¹Department of Gastroenterology and Hepatology, Medical University of Gdansk, Gdansk, Polen
²Centre for Functional Genomics and Bio-Chips, Institute for Biochemistry and Molecular Genetics, Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia
³Faculty of Computer and Information Science, University of Ljubljana, Ljubljana, Slovenia
⁴Hannover Health Science Campus, Hannover Medical School (MHH), Hannover, Germany
⁵Klinik für Gastroenterologie, Hepatologie und Transplantationsmedizin, Essen, Deutschland
⁶Institute of Clinical Chemistry and Clinical Pharmacology, University Hospital Bonn, Bonn, Germany
⁷Liver and Internal Medicine Unit, Department of General, Transplant and Liver Surgery, Medical University of Warsaw, Warsaw, Poland

Text

Background: Primary biliary cholangitis (PBC) is a chronic autoimmune cholestatic liver disease of unknown origin. Ursodeoxycholic acid (UDCA) is the first-line treatment for PBC; however, some patients exhibit insufficient biochemical response to UDCA, indicating a worse prognosis and reduced survival. Cholesterol plays diverse biological roles and is derived in humans from either de novo synthesis or intestinal absorption. This study aimed to analyze the serum sterol profile and its association with PBC phenotypes.

Patients and Methods: A total of 129 PBC patients (age 32–93 years, 119 women) were included. Of these, 44 had cirrhosis at diagnosis, and 13 had undergone liver transplantation for PBC prior to inclusion. Biochemical response to UDCA, assessed using Barcelona criteria, was available for 62 (48%) patients. Serum sterol levels were measured by gas chromatography/mass spectrometry (GC/MS). Machine learning (ML) algorithms were employed to develop a model predicting UDCA response.

Results: Cirrhotic PBC patients exhibited significantly higher levels of sitosterol, campesterol, and cholestanol compared to non-cirrhotic individuals (all P < 0.01). Cirrhotic patients also showed increased phytosterol:cholesterol ratios and reduced lathosterol:cholesterol ratios (all P < 0.01), while desmosterol:cholesterol ratios were similar between cirrhotic and non-cirrhotic groups (P > 0.05). Post-transplantation patients had elevated 7α-hydroxycholesterol levels, likely reflecting increased radical oxidation. An ML Support Vector Classifier model incorporating 24-hydroxycholesterol, 27-hydroxycholesterol, campesterol, cholesterol, desmosterol, lanosterol, and stigmasterol demonstrated high accuracy, precision, and sensitivity (> 80%) in predicting UDCA non-response.

Conclusions: PBC patients with liver cirrhosis are characterized by reduced cholesterol synthesis and increased sterol absorption compared to patients without cirrhosis. Biochemical response to UDCA therapy is associated with serum levels of specific sterols, providing potential insights into disease management and prognosis.