Affinity Proteomics-Based Non-Invasive Detection of Clinically Significant Liver Disease

<h4>Background</h4>Non-invasive biomarkers predicting major adverse liver outcomes (MALO) are urgently needed.<h4>Aims</h4>We assessed a novel, proximity extension assay-based high-throughput targeted proteomics.<h4>Methods</h4>Plasma proteomic data (> 2900 proteins) and clinical information were accessed from the population-based UK Biobank cohort, comprising ~52,000 individuals with a median follow-up of > 10 years, including obese (~12,700) and diabetic (~1500) participants. Validation cohorts comprised 287 participants with severe alpha1-antitrypsin deficiency (Pi*ZZ genotype), and 960 people living with HIV, who underwent liver stiffness measurement (LSM) via transient elastography. Selected proteomic parameters were compared to routine measurements. Bayes-moderated linear models (covariates: age and sex) assessed the differential abundance. Logistic regression was used to develop and validate a prognostic score.<h4>Results</h4>Routine gamma-glutamyltransferase (GGT) and aspartate aminotransferase (AST) strongly correlated with proteomic measurements (r = 0.92 and r = 0.71, respectively). Similarly, proteomic-based thrombospondin-2 levels correlated with immunoassay-based values (r = 0.85). Twenty proteins were consistently associated with future MALOs/increased liver stiffness in all cohorts. A novel five-component proteomic score derived from the UK Biobank cohort demonstrated superior predictive power for MALOs (AUROC = 0.84) compared to AST-to-platelet-ratio index (AUROC = 0.73) and Fibrosis-4 index (AUROC = 0.72), with stable performance in obese and diabetic subcohorts. In people living with HIV and alpha1-antitrypsin deficiency patients, all five components increased across fibrosis stages and the proteomic score outperformed AST-to-platelet-ratio/Fibrosis-4 index in predicting significant LSM-based liver fibrosis.<h4>Conclusions</h4>We identify a new proteomic score comprising epithelial/hepatic stellate cell markers that yields a robust predictive performance across different liver disease aetiologies. Thereby, we demonstrate the usefulness of emerging proteomic techniques in hepatology.<h4>Trial registration</h4>The registry study is listed at ClinicalTrials.gov (NCT02929940).