Granzymes are molecular weapons these cells use to destroy targets or coordinate immune responses. Granzyme B–producing cells act as direct destroyers of diseased cells. Granzyme K–producing cells are less well studied and may work more like alarms, calling in immune backup.

Can the ratio predict future illness?

To find out, the researchers turned to the UK Biobank, a database that tracked 500,000 initially healthy participants aged 40–69 for up to 15 years, along with blood protein measurements.

Because the Biobank did not count CD8 T cells directly, the team first trained a computer model on a smaller dataset that included both cell counts and blood proteins. The model learned to recognize the protein signatures left behind when either granzyme B– or granzyme K–producing cells dominated. The researchers then applied it to baseline blood samples from 50,000 Biobank participants and calculated each person's position on the spectrum.

The 15-year medical records showed a clear pattern: healthy adults whose immune systems leaned heavily toward granzyme B cells at baseline had a higher risk of death ten years later than those with more granzyme K cells. The granzyme B–dominant group was also more likely to develop type 2 diabetes, hypertension, liver disease and renal failure, among other chronic conditions, over the following decade.

"A higher level of granzyme B cells in a healthy state means the immune system might be already reacting to something," said co-author Marina Terekhova, an instructor in pathology and immunology at WashU Medicine. "It isn't a formal disease diagnosis, but it may indicate the body is off track."

Why do immune systems age differently?

The findings suggest a healthy immune system gradually shifts toward granzyme K cells over time, while one that leans toward granzyme B cells signals unhealthy aging. Because some individuals reach the granzyme B state much earlier in life, their immune systems follow different trajectories — explaining why people born in the same year can face different risks for chronic disease and death decades later.

The study is observational. It shows an association between the cell ratio and later health outcomes, not a proven cause, and the protein-model approach is an indirect measure of cell counts.

Will this become a blood test?

Mapping a patient onto the immune aging spectrum today would require complex and costly specialized technology. Artyomov and his lab are now adapting the research into a simple blood test that standard laboratory equipment could process, which could make early immune health monitoring part of routine preventive care.

"We can think of it like checking a car's oil level before the check engine light ever turns on," Artyomov said. "Our goal is to develop an early-warning diagnostic that alerts physicians to subclinical immune stress, prompting more specialized testing that could allow for intervention long before chronic disease symptoms appear."

The paper, "Single-cell analyses of global cohorts outline determinants of immune aging and link the ratio of GZMK⁺ to GZMB⁺ Tem cells to health trajectories," appears in Immunity (DOI: 10.1016/j.immuni.2026.09.012).