccination schedule without specific contraindications (recommendation 22, grade 1C), postponing it if there is severe coagulopathy or progression to liver failure, with live attenuated vaccines 1 year after surgery and complete vaccination before transplantation [5]. With high-dose corticosteroids (prednisone 20 mg/day or more, or above 2 mg/kg/day below 10 kg) for 14 days or more, inactivated vaccines are given at least 2 weeks before or after stopping, and live vaccines at least 4 weeks before or 4 weeks after; with a high dose for less than 14 days, inactivated vaccines are continued and live vaccines are given 2 weeks after stopping; before KPE, inactivated vaccines are given at least 2 days before and live vaccines at least 21 days before; before transplantation, inactivated vaccines at least 2 weeks before and live vaccines at least 4 weeks before [5]. The French PNDS recommends all vaccines in the schedule, with attention to those against hepatitis B and hepatitis A, pneumococcal, meningococcal and human papillomavirus vaccines, varicella vaccine, annual influenza vaccine in children with biliary cirrhosis, especially those on the transplant waiting list, and live attenuated vaccines (measles–mumps–rubella and varicella) before transplantation, in agreement with the transplant team, because they are classically contraindicated afterwards [38]. In BASM with asplenia, the pneumococcal and meningococcal vaccines of section 7.6 are added [41].

7.9. Agents and situations to avoid, perioperative care, pregnancy and transition

Ursodeoxycholic acid is contraindicated when KPE has failed or has not restored good bile flow, and the oral suspension contains benzoate, which may worsen neonatal jaundice [95]. Excess iron is hepatotoxic, and supplementation should follow reference daily intakes [5]. Rifampicin used for pruritus requires monitoring for drug-induced hepatitis [36]. High-dose corticosteroids constrain the schedule of live vaccines (section 7.8) [5]. Adult liver disease indices should not decide transplantation in these patients [2]. Perioperatively, intravenous vitamin K before any surgery, prior correction of significant heart disease and, in heterotaxy, postponement of an elective Ladd procedure until cardiac repair are the main rules [3, 5, 41]. In portal hypertension, the French PNDS contraindicates as a matter of principle aspirin and all non-steroidal anti-inflammatory drugs, a list of which should be given to the parents; advises against gastrostomy, because of the risk of peristomal varices; advises against large-volume paracentesis except for significant dyspnoea, because of the risk of volaemic changes and ischaemic hepatic necrosis; considers long-term beta-blockers without demonstrated efficacy in secondary prevention of variceal haemorrhage in children with BA and long-term proton pump inhibitors without indication in the prevention of variceal rupture; and requires early correction of any hypovolaemia, which favours ischaemic necrosis [38].

In France, after diagnosis, the doctor at the reference or competence centre provides and fills in an «Atrésie des voies biliaires» emergency card, which flags the risk of bacterial cholangitis and the contraindications to aspirin and non-steroidal anti-inflammatory drugs, records the type of KPE (hepatoportoenterostomy, hepatoportocholecystostomy or other), transplant listing, portal hypertension, oesophageal varices and hypersplenism and the team's contact details, and states that unexplained fever above 38.5 °C with clinical or biochemical worsening of cholestasis should be treated as cholangitis, with admission and empirical antibiotic therapy, for example piperacillin–tazobactam 80 to 100 mg/kg every 8 hours [38]. Haemorrhage from ruptured varices is an emergency requiring admission to a centre with endoscopy and intensive care: the source (nasogastric tube and lavage to demonstrate active upper gastrointestinal haemorrhage), volume and tolerance are assessed; resuscitation is instituted (continuous monitoring, peripheral and central venous access, volume expansion and red cell concentrate transfusion if necessary); the cause is treated with cold gastric lavage (with Lévy's potion, which combines an aluminium phosphate gel and tranexamic acid, or with saline at 4 °C), intravenous proton pump inhibitor and octreotide and, if necessary, a Blakemore tube; endoscopy is performed as soon as haemodynamics stabilise; and, in cirrhosis, broad-spectrum intravenous antibiotic therapy for at least 48 hours. These are followed by prevention of encephalopathy in hepatocellular failure (oral lactulose, protein restriction), paracentesis of ascites to exclude infection, albumin to prevent ascites from haemorrhage-induced hypoalbuminaemia and prevention of hepatorenal syndrome (renal monitoring, correction of hypovolaemia, albumin, avoidance of nephrotoxic drugs and discontinuation of diuretics) [38]. The PNDS does not give doses for these drugs. The 2026 APSPGHAN statement proposes intravenous octreotide as a bolus of about 1 µg/kg (maximum 50 µg) followed by an infusion of 1 to 2 µg/kg/h (maximum 50 µg/h), maintained for 2 to 5 days after endoscopy, and accepts that proton pump inhibitors speed ulcer healing after endoscopic treatment, without specific paediatric studies [114]. As a label comparison, intravenous esomeprazole is authorised in the United States, from 1 month to 17 years, only for erosive oesophagitis, with 0.5 mg/kg/day under 1 year, 10 mg/day from 1 to 17 years below 55 kg and 20 mg/day from 55 kg, for up to 10 days, so use in variceal haemorrhage is extrapolated; the lactulose label, with scant paediatric information, gives for portosystemic encephalopathy 2.5 to 10 mL/day in divided doses in infants and 40 to 90 mL/day in older children and adolescents, with the aim of 2 to 3 soft stools daily [116, 117].

The 2026 Japanese guideline recommends multidisciplinary management of pregnancy and delivery in native liver survivors, to anticipate deterioration in general condition or liver function and diagnose complications early (strong recommendation, evidence C, agreement of 76%); the sources consulted provide no quantitative data on the course of pregnancy in these women or in transplant recipients [57]; the 2024 British guideline on the transition of paediatric liver diseases to adulthood, the first of its kind, includes BA and addresses sexual health and fertility, but was read only in abstract [118]. Transition requires lifelong multidisciplinary follow-up, because complications may appear more than 20 years after KPE, and a Japanese guide to self-management and transition for adolescents and young adults with BA was published in 2026 [2, 5, 119]. In France, the PNDS refers therapeutic education to programmes that vary between centres and adopts the 2016 FILFOIE network recommendations for transition in three periods: preparation from 14 to 18 years (informing early about the transfer, usually at around 18 years; consultations that promote autonomy; therapeutic education sessions; annual assessment of knowledge by questionnaire; multidisciplinary consultations; choice of the adult doctor at least 1 year beforehand; choice of a general practitioner at 16 years, as French regulations require), transfer from 16 to 20 years (during a period of clinical, psychological and social stability, with joint or alternating consultations, a referral letter and confirmation of the first contact, to avoid loss to follow-up) and reception in the adult service (sufficient consultation time, introduction of the team and information to the general practitioner and paediatrician, including about missed appointments). Education of the adolescent covers the disease, complications, follow-up, adherence and risk behaviours, such as alcohol and other substances. Adults with native liver remain at high risk of cholangitis and of portal hypertension with variceal haemorrhage, complications that justify discussing transplantation at a young age even without hepatocellular failure, and growth, puberty and neurodevelopment deserve attention; those transplanted in childhood require monitoring for immunosuppressant toxicity and the risk of poor adherence and graft loss. Disclosure of the diagnosis should explain the disease, its consequences and the treatment plan in a way adapted to the family, take account of the psychological impact and provide information on the long-term condition scheme, the emergency card, sources of information and the patient association [38].

7.10. Guideline recommendations and divergences

Table sources: [2, 3, 5, 38, 42, 44, 47, 57, 73, 79, 88, 91, 92, 95, 96, 114, 118].

The 1st edition of the Japanese guideline, published in an abridged English version in 2020–2021, was the first clinical practice guideline dedicated to BA developed according to the 2014 MINDS manual (Medical Information Network Distribution Service): a writing group of 36 JBAS specialists, which brings together 101 hospitals where most Japanese patients are treated, formulated 25 clinical questions, the systematic review included the PubMed and Japan Medical Abstracts Society literature up to July 2014, evidence was graded from A (strong) to D (very weak) and each recommendation required more than 70% of votes in up to three Delphi rounds, failing which it was left without a recommendation [79]. In diagnosis, it proposed screening by stool colour card (weak, C, 90%), made no recommendation on examining infants with card colour no. 4 (52% agreement), recommended measuring direct bilirubin in the infant with prolonged jaundice and hepatomegaly, abnormal stools or dark urine (strong, B, 72%), performing intraoperative cholangiography (strong, C, 87%) and considering the histology of the KPE specimen in the therapeutic strategy (strong, C, 82%) [79]. In treatment, it recommended intravenous vitamin K before surgery (strong, D, 78%), proposed KPE in the first 30 days (weak, C, 94%), took no position on corticosteroids (58%, evidence B), proposed intravenous antibiotics for 2–4 weeks followed by the oral route (weak, D, 74%), postoperative ursodeoxycholic acid (weak, D, 78%) and repeat KPE when jaundice clears only transiently (weak, C, 76%) [79]. In complications, it recommended considering cholangitis prophylaxis with co-trimoxazole (strong, C, 78%), proposed broad-spectrum antibiotics followed by therapy directed at the isolated organism (weak, C, 81%) and percutaneous transhepatic biliary drainage as short-term palliation of intrahepatic biliary cysts (weak, D, 84%), recommended immediate endoscopic treatment of variceal haemorrhage, by ligation or, if impossible, sclerotherapy, after stabilisation with vasoactive drugs (strong, C, 71%), proposed periodic oximetry to detect hepatopulmonary syndrome (weak, D, 87%) and recommended echocardiography in portal hypertension to look for latent pulmonary hypertension (strong, B, 84%) [79]. In prognosis, it proposed transplantation for growth failure in native liver survivors, especially between 1 and 5 years (weak, C, 74%), recommended multidisciplinary management of pregnancy (strong, C, 77%), regular imaging following the ALARA principle (as low as reasonably achievable) when radiation is involved (strong, C, 74%), prophylactic treatment of oesophagogastric varices (weak, C, 74%) and treatment of hypersplenism (strong, D, 76%), and made no recommendation on the age for transplantation after KPE or on primary transplantation with a PELD (pediatric end-stage liver disease) of 10 or more [79]. Compared with the 2nd edition, the 1st already set 30 days for KPE, although with lower strength, and differed mainly on biopsy, corticosteroids and hypersplenism [57, 79].

8. Reproductive Options and Family Counselling

8.1. Isolated biliary atresia

Isolated BA behaves as a multifactorial disease: familial aggregation is rare, twins are usually discordant, including monozygotic twins, and the variants identified do not follow Mendelian patterns [1, 4, 7]. The French PNDS (French National Diagnosis and Care Protocol) summarises this position: BA is not hereditary, familial cases are exceptional, twin studies describe more discordant pairs, including monozygotic ones, than concordant pairs, and any genetic predisposition is not Mendelian; a genetic contribution is more plausible in the syndromic forms and an environmental one in the isolated forms, and the process begins in foetal life in both [38]. Familial cases exist, but are few: a 1988 review counted 11 familial occurrences of extrahepatic BA in the literature and added two families with two affected siblings; sibships with 2 or more affected siblings have been described (5 in a 1964 review, some with renal and cardiac malformations), three cases in one family in 1932, two siblings born to consanguineous parents in 1978 and two siblings, one with extrahepatic BA and the other with intrahepatic BA, in 1999; discordance in at least three pairs of monozygotic twins supports a polygenic or multifactorial basis [22]. A recurrent intrauterine infection, such as listeriosis in successive pregnancies, is a non-genetic explanation to consider when two siblings are affected [22]. Once genetic causes of bile duct atresia have been excluded, the recurrence risk for siblings is low, but the sources consulted do not provide a validated numerical empirical risk, and the absence of such a figure does not mean zero risk [40]. The genetic contribution is polygenic, with common variants of modest effect (OR 1.4 to 2.7) and rare variants of reduced penetrance, so no known susceptibility genotype allows the disease to be predicted in a sibling or child [2]. Modifiable maternal factors with a described association, especially pregestational diabetes, justify optimising glycaemic control before a new pregnancy, although a causal relationship has not been demonstrated [2, 3]. Adult survivors who wish to have children should be informed that there are, in the sources consulted, no data on recurrence in offspring.

8.2. Syndromic and monogenic forms

In BASM, counselling depends on whether or not a cause is found. Biallelic PKD1L1 variants follow autosomal recessive inheritance, with a 25% risk for each sibling of having two variants, but expression is variable and penetrance incomplete, so the same genotype may give situs inversus, heterotaxy with or without BA or a minimal phenotype [7, 9]. In families with a laterality defect without a specific diagnosis, the Oxford Desk Reference gives an empirical sibling recurrence risk of 5% without consanguinity or family history, and of 25% with consanguinity [64]. In X-linked heterotaxy due to ZIC3, the recurrence risk depends on whether the mother is heterozygous, and some heterozygous women have situs inversus totalis, which makes ZIC3 testing important for recurrence risk [11, 41]. CFC1 and FOXA2 variants have been found in parents with situs inversus without BA, which illustrates incomplete penetrance and precludes using the parental phenotype to exclude transmission [3, 7]. Polysplenia and asplenia may occur in different members of the same family as expressions of the same defect [60]. Maternal diabetes is strongly associated with BASM and left isomerism, which reinforces preconception counselling in diabetic women [3, 64].

In X-linked heterotaxy due to ZIC3, none of the 8 living obligate carriers in the original family had signs of heterotaxy, but incomplete penetrance has been described in a male carrier and skewed X inactivation explained an affected sister, so the phenotype of relatives is not enough to exclude carrier status [19, 28]. In HTX2 due to CFC1, the variants described in 2000 came from parents without a phenotype [27].

In Mitchell-Riley syndrome, inheritance is autosomal recessive (25% risk for each sibling), and parents heterozygous for truncating RFX6 variants may have MODY-type diabetes of reduced penetrance, which should be looked for [12]. In chromosomal syndromes (cat eye, trisomy 18, Emanuel) and in dominant or X-linked monogenic syndromes (Kabuki, Hardikar), the risk depends on the mechanism and on whether the alteration is present in the parents [3, 66]. In cat eye syndrome, the extra chromosome usually arises de novo, but is transmitted by both sexes and parental mosaicism may go unnoticed in a parent without a phenotype or with minimal signs (preauricular pits, downslanting palpebral fissures); after the birth of an affected child, karyotyping of both parents is recommended, and even with a normal lymphocyte karyotype a small risk persists because of hidden mosaicism, including germline mosaicism, for which there is no numerical estimate; an affected person with preserved fertility has a risk close to 50% for each child, and early post-zygotic loss of the marker explains the frequent mosaicism in familial forms [21]. Hardikar and Kabuki 1 are, in the OMIM series, almost always de novo, and in Zimmermann-Laband rare germline mosaicism is accepted [13, 15, 16]. Among the genetic cholestatic diseases in the differential diagnosis, most are autosomal recessive, with a 25% risk for siblings [36].

8.3. Cascade testing, prenatal and preimplantation diagnosis

Cascade testing only makes sense when a monogenic or chromosomal cause has been identified in the proband; in that case it allows carriers to be identified and prenatal diagnosis or preimplantation genetic testing to be offered [40]. In the absence of an identified cause, a new pregnancy is followed by anomaly ultrasound and, in families with laterality defects, by foetal echocardiography at 14, 19 and 22 weeks [64]. Prenatal ultrasound may reveal cystic BA or the anomalies of BASM, but its sensitivity and specificity do not allow it to be used alone for diagnosis, and newborns with suspicious findings should be assessed soon after birth [2, 5]. In a sibling born after a case of BA, early measurement of direct bilirubin and observation of stool colour are the practical measures for early detection [5, 76].

8.4. Counselling of minors and the family

Genetic testing of minors is justified when it changes surveillance or treatment, as in heterotaxy, in which the consensus recommends genetic assessment of all patients, or in RFX6 diabetes; polygenic susceptibility variants, of modest effect and with heterogeneity between populations, have no demonstrated individual predictive value [2, 12, 41]. Counselling should explain to parents that, according to the available data, the injury begins before birth, with conjugated bilirubin already raised in the first days of life, and that no means of prevention is known [2]. In adolescence, transition to adult care includes information about the disease itself, sexual health and fertility [118].

9. Special Considerations: Population Diversity and Epidemiology

9.1. Incidence and what each figure measures

Estimates of BA incidence come from national registries, reference-centre series and congenital anomaly registries, which differ in coverage, denominator and case definition. The 2024 review gives 1:14,000 to 1:22,000 live births in Western countries with national data (the Netherlands, France, Croatia, Canada, the United States, England and Wales, Switzerland, Finland and Sweden) and 1:5000 to 1:9000 in East Asia (Taiwan, Japan, Korea, Shanghai and Hong Kong); in New Zealand, incidence was 1:5000 in Māori and 1:16,000 in those of European descent [2]. Other sources give similar values: about 1:15,000 to 1:20,000 in the United Kingdom, Europe and North America, with possible urban «hot spots» such as central London (up to 1:10,000), and 1:18,000 in children of European ancestry [37, 52, 53]. In Saudi Arabia, despite high consanguinity, overall incidence was low (1:44,000) [3]. The 2017 NASPGHAN/ESPGHAN guideline summarised estimates at about 1:6000 live births in Taiwan, 1:12,000 in the United States, 1:19,000 in Canada and 1:18,000 in Europe, and distinguished the non-syndromic form (84%), BA with at least one malformation without a laterality defect (6%) and syndromic BA with laterality defects (10%), the last two with cardiovascular (16%) and digestive (14%) anomalies [42]. OMIM cites an approximate worldwide incidence of 1:10,000 live births, with 1.46 per 10,000 in Taiwan against 0.65 to 0.85 per 10,000 in the United States, and FISPGHAN 1 to 3 per 10,000 in East Asia and French Polynesia against about 0.5 per 10,000 in Europe and North America; these are aggregations of studies with different methods, not mutually comparable estimates [22, 73]. There are no published national data from Africa, South America, India or Australia [2]. The French PNDS (French National Diagnosis and Care Protocol) estimates incidence at about 1:20,000 live births in Europe and at 1:17,000 in France, attributing the latter value to the 1986–2002 national study, and describes higher values in Hawaii (up to 1:10,000) and in the Polynesian population of French Polynesia (about 1:3000), one of the highest incidences in the world, without apparent temporal variation in each region [38]. The national study cited included 743 patients resident in France, 472 born from 1986 to 1996 and 271 from 1997 to 2002, of whom 650 (87.5%) were born in metropolitan France, 67 (9.0%) in the overseas territories and 26 (3.5%) abroad, with a male:female ratio of 0.82; in metropolitan France, incidence was 5.12 per 100,000 live births (95% CI 4.63–5.61) in 1986–1996 and 5.09 per 100,000 (95% CI 4.45–5.73) in 1997–2002, with no significant difference between periods, corresponding to about 1:19,500 [39]. That publication does not present the value of 1:17,000 nor an incidence for the whole of French territory including the overseas territories, so this article uses the published metropolitan estimate. The earlier national study, of those born from 1986 to 1996, identified 461 patients, 421 in metropolitan France, with an incidence of 5.12 per 100,000 live births (95% CI 4.63–5.61), about 1:19,500, without regional variation, and 40 in the overseas territories, with 29.4 per 100,000 in French Polynesia, 5.7 times higher; no seasonality or temporal or spatiotemporal clustering was demonstrated [120]. The Polynesian value, consistent with the 32 per 100,000 in Orphanet, indicates a population at particularly high risk, although the sources consulted do not identify its genetic basis.

Table source: Orphanet/Orphadata, isolated BA [1]. EUROCAT values by country depend on the regional registry and year, are not national prevalences, and the variation between 1.6 and 16.6 per 100,000 mainly reflects differences in coverage and definition. The overall class assigned by Orphanet is 1 to 5 per 10,000 worldwide and 1 to 9 per 100,000 in Europe, and the prevalence of BASM is considered unknown [1, 18].

9.2. Sex, ethnicity, season and demographic factors

BA is more frequent in females, with a risk ratio of about 1.4 in the West and in Asia and 1.8 in Japan, but the female:male ratio is 1.1:1 in China and 0.7:1 in Sweden; in BASM it is 2:1 in England and Wales, and similar in Canada [2]. The risk is 2 to 3 times higher in Asian countries and in people of Asian descent, and incidence is higher in non-white children in England and Wales, Atlanta and New York [2]. However, in the London metropolitan area incidence was 7.02 per 100,000 in white Caucasians and 7.74 per 100,000 in non-whites (P=0.590), with an equal proportion of BASM, which indicates that the geographical disparity is not explained by ethnicity, a conclusion similar to that of a Canadian study [3]. The larger studies show no consistent seasonality; in Taiwan, incidence fell without significant correlation with rotavirus vaccine coverage, and a European study associated the strictest COVID-19 pandemic lockdowns with a greater fall in incidence [2].

9.3. Distribution of syndromic forms

BASM accounts for about 10% of cases in European and North American series and less than 3% in India, China and Japan, with no data from sub-Saharan Africa; FISPGHAN gives 10% to 15% for syndromic forms in North America and Europe and less than 1% to 3% in Asia and South America, values that fall with a strict definition of BASM and rise when other malformations are included [73]; since isolated BA is about 3 times more frequent in Asia, the absolute incidence of BASM may be similar between regions [2, 3]. CMV-associated BA is more frequent in Asia, and cystic BA accounts for 5% to 10% of series [2, 37].

Table source: Davenport 2025, Table 2 [3].

9.4. Brazil and Portugal

In Brazil, a study of 6 centres between 1982 and 2008, with 513 patients, showed KPE in 76.4% at 82.6 ± 32.8 days, with later operation in the Northeast than in the South and Southeast; 46.6% were transplanted (69% after KPE and 31% as primary transplantation), overall survival was 67.6%, increasing by decade, 4-year survival after KPE, with or without transplantation, was 80%, 77.7% and 60.5% for KPE up to 60 days, between 61 and 90 days and after 90 days, and 4-year native liver survival was 36.8% (54%, 33.3% and 26.6% in the same groups) [121]. The authors concluded that late referral remained the main problem in Brazil [121]. The Brazilian Society of Paediatrics, with the Ministry of Health, launched the «Alerta Amarelo» (Yellow Alert) campaign and the Child Health Record (Caderneta de Saúde da Criança) includes a stool colour scale [74]. In a Brazilian transplant centre, between 2013 and 2023, 83% of transplants for BA used a living donor [93]. In Portugal, the sources consulted only document the inclusion of the stool colour card in the digital version of the Child and Youth Health Record (Boletim de Saúde Infantil e Juvenil), and include no national data on incidence or on KPE outcomes [75].

9.5. Special populations

Preterm infants have twice (under 37 weeks) or five times (under 32 weeks) the incidence, more associated malformations and later KPE, with similar outcomes; those small for gestational age have about a 5-fold higher risk [2]. Children of mothers with pregestational diabetes, especially in BASM, are another risk group [2, 3]. In children with heterotaxy and heart disease, BA occurs in about 6% [41]. In high-income countries without centralisation, such as Germany, the jaundice clearance rate is around 24% to 25%, and in Brazil the Northeast operated later, which shows that the organisation of care is a population-level determinant of prognosis [2, 37, 121].

9.6. Reference networks and patient organisations

In the European Union, BA is covered by ERN RARE-LIVER, the European Reference Network for rare liver diseases, whose research network BARD (Biliary Atresia and Related Diseases) produced the 2022 position statement on initial treatment, and paediatric transplantation by ERN TransplantChild [47–49]. In the United States and Canada, ChiLDReN (Childhood Liver Disease Research Network) conducted the START, PRIME, PROBE and BASIC studies [7, 44, 122]. The Children's Liver Disease Foundation, in the United Kingdom, supports families of children with liver disease, and the Children's Heart Federation families with heart disease, including heterotaxy [40, 64]. In France, BA belongs to the health network for rare liver diseases of adults and children (FILFOIE, https://www.filfoie.com/) and the reference centre for biliary atresia and genetic cholestasis (AVB-CG), with its coordinating site at Bicêtre Hospital (AP-HP, Le Kremlin-Bicêtre) and constituent sites in Lyon (Hôpital Femme-Mère-Enfant), Marseille (Hôpital de La Timone), Paris (Hôpital Necker-Enfants malades) and Toulouse (Hôpital des Enfants); the paediatric competence centres are in Bordeaux, Caen, Lille, Montpellier, Rennes, Saint-Denis de La Réunion, Strasbourg and Tours, and the adult hepatology centres at Hôpital Beaujon (Clichy) and Hôpital Paul-Brousse (Villejuif). The disease gives access to the long-term condition scheme, with full coverage of care (ALD 6, active chronic liver diseases and cirrhosis, or ALD 28 for organ transplantation). The patient association Association Maladies du Foie depuis l'Enfance (AMFE, Malakoff; https://amfe.fr/) brings together parents, young people and families, collaborates with the centres, supports and co-funds research and accompanies families, including bereaved ones [38]. The sources consulted did not identify patient associations specific to BA in Portugal or Brazil.

10. Illustrative Clinical Case

10.1. Cystic biliary atresia with a de novo KDM6A variant (Masui et al., 2019)

Masui and colleagues described, in Japan, a boy born at 39 weeks by vaginal delivery, weighing 3330 g and with no relevant family history, in whom prenatal ultrasound at 16 weeks had shown a 37 × 40 mm cystic lesion beneath the liver, with a differential diagnosis between choledochal cyst and cystic BA; the foetal karyotype was 46,XY (figure 1 of the article: https://pmc.ncbi.nlm.nih.gov/articles/PMC6694366/figure/Fig1/) [72]. On the 1st day of life jaundice and acholic stools appeared, with total bilirubin of 6.24 mg/dL, direct bilirubin of 4.53 mg/dL, AST of 124 IU/L, ALT of 65 IU/L, GGT of 47 IU/L (a low-normal value, unusual in BA) and total bile acids of 139.2 µmol/L; tests for CMV and herpes simplex virus were negative, and postnatal ultrasound showed a 43 × 32 mm cystic mass at the liver hilum without dilatation of the intrahepatic ducts [72].

At 4 days of life, laparotomy with cholangiography showed a «tree-like» pattern of the intrahepatic ductules, according to the Nio classification, and a cystic common bile duct without passage of contrast into the intestine, with complete occlusion of the distal common bile duct, which confirmed cystic BA (figure 2: https://pmc.ncbi.nlm.nih.gov/articles/PMC6694366/figure/Fig2/); a Roux-en-Y hepaticojejunostomy was performed, and jaundice cleared 2 months after surgery [72]. The course revealed a broad forehead, hypertelorism, a small pointed chin, large anteverted ears and foetal finger pads (figure 3: https://pmc.ncbi.nlm.nih.gov/articles/PMC6694366/figure/Fig3/), opisthotonus with normal brain MRI, pulmonary artery stenosis, bicuspid aortic valve and posterior embryotoxon [72]. Liver biopsy showed mild portal inflammation and fibrosis with paucity of interlobular ducts, demonstrated by immunohistochemistry for cytokeratin 7 (figure 4: https://pmc.ncbi.nlm.nih.gov/articles/PMC6694366/figure/Fig4/) [72].

The combination of ductal paucity, pulmonary stenosis, posterior embryotoxon and facies suggested Alagille syndrome, but JAG1 and NOTCH2 sequencing was negative; trio exome sequencing identified a heterozygous de novo nonsense variant in exon 29 of KDM6A, c.3835C>T p.(Arg1279Ter) (figure 5: https://pmc.ncbi.nlm.nih.gov/articles/PMC6694366/figure/Fig5/) [72]. The article does not state the transcript; the numbering corresponds to NM_021140.4, and on the MANE transcript the same variant is NM_001291415.2:c.3991C>T p.(Arg1331Ter) (NC_000023.11:g.45090821C>T), according to validation in VariantValidator [62]. The geneticists did not confirm Kabuki syndrome because the typical facies was absent, and at 2 years liver parameters were normal [72]. The authors recall that posterior embryotoxon occurs in about 22% of the general population and that loss of intrahepatic ducts has been described in 20% and 16.2% of extrahepatic BA in two series, and propose a link between KDM6A and the Notch pathway [72].

10.2. Lessons from the case

The case brings together several points of this article. Cystic BA can be suspected before birth and operated on in the first days, with a good prognosis after resection of the cyst, as described in section 3.4 [2, 37]. The combination of BA, heart disease and posterior embryotoxon is not enough to diagnose Alagille syndrome, and ductal paucity may accompany BA, so testing of JAG1 and NOTCH2 and, if negative, trio exome sequencing are decisive [50, 72]. The link between BA and KDM6A Kabuki syndrome adds to the other reports in which the diagnosis of BA preceded that of the syndrome [3]. The technique used, hepaticojejunostomy in a cyst with a visible intrahepatic tree, corresponds to the treatment of cystic BA and should not be generalised to type III forms, in which KPE is the rule [2, 5].

10.3. Complementary case: biliary atresia splenic malformation syndrome

Alhashmi and colleagues described a girl born at 37 weeks weighing 3860 g, in whom prenatal ultrasound had suggested cystic adenomatoid malformation, BA, polysplenia, absent IVC and oligohydramnios [108]. The stools, initially bile-stained, became creamy and acholic at 2 weeks, with total bilirubin of 7.89 mg/dL, direct bilirubin of 6.48 mg/dL and GGT of 438 U/L (995 U/L in the 1st week) and negative TORCH serologies; ultrasound showed no gallbladder or common bile duct and confirmed polysplenia and interruption of the IVC with azygos continuation, the echocardiogram showed ventricular and atrial septal defects, and MR cholangiography confirmed absence of the gallbladder and accessory spleens up to 2 cm [108]. She underwent KPE at 3 weeks, with a 30 to 40 cm Roux loop and correction of malrotation by the Ladd procedure at the same operation, followed by corticosteroids; at 4 months, total bilirubin was 2.39 mg/dL, direct bilirubin 1.78 mg/dL and GGT 351 U/L [108]. The figures of the article (MR cholangiography and operative view) were not consulted individually. The case illustrates prenatal detection of the components of BASM, early KPE and simultaneous correction of malrotation, but provides no information on long-term outcome, and the use of corticosteroids corresponds to a centre's choice on which the guidelines diverge (section 7.5) [5, 44, 108].

11. Prognosis and Long-Term Follow-Up

11.1. Portoenterostomy outcomes by country

Table sources: [2, 39, 121].

The rates are not directly comparable, because jaundice clearance is defined with different thresholds and the cohorts differ in period, age at KPE and proportion of syndromic forms. Even so, they show that centralisation and early KPE are associated with the best outcomes, and the 2024 review proposes as a benchmark a jaundice clearance of 50% to 70% in experienced centres [2]. In the 20-year follow-up cited by the international guideline, native liver survival was 51.3% at 5 years and 46.5% at 10 years, and overall survival 91.5% and 90.5% [5]. The French PNDS (French National Diagnosis and Care Protocol) on bile acid synthesis defects (2019) states that, with KPE before 30 days, about 40% of operated children can expect to be alive with their native liver at 20 years, and that after 45 days the probability of success falls and postoperative complications increase, without stating the source cohort [88]. Expected native liver survival is 45% to 65% at 2 years and about 25% in early adulthood, so at least three quarters of patients will need transplantation during their lifetime; more than half are transplanted before 5 years, and 60% to 75% by 18 years [2, 4, 47]. Almost half of native liver survivors at 2 years end up transplanted before adulthood [2].

11.2. Prognostic factors

Age at KPE is the most important modifiable factor: in isolated BA, jaundice clearance exceeds 70% when KPE is performed before 30 days and falls below 40% after 100 days, and in Brazil 4-year native liver survival fell from 54% (KPE up to 60 days) to 26.6% (after 90 days) [2, 121]. In a French series of 743 infants cited by the 2017 guideline, native liver survival was 57.1%, 37.9%, 32.4% and 28.5% at 2, 5, 10 and 15 years, decreasing with age at KPE between under 45 and 90 days, and the authors estimated that operating on all before 46 days would avoid 5.7% of annual liver transplants in children under 16 in France; even after 90 days, some series describe effective drainage in 13% to 35% [42]. FISPGHAN summarises the course after KPE as about 50% transplanted before 2 years and about 75% by 20 years, and cites a meta-analysis that confirmed the benefit of surgery before 30 days on native liver survival at 5, 10 and 20 years [73]. The clinical form matters: cystic BA has the best outcomes, and BASM and CMV-associated BA the worst [2]. In BASM, the French national series found 20-year native liver survival of 15% against 28% in the others, and the 2006 English series 32% at 10 years; in BA with heart disease, mortality was 40% and 5-year native liver survival 54% [3]. A systematic review of 195 patients with BASM found a response to KPE similar to that of the isolated form, with mortality dominated by heart disease [3]. Among preoperative markers, the APRI index predicts native liver survival by quartiles, a GGT below 300 IU/L was unfavourable in a Chinese series, Scheuer fibrosis 3–4 was associated with a higher risk of transplantation (HR 1.73; 95% CI 1.21–2.48) in a study with 316 biopsies in which no other histological feature was related to jaundice clearance, and serum FGF19 above 109 pg/mL was associated with markedly reduced native liver survival [2]. Rising IL-8 was associated with transplantation or death 1 year after KPE, and IgM anti-cholangiocyte autoantibodies with worse outcome [2]. After KPE, total bilirubin above 100 µmol/L at 3 months defines failure and indicates transplant assessment [5].

The French national study of 1986–2002 reviewed, in the 45 paediatric centres involved, the records of all children with BA born in that period and resident in France, with histology of the remnant available in 584 (78%) and a median follow-up of survivors of 7 years (0.2 to 18.1), and compared cohort A (472 born from 1986 to 1996) with cohort B (271 born from 1997 to 2002) [39]. KPE or a variant was performed in 695 children (93.5%), at a median age of 60 days (12 to 180), 61 days in cohort A and 57 in cohort B, a non-significant difference despite information campaigns for paediatricians; jaundice cleared (bilirubin of 20 µmol/L or less) in 242 of 668 (36.2%), 34.4% and 39.5% [39]. Native liver survival 4 years after KPE was 40.6% (40.1% and 42.7% in the two cohorts); with KPE up to 45 days it was 46.5% and 51.2%, against 37.5% and 39.8% after that age, and in anatomical types 1 and 2 it was 71.7% and 66.2%, against 36.0% and 43.2% in types 3 and 4 [39]. In the multivariate analysis of cohort A, independent factors were KPE after 45 days (risk ratio 1.43; 95% CI 1.07–1.91), types 3 and 4 (2.08), polysplenia (1.89; 95% CI 1.29–2.77) and centre caseload (3 to 5 KPEs per year, 2.08; 2 or fewer, 1.35, against more than 20); in cohort B, only polysplenia remained significant (1.87; 95% CI 1.05–3.34), with 4-year native liver survival of 18.4% and 19.2% in patients with polysplenia in the two cohorts [39]. The centre effect was the main result: native liver survival 4 years after KPE was 47.8% and 47.7% in the centre with more than 20 KPEs per year, remained at 33.9% and 33.4% in centres with 2 or fewer (27 and 18 centres) and rose from 30.9% to 44.5% (P=0.04) in centres with 3 to 5; the number of centres performing KPE fell from 30 to 22, but those with the smallest caseload still performed 28.6% of the KPEs in cohort B, against 37.0% in cohort A [39].

The 48 patients (6.5%) who did not undergo KPE, because of late diagnosis in 30, contraindication due to a multiple malformation syndrome in 2 and other or unknown reasons in 16, had native liver survival of 58.4% at 1 year, 28.4% at 2 years, 10.9% at 3 years and 2.2% at 5 years; 4-year overall survival of those referred for primary transplantation rose from 53.3% to 93.3% between the cohorts [39]. A total of 386 children were transplanted (456 transplants, with 62 second and 8 third transplants), at a median age at first transplant of 18.6 months (5.2 to 172.1); survival 4 years after transplantation rose from 75.1% to 88.8% (P=0.006), and in cohort A independent factors were liver failure before transplantation (prothrombin time of 50% or less; risk ratio 1.94; 95% CI 1.06–3.55) and a centre with fewer than 5 paediatric transplants per year (1.92; 95% CI 1.09–3.38), with no influence of graft type [39]. Split grafts rose from 14.4% to 33.4% and living-donor grafts from 6.2% to 14.6% of first transplants, and deaths on the waiting list fell from 8.5% to 2.1% of transplants (P=0.02) [39]. Overall 4-year survival rose from 73.6% to 87.1% (P<0.001) and observed mortality fell from 30.5% to 12.2%; in cohort A, overall 4-year survival was 97.8% after KPE with jaundice clearance, 65.2% after failed KPE and 50% without KPE, and in the multivariate analysis KPE failure (risk ratio 4.9; 95% CI 2.9–8.3), absence of KPE (6.9; 95% CI 3.5–13.6) and treatment in a centre with 3 to 5 (2.0) or with 2 or fewer new patients per year (2.4) increased the risk of death [39]. Eleven children operated on with a wrong diagnosis of BA (1.6%) were excluded from the study: 5 had Alagille syndrome, 2 alpha-1 antitrypsin deficiency, 2 cystic fibrosis, 1 choledochal cyst and 1 ischaemic cholangiopathy [39].

In the French BA observatory (1986–2015), overall survival was 87% at 5 years and 76% at 30 years, and 26% of operated children reached 20 years with their native liver; 5-year native liver survival was 56% with KPE before 31 days, 44% between 31 and 60 days, 36% between 61 and 90 days and 33% after 90 days, and about 40% of those operated on before 30 days are alive with native liver at 20 years. The PNDS separates prognostic factors related to the disease, which are not modifiable (associated malformations such as heart disease, polysplenia syndrome, anatomical type and histological lesions of the remnant, degree of fibrosis and portal pressure at KPE), from those related to care (age at KPE, centre experience, KPE and transplantation in the same centre, early referral to the transplant centre and access to paediatric, split or living-donor grafts). Portal pressure measured at KPE correlates with jaundice clearance and with the risk of threatening portal hypertension even with normal bilirubin. Even with restored flow, cirrhosis develops progressively in almost all patients, because of the intrahepatic involvement already present at diagnosis, and rare cases of hepatocellular carcinoma have been described [38].

11.3. Long-term complications in native liver survivors

Table source: 2024 review, Table 4 [2].

Hepatopulmonary syndrome (5% to 7% of children with native liver) and portopulmonary hypertension (less than 1%) are indications for transplant listing before irreversible cardiopulmonary complications [2]. Late complications, such as cholangitis, portal hypertension, hepatic deterioration and neoplasia, may appear more than 20 years after KPE, and follow-up therefore continues even when jaundice has cleared (recommendation 19, grade 1A) [5]. The Japanese guideline recommends screening for liver tumours during prolonged follow-up of native liver survivors, with tumour markers and imaging (strong recommendation, evidence C, agreement of 88%), because, although rare, tumours have been described, and cholangiocarcinoma has a particularly poor prognosis; it also proposes looking for varices from childhood by methods suited to the patient, with less invasive imaging and laboratory tests as a possible alternative, and treating oesophageal varices prophylactically, a measure associated with fewer complications than treatment after haemorrhage but which, in childhood, is understood as a bridge to transplantation [57].

For the French PNDS, portal hypertension appears in at least two thirds of children after KPE, even with regression of cholestasis. Primary prophylaxis of variceal haemorrhage, adopted in France and Europe, looks by upper gastrointestinal endoscopy, or by oesophageal capsule in older children, for high-risk varices (grade II with red signs, grade III and subcardial varices), located mainly in the oesophagus, stomach, Roux-en-Y loop and rectum; the reference treatment is band ligation, generally possible above 10 kg, and sclerotherapy is reserved for specialised centres. There are no firm criteria for the first endoscopy: in KPE failure it is discussed between 6 and 18 months, depending on the severity of portal hypertension and the transplant plan, and in success it is decided by signs of cirrhosis and portal hypertension (splenomegaly, thrombocytopenia, bilirubin and serum bile acids, portosystemic shunts, ascites, liver and spleen elastography); serial elastography may help to follow fibrosis and portal hypertension, but the data are insufficient to recommend it routinely. Exceptionally, refractory portal hypertension or very frequent ligations justify surgical or radiological portosystemic shunting, and hypersplenism splenic arterial embolisation or partial splenectomy, decided in a specialised centre; in KPE failure, portal hypertension is treated by transplantation, with sclerotherapy, ligation or shunting while waiting. Gastrointestinal haemorrhage may come from oesophagogastric varices, gastropathy or intestinal varices, and hypochromic microcytic iron-deficiency anaemia may reveal occult bleeding; secondary prevention rests on endoscopic ligation or sclerotherapy, and threatening haemorrhage persisting despite endoscopic treatment indicates transplantation with national priority [38]. The 2026 APSPGHAN statement acknowledges that the evidence does not define the optimal age for the first endoscopy in BA and cites the proposal to perform it at 9 to 12 months, especially after unsuccessful KPE or with signs of portal hypertension, generally followed by annual surveillance (agreement of 76.9%, low certainty) [114].

Large intrahepatic biliary cavities may appear months or years after KPE, even with resolved jaundice, become infected or compress the portal vein, and their chronic infection, or chronic cholangitis, may justify transplantation despite a successful KPE. Ascites, often with hypoalbuminaemia, generally signals progression to hepatocellular failure; it requires exclusion of infection (diagnostic paracentesis) and portal vein thrombosis, rare in BA, and a search for protein loss through glomerulopathy (proteinuria) or protein-losing enteropathy (faecal alpha-1 antitrypsin clearance), and is treated with fluid and salt restriction, albumin infusions and oral diuretics (spironolactone, furosemide), with monitoring for hyponatraemia and hyperkalaemia [38]. The PNDS does not give doses: the British SmPC of spironolactone (Aldactone®) gives for children a starting dose of 1 to 3 mg/kg/day in divided doses, adjusted to response and tolerance, and contraindicates it in moderate to severe renal impairment and hyperkalaemia; the US label for oral furosemide gives 2 mg/kg as a single dose, increasable by 1 to 2 mg/kg with a minimum interval of 6 to 8 hours, not exceeding 6 mg/kg; and the US human albumin label consulted (Albuminex®) has no paediatric data and gives only adult doses, which cannot be transposed directly to infants [123–125].

Hepatopulmonary syndrome and portopulmonary hypertension may appear even with resolved jaundice, do not regress spontaneously and are indications for transplantation; they are not specific to BA and are attributed to the arrival in the lungs, through portosystemic shunts, of intestinal vasoactive substances not cleared by the liver. Hepatopulmonary syndrome causes hypoxia, cyanosis, dyspnoea and digital clubbing, is confirmed by lung perfusion scintigraphy and contrast echocardiography with microbubbles and, if severe, complicates respiratory management at transplantation and reduces post-transplant survival; portopulmonary hypertension causes exertional dyspnoea, faintness or sudden death, is confirmed by echocardiography and, if necessary, catheterisation, and when severe may contraindicate transplantation, so it should be detected early and treated in a specialised centre until grafting. Transplantation reverses the shunts and portopulmonary hypertension that is not yet fixed. Acute ischaemic necrosis of the liver, favoured by gastrointestinal haemorrhage, hypovolaemia or sepsis, is suspected in the presence of abdominal pain, fever, rising aminotransferases and the appearance or worsening of hepatocellular failure; some children recover, but transplantation with national priority and urgent transfer to a paediatric transplant centre should be considered, and an arterial resistance index above 1 on Doppler ultrasound signals particular risk. Hepatocellular carcinoma and high-grade dysplasia have been described from the first years of life, and exceptionally hepatoblastoma and cholangiocarcinoma; the PNDS requires screening throughout pre-transplant follow-up with annual liver ultrasound and six-monthly alpha-fetoprotein, investigation of suspicious lesions by magnetic resonance imaging or contrast-enhanced ultrasound and, if suspicion persists, discussion of biopsy or transplantation, indicated in confirmed carcinoma and, to a lesser extent, in high-grade dysplasia; sometimes the lesion is only discovered in the explanted liver [38].

11.4. Surveillance table

12. Summary and Concluding Remarks

Biliary atresia is a progressive obliterative cholangiopathy that begins before birth, manifests as persistent jaundice with acholic stools and dark urine and is the most frequent identifiable cause of neonatal cholestasis and the leading indication for liver transplantation in childhood. It is not a homogeneous disease: the isolated form, the majority, results from polygenic susceptibility with common variants in ADD3, EFEMP1 and ciliogenesis genes and with rare variants in ciliary genes, on which perinatal, immune and environmental factors act; the biliary atresia splenic malformation syndrome reflects a combined disturbance of laterality and biliary morphogenesis, with PKD1L1 as the most consistent gene and a strong association with maternal diabetes; cystic BA and the cytomegalovirus-associated form are distinguished by the timing and mechanism of injury; and BA appears as an occasional finding in syndromes such as cat eye, Kabuki, Hardikar and Mitchell-Riley.

Diagnosis depends on recognising that all jaundice after 2 weeks requires direct bilirubin, on immediate referral in the presence of cholestasis and on confirming the disease by intraoperative cholangiography; the stool colour card and neonatal direct bilirubin allow diagnosis to be brought forward, and GGT, ultrasound, MMP-7 and biopsy guide the decision. Genetic testing neither confirms nor excludes BA, but is essential for the differential diagnosis with Alagille syndrome and monogenic intrahepatic cholestatic diseases, and for characterising the syndromic forms and the recurrence risk.

The central treatment remains Kasai portoenterostomy, whose outcome depends on age at surgery and on the organisation of care in reference centres. Corticosteroids, ursodeoxycholic acid and antibiotic prophylaxis are used with limited evidence and with divergence between guidelines, and the 2024 French national protocol is the only guideline that sets a dose of ursodeoxycholic acid and that prefers periodic intramuscular vitamin supplementation while jaundice persists; antivirals show a favourable signal in the cytomegalovirus-associated form; and ileal bile acid transporter inhibitors did not improve outcomes in the EMBARK and BOLD trials. As at least three quarters of patients will need transplantation, and complications may appear decades later, follow-up is multidisciplinary and lifelong, with surveillance of nutrition, portal hypertension, cholangitis, pulmonary function, liver neoplasia, development and transition to adult care. In the syndromic forms, management of asplenia, malrotation, heart disease and neonatal diabetes is added, together with genetic counselling directed at the identified cause.

13. References

1. Orphanet. Isolated biliary atresia (ORPHA:30391). https://www.orpha.net/en/disease/detail/30391
2. Tam PKH, Wells RG, Tang CSM, Lui VCH, Hukkinen M, Luque CD, et al. Biliary atresia. Nat Rev Dis Primers. 2024;10(1):47. doi:10.1038/s41572-024-00533-x. PMID: 38992031; PMC11956545.
3. Davenport M. Syndromic variants of biliary atresia. World J Pediatr Surg. 2025;8(3):e001040. doi:10.1136/wjps-2025-001040. PMID: 40519534; PMC12161394.
4. Meng Y, Yang Q, Liu S, Ke X, Zhan J. Genetic background and biliary atresia. World J Pediatr Surg. 2025;8(3):e001023. doi:10.1136/wjps-2025-001023. PMID: 40519535; PMC12161372.
5. Ge L, Yang J, Li T, Zhu T, Davenport M, Petersen C, et al. Diagnostic and management guidelines for biliary atresia in 2025. Hepatobiliary Surg Nutr. 2026;15(4):103. doi:10.21037/hbsn-2025-97. PMID: 42569433; PMC13449986.
6. Firth HV, Hurst JA. Oesophageal and intestinal atresia (including tracheo-oesophageal fistula). In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 266–269.
7. Berauer JP, Mezina AI, Okou DT, Sabo A, Muzny DM, Gibbs RA, et al. Identification of Polycystic Kidney Disease 1 Like 1 Gene Variants in Children With Biliary Atresia Splenic Malformation Syndrome. Hepatology. 2019;70(3):899-910. doi:10.1002/hep.30515. PMID: 30664273; PMC6642859.
8. Lam WY, Tang CS, So MT, Yue H, Hsu JS, Chung PH, et al. Identification of a wide spectrum of ciliary gene mutations in nonsyndromic biliary atresia patients implicates ciliary dysfunction as a novel disease mechanism. EBioMedicine. 2021;71:103530. doi:10.1016/j.ebiom.2021.103530. PMID: 34455394; PMC8403738.
9. GeneGuidance. PKD1L1 — clinical gene curation (visceral heterotaxy 8). Version of 2026-09-26.
10. GeneGuidance. CFC1 — clinical gene curation (visceral heterotaxy 2). Version of 2026-09-26.
11. GeneGuidance. ZIC3 — clinical gene curation. Version of 2026-09-26.
12. GeneGuidance. RFX6 — clinical gene curation (Mitchell-Riley syndrome). Version of 2026-09-26.
13. OMIM® Online Mendelian Inheritance in Man. #147920 KABUKI SYNDROME 1; KABUK1. Johns Hopkins University; last edited 16/07/2025. https://omim.org/entry/147920. Accessed 29 October 2026.
14. OMIM® Online Mendelian Inheritance in Man. #300867 KABUKI SYNDROME 2; KABUK2. Johns Hopkins University; last edited 16/07/2025. https://omim.org/entry/300867. Accessed 29 October 2026.
15. OMIM® Online Mendelian Inheritance in Man. #301068 HARDIKAR SYNDROME; HDKR. Johns Hopkins University; last edited 01/07/2024. https://omim.org/entry/301068. Accessed 29 October 2026.
16. OMIM® Online Mendelian Inheritance in Man. #135500 ZIMMERMANN-LABAND SYNDROME 1; ZLS1. Johns Hopkins University; last edited 21/03/2022. https://omim.org/entry/135500. Accessed 29 October 2026.
17. OMIM® Online Mendelian Inheritance in Man. *612659 REGULATORY FACTOR X, 6; RFX6. Johns Hopkins University; last edited 21/04/2025. https://omim.org/entry/612659. Accessed 29 October 2026.
18. Orphanet. Biliary atresia with splenic malformation syndrome (ORPHA:244283). https://www.orpha.net/en/disease/detail/244283
19. OMIM® Online Mendelian Inheritance in Man. #306955 HETEROTAXY, VISCERAL, 1, X-LINKED; HTX1. Johns Hopkins University; last edited 10/08/2026. https://omim.org/entry/306955. Accessed 29 October 2026.
20. OMIM® Online Mendelian Inheritance in Man. #605376 HETEROTAXY, VISCERAL, 2, AUTOSOMAL; HTX2. Johns Hopkins University; last edited 08/01/2024. https://omim.org/entry/605376. Accessed 29 October 2026.
21. OMIM® Online Mendelian Inheritance in Man. #115470 CAT EYE SYNDROME; CES. Johns Hopkins University; last edited 22/04/2022. https://omim.org/entry/115470. Accessed 29 October 2026.
22. OMIM® Online Mendelian Inheritance in Man. %210500 BILIARY ATRESIA, EXTRAHEPATIC; EHBA. Johns Hopkins University; last edited 24/02/2026. https://omim.org/entry/210500. Accessed 29 October 2026.
23. OMIM® Online Mendelian Inheritance in Man. #617205 HETEROTAXY, VISCERAL, 8, AUTOSOMAL; HTX8. Johns Hopkins University; last edited 24/01/2025. https://omim.org/entry/617205. Accessed 29 October 2026.
24. OMIM® Online Mendelian Inheritance in Man. #615710 MITCHELL-RILEY SYNDROME; MTCHRS. Johns Hopkins University; last edited 22/11/2024. https://omim.org/entry/615710. Accessed 29 October 2026.
25. OMIM® Online Mendelian Inheritance in Man. %601346 MARTINEZ-FRIAS SYNDROME. Johns Hopkins University; last edited 10/04/2014. https://omim.org/entry/601346. Accessed 29 October 2026.
26. OMIM® Online Mendelian Inheritance in Man. *609721 POLYCYSTIN 1-LIKE 1; PKD1L1. Johns Hopkins University; last edited 04/08/2023. https://omim.org/entry/609721. Accessed 29 October 2026.
27. OMIM® Online Mendelian Inheritance in Man. *605194 CRYPTIC, EGF-CFC FAMILY MEMBER 1; CFC1. Johns Hopkins University; last edited 07/10/2025. https://omim.org/entry/605194. Accessed 29 October 2026.
28. OMIM® Online Mendelian Inheritance in Man. *300265 ZIC FAMILY, MEMBER 3; ZIC3. Johns Hopkins University; last edited 05/01/2026. https://omim.org/entry/300265. Accessed 29 October 2026.
29. OMIM® Online Mendelian Inheritance in Man. *601568 ADDUCIN 3; ADD3. Johns Hopkins University; last edited 15/02/2024. https://omim.org/entry/601568. Accessed 29 October 2026.
30. OMIM® Online Mendelian Inheritance in Man. *600395 GLYPICAN 1; GPC1. Johns Hopkins University; last edited 21/01/2023. https://omim.org/entry/600395. Accessed 29 October 2026.
31. OMIM® Online Mendelian Inheritance in Man. *601548 EGF-CONTAINING FIBULIN-LIKE EXTRACELLULAR MATRIX PROTEIN 1; EFEMP1. Johns Hopkins University; last edited 06/05/2024. https://omim.org/entry/601548. Accessed 29 October 2026.
32. Orphanet. Biliary atresia and associated disorders (ORPHA:498345). https://www.orpha.net/en/disease/detail/498345
33. Orphanet. Syndromic biliary atresia (ORPHA:498350). https://www.orpha.net/en/disease/detail/498350
34. Orphanet. Martínez-Frías syndrome (ORPHA:137862). https://www.orpha.net/en/disease/detail/137862
35. Orphanet. Hypoplastic pancreas-intestinal atresia-hypoplastic gallbladder syndrome (ORPHA:293864). https://www.orpha.net/en/disease/detail/293864
36. Amendola M, Squires JE. Pediatric Genetic Cholestatic Liver Disease Overview. 2022 Sep 15 [revised 2023 May 25]. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–. https://www.ncbi.nlm.nih.gov/books/NBK584020/
37. Davenport M. Updates in Biliary Atresia: Aetiology, Diagnosis and Surgery. Children (Basel). 2025;12(1):95. doi:10.3390/children12010095. PMID: 39857926; PMC11764285.
38. Haute Autorité de Santé. Protocole National de Diagnostic et de Soins (PNDS): Atrésie des voies biliaires. Centre de référence de l'atrésie des voies biliaires et cholestases génétiques. Saint-Denis La Plaine: HAS; 2024 Dec 2 [text of November 2024; cited 2026 Nov 6]. Available from: https://www.has-sante.fr/jcms/p_3563486
39. Serinet MO, Broué P, Jacquemin E, Lachaux A, Sarles J, Gottrand F, et al. Management of patients with biliary atresia in France: results of a decentralized policy 1986-2002. Hepatology. 2006;44(1):75-84. doi:10.1002/hep.21219. PMID: 16799986.
40. Firth HV, Hurst JA. Prolonged neonatal jaundice and jaundice in infants below 6 months. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 288–290.
41. Saba TG, Geddes GC, Ware SM, Schidlow DN, Del Nido PJ, Rubalcava NS, et al. A multi-disciplinary, comprehensive approach to management of children with heterotaxy. Orphanet J Rare Dis. 2022;17(1):351. doi:10.1186/s13023-022-02515-2. PMID: 36085154; PMC9463860.
42. Fawaz R, Baumann U, Ekong U, Fischler B, Hadzic N, Mack CL, et al. Guideline for the Evaluation of Cholestatic Jaundice in Infants: Joint Recommendations of the North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition and the European Society for Pediatric Gastroenterology, Hepatology, and Nutrition. J Pediatr Gastroenterol Nutr. 2017;64(1):154-168. doi:10.1097/MPG.0000000000001334. PMID: 27429428.
43. Hoskins BJ, Pramparo T, Gough E, Ponte A, Dutta R, Karnsakul W. Clinical utility of an evolving cholestasis gene panel in 10,000 children and adults. Front Pediatr. 2026;14:1841006. doi:10.3389/fped.2026.1841006. PMID: 42232620; PMC13222966.
44. Bezerra JA, Spino C, Magee JC, Shneider BL, Rosenthal P, Wang KS, et al. Use of corticosteroids after hepatoportoenterostomy for bile drainage in infants with biliary atresia: the START randomized clinical trial. JAMA. 2014;311(17):1750-9. doi:10.1001/jama.2014.2623. PMID: 24794368; PMC4303045.
45. ClinicalTrials.gov. Evaluation of Maralixibat in Biliary Atresia Response Post-Kasai (EMBARK). NCT04524390 [results; last updated 2025 Mar 19; cited 2026 Oct 28]. Available from: https://clinicaltrials.gov/study/NCT04524390
46. Ipsen. Ipsen provides update on Phase III BOLD trial in biliary atresia [press release]. Paris: Ipsen; 2026 Jul 24 [cited 2026 Oct 28]. Available from: https://www.biospace.com/press-releases/ipsen-provides-update-on-phase-iii-bold-trial-in-biliary-atresia
47. Davenport M, Madadi-Sanjani O, Chardot C, Verkade HJ, Karpen SJ, Petersen C. Surgical and Medical Aspects of the Initial Treatment of Biliary Atresia: Position Paper. J Clin Med. 2022;11(21):6601. doi:10.3390/jcm11216601. PMID: 36362829; PMC9656543.
48. European Reference Network on Rare Hepatological Diseases (ERN RARE-LIVER) [cited 2026 Oct 28]. Available from: https://rare-liver.eu/
49. European Reference Network on Transplantation in Children (ERN TransplantChild) [cited 2026 Oct 28]. Available from: https://www.transplantchild.eu/
50. Spinner NB, Loomes KM, Krantz ID, Gilbert MA. Alagille Syndrome. 2000 May 19 [updated 2024 Jan 4]. In: GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–. https://www.ncbi.nlm.nih.gov/books/NBK1273/
51. Davenport M, Savage M, Mowat AP, Howard ER. Biliary atresia splenic malformation syndrome: an etiologic and prognostic subgroup. Surgery. 1993;113(6):662-8. PMID: 8506525.
52. Chen Y, Gilbert MA, Grochowski CM, McEldrew D, Llewellyn J, Waisbourd-Zinman O, et al. A genome-wide association study identifies a susceptibility locus for biliary atresia on 2p16.1 within the gene EFEMP1. PLoS Genet. 2018;14(8):e1007532. doi:10.1371/journal.pgen.1007532. PMID: 30102696; PMC6107291.
53. Glessner JT, Ningappa MB, Ngo KA, Zahid M, So J, Higgs BW, et al. Biliary atresia is associated with polygenic susceptibility in ciliogenesis and planar polarity effector genes. J Hepatol. 2023;79(6):1385-1395. doi:10.1016/j.jhep.2023.07.039. PMID: 37572794; PMC10729795.
54. Sok P, Sabo A, Almli LM, Jenkins MM, Nembhard WN, Agopian AJ, et al. Exome-wide assessment of isolated biliary atresia: A report from the National Birth Defects Prevention Study using child-parent trios and a case-control design to identify novel rare variants. Am J Med Genet A. 2023;191(6):1546-1556. doi:10.1002/ajmg.a.63185. PMID: 36942736; PMC10947986.
55. Ha JL, Ye Y, Liu H, Wu Z, Wai AW, Yan N, et al. Loss of function of Adducin 3 (ADD3) causes abnormal development and impaired barrier function of human and mouse bile duct cells resulting in increased incidence and severity of Biliary Atresia. EBioMedicine. 2025;122:106052. doi:10.1016/j.ebiom.2025.106052. PMID: 41297070; PMC12686720.
56. Chu CY, Tai CS, Hsu CT, Chang KC, Chen HL, Ni YH, et al. Loci Associated With Susceptibility to Biliary Atresia-A Genome-Wide Association Study in Taiwan. Pediatr Int. 2026;68(1):e70521. doi:10.1111/ped.70521. PMID: 42622066; PMC13491567.
57. Okubo R, Kuroda T, Abukawa D, Nio M, Suzuki T, Uchida H, et al. Clinical Practice Guidelines for Biliary Atresia in Japan, 2nd Edition. J Hepatobiliary Pancreat Sci. 2026;33(6):433-441. doi:10.1002/jhbp.70101. PMID: 41837546.
58. Jones KL, Jones MC, del Campo M. Trisomy 18 syndrome. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 8–13.
59. Jones KL, Jones MC, del Campo M. Kabuki syndrome. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 138–141.
60. Jones KL, Jones MC, del Campo M. Laterality sequences. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 866–869.
61. GeneGuidance. ADD3, EFEMP1, KIF3B, PKHD1L1, GPC1, AFAP1, MAN1A2, ARF6 — identifiers and clinical gene curation. Version of 2026-09-26 [accessed 2026-10-28].
62. Freeman PJ, Hart RK, Gretton LJ, Brookes AJ, Dalgleish R. VariantValidator: Accurate validation, mapping, and formatting of sequence variation descriptions. Hum Mutat. 2018;39(1):61-68. doi:10.1002/humu.23348. PMID: 28967166.
63. National Center for Biotechnology Information. ClinVar; VCV000235796, NM_138295.5(PKD1L1):c.6473+2_6473+3del [cited 2026 Oct 28]. Available from: https://www.ncbi.nlm.nih.gov/clinvar/variation/235796/
64. Firth HV, Hurst JA. Laterality disorders including heterotaxy and isomerism. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 206–207.
65. Parolini F, Hadzic N, Davenport M. Adjuvant therapy of cytomegalovirus IgM + ve associated biliary atresia: Prima facie evidence of effect. J Pediatr Surg. 2019;54(9):1941-1945. doi:10.1016/j.jpedsurg.2018.12.014. PMID: 30772005.
66. Jones KL, Jones MC, del Campo M. Cat eye syndrome. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 60–61.
67. Emanuel BS, Zackai EH, Medne L. Emanuel Syndrome. 2007 Apr 20 [updated 2017 Aug 31]. In: GeneReviews® [Internet]. https://www.ncbi.nlm.nih.gov/books/NBK1263/
68. Nisbet AF, Hathaway ER, Kalish JM. Simpson-Golabi-Behmel Syndrome Type 1. 2006 Dec 19 [updated 2024 Nov 14]. In: GeneReviews® [Internet]. https://www.ncbi.nlm.nih.gov/books/NBK1219/
69. Coman D, Kranc KR, Christodoulou J. Fumarate Hydratase Deficiency. 2006 Jul 5 [updated 2026 Jun 9]. In: GeneReviews® [Internet]. https://www.ncbi.nlm.nih.gov/books/NBK1506/
70. Mitchel MW, Moreno-De-Luca D, Myers SM, Levy RV, Turner S, Ledbetter DH, Martin CL. 17q12 Recurrent Deletion Syndrome. 2016 Dec 8 [updated 2025 Aug 14]. In: GeneReviews® [Internet]. https://www.ncbi.nlm.nih.gov/books/NBK401562/
71. De León DD, Pinney SE. Permanent Neonatal Diabetes Mellitus. 2008 Feb 8 [updated 2025 Sep 25]. In: GeneReviews® [Internet]. https://www.ncbi.nlm.nih.gov/books/NBK1447/
72. Masui D, Fukahori S, Mizuochi T, Watanabe Y, Fukui K, Ishii S, et al. Cystic biliary atresia with paucity of bile ducts and gene mutation in KDM6A: a case report. Surg Case Rep. 2019;5(1):132. doi:10.1186/s40792-019-0688-4. PMID: 31414320; PMC6694366.
73. Chen HL, Taylor SA, Lee WS, Ciocca M, El-Guindi MA, Yachha SK, et al. Diagnostic approaches for infants with cholestatic liver diseases: Position paper and perspectives of the Federation of International Societies of Pediatric Gastroenterology, Hepatology, and Nutrition. J Pediatr Gastroenterol Nutr. 2025;81(5):1360-1377. doi:10.1002/jpn3.70207. PMID: 40977420.
74. Sociedade Brasileira de Pediatria. Alerta Amarelo: ações da SBP ajudam na prevenção, diagnóstico precoce e tratamento da colestase em lactentes. 2017 Sep 19 [cited 2026 Oct 28]. Available from: https://www.sbp.com.br/alerta-amarelo-acoes-da-sbp-ajudam-na-prevencao-diagnostico-precoce-e-tratamento-da-colestase-em-lactentes/
75. Ashworth J, Tavares M, Santos Silva E, Lopes AI. The Stool Color Card as a Screening Tool for Biliary Atresia in the Digital Version of the Portuguese Child and Youth Health Booklet. Acta Med Port. 2021;34(9):632-633. doi:10.20344/amp.16679. PMID: 34274031.
76. Harpavat S, Garcia-Prats JA, Anaya C, Brandt ML, Lupo PJ, Finegold MJ, et al. Diagnostic Yield of Newborn Screening for Biliary Atresia Using Direct or Conjugated Bilirubin Measurements. JAMA. 2020;323(12):1141-1150. doi:10.1001/jama.2020.0837. PMID: 32207797; PMC7093763.
77. Ranucci G, Della Corte C, Alberti D, Bondioni MP, Boroni G, Calvo PL, et al. Diagnostic approach to neonatal and infantile cholestasis: A position paper by the SIGENP liver disease working group. Dig Liver Dis. 2022;54(1):40-53. doi:10.1016/j.dld.2021.09.011. PMID: 34688573.
78. Mercadal-Hally M, Loverdos I, Reyes-Andrade J, Martín SHC, Álvare AM, Vegasa-Álvarez AM, et al. Assessment and Management of Cholestasis: Consensus Document of the Spanish Society of Pediatric Gastroenterology, Hepatology and Nutrition (SEGHNP), the Spanish Association of Primary Care Pediatrics (AEPap) and the Spanish Society of Primary Care Pediatrics (SEPEAP). An Pediatr (Engl Ed). 2025;103(6):504034. doi:10.1016/j.anpede.2025.504034. PMID: 41350188.
79. Ando H, Inomata Y, Iwanaka T, Kuroda T, Nio M, Matsui A, et al. Clinical practice guidelines for biliary atresia in Japan: A secondary publication of the abbreviated version translated into English. J Hepatobiliary Pancreat Sci. 2021;28(1):55-61. doi:10.1002/jhbp.816. PMID: 32780928.
80. Bull LN, Morotti R, Squires JE. ATP8B1 Deficiency. 2001 Oct 15 [updated 2021 Sep 9]. In: GeneReviews® [Internet]. https://www.ncbi.nlm.nih.gov/books/NBK1297/
81. Zariwala MA, Despotes KA, Davis SD. Primary Ciliary Dyskinesia. 2007 Jan 24 [updated 2025 May 22]. In: GeneReviews® [Internet]. https://www.ncbi.nlm.nih.gov/books/NBK1122/
82. Nyhan WL, Hoffmann GF, Al-Aqeel AI, Barshop BA. Alpha-antitrypsin deficiency. In: Atlas of Inherited Metabolic Diseases. 4th ed. Boca Raton: CRC Press; 2020. p. 807–814.
83. Savant A, Bojanowski CM, Upadia J. Cystic Fibrosis. 2001 Mar 26 [updated 2026 Oct 1]. In: GeneReviews® [Internet]. https://www.ncbi.nlm.nih.gov/books/NBK1250/
84. Song YZ, Oishi K, Saheki T. Citrin Deficiency. 2005 Sep 16 [updated 2025 Mar 20]. In: GeneReviews® [Internet]. https://www.ncbi.nlm.nih.gov/books/NBK1181/
85. Bremova-Ertl T, Patterson M. Niemann-Pick Disease Type C. 2000 Jan 26 [updated 2025 Nov 20]. In: GeneReviews® [Internet]. https://www.ncbi.nlm.nih.gov/books/NBK1296/
86. Stoller JK, Hupertz V, Aboussouan LS. Alpha-1 Antitrypsin Deficiency. 2006 Oct 27 [updated 2023 Jun 1]. In: GeneReviews® [Internet]. https://www.ncbi.nlm.nih.gov/books/NBK1519/
87. El-Hattab AW, Scaglia F. Deoxyguanosine Kinase Deficiency. 2009 Jun 18 [updated 2023 Feb 9]. In: GeneReviews® [Internet]. https://www.ncbi.nlm.nih.gov/books/NBK7040/
88. Haute Autorité de Santé. Protocole National de Diagnostic et de Soins (PNDS): Déficits de synthèse des acides biliaires primaires (déficit en 3β-hydroxy-C27-stéroïde déshydrogénase/isomérase et en Δ4-3-oxostéroïde 5β-réductase). Centre de référence de l'atrésie des voies biliaires et cholestases génétiques. Saint-Denis La Plaine: HAS; 2019 Nov 27 [text of October 2019; cited 2026 Nov 6]. Available from: https://www.has-sante.fr/jcms/p_3121131
89. Jones KL, Jones MC, del Campo M. Alagille syndrome. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 822–825.
90. Ho SKL, Leung LT, Luk HM, Lo IFM. Strømme Syndrome. 2022 Nov 10. In: GeneReviews® [Internet]. https://www.ncbi.nlm.nih.gov/books/NBK586170/
91. THERAVIA. Orphacol (cholic acid) 50 mg and 250 mg hard capsules: summary of product characteristics (EPAR product information). Amsterdam: EMA; first authorisation 2013 Sep 12, renewal 2019 Apr 24 [cited 2026 Nov 6]. Available from: https://www.ema.europa.eu/en/documents/product-information/orphacol-epar-product-information_en.pdf
92. U.S. Food and Drug Administration. CHOLBAM (cholic acid) capsules prescribing information. Mirum Pharmaceuticals; NDA 205750, initial approval 2015 Mar 17; revised 2026-01-26. Available from: https://api.fda.gov/drug/label.json?search=openfda.brand_name:%22CHOLBAM%22
93. Utz Melere M, Sanha V, Farina M, da Silva CS, Nader L, Trein C, et al. Primary liver transplantation vs transplant after Kasai portoenterostomy in children with biliary atresia: A retrospective Brazilian single-center cohort. World J Transplant. 2024;14(1):88734. doi:10.5500/wjt.v14.i1.88734. PMID: 38576759; PMC10989469.
94. Davenport M, Parsons C, Tizzard S, Hadzic N. Steroids in biliary atresia: single surgeon, single centre, prospective study. J Hepatol. 2013;59(5):1054-8. doi:10.1016/j.jhep.2013.06.012. PMID: 23811305.
95. Dr. Falk Pharma UK. Ursofalk 250 mg/5 ml Suspension — Summary of Product Characteristics [text revised 11/2024; updated on eMC 2025 Feb 3; cited 2026 Oct 29]. Available from: https://www.medicines.org.uk/emc/product/146/smpc
96. Dr. Falk Pharma Portugal, Sociedade Unipessoal, Lda. Ursofalk 250 mg cápsulas e Ursofalk 500 mg comprimidos revestidos por película (ácido ursodesoxicólico): resumo das características do medicamento [text revised: November 2024]. INFARMED, Infomed (med_id 8950 e 582526). Available from: https://extranet.infarmed.pt/INFOMED-fo/download-ficheiro.xhtml?med_id=8950&tipo_doc=rcm. Accessed 29 October 2026.
97. U.S. Food and Drug Administration. VALCYTE (valganciclovir) prescribing information. Genentech; revised 2025-12-08 (set id 4c517a39-2ded-4c5a-8d56-276853414b31). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=4c517a39-2ded-4c5a-8d56-276853414b31
98. Mack CL, Spino C, Alonso EM, Bezerra JA, Moore J, Goodhue C, et al. A Phase I/IIa Trial of Intravenous Immunoglobulin Following Portoenterostomy in Biliary Atresia. J Pediatr Gastroenterol Nutr. 2019;68(4):495-501. doi:10.1097/MPG.0000000000002256. PMID: 30664564; PMC6428610.
99. ClinicalTrials.gov. Search of interventional trials in biliary atresia updated since 2024-06-01 [cited 2026 Oct 28]. Available from: https://clinicaltrials.gov/
100. U.S. Food and Drug Administration. ZOSYN (piperacillin and tazobactam) in Galaxy containers prescribing information. Wyeth Pharmaceuticals (Pfizer); revised 2023-03-15 (set id 8ab9d90d-2f7f-4599-a44f-3d0a8f8ab623). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=8ab9d90d-2f7f-4599-a44f-3d0a8f8ab623
101. U.S. Food and Drug Administration. Amikacin sulfate injection prescribing information. Hikma Pharmaceuticals USA; revised 2024-04-08 (set id 0b56f6df-a05d-4520-8bf0-d7cefe20f6ad). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=0b56f6df-a05d-4520-8bf0-d7cefe20f6ad
102. U.S. Food and Drug Administration. Ciprofloxacin tablets prescribing information. Umasuto; revised 2025-03-06 (set id 0c355077-7361-41d0-9e82-eb31beaf5daa). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=0c355077-7361-41d0-9e82-eb31beaf5daa
103. European Medicines Agency. Vedrop 50 mg/ml oral solution (tocofersolan): EPAR, Annex I — summary of product characteristics [first authorisation 2009 Jul 24; renewal 2014 Apr 23; cited 2026 Nov 2]. Available from: https://www.ema.europa.eu/en/medicines/human/EPAR/vedrop
104. U.S. Food and Drug Administration. BYLVAY (odevixibat) prescribing information. Ipsen; revised 2025-12-29 (set id 0935cd31-3251-4a0a-ac23-25ca4be12be9). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=0935cd31-3251-4a0a-ac23-25ca4be12be9
105. U.S. Food and Drug Administration. LIVMARLI (maralixibat) prescribing information. Mirum; revised 2025-04-10 (set id 64000394-1ef6-4e76-8ba8-11f25ba1b167). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=64000394-1ef6-4e76-8ba8-11f25ba1b167
106. U.S. Food and Drug Administration. RIFAMPIN prescribing information. Epic Pharma; revised 2023-12-20 (set id 0391f1e1-70d4-4b15-8b58-055bb6a385b8). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=0391f1e1-70d4-4b15-8b58-055bb6a385b8
107. U.S. Food and Drug Administration. Amoxicillin prescribing information. A-S Medication Solutions; revised 2023-12-18 (set id 00fbd46e-05fd-4f8a-9f59-a7a4d01c8e54). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=00fbd46e-05fd-4f8a-9f59-a7a4d01c8e54
108. Alhashmi H, Chawshly E, Çelebi S. Biliary atresia-splenic malformation (BASM) syndrome: A case report. Int J Surg Case Rep. 2024;121:109937. doi:10.1016/j.ijscr.2024.109937. PMID: 38909387; PMC11254726.
109. ClinicalTrials.gov. Efficacy and Safety of Odevixibat in Children With Biliary Atresia Who Have Undergone a Kasai HPE (BOLD). NCT04336722 [last updated 2026 Jul 2; cited 2026 Oct 28]. Available from: https://clinicaltrials.gov/study/NCT04336722
110. HCPLive. Odevixibat fails to improve native liver survival in phase 3 BOLD trial for biliary atresia. 2026 Jul 28 [cited 2026 Oct 28]. Available from: https://www.hcplive.com/view/odevixibat-fails-to-improve-native-liver-survival-in-phase-3-bold-trial-for-biliary-atresia
111. ClinicalTrials.gov. Obeticholic Acid in Pediatric Subjects With Biliary Atresia (CARE), NCT05321524; and Study to Assess ... Obeticholic Acid ... in Pediatric Participants With Biliary Atresia, NCT06121375 [cited 2026 Oct 28]. Available from: https://clinicaltrials.gov/study/NCT05321524 ; https://clinicaltrials.gov/study/NCT06121375
112. European Medicines Agency. Bylvay (odevixibat): EPAR [page updated 2026 May 8; cited 2026 Oct 28]. Available from: https://www.ema.europa.eu/en/medicines/human/EPAR/bylvay
113. European Medicines Agency. Livmarli (maralixibat): EPAR [page updated 2026 Apr 17; cited 2026 Oct 28]. Available from: https://www.ema.europa.eu/en/medicines/human/EPAR/livmarli
114. Ukarapol N, Poddar U, Lui TYS, Lee YJ, Lopez RN, Prachuapthunyachart S, et al. APSPGHAN Statement on Endoscopic Variceal Bleeding Hemostasis. Pediatr Gastroenterol Hepatol Nutr. 2026;29(1):1-21. doi:10.5223/pghn.2026.29.1.1. PMID: 41567251; PMC12816376.
115. Langreen S, Pauer N, Pfister ED, Junge N, Baumann U, Madadi-Sanjani O, et al. Rectal budesonide: A potential game changer after Kasai hepatoportoenterostomy. J Pediatr Gastroenterol Nutr. 2025;81(3):626-633. doi:10.1002/jpn3.70147. PMID: 40600375; PMC12408950.
116. U.S. Food and Drug Administration. Esomeprazole sodium for injection prescribing information. Slate Run Pharmaceuticals; revised 2026-06-22 (set id 23395673-116d-4a8e-b0ce-59f9ca5b3daf). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=23395673-116d-4a8e-b0ce-59f9ca5b3daf
117. U.S. Food and Drug Administration. Lactulose solution prescribing information. Kesin Pharma; revised 2026-08-03 (set id 163d5093-e35c-0bbf-e063-6294a90a803e). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=163d5093-e35c-0bbf-e063-6294a90a803e
118. Joshi D, Nayagam J, Clay L, Yerlett J, Claridge L, Day J, et al. UK guideline on the transition and management of childhood liver diseases in adulthood. Aliment Pharmacol Ther. 2024;59(7):812-842. doi:10.1111/apt.17904. PMID: 38385884.
119. Hiratsuka K, Watayo H, Nakamura N. Development and clinical implementation of a care guideline to support self-management and responsibility transition in adolescents and young adults with biliary atresia. Health Care Transit. 2026;4:100150. doi:10.1016/j.hctj.2026.100150. PMID: 42569562; PMC13450250.
120. Chardot C, Carton M, Spire-Bendelac N, Le Pommelet C, Golmard JL, Auvert B. Epidemiology of biliary atresia in France: a national study 1986-96. J Hepatol. 1999;31(6):1006-13. doi:10.1016/s0168-8278(99)80312-2. PMID: 10604573.
121. Carvalho Ed, Santos JL, Silveira TR, Kieling CO, Silva LR, Porta G, et al. Biliary atresia: the Brazilian experience. J Pediatr (Rio J). 2010;86(6):473-9. doi:10.2223/JPED.2054. PMID: 21140036.
122. Childhood Liver Disease Research Network (ChiLDReN) [cited 2026 Oct 28]. Available from: https://childrennetwork.org/
123. Pfizer Limited. Aldactone 25 mg Film-Coated Tablets — Summary of Product Characteristics [text revised 01/2026; updated on eMC 2026 Feb 2; cited 2026 Nov 2]. Available from: https://www.medicines.org.uk/emc/product/1619/smpc
124. U.S. Food and Drug Administration. Furosemide tablets prescribing information. Bryant Ranch Prepack; revised 2023-03-01 (set id 01a5f094-b473-4e46-9e61-69d5ec6dd766). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=01a5f094-b473-4e46-9e61-69d5ec6dd766
125. U.S. Food and Drug Administration. ALBUMINEX (albumin human) prescribing information. BPL; revised 2024-12-31 (set id 167f48d0-8ae5-4f7e-b050-9673dad78acc). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=167f48d0-8ae5-4f7e-b050-9673dad78acc