An immune cell atlas of 2609 people identifies a blood marker whose association with mortality is comparable to that of male sex: a lifelong herpesvirus carried by most adults drives the marker, while age has little effect
An immune cell atlas of 2609 people identifies a blood marker whose association with mortality is comparable to that of male sex: a lifelong herpesvirus carried by most adults drives the marker, while age has little effect
On October 9, a team from Washington University in St. Louis, King’s College London and several other institutions published a paper in Immunity. They combined single-cell blood data from 2609 people across four continents, covering 12,4 million immune cells, and found that killer immune cells age differently in people of the same age. They then tested a protein-based score derived from this finding in 48 620 UK Biobank participants followed for 17 years.
Earlier studies of human immune aging showed limited agreement across laboratories, so the team compared eight independent cohorts and retained only changes found in at least five. The most consistent signal was a familiar process, now measured more precisely: the thymus, an organ behind the breastbone that produces new T-cells, becomes less active with age, leaving fewer new T-cells in the blood. An “immune clock” based on this process estimates a person’s age with an error of about eight years.
The second finding was largely independent of age. Memory killer T-cells (CD8 lymphocytes that kill infected and tumor cells) include two groups: one uses the enzyme granzyme K, the other granzyme B. One of the study’s leaders is Maxim Artyomov, an immunologist at Washington University. His laboratory first observed this distinction in a single cohort in 2023 and developed a way to measure T-cell output from the thymus using the protein CD38 in 2024. In young people, the two groups are present in roughly equal proportions, and age itself changes their ratio only slightly: it rises by a factor of 1,2 from age 18 to 64, then falls to 0,64 times its previous level by age 97. This represents a dimension of variation between people that is distinct from age. Cytomegalovirus, a lifelong and usually asymptomatic herpesvirus carried by most adults, shifts the balance sharply toward granzyme B. People with chronic cardiovascular disease and lupus showed a greater shift even in the absence of the virus.
This finding highlights a specific difficulty in applying mouse results to humans. In old mice, granzyme K cells represent the terminal, exhausted stage of T-cell development and mark inflammatory aging in tissues: transferring these cells into young mice impaired memory, while blocking the enzyme improved memory in old mice. Human blood shows the reverse pattern: cells at an intermediate stage carry granzyme K, while terminal-stage cells, which are the most inflammatory, carry granzyme B. The human counterparts of the harmful mouse cells are found in tissues, such as tumors, rather than in blood. The authors consider the reason for this difference unresolved.
Counting these cells directly requires expensive single-cell sequencing, which is impractical for large cohorts. The team therefore identified blood plasma proteins associated with the proportion of granzyme B cells and used them to train a simple model. When they applied it to 48 620 UK Biobank participants, they found that among people aged 60 to 70, those in the highest quarter of the score had a 43% higher risk of death from any cause than those in the lowest quarter. This was comparable to the difference between men and women in the same model, in which men had a 66% higher risk. The association also held in younger people, independently of sex. People with higher scores were also more likely to develop sepsis, type 2 diabetes, gout, liver disease and kidney failure. The score was not associated with dementia or Parkinson’s disease.
The atlas links immune profiles to mortality risk, but there is currently no drug or treatment protocol that corrects the granzyme balance. Five of the paper’s coauthors, including Artyomov, have already filed a provisional patent application for the protein test itself.