The Human Pan Disease Atlas article is now available in the latest issue of Science released December 18. The article describes how a next-generation targeted proteomics assay was used to analyze the blood profiles of thousands of patients representing most major disease classes, and to assess the stability and variability of protein profiles in healthy adults as well as for the child to adult development.

Next-generation plasma profiling was used to investigate the proteome profiles in blood from more than 8000 individuals covering both healthy longitudinal samples and patients representing 59 different diseases. The strategy based on a proximity extension assay set-up allowed 5000 proteins to be simultaneously measured with high sensitivity from a small drop of blood.

Disease cohorts including cancers and infectious, autoimmune, cardiovascular, and metabolic diseases was used to investigate the distinct and shared proteomics signatures across diseases. This multi-comparison approach revealed that even if many protein profiles showed disease-specific abundances, several diseases share a large number of differentially abundant proteins, particularly among liver-related and infectious diseases. A pan-disease strategy is therefore suggested as a useful complement to avoid the reproducibility issues common in classic case-control studies.

The effects on the blood proteome during childhood development, and how the blood proteome reflects the biological age was also investigated. The dynamic changes in the plasma proteome from childhood to adulthood was investigated by using a longitudinal cohort where samples were collected from the same 100 individuals at 4, 8, 16, and 24 years of age. The results show a great influence of age and sex on protein concentrations during development with many proteins having significantly lower abundances with increasing age and the other way around. For proteins with increasing concentrations the data suggests that adult-like concentrations are typically reached by age 16.

To investigate the relationship between chronological and biological age models from a machine learning approach was used to predict biological age on the basis of blood profiles. By using the 50 most age-related proteins the chronological age could in most cases be accurately predicted.

Read the article