Reference data reversed an inferred age link in human blood
In 981 blood donors, all 80 references covering six cell types reversed an immune cell’s estimated age association compared with data carrying known cell labels.
Limit: References averaged across cell states, missing age-related shifts within CD8 T cells
Edited by Boris Djordjevic · AgingResearch.News
An observational analysis of 981 blood donors tested how reference data from individual cells affected estimates of cell proportions in mixed-cell samples. Researchers use these references to separate a sample’s combined signals into contributions from different cell types.
For CD8 T cells, a type of immune cell, the inferred association with age ran opposite to the association in data with known cell labels in all 80 references covering six cell types. These cells shifted between states with age, which references averaging across states could not track. The reversal also occurred with three other estimation methods, pointing to the reference rather than a single method as its source. The researchers concluded that removing this systematic error required references that distinguished the shifting cell states.
Why it matters
The work addresses a problem for aging research: whether estimated differences in immune-cell proportions reflect age-related biology or errors in the reference data.
Caveats
This observational analysis tested errors in estimates, not whether immune-cell proportions cause aging. The paper is a preprint and has not yet been peer-reviewed.
The paper
University of Edinburgh
Research Square, Oct 26, 2026, Open access, Preprint, not peer-reviewed
