Proteomic profiling of liver and plasma uncovers biomarkers and mechanisms in patients with primary biliary cholangitis
Communications Medicine, 2026
Kellemann J., Jensen A., Nielsen A., Werge M., Rashu E., Hetland L., Thing M., Nabilou P., Burisch J., Junker A., Hobolth L., Mortensen C., Tofteng F., Bendtsen F., Kjeldsen S., Møller S., Vyberg M., Serizawa R., Winther-Sørensen M., Ytting H., Gluud L., Wewer Albrechtsen N.
| Disease area | Application area | Sample type | Products |
|---|---|---|---|
Immunological & Inflammatory Diseases Hepatology | Cross-platform Validation | Plasma |  Olink Explore 3072/384 |
Abstract
Background
Primary biliary cholangitis (PBC) is a rare autoimmune liver disease that can progress to fibrosis and cirrhosis. As molecular biomarkers for PBC remain limited, we performed integrated liver and plasma proteomic analyses to identify disease-associated proteins, pathways, and candidate biomarkers.
Methods
In this cross-sectional study, we recruited 15 patients with newly diagnosed PBC, 17 patients with metabolic dysfunction-associated steatotic liver disease (MASLD), and 19 healthy controls. Untargeted mass spectrometry-based proteomic analyses were performed on liver tissue and matched plasma samples. Differential expression and pathway enrichment analyses were used to characterise disease-associated proteomic alterations. As proof of concept, logistic regression-based machine learning models were used to evaluate whether proteomic signatures could distinguish PBC from MASLD and healthy controls. Selected proteins were evaluated in an independent United Kingdom Biobank (UKB) cohort.
Results
Here we show that 1917 liver and 102 plasma proteins are significantly altered in PBC compared with healthy controls, including 26 altered across tissues. Differentially abundant proteins are enriched in immune activation, biliary epithelial cell function, hepatic injury, and metabolic pathways. Plasma proteomics-based machine-learning models distinguish PBC from MASLD and healthy controls (area under the receiver operating characteristic curve 0.96 ± 0.06). Of the 20 proteins identified by the classification model, 12 are available in the UKB dataset, of which seven are significantly associated with PBC after adjustment for age, sex and BMI.
Conclusion
Integrated liver and plasma proteomic analyses identify molecular pathways and circulating proteins associated with PBC. Further validation in larger, independent, age- and sex-matched cohorts is warranted.