Abstract
SPATIAL TRANSCRIPTOMICS IDENTIFIES IMMUNE EXPANSION OF SCAR REGIONS AND REDUCED HEPATOCYTE ZONATION IN LIVERS FROM PATIENTS WITH BILIARY ATRESIA
Background: Multiple metabolic and inflammatory pathways have been associated with patient outcome in biliary atresia (BA), however, the spatial relevance of these pathways among BA and non-BA disease controls has not been well defined. In the present study, we overcome this gap in knowledge and define the hepatic spatial transcriptome in pediatric cholestasis.
Methods: VISIUM Spatial transcriptomics was performed on frozen liver tissue obtained at the time of pediatric liver transplantation from 4 groups: 1) BA patients < 2 years of age (BA1, n = 3), 2) BA patients > 2 years of age (BA2, n = 4), 3) non-BA cholestatic patients (nonBA, n = 4), and 4) non-diseased pediatric donor liver (NL, n = 3). Published single-cell datasets were used to infer cell-type composition of hepatocyte and scar regions. Standard statistical tools were used to integrate and identify differential gene expression (DE), and tissue composition within and between samples. Metabolic pathways were modeled from gene expression using the compass algorithm.
Results: BA1 patients had higher direct bilirubin level at the time of sample collection with median of 20.4 mg/dL (IQR 11.7-NA) compared to BA2 and nonBA groups with medians of 2.0 mg/dL (IQR 0.6-2.2) and 4.8 mg/dL (IQR 1.3-11.7) respectively (p = 0.004). This was associated with an increase in bile acid synthesis pathways and a trend towards greater cholangiocyte signature (Figure 1) in BA1 compared to BA2. In contrast, BA2 had higher immune infiltration in scar regions (Figure 1A) and arginine and proline metabolism. BA1 scars were most unique with 662 DE genes, whereas BA2 and nonBA scars had fewer than 50 DE genes. More specific comparison among the scar signatures of BA patients demonstrated distinct scar-associated clusters involved in angiogenesis and T cell activation in BA1 versus enriched processes of translation and peptide metabolic processes in BA2. Both BA groups showed reduced hepatocyte zonation as demonstrated by lower portal hepatocyte gene signature than nonBA and NL groups (Figure 1B).
Conclusion: We identify distinct spatial transcriptomes in children with cholestatic liver disease that differ by etiology and disease severity. BA patients requiring earlier transplant (BA1) exhibited a higher bile acid metabolic signature and greater heterogeneity within the scar region than older BA patients (BA2). These findings help identify aberrant immune-metabolic patterns that may contribute to disease progression in BA.