Introduction: Patients born with congenital porto-systemic shunts have been shown to have a high risk of benign and malignant liver tumors in otherwise healthy livers. This study aimed to evaluate the genetic landscape of liver tumors in patients with congenital porto-systemic shunts (CPSS) and correlate genotype with histological findings. Methods: Nodules from patients with CPSS and sporadic pediatric focal nodular hyperplasia (FNH) or FNH-like nodules were evaluated histologically and sequenced for a panel of 50 genes using next-generation sequencing. Results: Thirty-eight nodules from 17 patients with CPSS were histologically classified as hepatoblastomas (n = 2), hepatocellular carcinomas (n = 4), HNF-1α-inactivated hepatocellular adenomas (HCAs) (n = 2), β-catenin-activated HCAs (n = 5), unclassified HCAs (n = 9), and FNH-like nodules (n = 16). CTNNB1 variants were detected in 26/38 nodules (68%) across different histological categories (2/2 hepatoblastomas, 4/4 HCCs, 10/16 HCAs, 10/16 FNH-like nodules), but not in sporadic FNH or FNH-like nodules (0/10). Less frequent variants were identified in APOB, GNAS, HNF1A, SERPINA1, MAML2, PTCH1, G6PC, KMT2C, DICER1, AXIN1, IL6ST and the promoter region of TERT. Germline variants were identified in AXIN1, HFE, SERPINA1, and ZNF521. CTNNB1 variants affecting amino acid positions 32 and 33 are more common in malignant tumors. Multiple CTNNB1 variants were identified in 6/7 (86%) of patients with multiple nodules, but no intratumoral variation was found. Discussion: CPSS is strongly associated with nodules containing variants in CTNNB1, irrespective of the histological category. Areas in background liver containing these variants were also identified, and different variants could be identified in individual patients. The high proportion of CTNNB1 variants may explain the higher malignant potential of benign tumors found in CPSS.

Congenital porto-systemic shunts (CPSSs) are rare congenital malformations that result in shunting of blood from the portal vein to the systemic circulation, depriving the liver of portal venous inflow. The development of liver tumors is strongly associated with CPSS, particularly in patients with type 1 shunts, with complete bypassing of the portal system to the systemic circulation [1]. The tumors that have been observed include benign focal nodular hyperplasia (FNH), hepatocellular adenomas (HCAs) and their malignant counterparts, hepatocellular carcinomas (HCCs), and hepatoblastomas (HBs). Atypical nodules, such as FNH-like nodules or well-differentiated hepatocellular neoplasms of uncertain malignant potential, have been observed in patients with vascular hepatic pathologies, including CPSS and Budd-Chiari syndrome [2‒4]. Nodules in patients with CPSS are associated with radiologically atypical features with unpredictable evolution, even when they appear to belong to a low-risk category, such as FNH [1, 4‒6]. CPSS is of particular interest, as patients have livers without any confounding pathology that could increase their risk of developing liver tumors. They are a natural example of hepatic tumor formation due to the impact of hepatic deprivation of portal flow and compensatory increase in arterial flow.

In this study, we aimed to evaluate the genetic mutational landscape of tumors in the livers of patients with CPSS and to correlate it with histological appearance to further understand the evolution of the tumors in the CPSS patient cohort. Histology of background liver has been previously reported along with clinical outcomes for CPSS patients [1, 7, 8].

A retrospective single-center study of patients with CPSS was performed for patients presenting between January 1990 and January 2020. All patients were reviewed for the presence of macroscopic or radiologically detected liver nodules and the availability of tissue samples for histological evaluation and DNA isolation. Ethical approval for this study was granted by the Health Research Authority of the United Kingdom (16/EM/0342). Selected samples were retrieved from the King’s Pediatric Liver Tissue Biobank (18/WA/0009). We have previously outlined the clinical outcomes in our patient cohort, as well as histological analysis of the background liver in patients with CPSS [1, 7, 8].

Histology

All liver nodules in patients with CPSS and available background liver were assessed using a panel of stains (hematoxylin and eosin, periodic acid-Schiff, orcein, Perls, reticulin, Sirius red), and immunohistochemistry was performed for glutamine synthetase (GS), β-catenin, and Ki67. The tumors were reviewed by two histopathologists (Y.Z. and M.D.) and classified according to the WHO Classification of Tumors Digestive System Tumors. Lesions that had morphological features of FNH but lacked map-like expression of GS (typical for FNH) were classified as FNH-like nodules [2‒4]. Descriptions of GS staining were standardized using the classification proposed by Sempoux et al. [9] and examples are given in online supplementary material 5.0 (for all online suppl. material, see https://doi.org/10.1159/000543217). Only diffuse homogeneous expression of GS was counted for β-catenin-activated HCAs. For the classification of tumor types, the panel of immunohistochemical staining was expanded to include one or more of the following: glypican 3, serum amyloid-A, C-reactive protein, CK7, liver fatty acid-binding protein, and CD34.

The proliferation rate was calculated as the number of Ki67-positive lesional hepatocytes (excluding positive hematolymphoid cells) per ×200 magnification field averaged over at least 2 separate ×200 fields. Fields not entirely occupied by lesional hepatocytes, such as the presence of large vessels or fibrous stroma, were avoided. Ten healthy control liver transplant donors in the pediatric age group and 5 pediatric control patients with FNHs (non-CPSS) and available background liver tissue were used as healthy controls for the proliferation rate.

Next-Generation Sequencing for Somatic and Germline Variants

Tumor and background liver DNA samples were extracted from formalin-fixed paraffin-embedded (FFPE) samples of CPSS patients using a QIAamp DNA FFPE Tissue Kit (Qiagen, USA) according to the manufacturer’s instructions. Furthermore, control tumors tissue was also sequenced by the same method. The DNA concentration was quantified using a Qubit fluorimeter, and the Agilent Genomic TapeStation 4200 was used for quantification and to assess the degradation of DNA and the need for re-extraction. A custom next-generation sequencing (NGS) panel of 50 genes implicated in hepatocellular tumorigenesis was designed using Agilent SureDesign for Agilent HaloPlex library preparation. The full panel of sequenced genes is shown in online supplementary material 1.0. Amplicon libraries were pooled and sequenced on an Illumina MiSeq sequencer, according to the manufacturer’s instructions.

FastQ processing and variant calling were performed using Agilent SureCall software with a minimum variant frequency of 10% and a minimum of 10 forward and reverse reads. Variants with 5–10% frequency and a minimum of 10 forward and reverse reads were also noted as low-frequency variants. All CPSS samples and control nodule samples underwent NGS of the panel of 50 genes. To determine which variants were somatic, background liver tissue was used to confirm the absence of a variant. Variants which were identified in the background liver and nodules of frequency of at least 40% have been reported as germline.

Targeted Deep CTNNB1 NGS

Detected variants were confirmed using targeted deep CTNNB1 NGS. Furthermore, targeted deep NGS was used in micronodules and larger nodules with confirmed CTNNB1 variants where nodules were subdivided to look for intranodular variant heterogeneity. A custom-designed library preparation method employing molecular barcodes targeted the genomic region encoding amino acids (AA) 32–45 and 30 bases on either side, which was adapted from Peng et al. [10]. The full methodology and primer design are provided in online supplementary material 2.0. FastQ processing and demultiplexing of molecular barcodes were performed on a Linux Command Line Interface using Burrows-Wheeler Aligner, Agilent Genomics NextGen Toolkit (AGeNT), Samtools, and Agilent SureCall software; this is described fully in online supplementary material 3.0. The minimum variant call frequency was 1%, with a minimum of 2 reads. This library preparation method was used for samples gathered by laser capture microdissection (LCM), as well as to resequence and confirm variants of CTNNB1 in previously positive tumor DNA with the same minimum variant call parameters as in the NGS protocol.

Laser Capture Microdissection of Micronodules and Intratumoral Variation

Several areas of interest became apparent in the background liver, associated with the elevated expression of GS. LCM using a Leica Laser Microdissection Scope (LMD 6500) was performed to isolate areas with increased GS expression. DNA was extracted as previously described in the NGS methods and sequenced as described in the targeted deep CTNNB1 NGS and online supplementary material 2.0.

Tumors were also examined for uniform or differing CTNNB1 variants across the same tumor. Different areas or micronodules within the same tumor were dissected using LCM and sequenced separately.

Tumor Histological Classification

Forty-seven patients with CPSS were eligible for inclusion in the study. Seventeen patients had tumor histology samples available for analysis: 10 had a single tumor and 7 had multiple tumors. Median age at resection or biopsy in the 17 included CPSS patients was 14.8 (IQR 7.8–24.1) years. Of these 17 patients, 38 had FFPE tissue available. The specimens available for review consisted of the products of 7 core needle biopsies, 2 wedge biopsies, and 29 complete nodules from surgical resections.

The clinical course, treatment, and outcomes of all CPSS cases at our center have been published, and a brief synopsis of the most relevant clinical information follows [1]. 28/47 (60%) of patients with CPSS developed liver nodules radiologically detected at a median age of 9.5 (IQR 4–17) years. Malignant liver tumors affected 7/47 (15%) patients at a median age of 11 years (IQR 5–31) years but were seen as young as 2 years. The most important association of nodule formation was with patients who had type 1 CPSS (without detectible intrahepatic portal venous inflow). Shunt closure and restoration of intrahepatic portal flow are associated with improved nodule size. In 11 patients with nodules at the time of shunt closure, a decrease in size was observed in 4 (36%) patients, and nodules resolved completely in 3 (27%) patients. Nodules were predominantly contrast enhancing in the 28 patients with nodules: 20 (71%) displaying contrast enhancement on CT or MRI, 3 (11%) who did not display enhancement post contrast, and 5 (18%) who had MRIs without contrast. Regarding significant risk factors for hepatic neoplasm formation, no patient was detected of having been infected by hepatitis viruses. A comprehensive review of background liver of our cohort has been previously published, but no patient was found to have a cirrhotic liver but mild or patchy slender incomplete porto-portal septa were commonly seen [8].

The 38 tumors available were classified histologically as follows: 2 HBs (1 mixed epithelial and mesenchymal and 1 fetal subtype), 4 HCCs, 5 β-catenin activated HCAs, 9 unclassified HCAs, 2 HNF-1α-inactivated HCAs, and 16 FNH-like nodules. FNH-like nodules displayed features of lobulated nodules with variably broad fibrous septa (typically a central scar) containing reactive ductules, large arteries, and lymphocytic infiltrates. None of the cases showed typical map-like GS expression. HCAs consist of homogeneous or multinodular proliferation of mature hepatocytes without fibrous septa or a central scar. Some HCAs show features of steatosis, sinusoidal dilatation, hemorrhage, or small portal tract-like structures with reactive ductules. HCAs with a complete loss of liver fatty acid binding protein expression were classified as the HNF-1α-inactivated type, HCAs with diffuse homogeneous GS expression were determined as the β-catenin activated type, and the remainder without characteristic features were unclassified.

CTNNB1 Variants Occurred at High Frequency across Tumor Classification

One HCC sample was not successfully sequenced due to the degree of DNA degradation in the FFPE sample despite repeated extractions; the remaining 37 nodules were sequenced. Median DNA integrity number as quantified by TapeStation was 4.7 (IQR 4.1–5.8). Figure 1 summarizes the somatic and germline variants identified according to the tumor histology. A predisposition for heterozygous somatic variants in CTNNB1 was evident, as they were present in 26/37 (70%) sequenced nodules. Other heterozygous somatic variants were infrequent, with variants in APOB, GNAS, HNF1A, SERPINA1, and the promoter region of TERT found to be mutated in two tumors each. Heterozygous somatic variants of MALM2, PTCH1, G6PC, KMT2X, DICER1, AXIN1, and IL6ST were identified each in one tumor. A complete list of variants is provided in the online supplementary material 4.0.

Fig. 1.

Somatic and germline variants identified in CPSS tumors according to tumor histology and variant type.

Fig. 1.

Somatic and germline variants identified in CPSS tumors according to tumor histology and variant type.

Close modal

Promoter variants of the TERT gene in the hot spot loci of −124 or −148 from the ATG (chromosome 5 positions 1,295,228 and 1,295,250) were identified in 2 HCCs in adult patients. These tumors had concurrent CTNNB1 variants. The proportion of malignant tumors (HCC and HB) with multiple variants was 3/5 (60%), which was significantly higher than that observed in benign tumors (4/32 [13%], p = 0.04).

Germline Variants in CPSS Patients

The germline variants identified in the background liver and tumors of the patients are listed in Table 1. A combination of heterozygous germline variants in AXIN1 and HFE was observed in one patient who had two FNH-like nodules and HCCs. In this patient, one FNH-like lesion had no CTNNB1 variant, while the second had p.Ser33Pro, and a high allele frequency AXIN1 variant (variant allele frequency of 0.8 vs. 0.5 in the background liver), suggesting loss of heterozygosity. HCC had a missense CTNNB1 variant, leading to p.Asp32Gly.

Table 1.

Germline variants identified and confirmed in both the background liver and tumor of patients

ChromosomePositionBase changeSubstitutionPatients with variant
AXIN1 16 347,063 C>T p.Gly650Ser 
HFE 26,091,645 C>A p.Phe125Lys 
SERPINA1 14 94,844,947 C>T p.Glu366Lys 
ZNF521 18 22,901,998 G>A p.His65Pro 
ChromosomePositionBase changeSubstitutionPatients with variant
AXIN1 16 347,063 C>T p.Gly650Ser 
HFE 26,091,645 C>A p.Phe125Lys 
SERPINA1 14 94,844,947 C>T p.Glu366Lys 
ZNF521 18 22,901,998 G>A p.His65Pro 

One patient was a heterozygous carrier of the Z allele for alpha-1 antitrypsin (p.Glu366Lys missense variant in the gene SERPINA1), which has not been shown to contribute to tumor formation. Heterozygous HFE germline variants associated with the autosomal recessive inheritance of hemochromatosis have been observed. None of the patients in our cohort had hemochromatosis or alpha-1 antitrypsin deficiency. A missense variant was found in ZNF521 causing p.His65Pro, not been previously reported in hepatocellular tumors.

CTNNB1 Variants Are a Feature of Nodular Lesions in CPSS

Studies of adult FNHs without CPSS have not shown evidence of variants in CTNNB1, but to date these tumors have not been investigated in children [11, 12]. Ten control sporadic FNH tumors in pediatric patients were selected based on their histology for comparison with CPSS tumors for both GS staining and genotype. This study aimed to determine whether CTNNB1 variants and atypical GS staining are features of pediatric FNHs or tumors alone in patients with CPSS. The 10 control patients were otherwise healthy, without liver disease, vascular anomalies, or any other confounding pathology.

The median age at biopsy or excision of the control nodules was 7.3 (IQR 5–10.4) years compared to 15.3 (IQR 12–16.5) years in the 10 CPSS patients with FNH-like nodules (p = 0.006). 8/10 (80%) of the control FNHs showed typical map-like GS staining, two did not show increased GS staining, and none of the tumors had abnormally increased β-catenin expression. The 2 with atypical GS were reclassified due to GS expression as FNH-like nodules. There was a significant difference in the proportion of maps such as GS staining in the control vs. CPSS group (p < 0.001), and there were 3/16 (19%) FNH-like nodules with increased β-catenin staining (nuclear) in the CPSS group, although the comparison did not reach significance (p = 0.26). No variants were identified in the control group compared to 10/16 (63%) CTNNB1-mutated FNH-like nodules had a strong association with CPSS (p = 0.003) but not in pediatric FNHs.

CTNNB1 Variants Causing Substitution in AA 32 and 33 Are Associated with Malignant Tumors

Figure 2 classifies the missense CTNNB1 variant according to both histological and AA changes. With the exception of p.Lys335Ile in exon 7, all missense mutations were observed in exon 3. Fewer variants affecting codons 32 and 33 occurred, and these were associated with malignant tumors (HB, HCC) in 3 cases and once observed in an FNH-like nodule. p.Asp32Gly, p.Ser33Cys, and p.Ser33Pro accounted for 3/5 (60%) variants identified in malignant tumors, which was a significantly higher proportion than that in benign tumors 1/19 (p = 0.02). Importantly, the two patients with p.Ser45Phe or p.Ser45Pro CTNNB1 variants that occurred in HCCs were both adults between 30 and 50 years of age who had both acquired mutations in the promoter region of TERT in addition to the lower risk CTNNB1 variant.

Fig. 2.

Tumor histology type and proportion of CTNNB1-mutated tumors, with a breakdown of the number of tumors with particular AA changes.

Fig. 2.

Tumor histology type and proportion of CTNNB1-mutated tumors, with a breakdown of the number of tumors with particular AA changes.

Close modal

Genotype-Phenotype Correlation of CTNNB1 Variants and Glutamine Synthetase and β-Catenin Immunohistochemistry

All CPSS background livers and tumors were stained for GS and β-catenin. Background liver showed perivenular GS staining and membranous β-catenin staining in all samples with a few foci of increased expression, and these areas of interest were investigated separately and are discussed later as micronodules. Figure 3 outlines the pattern of GS and β-catenin staining according to both histology and CTNNB1 AA substitutions. Abnormally increased GS staining was more sensitive (96%) for variants in CTNNB1 when compared to β-catenin staining (35%), although β-catenin was slightly more specific (72%) than GS (64%).

Fig. 3.

Histogram showing GS and β-catenin staining according to tumor histological classification (a) and particular CTNNB1 variant present (b).

Fig. 3.

Histogram showing GS and β-catenin staining according to tumor histological classification (a) and particular CTNNB1 variant present (b).

Close modal

Diffuse homogeneous GS staining patterns were observed more frequently in the nodules (9/14, 64%) with variants causing AA substitutions at positions 32, 33, 36, and 41 than in those with AA substitutions at position 45 or 335 (1/11, 9%, p = 0.01). However, a diffuse homogeneous GS staining pattern was not significantly associated with malignant tumors, as 9/32 (28%) benign tumors had this pattern compared to 3/5 (60%) malignant tumors (p = 0.30). Conversely, abnormally increased β-catenin staining was associated with malignant tumors; 5/5 (100%) had nuclear (n = 4) or cytoplasmic (n = 1) staining, compared to 7/32 (19%) benign tumors with nuclear (n = 6) and cytoplasmic (n = 1) staining (p = 0.002).

There was significant variability in GS expression, β-catenin expression, and genotype for FNH-like nodules in children with CPSS, showing that in this clinical context, they are a diverse group of tumors. GS staining patterns were as follows: not increased (n = 4), focal patchy (n = 4), diffuse heterogeneous (n = 4), and diffuse homogeneous (n = 4). Four with no CTNNB1 variants did not have increased GS or β-catenin staining, and one with a low-frequency p.His36Pro variant showed diffuse heterogeneous GS and membranous β-catenin staining. Six CPSS FNHs had variants affecting AA 45 (5 with p.Ser45Phe substitution, 1 p.Ser45Pro), 3 of which had focal patchy GS expression, 3 had diffuse heterogeneous, and all had membranous expression of β-catenin. Two FNH-like nodules with a p.Thr41Ala substitution had diffuse homogeneous GS with nuclear staining, and one had membranous β-catenin staining. Those without variants tended to show normal GS and β-catenin expression, those affecting AA 45 showed a mild to moderate increase in Wnt pathway activation, and those affecting AA 41 and 33 showed a moderate to severe increase.

In one patient, a biopsy was performed at the time of shunt closure from a 5 cm tumor in the left lobe of the liver with 3 smaller (unbiopsied) nodules. The tumor showed diffuse homogeneous GS and nuclear β-catenin expression in an FNH-like nodule with a p.Ser33Pro CTNNB1 variant. Three months after shunt closure, a second laparotomy and left hepatectomy was performed, and at this point, the initial tumor could no longer be palpated as it had regressed in size. Two of the radiologically noted tumors regressed completely, and 2 tumors were detectable on histological sectioning of the specimen. One was an FNH-like nodule with diffuse homogeneous GS staining and nuclear β-catenin staining, with no detectable variants. The second was a nodular area with a 4-mm focus of HCC with a p.Asp32Gly variant. The initial variant on biopsy at the time of shunt closure was not detected, which may imply that the nodule had regressed completely.

The HCAs displayed a similar genotype-phenotype correlation. Six with no CTNNB1 variant showed no increase in GS or β-catenin staining in 3, only focal patchy GS increase in 2, and 1 atypical β-catenin-activated tumor with nuclear β-catenin staining occurred in a patient with multiple other tumors with confirmed variants in their explanted liver. The 2 unclassified HCAs with p.Ser45Pro variants did not show increased β-catenin expression, 1 with diffuse heterogeneity, and 1 with focal patchy GS staining. Seven had p.Thr41Ala variants with variable phenotypes: 3 diffuse heterogeneous GS with membranous β-catenin expression, 1 with diffuse heterogeneous GS but cytoplasmic β-catenin expression, and 3 with diffuse homogeneous GS (2 with nuclear and 1 with membranous β-catenin expression). The 1 β-catenin-activated p.His36Pro with diffuse homogeneous GS had membranous β-catenin expression.

CTNNB1 Variants Identified in Background Non-Tumoral Liver Tissue

Four areas of increased GS expression were identified in the background liver tissue of 4 patients. In 3 patients, these areas were less than 1 mm in diameter. Representative images are shown in Figure 4 and online supplementary material 6.0. In the fourth patient, increased GS expression was observed in the liver parenchyma adjacent to an area of an FNH-like nodule in which GS expression was not increased, and without a CTNNB1 variant. These areas of increased GS expression were LCM and sequenced for CTNNB1 variants. In 2 of 4 of these lesions, somatic variants were identified with p.Thr41Ala substitutions. Other areas of the background liver in the matched patients did not show any evidence of CTNNB1 variants. These are likely to represent areas harboring sporadic variants that were identified incidentally on the histological review of background liver and were not of a size that could be identified radiologically. These are presumed to represent early stages of neoplastic evolution as part of a field change.

Fig. 4.

An example of a <1 mm micronodule in the background liver of CPSS patients seen at ×100 magnification. a H&E. b GS showing dense. c β-Catenin staining. d Ki67 staining showing few cells positive in the same area with GS and β-catenin positivity.

Fig. 4.

An example of a <1 mm micronodule in the background liver of CPSS patients seen at ×100 magnification. a H&E. b GS showing dense. c β-Catenin staining. d Ki67 staining showing few cells positive in the same area with GS and β-catenin positivity.

Close modal

Proliferation Rates Correlates with Tumor Classification rather than Presence of CTNNB1 Variant

The proliferative rate was calculated on 33/37 (89%) nodules – there was insufficient tissue to perform Ki67 staining in 4/37 (11%). Figure 5 shows the proliferative rate according to the nodule classification (a) as well as AA affected by substitution (b). The mean proliferative rate per ×200 field was similar in the background liver of CPSS patients 1.1 (IQR 0–1.5), compared to 1.2 (IQR 0–1.6) in controls. There was no significant increase in the mean proliferative rage when comparing benign lesions to background liver. Both β-catenin-activated HCAs (median 5.5, IQR 3.8–8.3) and HCC/HB (median 6.5, IQR 2.5–7.5) had a significantly increased proliferative rate compared to background liver (p = 0.03). The remaining types of benign tumors did not have a significantly higher proliferation rate than background liver, nor did the CPSS background liver compared to healthy control background liver.

Fig. 5.

The average rate of proliferation and standard deviation counted as Ki67-positive hepatocytes per ×200 field according to tissue histological classification on a logarithmic scale (a) and presence of specific AA substitutions (b). HC, healthy control; CPSS BL, CPSS background liver. One-tailed Mann-Whitney U tests were performed compared against CPSS background liver.

Fig. 5.

The average rate of proliferation and standard deviation counted as Ki67-positive hepatocytes per ×200 field according to tissue histological classification on a logarithmic scale (a) and presence of specific AA substitutions (b). HC, healthy control; CPSS BL, CPSS background liver. One-tailed Mann-Whitney U tests were performed compared against CPSS background liver.

Close modal

Median proliferative rate in benign tumors of CPSS patients with confirmed CTNNB1 variants was 0.5 (IQR 0–1.5), and like that observed in the background liver of CPSS patients (median 0.5, IQR 0–1.5; p = 0.65) and benign tumors without variants 2 (IQR 0.5–3; p = 0.17). However, the proliferative rate was significantly increased in tumors with substitutions of AA 32 or 33 (median 16.8, IQR 7.3–41.6) when compared to background liver (p = 0.004). It is unclear if this is due to the confounding factor that 3/4 (75%) of these variants were found in malignant tumors. Proliferation rates were also significantly increased when comparing variants affecting AA 32 or 33 to all other confirmed CTNNB1 variants (median 0.8, IQR 0–2.3; p = 0.004).

No Intratumoral Variation of CTNNB1 but Multiple Variants Identified in Livers with Multiple Nodules

Seven patients had multiple tumors sequenced and 6 (86%) had confirmed CTNNB1 variants. In these 6 patients, all were found to have 2–3 different CTNNB1 variants amongst 2–8 tumors sequenced per patient. The histological classification and variants identified are shown in online supplementary material 7.0.

In 7 tumors with confirmed CTNNB1 variants and adequate tissue, separate areas of the lesion were sequenced for CTNNB1 to investigate if there was any intratumoral variability in CTNNB1 variants. The individual variants were confirmed and no other CTNNB1 variants were identified. An example of one of the lesions and the sample areas is shown in online supplementary material 8.0. Thus, intrahepatic genetic variability existed between different tumors but we did not detect any intratumoral variability.

We have investigated by NGS the mutational landscape of a large series of hepatocellular tumors arisen in the livers of patients with CPSS and found that variants in CTNNB1 are the dominant molecular feature, whereas somatic and germline variants in other genes are comparatively less frequent in this setting. We identified CTNNB1 variants in the entire spectrum of tumors, ranging from FNH-like nodules to HCC and HB, and in incidental microscopic hepatocellular foci in the background liver. We also found an association between specific CTNNB1 variants and the lesional proliferative rate, although these were more frequent in malignant tumors. This high rate of CTNNB1 variants may account for the abnormal and unpredictable evolution of tumors in patients with CPSS, irrespective of their radiological or histological appearance. This emphasizes the need for in-depth investigations of tumors in CPSS patients, at a multidisciplinary level and to be based on close correlation between imaging, histological, immunohistochemical, and molecular features [1]. The presence of CTNNB1 variants in microscopic foci well below the imaging detection threshold argues in favors of close regular monitoring.

The precise role of CTNNB1 variants in tumor pathogenesis in patients with CPSS remains unclear. However, its association with a higher risk of malignant progression in HCAs is well documented, and these variants are present in approximately one-third of HCCs [11, 13]. CTNNB1 variants in exon 3 activate the Wnt pathway by affecting the β-transducin repeat-containing protein binding site of β-catenin, thereby inhibiting its breakdown. β-Catenin, a key regulator of the canonical Wnt pathway, translocates to cell nuclei and increases target gene expression including GS and the pro-proliferative genes c-Myc and cyclin D1. The presence of a CTNNB1 variant likely provides a survival advantage to hepatocytes in livers affected by CPSS and contributes to tumor formation, although the mechanism is unclear. A higher proportion of arterial blood is relatively depleted of nutrients compared to portal blood. This may partly mimic the microenvironment in zone 3 of normal liver sinusoids near the central veins, which display higher expression of GS and where nutrients are at lower concentrations when compared to zone 1. The higher oxygen partial pressures of arterial blood may also create an environment that favors cells with CTNNB1 variants. The lack of genotypic variability within the tumors studied implies the acquisition of the somatic variant early in tumor growth. Furthermore, finding different CTNNB1 variants in single patients with multiple nodules demonstrated the strength of the association between CPSS and somatic CTNNB1 variants. Different areas of the liver are affected in parallel even at pediatric ages; we are unaware of any other clinical diagnosis that has shown a strong association with CTNNB1 variants.

Increased arterial supply to the liver in CPSS independently contributed to nodule formation. In FNHs, hypertrophied arterioles are a characteristic feature that is believed to cause locally elevated sinusoidal pressures, resulting in hepatocellular hyperplasia and FNH formation [14]. Hypertrophied arterioles at a microscopic level and increased total hepatic arterial supply have been shown to be features of CPSS, compensating for the decreased portal inflow [7, 8]. The presence of FNH-like nodules without variants supports this hypothesis. The resolution of tumors observed after shunt closure and re-establishment of normal portal/arterial proportional inflow is particularly interesting. Our previously reported cohort included 11 patients with nodules at the time of the shunt closure. Four of 11 (36%) decreased in size and 3 (27%) resolved completely after shunt closure. [1] We further present a case in which a tumor with a confirmed variant was not detectible on hepatectomy 3 months after shunt closure and other existing tumors had evolved, some toward resolution and one to an HCC. It is possible that the underlying vascular anomaly and the resulting dominant hepatic arterial supply may act synergistically with mutational events in CTNNB1 to drive tumor evolution.

An unbalanced portal/arterial supply is characteristic but not exclusive to CPSS livers. For example, it can be observed in cirrhosis and Budd-Chiari syndrome, both of which are associated with the formation of liver tumors and FNH-like nodules [13]. In cirrhotic portal hypertension, increased resistance in the hepatic portal circulation reduces portal flow through the liver. In addition, disruption of the normal hepatic architecture by fibrous septa and bridging portal tracts has been shown by Tanaka and Wanless to be associated with intrahepatic portal venule occlusion, affecting 25%–36% of medium and large portal venous branches [15, 16]. These microenvironments in portions of cirrhotic livers have similarities with respect to the proportion of portal and arterial blood in CPSS livers. There are also common features between tumors found in patients with CPSS and atypical FNH-like nodules observed in cirrhotic livers. In both cases, GS staining atypical for FNHs was described [17, 18]. Somatic variants of CTNNB1 are identified in 27–37% of HCCs arising from cirrhosis [19‒21]. The cirrhotic population is highly heterogeneous with confounding risk factors for the development of liver tumors. However, it is possible that the association between portal deprivation and the presence of CTNNB1 variants is shared in both CPSS and cirrhotic livers.

The germline variants that were observed in CPSS patients have varying significance, with the majority not clearly contributing to malignancy (HFE, SERPINA1, and ZNF521). Loss of heterozygosity, with one somatic and one germline AXIN1 variant, was observed in one FNH-like nodule. AXIN1 is a protein required for the degradation of β-catenin, and thus, its absence would have a proliferative effect through overactivation of the Wnt pathway. Heterozygous AXIN1 variants have been identified in malignant liver tumors with uncertain significance; however, the loss of heterozygosity observed in one tumor is considered pathogenic in hepatic malignancies [22, 23]. Conversely, the HFE variants that were identified are not thought to have an impact on malignant transformation, as previous studies have shown a lack of association between HCC and HFE variants, albeit in patients with cirrhosis [24]. The p.Glu366Lys variant of SERPINA1 is a well-documented Z allele for alpha-1 antitrypsin deficiency. Heterozygosity has been reported to be a risk factor for cholangiocarcinomas, but is potentially protective against the development of HCCs in cirrhotic patients [25, 26].

FNHs in patients with no liver pathology have not previously been associated with variants or elevated malignant potential [27]. However, we have demonstrated that in CPSS, FNH-like nodules are commonly mutated. In some cases, the initial histological classification prior to systematic examination of GS staining was FNH; however, the lack of map-like GS staining led to an altered classification. We further found that in 2 control FNHs, nodules were atypical and thus should be classified as FNH-like, but the lack of increased GS expression correlated accurately with no CTNNB1 variant. FNH-like nodules in CPSS may clinically evolve more like HCAs with CTNNB1 variants, which are thought to lead to a higher rate of malignant transformation [9, 12, 15].

The genotypic-phenotypic correlation of CTNNB1 variants with nodule histological classification and staining patterns of GS and β-catenin in HCAs and HCCs has been well studied [28]. We have added to the body of knowledge by showing that, in the clinical context of CPSS, genotypic-phenotypic correlation also extends to FNH-like nodules. We have confirmed previous findings of GS and β-catenin staining patterns and variants observed in CTNNB1-mutated HCAs [9]. Variants affecting AA 45 were more commonly associated with a focal patchy pattern, and diffuse homogeneous patterns are a feature of variants affecting AA 32–41 of exon 3. We found that variants affecting AA 32 and 33 correlated with malignant lesions, which is similar to findings from a series of adult hepatocellular tumors [9, 28].

CPSS are complex patients with regard to radiological investigations, histological examination, DNA sequencing, and hepatological and surgical care. Radiologically, nodules display arterial enhancement, making it difficult to rule out malignancy. Histologically benign nodules may harbor CTNNB1 variants that are closely associated with malignant tumors. Clinically, these patients may have multisystem complications (e.g., hepatopulmonary syndrome) requiring additional expertise. These peculiarities may justify more aggressive investigation and treatment of nodules in patients with CPSS and advocate for a dedicated multidisciplinary approach to their detection, monitoring, and treatment.

Variants in CTNNB1 are common in hepatocellular lesions arising in the livers of patients with CPSS, irrespective of their radiological and/or histological appearance, and are also present in background microscopic foci. This may account for the unpredictable behavior of these lesions and emphasize the need for close monitoring.

Written informed consent for this study was not obtained as it was not required. This study has been granted ethical approval for the use of retrospective samples without requiring written informed consent by the UK Health Research Authority (Research Ethics Committee East Midlands Approval No. 16/EM/0342) and for samples from Pediatric Liver Tissue Biobank for selected samples (Research Ethics Committee Wales Approval No. 18/WA/0009).

The authors have no conflicts of interest to declare.

This study was supported by a joint grant from the Children’s Liver Disease Foundation and the British Society of Pediatric Gastroenterology, Hepatology, and Nutrition, King’s Health Partners, and Children’s Surgery at King’s College Hospital.

Athanasios Tyraskis: experimental design and execution, data gathering, data analysis and interpretation, and writing of manuscript tables and figures. Riley Cook: experimental design and analysis. Sandra Strautnieks: experimental design (next-generation sequencing), data gathering, and manuscript editing. Pierre Foskett: experimental design (next-generation sequencing), data gathering, and manuscript editing. Claudio De Vito: experimental design (next-generation sequencing and histology), and data gathering and analysis (histology). Maesha Deheragoda: data gathering (histology). Yoh Zen: study design, data gathering (histology), and manuscript editing. Alberto Quaglia: concept generation, experimental design, data gathering (histology), and manuscript editing. Mark Davenport and Nigel Heaton: concept generation, data analysis (clinical), and manuscript editing. Richard Thompson: concept generation, experimental design (next-generation sequencing), and manuscript editing.

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants and the requirement be the Research Ethics Committee to guard and store such date hospital computers. Noncompromising data are available to the corresponding and first author (A.T. and R.J.T.) upon reasonable request.

1.
Tyraskis
A
,
Davenport
M
,
Deganello
A
,
Sellars
M
,
De Vito
C
,
Kane
P
, et al
.
Complications of congenital porto-systemic shunts: liver tumors are affected by shunt severity, but pulmonary and neurocognitive associations are not
.
Hepatol Int
.
2022
;
16
(
4
):
918
25
.
2.
Bedossa
P
,
Burt
AD
,
Brunt
EM
,
Callea
F
,
Clouston
AD
,
Dienes
HP
, et al
.
Well-differentiated hepatocellular neoplasm of uncertain malignant potential: proposal for a new diagnostic category
.
Hum Pathol
.
2014
;
45
(
3
):
658
60
.
3.
Evason
KJ
,
Grenert
JP
,
Ferrell
LD
,
Kakar
S
.
Atypical hepatocellular adenoma-like neoplasms with β-catenin activation show cytogenetic alterations similar to well-differentiated hepatocellular carcinomas
.
Hum Pathol
.
2013
;
44
(
5
):
750
8
.
4.
Umetsu
SE
,
Joseph
NM
,
Cho
SJ
,
Morotti
R
,
Deshpande
V
,
Jain
D
, et al
.
Focal nodular hyperplasia-like nodules arising in the setting of hepatic vascular disorders with portosystemic shunting show β-catenin activation
.
Hum Pathol
.
2023
;
142
:
20
6
.
5.
Sorkin
T
,
Strautnieks
S
,
Foskett
P
,
Peddu
P
,
Thompson
RJ
,
Heaton
N
, et al
.
Multiple β-catenin mutations in hepatocellular lesions arising in Abernethy malformation
.
Hum Pathol
.
2016
;
53
:
153
8
.
6.
Sanada
Y
,
Mizuta
K
,
Niki
T
,
Tashiro
M
,
Hirata
Y
,
Okada
N
, et al
.
Hepatocellular nodules resulting from congenital extrahepatic porto-systemic shunts can differentiate into potentially malignant hepatocellular adenomas
.
J Hepatobiliary Pancreat Sci
.
2015
;
22
(
10
):
746
56
.
7.
Tyraskis
A
,
Deganello
A
,
Sellars
M
,
De Vito
C
,
Thompson
R
,
Quaglia
A
, et al
.
Portal venous deprivation in patients with porto-systemic shunts and its effect on liver tumors
.
J Pediatr Surg
.
2020
;
55
(
4
):
651
4
.
8.
de Vito
C
,
Tyraskis
A
,
Davenport
M
,
Thompson
R
,
Heaton
N
,
Quaglia
A
.
Histopathology of livers in patients with congenital porto-systemic shunts (Abernethy malformation): a case series of 22 patients
.
Virchows Arch
.
2019
;
474
(
1
):
47
57
.
9.
Sempoux
C
,
Gouw
ASH
,
Dunet
V
,
Paradis
V
,
Balabaud
C
,
Bioulac-Sage
P
.
Predictive patterns of glutamine synthetase immunohistochemical staining in CTNNB1-mutated hepatocellular adenomas
.
Am J Surg Pathol
.
2021
;
45
(
4
):
477
87
.
10.
Peng
Q
,
Vijaya
SR
,
Lewis
M
,
Randad
P
,
Wang
Y
.
Reducing amplification artifacts in high multiplex amplicon sequencing by using molecular barcodes
.
BMC Genomics
.
2015
;
16
(
1
):
1
12
.
11.
Bioulac-Sage
P
,
Rebouissou
S
,
Sa Cunha
A
,
Jeannot
E
,
Lepreux
S
,
Blanc
JF
, et al
.
Clinical, morphologic, and molecular features defining so-called telangiectatic focal nodular hyperplasias of the liver
.
Gastroenterology
.
2005
;
128
(
5
):
1211
8
.
12.
Chen
YW
,
Jeng
YM
,
Yeh
SH
,
Chen
PJ
.
P53 gene and Wnt signaling in benign neoplasms: beta-catenin mutations in hepatic adenoma but not in focal nodular hyperplasia
.
Hepatology
.
2002
;
36
(
4 Pt 1
):
927
35
.
13.
Cazals-Hatem
D
,
Vilgrain
V
,
Genin
P
,
Denninger
MH
,
Durand
F
,
Belghiti
J
, et al
.
Arterial and portal circulation and parenchymal changes in Budd-Chiari syndrome: a study in 17 explanted livers
.
Hepatology
.
2003
;
37
(
3
):
510
9
.
14.
Wanless
I
.
The pathogenesis of focal nodular hyperplasia of the liver
.
J Gastro Hepatol
.
2004
;
19
(
s7
):
s342
3
.
15.
Tanaka
M
,
Wanless
IR
.
Pathology of the liver in Budd-Chiari syndrome: portal vein thrombosis and the histogenesis of veno-centric cirrhosis, veno-portal cirrhosis, and large regenerative nodules
.
Hepatology
.
1998
;
27
(
2
):
488
96
.
16.
Wanless
IR
,
Wong
F
,
Blendis
LM
,
Greig
P
,
Heathcote
JE
,
Levy
G
.
Hepatic and portal vein thrombosis in cirrhosis: possible role in development of parenchymal extinction and portal hypertension
.
Hepatology
.
1995
;
21
(
5
):
1238
47
.
17.
Quaglia
A
,
Prasad
N
,
Nozza
P
,
Davies
S
,
Grasso
A
,
Burroughs
A
, et al
.
Focal nodular hyperplasia-like areas in cirrhosis
.
J Hepatol
.
2000
;
32
:
158
.
18.
Rebouissou
S
,
Couchy
G
,
Libbrecht
L
,
Balabaud
C
,
Imbeaud
S
,
Auffray
C
, et al
.
The beta-catenin pathway is activated in focal nodular hyperplasia but not in cirrhotic FNH-like nodules
.
J Hepatol
.
2008
;
49
(
1
):
61
71
.
19.
Cancer Genome Atlas Research Network
;
Electronic address: wheeler@bcmedu
;
Cancer Genome Atlas Research Network
.
Comprehensive and integrative genomic characterization of hepatocellular carcinoma
.
Cell
.
2017
;
169
(
7
):
1327
41.e23
.
20.
Totoki
Y
,
Tatsuno
K
,
Covington
KR
,
Ueda
H
,
Creighton
CJ
,
Kato
M
, et al
.
Trans-ancestry mutational landscape of hepatocellular carcinoma genomes
.
Nat Genet
.
2014
;
46
(
12
):
1267
73
.
21.
Schulze
K
,
Imbeaud
S
,
Letouzé
E
,
Alexandrov
LB
,
Calderaro
J
,
Rebouissou
S
, et al
.
Exome sequencing of hepatocellular carcinomas identifies new mutational signatures and potential therapeutic targets
.
Nat Genet
.
2015
;
47
(
5
):
505
11
.
22.
Satoh
S
,
Daigo
Y
,
Furukawa
Y
,
Kato
T
,
Miwa
N
,
Nishiwaki
T
, et al
.
AXIN1 mutations in hepatocellular carcinomas, and growth suppression in cancer cells by virus-mediated transfer of AXIN1
.
Nat Genet
.
2000
;
24
(
3
):
245
50
.
23.
Taniguchi
K
,
Roberts
LR
,
Aderca
IN
,
Dong
X
,
Qian
C
,
Murphy
LM
, et al
.
Mutational spectrum of beta-catenin, AXIN1, and AXIN2 in hepatocellular carcinomas and hepatoblastomas
.
Oncogene
.
2002
;
21
(
31
):
4863
71
.
24.
Boige
V
,
Castéra
L
,
De Roux
N
,
Ganne-Carrié
N
,
Ducot
B
,
Pelletier
G
, et al
.
Lack of association between HFE gene mutations and hepatocellular carcinoma in patients with cirrhosis
.
Gut
.
2003
;
52
(
8
):
1178
81
.
25.
Mihalache
F
,
Höblinger
A
,
Grünhage
F
,
Krawczyk
M
,
Gärtner
BC
,
Acalovschi
M
, et al
.
Heterozygosity for the alpha1-antitrypsin Z allele may confer genetic risk of cholangiocarcinoma
.
Aliment Pharmacol Ther
.
2011
;
33
(
3
):
389
94
.
26.
Rabekova
Z
,
Frankova
S
,
Jirsa
M
,
Neroldova
M
,
Lunova
M
,
Fabian
O
, et al
.
Alpha-1 antitrypsin and hepatocellular carcinoma in liver cirrhosis: SERPINA1 MZ or MS genotype carriage decreases the risk
.
Int J Mol Sci
.
2021
;
22
(
19
):
10560
.
27.
Perrakis
A
,
Demir
R
,
Müller
V
,
Mulsow
J
,
Aydin
Ü
,
Alibek
S
, et al
.
Management of the focal nodular hyperplasia of the liver: evaluation of the surgical treatment comparing with observation only
.
Am J Surg
.
2012
;
204
(
5
):
689
96
.
28.
Rebouissou
S
,
Franconi
A
,
Calderaro
J
,
Letouzé
E
,
Imbeaud
S
,
Pilati
C
, et al
.
Genotype-phenotype correlation of CTNNB1 mutations reveals different ß-catenin activity associated with liver tumor progression
.
Hepatology
.
2016
;
64
(
6
):
2047
61
.