Abstract / Summary
Advanced fibrosis is a major determinant of liver–related mortality in metabolic dysfunction–associated steatohepatitis (MASH), yet current non–invasive tests lack specificity for fibrosis staging and provide limited resolution of liver–proximal disease biology. We investigated whether circulating extracellular vesicle (EV)–derived proteomics could identify a validated biomarker signature of advanced fibrosis in MASH and provide biological insight into underlying disease mechanisms. Plasma EVs were isolated from 222 adult participants in the NIDDK NASH Clinical Research Network with biopsy–confirmed MASLD. Participants (mean age 52 years; mean BMI 33 kg/m 2 ) included individuals with no fibrosis (F0) and advanced fibrosis (F3–F4), representing the extremes of the histologic fibrosis spectrum. A discovery cohort (n = 112) matched for age, sex, and BMI, was used for the initial signature identification, followed by testing in an independent, unmatched validation cohort (n = 110). Unbiased high–resolution mass spectrometry (DIA–PASEF) was used for EV proteomic profiling, followed by differential abundance analysis and machine learning–based classifier development using EV proteins alone or in combination with clinical variables. We identified and validated an 11–protein EV signature associated with advanced fibrosis (F3–F4). Proteins including QSOX1, FCGR3A, PIGR, and TGFBI, showed abundance patterns that tracked with fibrosis severity in the sampled stages, consistent with fibrosis–associated disease biology. EV cargo showed significantly greater enrichment for liver–associated signals compared with bulk plasma/serum proteomic datasets. In the validation cohort, an Elastic Net model integrating EV proteins with clinical variables achieved an AUROC of 0.807 (95% CI: 0.722–0.893), compared with 0.645 for the corresponding clinical model. Circulating EV proteomics identified and independently validated a signature associated with advanced fibrosis in MASH and captured liver–enriched signals that were less apparent in bulk plasma/serum analyses. These findings support EV proteomics as a complementary approach for fibrosis assessment and as a potential source of biologically informative signals associated with liver disease.