In collaboration with Odense University Hospital in Denmark, the Mann group at Novo Nordisk Foundation Center for Protein Research have developed a machine learning model based on their biomarker panel, which for the first time outperforms existing tests, laying the foundation for a generic proteomics liver health assessment.
METHODS
They applied high-throughput MS-based proteomics technology to profile paired liver and plasma samples from patients in a large clinically well-characterized cohort of alcohol-related liver disease and matched healthy controls. Proteome regulation in the liver and plasma revealed changes at pathway and biological processes levels. Integrated liver and plasma proteomics helps to capture disease stage-relevant protein signatures in the bloodstream which are concordant with the liver. Lastly, they built a machine learning model to identify early stages of liver fibrosis, inflammation and steatosis. This represents the largest and most detailed characterization of proteome changes in liver disease.
CLINICAL PERSPECTIVES
By integrating liver and plasma proteomics, together with clinical data on patients and healthy controls in the cohort they were able to define plasma biomarker panels for different hepatic lesions of alcohol-related liver disease. When combined with machine learning, the plasma biomarker panels yielded predictive models that either performed as well as existing diagnostic strategies, or outperformed them, including in the context of predicting risk of future adverse liver events. Their study indicates that screening high risk individuals by proteomics could predict disease progression from an asymptomatic stage to symptomatic, therefore informing treatment options and intervention opportunities.
FUTURE DIRECTIONS
With ongoing improvements in technology, they expect an increase in performance of their model. Targeted or ‘global targeted’ MS-based assays could be developed to retain the full specificity of this technology. An additional benefit of plasma proteomic profiling is its generic nature, meaning that it provides additional information apart from the targeted panel.
→ Read full publication on biorxiv
→ Fore more data using the Evosep One technology visit our Literature Room
LET’S STAY IN TOUCH
Join our monthly newsletter to stay tuned on new data and publications using Evosep One technology.