Abstract
Non-invasive biomarkers of non-alcoholic fatty liver disease (NAFLD) supporting diagnosis and monitoring disease progression are urgently needed. The present study aimed to establish a bioinformatics pipeline capable of defining and validating NAFLD biomarker candidates based on paired hepatic global gene expression and plasma bioanalysis from individuals representing different stages of histologically confirmed NAFLD (no/mild, moderate, more advanced NAFLD). Liver secretome gene signatures were generated in a patient cohort of 26 severely obese individuals with the majority having no or mild fibrosis. To this end, global gene expression changes were compared between individuals with no/mild NAFLD and moderate/advanced NAFLD with subsequent filtering for candidate gene products with liver-selective expression and secretion. Four candidate genes, including LPA (lipoprotein A), IGFBP-1 (insulin-like growth factor-binding protein 1), SERPINF2 (serpin family F member 2) and MAT1A (methionine adenosyltransferase 1A), were differentially expressed in moderate/advanced NAFLD, which was confirmed in three independent RNA sequencing datasets from large, publicly available NAFLD studies. The corresponding gene products were quantified in plasma samples but could not discriminate among different grades of NAFLD based on NAFLD activity score. Conclusion: We demonstrate a novel approach based on the liver transcriptome allowing for identification of secreted hepatic gene products as potential circulating diagnostic biomarkers of NAFLD. Using this approach in larger NAFLD patient cohorts may yield potential circulating biomarkers for NAFLD severity.