Background and Objectives: Non-alcoholic fatty liver disease (NAFLD) is a prevalent
chronic liver disorder with increasing global impact, yet reliable, non-invasive
biomarkers for diagnosis and disease monitoring remain limited. Extracellular
vesicles (EVs), small heterogeneous membrane-bound particles carrying proteins,
nucleic acids, and lipids, are emerging as promising candidates for biomarker
discovery. However, the lipid composition of urinary EVs and their potential
role in NAFLD, especially in non-alcoholic steatohepatitis (NASH), has not been
comprehensively explored. This study aimed to investigate the lipid molecular
profiles of urinary EVs and evaluate their utility as non-invasive biomarkers for
NASH detection.
Methods: Urinary EVs were isolated from 43 patients with non-alcoholic fatty liver
(NAFL) and 40 patients with biopsy-confirmed NASH using the EXODUS method
for purification. Lipidomic profiling was conducted via liquid chromatographytandem
mass spectrometry (LC-MS/MS). Comparative analyses of lipidomic
patterns between NAFL and NASH cohorts were performed, and machine learning
algorithms were applied to identify a panel of lipid markers predictive of NASH.
Results: Analysis revealed 422 distinct lipid species within urinary EVs, including
sterol lipids, fatty acyl lipids, glycerides, glycerophospholipids, and sphingolipids.
Machine learning and random forest modeling identified a four-lipid signature—
FFA (18:0), LPC (22:6/0:0), FFA (18:1), and PI (16:0/18:1)—capable of distinguishing
NASH from NAFL with a high diagnostic accuracy (AUC = 92.3%). These lipid species
were closely associated with NASH pathogenesis and progression.
Conclusion: Urinary EV lipidomics offers a promising, non-invasive approach for
NASH biomarker discovery. The identified lipid signature demonstrates potential
for clinical application in early diagnosis and monitoring of disease progression,
providing novel insights into the molecular mechanisms of NAFLD and NASH.