NEW YORK – A team led by investigators at Fudan University in China has assembled a proteomic atlas focused on blood plasma proteins with apparent ties to a wide range of traits and conditions, unearthing potential blood biomarkers and potential drug targets or treatment approaches.
"[O]ur study symbolizes major strides toward achieving a comprehensive understanding of the plasma proteomic atlas for human health and disease, with clinically actionable insights to integrate the advantages of the proteome across disease diagnosis, prediction, and treatment," co-senior and co-corresponding author Jin-Tai Yu, a neurology and neurological disorders researcher with Fudan University, and colleagues wrote in the journal Cell on Friday.
Using electronic health record data and plasma proteomic profiles generated with an Olink Explore Proximity Extension assay and next-generation sequencing by members of the UK Biobank project's Pharma Proteomics Project (PPP) consortium, the researchers searched for blood proteins associated with the presence or onset of 1,706 human traits or conditions in a randomized subset of 53,026 UKB participants followed for a median of 14.8 years.
The health features considered included 406 prevalent disease endpoints across 14 disease categories or "chapters," 660 incident disease endpoints from 13 disease chapters, and 986 health-related traits across 11 chapters.
"By implementing an analytical approach that incorporates comprehensive health-related phenotypes and assessing their associations with plasma protein levels uniformly, proteins exhibiting multiple significant associations were identified," the authors explained.
Based on data for 2,920 plasma proteins, investigators highlighted 168,100 ties between plasma proteins and disease, as well as 554,488 protein-trait associations. Most of the proteins seemed to coincide with multiple traits or conditions.
"Over 650 proteins were shared among at least 50 diseases, and over 1,000 showed sex and age heterogeneity," the authors reported, adding that blood plasma proteins "demonstrated promising potential in disease discrimination."
In an effort to dig into causal proteins behind the associations, the team went on to do proteome-by-phenome Mendelian randomization analyses based on published genome-wide association study and protein quantitative trait locus data.
Machine-learning methods helped the researchers put together prediction and diagnostic models, while an enrichment analysis done with the "Genome for repositioning drugs" (GREP) software led to 26 proposed drug targets along with clues for repurposing 37 approved drugs.
The team has compiled its plasma proteome-phenome data in an open-access atlas that is intended to help other teams continue coming up with blood-based protein biomarkers, methods for predicting disease development and trajectory, and identifying possible targets for future treatments.
"Moving forward, the research community will benefit from this open-access proteomics atlas to allow a deeper understanding of disease pathogenesis and promote the effective development of biomarkers, predictive models, and therapeutic targets," the authors concluded.