Clinical spectrum and genotype-phenotype correlation of ABCB4 mutations in children: Insights from a North Indian cohort

Abstract

BACKGROUND

Progressive familial intrahepatic cholestasis type 3, caused by mutations in the ABCB4 gene, is a rare genetic disorder. Although severe phenotypes due to biallelic mutations are well described, emerging data seem to suggest the clinical relevance of monoallelic variants.

AIM

To describe the clinical spectrum and genotype-phenotype correlation of ABCB4 mutations in children in a cohort of North Indian children.

METHODS

This is a retrospective analysis of a prospectively maintained database from a single tertiary care centre. Children (≤ 18 years) with ABCB4 mutations between January 2021 and March 2025 were analysed. The clinical presentation, laboratory investigations, genetic sequencing and outcomes were recorded. Patients were stratified into group 1 (homozygous/compound heterozygous) and group 2 (heterozygous). Variant pathogenicity was assessed using the American College of Medical Genetics guidelines and available predictive tools.

RESULTS

Of the 26 patients, 16 had biallelic mutations, and 10 had monoallelic mutations. Group 1 exhibited higher rates of positive family history (75% vs 30%, P = 0.04), ascites (43.2% vs 0%, P = 0.02), larger varices (40% vs 0%, P = 0.009), higher gamma glutamyl transferase levels (171 U/L vs 38 U/L, P = 0.007), and lower platelet counts (162 × 10⁹/L vs 415 × 10⁹/L, P = 0.007). Notably, two-thirds of patients in group 1 experienced disease progression, and one-third died during follow-up. Certain missense variants (e.g., c.2860T>C) and all nonsense variants were linked to rapid deterioration. Most children in group 2 had transient cholestasis with a good outcome, but two older children succumbed.

CONCLUSION

Mutations in the ABCB4 gene contribute significantly to pediatric chronic liver disease. Patients with severe biallelic mutations frequently experience a progressive disease course, whereas those with monoallelic mutations may progress slowly. Genetic testing for ABCB4 should be considered in children with cryptogenic chronic liver disease, especially those with high gamma-glutamyltransferase cholestasis and portal hypertension.

Key Words: ABCB4; Monoallelic mutation; Biallelic mutation; Progressive familial intrahepatic cholestasis type 3; Chronic liver disease; Decompensation; End-stage liver disease; Portal hypertension; Sclerosing cholangitis

Core Tip: Patients with ABCB4 mutations continue to pose an enigma for clinicians, and understanding of the varied manifestations continues to evolve. Children with biallelic mutations had a progressive disease with ascites, larger varices and decompensation. Children with monoallelic mutations fare better with significantly better liver function in follow-up; however, the risk of disease progression and decompensation remains and needs close monitoring. Protein-truncating mutations and missense mutations in highly conserved domains are associated with early disease progression. Early genetic testing and family screening may help in tailoring treatment and prognostication.

INTRODUCTION

Progressive familial intrahepatic cholestasis (PFIC) is a rare group of genetic cholestatic disorders, characterised by defective bile secretion. They present as infantile or childhood onset cholestasis, often resulting in end-stage liver disease (ESLD) in infancy or adulthood. The availability of accessible, high-throughput next-generation sequencing (NGS) has led to the discovery of several new PFIC subtypes[1,2]. PFIC type 3 (PFIC3) is one such disorder caused by a mutation in the ABCB4 gene. ABCB4 has two transmembrane domains and two cytoplasmic nucleotide-binding domains (NBDs), containing conserved sequences[3]. The ABCB4 gene encodes the multidrug resistance protein 3 (MDR3), which mediates the translocation of phosphatidylcholine to the outer leaflet of the bile canalicular membrane. This helps in neutralising the detergent and lithogenic properties of bile[4]. In the absence of this mechanism, toxic bile damages the cholangiocytes, leading eventually to cirrhosis[5]. Deleuze et al[6] were the first to demonstrate absent MDR3 mRNA expression via Northern blot analysis in two patients with high gamma-glutamyltransferase (GGT) cholestasis. Given the variable age at clinical presentation, overlap with other common paediatric chronic liver disease (CLD) and an inconsistent genotype-phenotype correlation, the ABCB4 mutation disorders remain a diagnostically and therapeutically challenging form of inherited liver disease.

With the availability of NGS, diagnosis of PFIC is often done in patients with liver diseases, but this has revealed a spectrum of phenotypic heterogeneity, with presentations that can closely mimic other common causes of CLD. Notably, studies have reported classical features of PFIC3 even in individuals harbouring heterozygous (HET) ABCB4 mutations, often affecting multiple family members, thereby complicating diagnostic and management strategies[7]. Hence, in this study, we aim to describe the clinical spectrum and genotype–phenotype correlation of ABCB4 mutations in children in a large cohort of North Indian children.

MATERIALS AND METHODS

Study population

This study was conducted in the Department of Paediatric Gastroenterology and Hepatology at one of the India's largest Paediatric tertiary care centre. We did a retrospective analysis of prospectively well-maintained outpatient and inpatient data on children (≤ 18 years) with genetically confirmed ABCB4 mutations, including homozygous, compound HET, and HET variants, diagnosed between January 2021 and March 2025. Children (≤ 18 years) with clinically suspected cholestatic or cryptogenic CLD in whom genetic testing identified an ABCB4 variant were included, while those with other causes of acute and CLDs [viral hepatitis, Wilson’s disease (WD), autoimmune hepatitis (AIH), other etiologies of neonatal/infantile cholestasis] were excluded.

Decompensation/ESLD was defined as the presence of ascites, hepatic encephalopathy, and/or gastrointestinal bleeding. Oesophageal varices were classified as small or large according to the Baveno VI consensus guidelines[8]. Disease status at last follow-up was categorized as: (1) Improved: Resolution of jaundice and amelioration of portal hypertension and biochemistries; (2) Stable: No further deterioration in liver function tests (LFTs) or coagulation parameters; and (3) Worsened: Development of new signs or symptoms of portal hypertension, uncorrectable coagulopathy, death, or requirement for liver transplantation (LT) at the most recent follow-up.

Investigations and management

As per unit protocol, patients underwent relevant investigations at the time of admission, including a complete blood count, LFT with GGT, serum bile acids, and a coagulation profile. Investigations, such as LFT and coagulation parameters, were repeated every three months during the follow-up. Monthly follow-ups were done in decompensated patients. As part of the aetiological evaluation for CLD, additional investigations – such as serum ceruloplasmin, 24-hour urinary copper, immunoglobulin G levels, and autoantibody panels for AIH were performed as indicated. When feasible, a liver biopsy and immunohistochemistry (IHC) for MDR3 expression were performed. Liver histopathology was analysed by a single observer specialising in pediatric liver histopathology and was blinded to genetic diagnosis. The fibrosis stage, ductular reaction, biliary metaplasia, lobular inflammation, and the presence of ductular and canalicular cholestasis, as well as giant cell transformation, were assessed (Supplementary Tables 1 and 2)[9-11]. Whole-exome sequencing was performed on all suspected patients during the study period (Supplementary Table 3). Parental testing for carrier state and screening asymptomatic siblings was not performed due to financial constraints. Supportive treatment was provided to all patients with fat- and water-soluble vitamin supplementation, management of cholestasis and pruritus through sequential addition of ursodeoxycholic acid (UDCA), Rifampicin, and opioid antagonist Naltrexone, if indicated. Complications such as ascites, variceal bleeding and encephalopathy were managed as per standard protocol. LT was offered in those with worsening Pediatric End-Stage Liver Disease scores.

Clinical details, laboratory investigations, and other relevant data were recorded using a structured proforma. Cases were stratified based on mutation status, into homozygous/compound HET and HET groups, which were subsequently analyzed and compared. Institutional ethical clearance was obtained prior to the commencement of the study (No. IEC-INT/2023/DM-1452).

Statistical analysis

Descriptive statistics were calculated, with data presented as means with standard deviations, medians with interquartile ranges, or as %, as appropriate. Comparisons between patients with homozygous/compound HET and HET ABCB4 mutations, as well as between those with and without disease progression, were performed using the χ² test or Fisher’s exact test for categorical variables, and the non-parametric Mann-Whitney U test for continuous variables. Cases with missing information were retained in the analysis when the absent variables had no impact on the assessment of study outcomes. All statistical analyses were conducted using Statistical Package for the Social Sciences version 27, with a two-tailed P < 0.05 considered statistically significant.

RESULTS

A total of 26 patients (61.5% male) with ABCB4 mutations were included during the study period: (1) 14 with homozygous; (2) 2 with compound HET; and (3) 10 with HET variants. For analysis, patients with homozygous and compound HET mutations were grouped as group 1, and those with HET mutations were grouped as group 2.

Homozygous/compound HET mutations (group 1)

This group comprised 16 children (11, 68.75% male); 14 were homozygous and 2 were compound HET. The median age at symptom onset and presentation was 54 months (range: 4-168 months), and 60 months (range: 4-172 months), respectively. Consanguinity was documented in 37.5% of cases, while 75% had a family history of liver disease and/or gallstone disease. At presentation, 7 patients (43.8%) had clinical signs of liver decompensation. Fifteen patients underwent upper gastrointestinal endoscopy; large esophageal varices were identified in 6 (40%). Additional presenting features are summarised in Tables 1 and 2.

Table 1 Comparison of clinical, laboratory and genetic parameters between bi-allelic and monoallelic ABCB4 mutations, median (interquartile range).

Parameters	Homozygous/compound HET mutation (n = 16)	HET mutation (n = 10)	P value
General characteristics
Male gender	62.5%	60%	1.00
Median age at symptom onset in months	54 (4, 168)	10.5 (5, 72)	0.40
Median age at presentation in months	60 (4, 172)	18 (11, 126)	0.42
Consanguinity	37.5%	10%	0.19
Positive family history	75%	30%	0.04
Sibling loss	43.8%	33.3%	0.69
Anthropometry
Weight Z-score at admission	-1.09 (-2.03, -0.21)	-1.39 (-2.61, 0.10)	0.97
Height Z-score at admission	-0.91 (-2.72, -0.52)	-2.1 (-3.33, 0.03)	0.57
Weight Z-score at last follow up	-0.42 (-1.29, -0.125)	-0.58 (-1.63, 1.31)	0.90
Height Z-score at last follow up	-0.37(-1.83, 0.33)	-0.76 (-2.23, 0.91)	0.86
Clinical parameters at admission
Jaundice	68.8%	90%	0.35
Pruritus	68.8%	80%	0.66
Ascites	43.2%	0%	0.02
Gastrointestinal bleed	12.5%	0%	0.5
Decompensation at admission	43.8%	0%	0.02
Esophageal varix			0.009
No varix	20%	87.5%
Small varix	40%	12.5%
Large varix	40%	0
Genetics
Missense variants	81.3%	80%	1.00
Recent American College of Medical Genetics class likely pathogenic or pathogenic	100%	30%	< 0.001
Laboratory parameters
First visit
Platelets (× 109/L)	162 (64, 230)	415 (167, 577)	0.007
TB (mg/dL)	2.58 (1.6, 6.7)	3.84 (1.59, 8.27)	0.56
DB (mg/dL)	1.3 (0.71, 4.1)	2.15 (1.08, 4.14)	0.59
AST (U/L)	198 (116, 287)	124.5 (90.75, 238)	0.29
ALT (U/L)	120 (64, 201)	64.5 (37.25, 118.25)	0.06
ALP (U/L)	318 (280, 550)	296 (205.25, 384)	0.15
GGT (U/L)	171 (94, 450)	38 (17, 148.5)	0.007
Prothrombin time	76 (57.5, 100)	100 (88, 100)	0.12
INR	1.27 (1, 1.58)	1.04 (0.97, 1.11)	0.10
Bile acids (µmol/L)	139.5 (33.4, 174.2)	206 (80, 335)	0.24
At last visit
TB (mg/dL)	3.84 (0.94, 15.09)	0.42 (0.28, 2.65)	0.003
DB (mg/dL)	1.87 (0.33, 8.36)	0.09 (0.09, 1.9)	0.007
AST (U/L)	247 (106.7, 462.2)	41 (35.2, 110.2)	0.004
ALT (U/L)	131 (53.5, 140.5)	33 (25.4, 55.5)	0.003
ALP (U/L)	329 (176, 719)	265 (169.5, 461)	0.61
GGT (U/L)	146 (95, 310)	11 (10.25, 13.5)	0.003
INR	1.20 (1.08, 1.69)	1.02 (1, 1.16)	0.01
Total duration of follow up in months	28.5 (3.7, 59)	23 (11.7, 60)	0.61
Outcomes	31.3%	20%	0.66
Progressive disease (death, decompensated disease, uncorrectable coagulopathy at last follow-up)	68.8%	20%	0.01
Mortality	31.3%	20%	0.34

Mann-Whitney U test was applied for non-parametric quantitative variables, and χ²/Fisher's exact test was applied for qualitative variables. ALP: Alkaline phosphatase; AST: Aspartate aminotransferse; ALT: Alanine aminotransferse; DB: Direct bilirubin; GGT: Gamma-glutamyltransferase; HET: Heterozygous; INR: International normalised ratio; TB: Total bilirubin.

Table 2 Clinical characteristics of patients with biallelic ABCB4 mutations (n = 16).

Patient	Gender	Age at symptom onset (months)	Age at first presentation (months)	Consanguinity	Family history	Anthropometry (weight/height)	Clinical features	Disease status at presentation	Esophageal varix	Type of ABCB4 mutation	Disease status at last follow-up	Issues at last follow up	Final outcome at last follow-up
P1	Male	22	24	Yes	Yes	-0.34Z, -0.55Z	HSM	C	Small varix	Homozygous	S	PH	Alive
P2 (sibling of P1)	Male	6	60	Yes	Yes	0.21Z, -0.52Z	J, P, HSM	C	Large varix	Homozygous	S	PH	Alive
P3	Male	6	9	No	Yes	NA	HSM, P	C	No varix	Homozygous	W	PH, INR > 1.5	Alive
P4	Male	6	24	No	No	-1.42Z, -0.87Z	P, HSM	C	No varix	Homozygous	S		Alive
P5	Male	96	96	No	Yes	-0.17Z, 0.09Z	A, HSM	D	Large varix	Homozygous	S	PH, recurrent acute kidney injury and nephrotic range proteinuria	Alive
P6 (sibling of P5)	Female	168	172	No	Yes	-0.75Z, -1.11Z	J, P, A, HSM, E, B	D	Large varix	Homozygous	W	PH, INR > 1.5	Expired post LT (primary graft non-function)
P7	Male	120	120	Yes	Yes	-3.59Z, -2.70Z	J, P, A, HSM	C	Large varix	Homozygous	W	PH, INR > 1.5, E	Expired after 3 years of (LT (acute rejection)
P8	Female	58	60	No	No	0.06Z, -0.72Z	J, P, HSM, A, B	D	Small varix	Homozygous	W	INR > 1.5	Alive
P9	Female	6	48	No	Yes	-2.9Z, -2.72Z	J, P, HSM	C	Small varix	Homozygous	W	PH, E	Expired
P10 (sibling of P9)	Male	12	34	No	Yes	-2.5Z, -3.14Z	J, P, HSM	C	Large varix	Homozygous	W	PH	Alive
P11	Female	84	108	Yes	Yes	-0.02Z, -0.91Z	J, P, A, HSM	D	Small varix	Homozygous	W	PH, INR > 1.5, MRCP-sclerosing cholangitis	Expired
P12	Female	120	144	No	Yes	NA	J, HSM	C	Small varix	Homozygous	W	PH, B, INR > 1.5	Alive
P13	Male	60	60	Yes	Yes	-1.36Z, -3.46Z	J, A, HSM	D	Small varix	Homozygous	S		Alive
P14	Male	80	83	No	No	-0.77, 0.35	J, P	C	Large varix	Homozygous	W	PH, INR > 1.5	Alive
P15	Male	4	4	No	Yes	NA	J, HSM	C	Not done	Compound HET	W	PH, INR > 1.5, E, acute-on-chronic liver failure	Expired
P16	Male	50	52	No	No	-1.94Z, -3.07Z	P, HSM, A	D	No varix	Compound HET	S	Nephrotic range proteinuria (carries additional X-linked FLNA mutation)	Alive

Small and large varix as per Baveno VI guidelines. P3: Progression of portal hypertension (gastric varix), international normalised ratio > 1.5. P8, P13 and P15: Short follow-up (< 6 months). A: Ascites; B: Gastrointestinal bleed; C: Compensated; D: Decompensated; E: Encephalopathy; H: Hepatomegaly; HET: Heterozygous; HSM: Hepatosplenomegaly; INR: International normalised ratio; J: Jaundice; LT: Liver transplantation; NA: Not available; P: Pruritis; PH: Portal Hypertension; S: Stable; W: Worsened.

Clinical details of individual patients with bi-allelic mutations are presented in Table 2, and their genetic mutations, along with their pathogenicity, are listed in Table 3. Among the 16 patients with bi-allelic mutations, a total of 18 variants were identified, of which 81.3% were missense mutations (Table 3 and Figure 1).

Figure 1 Schematic representation of ABCB4 variants, exons and domains in the present cohort. NBD: Nucleotide binding domain; TMD: Transmembrane domain.

Table 3 ABCB4 variants in biallelic mutation and their pathogenicity (n = 16).

Patient	Zygosity	Exon number	Mutation	Predicted effect	Domain	Type of mutation	Polyphen	SIFT	Mutation taster	GnomAD1	Updated ACMG class	ACMG criteria
P1	Ho	23	c.2908T>C	p.Phe970Leu	NBD2	MISSENSE	B	D	DC	0	LP	PM1, PM2, PP2
P2 (sibling of P1)	Ho	23	c.2908T>C	p.Phe970Leu	NBD2	MISSENSE	B	D	DC	0	LP	PM1, PM2, PP2
P3	Ho	23	c.2908T>C	p.Phe970Leu	NBD2	MISSENSE	B	D	DC	0	LP	PM1, PM2, PP2
P4	Ho	23	c.2908T>C	p.Phe970Leu	NBD2	MISSENSE	B	D	DC	0	LP	PM1, PM2, PP2
P5	Ho	23	c.2860G>A	p.Gly954Ser	Linker (NBD2 adjacent)	MISSENSE	PossD	D	DC	0	LP	PM1, PM2, PM5, PP2, PP3, PP5
P6 (sibling of P5)	Ho	23	c.2860G>A	p.Gly954Ser	Linker (NBD2 adjacent)	MISSENSE	PossD	D	DC	0	LP	PM1, PM2, PM5, PP2, PP3, PP5
P7	Ho	23	c.2860G>A	p.Gly954Ser	Linker (NBD2 adjacent)	MISSENSE	PossD	D	DC	0	LP	PM1, PM2, PM5, PP2, PP3, PP5
P8	Ho	23	c.2860G>A	p.Gly954Ser	Linker (NBD2 adjacent)	MISSENSE	PossD	D	DC	0	LP	PM1, PM2, PM5, PP2, PP3, PP5
P9	Ho	4	c.139C>T	p.Arg47Ter	TMD1	NONSENSE	-	-	DC	0	P	PVS1, PS4, PM2
P10 (sibling of P9)	Ho	4	c.139C>T	p.Arg47Ter	TMD1	NONSENSE	-	-	DC	0	P	PVS1, PS4, PM2
P11	Ho	25	c.3230C>T	p.Thr1077Met	NBD2	MISSENSE	ProbD	D	DC	0.005	LP	PP2, PP3, PM3
P12	Ho	28	c.3760G>A	p.Gly1254Ser	NBD2	MISSENSE	ProbD	D	DC	0	LP	PP2, PP3, PM3
P13	Ho	15	c.1783C>T	p.Arg595Ter	NBD2 start	NONSENSE	D	D	DC	0.001	P	PVS1, PM2, PP5
P14	Ho	6	c.431G>A	p.Arg144Gln	-	MISSENSE	ProbD	D	DC	0.0007	LP	PM1, PM2, PP2, PP3
P15	Co HET	10	c.1031C>G	p.Ala344Gly	NBD1	MISSENSE	PossD	D	DC	0	LP	PM2, PP2
		19	c.2362C>T	p.Arg788Trp	NBD2	MISSENSE	PossD	D	DC	0.016	LP	PM2, PP2
P16	Co HET	15	c.1783C>T	p.Arg595Ter	NBD2 start	NONSENSE	D	D	DC	0.001	P	PVS1, PM2, PP5
		23	c.2906G>A	p.Arg969His	NBD2	MISSENSE	B	-	Polymorphism	0.0001	LP	PM2, PP2

¹Allele frequency: (1) 0: Strengthens pathogenicity; (2) 0.0001: May support pathogenicity; (3) 0.001: Considered rare; (4) 1%-5%: Unlikely pathogenic; and (5) > 5%: Too common to cause rare disease.

ACMG: American College of Medical Genetics; B: Benign; Co HET: Compound heterozygous; D: Damaging; DC: Mutation taster 1 disease causing; Ho: Homozygous; LP: Likely pathogenic; NBD: Nucleotide binding domain 1; P: Pathogenic; ProbD: Probably damaging; PossD: Possibly damaging; TMD: Transmembrane domain.

Missense mutations: Patients with the homozygous p.Phe970Leu variant (P1-P4) had infantile onset (mean, 10 months) but maintained compensated disease, consistent with minimal ATPase disruption from this moderately conserved domain variant.

Patients with the homozygous p.Gly954Ser variant (P5-P8) had a later onset (111 ± 39 months) but rapid progression and poor outcomes despite treatment. The variant, in a conserved domain, likely disrupts adenosine triphosphate (ATP) binding and folding; two patients died, others developed uncorrectable coagulopathy, with some (P5 and P7) initially misdiagnosed as AIH.

Patients P11, 12, and 14 also had (different) missense mutations located in the highly conserved segments of ABCB4 (Figure 1) with overall poor outcomes (Tables 2 and 3).

Nonsense mutations: Siblings P9 and P10 with nonsense homozygous p.Arg47Ter mutations also had poor clinical outcomes. P13 with nonsense homozygous p.Arg595Ter mutation presented with decompensated CLD (Tables 2 and 3).

Compound HET mutations: Patients P15 and P16 carried compound HET ABCB4 mutations with features as described in Tables 2 and 3.

Follow-up: The group 1 patients were followed up for a median duration of 28.5 months (range: 2-124 months). All received standard-of-care management, including UDCA. Two patients (P4 and P9) required adjunctive rifampicin for intractable pruritus. Two-thirds of the patients from this group had disease worsening at the last follow-up, with one-third of patients succumbing to the disease (Tables 1 and 2).

HET mutations (group 2)

Ten children with HET ABCB4 mutations were identified during the study period. Their clinical features, laboratory parameters, and genetic variants are detailed in Tables 4 and 5. Most of them presented with cholestasis in infancy, with resolution in follow-up (P18, P19, P21, P22, P25, P26). One had a history of recurrent jaundice (P20), but overall a favourable outcome, and one was diagnosed during family screening (P17) (Tables 4 and 5).

Table 4 Clinical characteristics of patients with monoallelic ABCB4 mutations (n = 10).

Patient	Gender	Age at symptom onset (months)	Age at first presentation (months)	Consanguinity	Family history	Anthropometry (weight/height)	Clinical features	Disease status at presentation	Esophageal varix	Disease status at last follow-up	Issues at last follow up/salient features	Final outcome at last follow- up
P17 (sibling of P3)	Female	180	180	No	Yes	0.73Z, 0.87Z	H	C	No	S	Screening detected	Alive
P18 (sibling of P15)	Female	5	17	No	Yes	-0.46Z	J, P, H	C	Not done	I	Infantile onset, self-limiting course	Alive
P19	Male	5	19	No	Yes	-2.33Z, -4.06Z	J, P, HSM	C	No varix	I	Infantile onset, self-limiting course	Alive
P20	Female	48	72	No	No	0.1Z, 0.1Z	J, P, H	C	No varix	I	Recurrent jaundice	Alive
P21	Male	1	3	No	No	-3.43Z, -2.15Z	J, P, HSM	C	No varix	I	Infantile onset, self-limiting course	Alive
P22	Male	8	8	No	No	-3.2Z, -2.54Z	J, P, HSM	C	No varix	I	Infantile onset, self-limiting course	Alive
P23	Male	144	192	No	No	-2.42Z, -3.33Z	J, P, HSM	C	Large varix	W	PH, B, A, E, MRCP-sclerosing cholangitis	Expired
P24	Female	48	108	No	No	-2.39Z, -3.34Z	J, P, HSM	C	Small varix	W	PH, E, SNHL, additional heterozygous mutations in GJB6 and SPTB (both autosomal dominant)	Expired
P25	Male	11	13	No	No	-0.34Z, -1.63Z	J, P, H	C	Not done	I	Infantile onset, self-limiting course	Alive
P26	Male	10	12	No	No	0.11Z, -0.04Z	J, P, HSM	C	No varix	I	Infantile onset, self-limiting course	Alive

Small and large varix as per Baveno VI guidelines. A: Ascites; B: Gastrointestinal bleed; C: Compensated; E: Encephalopathy; H: Hepatomegaly; HSM: Hepatosplenomegaly; I: Improved; J: Jaundice; P: Pruritis; PH: Portal Hypertension; S: Stable; SNHL: Sensorineural Hearing Loss; W: Worsened.

Table 5 ABCB4 variants in monoallelic mutations and their pathogenicity (n = 10).

Patient	Zygosity	Exon Number	Mutation	Predicted effect	Domain	Type of mutation	Polyphen	SIFT	Mutation taster	GnomAD1	Updated ACMG class	ACMG criteria
P17 (sibling of P3)	HET	23	c.2908T>C	p.Phe970Leu	NBD2	MISSENSE	B	D	DC	0	VUS	PM1, PM2, PP2
P18 (sibling of P15)	HET	19	c.2362C>T	p.Arg788Trp	NBD2	MISSENSE	ProbD	D	DC	0.016	VUS	PM2, PP2
P19	HET	8	c.808G>C	p.Gly270Arg	TMD1/NBD1 boundary	MISSENSE	ProbD	D	DC	0.189 (THRESHOLD:0.1)	VUS	PS1, PP2, PP3, BS1
P20	HET	8	c.808G>C	p.Gly270Arg	TMD1/NBD1 boundary	MISSENSE	ProbD	D	DC	0.189 (THRESHOLD:0.1)	VUS	PS1, PP2, PP3, BS1
P21	HET	16	c.1963C>G	p.Pro655Ala	NBD2	MISSENSE	B	-	Polymorphism	0	VUS	PP2
P22	HET	14	c.1571C>A	p.Thr524Asn	TMD2	MISSENSE	ProbD	D	DC	0	LP	PS4, PM2, PP3, PP2
P23	HET	9	c.928G>A	p.Ala310Thr	NBD1	MISSENSE	PossD	-	-	-	VUS	-
P24	HET	13	c.1558C>T	p.Gln520Ter	Linker before TMD2	NONSENSE	-	-	-	0.0001	LP	PVS1, PM2
P25	HET	15	c.1650C>A	p.Asn550Lys	TMD2	MISSENSE	PossD	D	-	0.0004	VUS	PM1, PM2, PP2
P26	HET	15	c.1783C>T	p.Arg595Ter	NBD2 start	NONSENSE	-	-	DC	0.001	Pathogenic	PVS1, PM2, PP5

¹Allele frequency: (1) 0: Strengthens pathogenicity; (2) 0.0001: May support pathogenicity; (3) 0.001: Considered rare; (4) 1%-5%: Unlikely pathogenic; and (5) > 5%: Too common to cause rare disease.

ACMG: American College of Medical Genetics; B: Benign; D: Damaging; DC: Mutation taster 1 disease causing; HET: Heterozygous; LP: Likely pathogenic; NBD: Nucleotide binding domain 1; ProbD: Probably damaging; PossD: Possibly damaging; TMD: Transmembrane domain 1; VUS: Variant of unknown significance.

Children with older age at onset and poor outcome: Two patients (P23 and P24) had a steady worsening despite carrying a HET ABCB4 mutation. They both presented to us in the preadolescent phase.

Patient P23 (HET p.Ala310Thr) had progressive disease leading to death. This NBD1 variant likely impairs ATP binding and transport; initial treatment for autoimmune sclerosing cholangitis yielded no response to immunosuppressants or UDCA. Patient P24 (HET p.Gln520Ter) presented with compensated CLD and portal hypertension at 108 months but died within a year (Table 4).

Follow-up: The patients in group 2 had a median follow-up duration of 23 months (range, 3-129 months). All received UDCA, with most showing an overall favourable response; however, both patients presenting at an older age died (Table 4).

Comparison between group 1 (homozygous/compound HET mutations) and group 2 (HET mutations)

Demographic characteristics, clinical features and complications, liver histopathology, variant pathogenicity, and overall outcomes were compared between the two groups (Table 1).

Family history of liver disease and/or gallstone disease, decompensated liver disease, proportion of larger varices, median GGT and proportion of patients carrying pathogenic/Likely pathogenic mutations were significantly higher in group 1 (Table 1). Patients in group 2 showed improvement in LFTs in follow-up, with significantly fewer patients developing progressive disease (Table 1). Given the limited sample size of this cohort, the statistically significant P values reported in Table 1 should be interpreted as indicative of strong clinical trends rather than definitive evidence, and they require validation in larger studies.

Liver biopsy was performed in 25 patients. Histopathology findings did not differ significantly between the two groups (Figure 2 and Supplementary Table 4). MDR3 immunostaining was performed in 11 patients overall (in both groups), with nine showing negative staining; no intergroup differences were noted (Figure 3). Sixteen (61.5%) of histopathology samples were subjected to Rhodanine staining, of which twelve patients (75%: 10 biallelic and 2 monoallelic) had Rhodanine staining positivity (Figure 2D), indicating copper excess.

Figure 2 Histopathology of a case of Progressive familial intrahepatic cholestasis type 3. A: The lobular architecture was distorted with porto-portal and occasional slender porto-central bridging fibrosis with the formation of incomplete nodules (masson trichrome, 100 ×); B: The fibrous septa showed variable lymphocytic sprinkling and moderate ductular reaction (hematoxylin and eosin, 200 ×); C: The hepatic lobules showed hepatocanalicular cholestasis and cholestatic rosettes (thick black arrows) (hematoxylin and eosin, 400 ×); D: Periportal/periseptal hepatocytes showed copper retention visualized as tiny red specks (thin orange arrows) (Rhodanine, 400 ×).

Figure 3 Immunohistochemical panel of a case of progressive familial intrahepatic cholestasis type 3. A: CK7 immunostain showed ductular reaction and periseptal to panacinar biliary metaplasia (200 ×); B: Bile salt export pump immunostain showed retained canalicular expression (400 ×); C: Multidrug resistance protein 3 immunostain which showed complete loss of canalicular stain (400 ×); D: A case of primary sclerosing cholangitis (control case) showing retained multidrug resistance protein 3 immunostaining (400 ×).

DISCUSSION

PFIC3 is an autosomal recessive genetic disorder characterized by cholestasis and a highly variable clinical presentation and prognosis. Biallelic mutations in ABCB4 typically result in complete MDR3 deficiency, often leading to ESLD during childhood. However, HET missense variants in ABCB4 have been implicated in symptomatic cholestasis, thereby complicating the approach and management. NGS has helped to understand the genetic basis of cryptogenic CLD. In this context, we present our experience with pediatric patients harbouring both monoallelic and biallelic ABCB4 mutations.

Our cohort included 16 patients with biallelic ABCB4 mutations. Their clinical features and LFT profiles were largely consistent with previous studies (Table 6)[12-15]. The median age at symptom onset in our cohort was higher than that in earlier reports. However, a review of 118 cases of PFIC3 noted a mean age of disease onset of 6.5 years, with 58.4% of patients experiencing symptoms before 60 months of age, similar to that in our cohort[16].

Table 6 Phenotypic, genetic and laboratory characteristics of various pediatric studies with biallelic ABCB4 mutation.

Ref.	Year/country	Number of patients	Median age at onset of symptoms (months)	Median Age at presentation (years)	Pruritus (%)	Jaundice (%)	Hepatomegaly (%)	Splenomegaly (%)	Variants with missense mutation (%)	Laboratory parameters
										Gamma-glutamyltransferase (IU/L)	Alanine aminotransferse (IU/L)	Aspartate aminotransferse (IU/L)	Total bilirubin (mg/dL)
Schatz et al[12]	2018/Germany	26	5	4.8	100	61.5	84.6	96.1	77.1	320	212
Chen et al[13]	2022/China	13	36	-	30.7	53.8	100	77	61.5	197		130	1.1
Al-Hussaini et al[14]	2021/Saudi Arabia	25	10	2	92	12	80	-	81.4	202		149	1.27
Gonzales et al[15]	2023/France	38	11	3.3	79	29	82	68	71	150		-	0.73
Index study	2025/India	16	54	5	68.8	68.8	100	81.2	81.3	171	120.5	198	2.58

Our cohort replicates the typical presentation of biallelic ABCB4 mutations as classical PFIC3, which often progresses to ESLD. At presentation, nearly half of our patients had ascites and decompensation. Despite supportive therapy and UDCA, disease progression occurred in 68.8% of patients till the last follow-up (Table 1). Other pediatric studies have documented progression to ESLD and the need for LT in roughly 17%-50% of cases, highlighting the severity and rapid progression of the disease in children[12,13,15,17,18].

Prior to the widespread use of genomic sequencing, liver histopathology with IHC for MDR3 was used to diagnose PFIC3. Histological findings in PFIC3 typically include higher grades of portal-based fibrosis and ductular reaction, accompanied by mild to moderate lobular and portal inflammatory infiltrates – features that were consistently observed in our cohort and have been similarly reported in multiple studies (Figure 2)[14,15,18].

The absence of MDR3 staining on IHC serves as a distinguishing feature of PFIC3 compared to other PFIC subtypes. In our study, 9 of 11 patients in whom IHC was done demonstrated negative MDR3 staining (Figure 3). A complete absence of staining is usually seen in patients with biallelic null variants, while those with missense or monoallelic mutations demonstrate faint or even preserved MDR3 expression. This pattern has been corroborated by Gonzales et al[15] and Jacquemin et al[19]. Though absent MDR3 staining in null variants alone is expected, we noted absent staining in missense mutations also. This absence has been attributed to intracellular misprocessing of MDR3 and the nature of the missense mutation. Mutations in highly conserved areas may prevent MDR3 expression[19]. Most mutations in our cohort were located in such domains. (Tables 3 and 5). Weng et al[20] observed in their cohort that missense variants had similar levels of MDR3 protein compared to wild-type, except in one case, demonstrating the variability in MDR3 expression, which can explain the poor reliability of MDR3 IHC.

AIH and WD remain the most common etiologies of pediatric CLD in our region[21,22]. Before the genetic diagnosis, some of our patients received treatment for AIH. Positive Rhodanine staining on histopathology further complicates the diagnosis, given the higher prevalence of WD compared to ABCB4 mutations in patients with CLD requiring hospitalisation. The presence of pruritus, early onset portal hypertension with varices, elevated GGT, and IHC findings for MDR3, as well as a poor or absent response to standard therapies for AIH or WD, raised the suspicion of PFIC3, leading to a genetic workup.

Missense mutations were the most common in our study, which is similar to a systematic review of 118 patients with PFIC3, in which missense variants accounted for 76.9% of all mutations[16]. The nature of the mutation, the resultant protein alteration, and the affected functional domain seem to be critical determinants of disease severity and progression.

Few variants had unique clinical phenotypes. Homozygous p.Phe970Leu mutation affects a moderately conserved domain and is predicted to cause minimal disruption to ATPase function, which may explain the slower disease progression observed in our cohort compared to other homozygous mutations. The same variant has been described in a Saudi Arabian cohort, which reported a similarly indolent course[14]. Interestingly, all patients in our cohort with this specific mutation originated from the Kashmir Valley, suggesting a potential regional or ethnic clustering. Factors such as ancient human migration, endogamy, and founder effects in geographically or culturally isolated populations like those in Kashmir may contribute to this phenomenon.

The homozygous p.Phe954Ser mutation affects a highly conserved domain with a direct role in ATP binding and protein folding, thereby contributing significantly to disease severity[19,23]. Delaunay et al[3] previously reported a compound HET presentation of this variant in two patients who developed cholestasis, icterus, and hepatomegaly at the age of four years. One of these individuals required LT at 21 years of age, while the other maintained native liver function up to 18 years. After functional characterization, this variant was classified as a class III mutation, i.e. impaired activity without major effects on protein maturation[3]. However, our patients had pure homozygous mutations, which would probably significantly affect protein activity, thereby leading to rapid progression. Gonzales et al[15] also described this variant in two patients, one of whom required LT while the other did not, similar to the variable disease course in our cohort. This can be attributed to undetected genetic variations in other hepatocanalicular transporter genes, hormonal factors, or environmental exposures[13].

The rapid disease progression observed in our patients with nonsense mutations is consistent with the known mechanism of nonsense-mediated mRNA decay, which leads to the elimination of protein expression[19]. The localisation of variants within NBDs (Figure 1, Tables 3 and 5), which are evolutionarily conserved and functionally critical, likely accounts for the progressive disease course observed in affected individuals[24]. A functional characterization of these mutations (as performed by Delaunay et al[3]) could have explained the phenotypic variability, but was not performed in our cohort. A proteomics study by Guerrero et al[25] demonstrated that, irrespective of the type of homozygous mutation, there is disturbed phosphorylation of certain residues in MDR3 protein, thereby affecting the binding or hydrolysis of ATP, which is crucial for phosphatidyl choline translocation. There is an upregulation of class I and II HLA, along with immunoglobulins and tumor necrosis factor alpha, in PFIC 3 livers, emphasising a proinflammatory and profibrogenic state. There is also an impairment in tight and adherens junctions, along with the reconfiguration of the actin-myosin microfilament network, in PFIC3 hepatocytes, which affects cell polarity. Weng et al[20] showed reduced phosphatidylcholine secretion in HEK293 cell lines with ABCB4 mutations compared to MDR3 wild-type, thereby suggesting the pathogenic potential of missense mutations in vitro.

HET ABCB4 mutation: Implications

The reason why heterozygotes present clinically remains a matter of debate. HET ABCB4 variants are increasingly recognized as clinically relevant in pediatric cholestatic liver diseases. In our cohort, monoallelic ABCB4 mutations were identified in 10 children, accounting for 40% of all ABCB4 mutation-positive cases. These findings highlight the possible pathogenic potential of HET variants, which were previously regarded as benign or of uncertain significance.

The clinical spectrum among these patients was heterogeneous, ranging from transient infantile cholestasis to progressive liver disease and portal hypertension. Such clinical heterogeneity in patients with monoallelic ABCB4 mutations may be attributed to functional heterogeneity, as described by Gordo-Gilart et al[26] who noted reduced MDR3 protein levels, intracellular retention of the protein in the endoplasmic reticulum, and a decrease in phosphatidylcholine efflux activity to 18%-56% of normal.

Most patients experienced a self-limited cholestatic course and recurrent jaundice without progression, with normalization of liver biochemistry following treatment with UDCA (Table 4). This observation aligns with the findings of Hegarty et al[27], who reported that 48% of patients with monoallelic variants achieved long-term biochemical normalization and were subsequently discharged from follow-up.

One patient with a novel protein-truncating HET p.Gln520Ter mutation demonstrated absent MDR3 expression and experienced rapid clinical deterioration (Tables 4 and 5). Truncating variants located within transmembrane domains may lead to more severe disease phenotypes, even in HET carriers, due to their critical role in MDR3 function, as observed in our cohort[17].

PFIC 3 and sclerosing cholangitis

Two patients had MRCP features of sclerosing cholangitis. The coexistence of primary sclerosing cholangitis-like features and ABCB4 mutation has been previously reported in adults. These findings support the hypothesis that ABCB4 may act as a susceptibility gene modifier in immune-mediated cholangiopathies[28-32]. Hegarty et al[27] reported cholangiopathy on axial imaging in four patients with monoallelic ABCB4 mutations, attributing this to chronic alterations in bile composition that damage the biliary epithelium. This mechanism was further supported in an animal model study. Bile regurgitation from leaky ducts into the portal tracts initiates periductal inflammation and activates periductal fibrogenesis. Over time, this cascade leads to obliterative cholangitis, resulting from the atrophy and death of bile duct epithelial cells[33].

The reason why heterozygotes present clinically remains a matter of debate. The variant may result in an altered protein, which could interfere with the function of the wild-type gene product. A truncating monoallelic mutation may leave the wild-type allele incapable of producing a sufficient quantity of protein to sustain a normal phenotype. It is also possible that a second mutation is not picked up by conventional NGS. In some cases, the mutant protein may alter the behaviour of other common diseases.

The long-term implications of monoallelic ABCB4 variants extend into adulthood, with established associations with intrahepatic cholestasis of pregnancy, low phospholipid-associated cholelithiasis, drug-induced liver injury, primary sclerosing cholangitis and even cholangiocarcinoma[28,29,34-36]. According to the European Association for the Study of the Liver recommendations, we plan to follow up these children at risk at regular intervals (3-6 months) with serial biochemical monitoring, liver imaging when indicated, and comprehensive family counselling[37].

The difference in clinical course between those with bi-allelic and monoallelic mutations was expected, as bi-allelic mutations cause a significant reduction in functional MDR3 protein, resulting in severe disease with early progression. Response to treatment also varies, as those with bi-allelic mutations continued to have worsening liver functions despite continuing on UDCA, as observed by Wang et al[16] in their systematic review. The role of Rifampicin and bile acid sequestrants, such as cholestyramine, is not well studied. Few experimental medications focusing on class II mutations, which exhibit trafficking defects with intracellular retention, have been studied, including ABCC7/CFTR correctors and structural analogues of roscovitine, but they are still in the early phases of trial[38]. Drugs with chaperone-like activity, such as curcumin and 4-phenylbutyric acid, are proposed to enhance the targeting of misfolded and intracellularly trapped mutant proteins to the plasma membrane, leading to increased phospholipid efflux activity[39,40]. LT is the only effective treatment for ESLD.

Strengths

This is the first and largest pediatric series from a major tertiary care centre that comprehensively characterises a sizable, well-phenotyped cohort of Indian children with ABCB4 mutations, a population for which such data are scarce. The inclusion and comparative analysis of both biallelic and monoallelic cases provide important clinical insights. The attempt to correlate specific genotypes (e.g., p.Phe970Leu, p.Gly954Ser) with clinical outcomes, including the observation of potential regional clustering, is a significant contribution to the field. The clinical relevance of HET mutations with variable presentations underscores the importance of prospective monitoring to identify the evolving spectrum of liver manifestations in this population. The integration of histopathology (including Rhodanine staining) and genetics to differentiate from WD and AIH is clinically very relevant.

Limitations

As it was a single-centre, retrospective study, the sample size was small and consequently had low statistical power for group comparisons. Hence, the comparison has to be interpreted with caution. There is a potential selection bias, as the cohort is from a tertiary referral centre, likely seeing more severe cases. Parental and sibling genetic testing were not performed, limiting our ability to confirm inheritance patterns and zygosity, which would have strengthened the interpretation of variant pathogenicity. In monoallelic cases, there is a possibility of undetected second-hit mutations in non-coding regions or complex structural variations. Functional validation of the identified ABCB4 variants was not undertaken, which could have provided deeper insights into the molecular mechanisms underlying the observed phenotypes.

CONCLUSION

ABCB4 mutations, in both homozygous and HET states, are a significant cause of pediatric liver disease often misclassified as cryptogenic CLD, AIH, or WD. Clinical clues include pruritus, elevated GGT, early portal hypertension, and limited response to standard therapy, which should lead to genetic testing. Biallelic truncating or severe missense variants usually progress, whereas monoallelic, mild missense variants respond to therapy but still need long-term surveillance; monoallelic truncating or conserved-region variants may advance to ESLD. Early diagnosis, genotype-guided management, and family screening are central to patient care. Larger prospective studies with family segregation and functional assays are needed to better define genotype-phenotype associations and guide management.