A blood score for “zombie cell” signals was linked to death and disease in UK Biobank

As we age, some of our cells stop dividing for good but don’t die. These senescent cells, sometimes called “zombie cells”, stay active and release a mix of signalling proteins that can inflame nearby tissue and push other cells into the same state. This mix is called the SASP, short for senescence-associated secretory phenotype. Senescent cells are thought to contribute to many diseases of old age, and drugs that remove them (senolytics) are being tested.

The problem is measuring them. Senescent cells make up only a tiny share of the cells in any tissue, and you can’t take biopsies from everyone. Their SASP proteins, however, end up in the blood. A team at the UConn Center on Aging tried to turn a set of those blood proteins into a single number that reflects how many senescent cells a person carries.

What they did

From databases of lab experiments, the team chose 38 proteins that senescent cells of several types consistently release. They then used blood measurements of these proteins from about 51,000 participants of the UK Biobank, a large British study of adults recruited in middle age.

They trained an AI model, a kind of neural network, on 85% of these people (about 43,000) to compress the 38 protein levels into one SASP Score. Age was used to guide the training but is not part of the final score. They then tested the score in the remaining 15% (7,639 people), comparing it with health records over a mean follow-up of 13.3 years.

Finally, they applied the score to blood samples from MEDEX, a trial in which 585 people aged 65 to 84 with mild memory complaints had been randomly assigned to exercise, mindfulness training, both, or neither, measured at the start, after 6 months and after 18 months. These samples had been measured with a different laboratory method, so the model was first adjusted to the new data.

What they found

• It tracked age and other aging measures. In the test group, the score rose with age and correlated with other estimates of biological age. Higher scores went with being male, smoking, a larger waist relative to the hips, frailty, high blood pressure, weaker grip and slower walking.
• Higher score, higher risk of death. After accounting for age, sex, education, deprivation, smoking, drinking, blood pressure and body weight, each step of one standard deviation up the score was linked to about 41% higher risk of dying during follow-up.
• Linked to many diseases. A higher score was also linked to new cases of chronic kidney disease, heart failure, lung cancer, delirium, dementia and Parkinson’s disease, among others.
• Better than single proteins. The combined score predicted death and every disease studied except diabetes more strongly than any one of the 38 proteins alone. The five biggest contributors were GDF-15, CXCL-9, PGF, TNFRSF11B and CCL-20, which the authors tie to the cell’s power plants, immune responses and the control of cell death.
• Exercise: a hint, not a result. In MEDEX, the formal test found no significant difference between the exercise and non-exercise groups over 18 months. Looking within each group, the score rose over time in those who didn’t exercise but not significantly in those who did.

What to keep in mind

• Observational. The UK Biobank results show links, not causes. A high score doesn’t prove senescent cells caused the deaths or diseases.
• Not a pure senescence signal. The authors note the score may also pick up general inflammation or tissue damage, that it isn’t known which cells the proteins in the blood come from, and that there is no agreement on which SASP proteins are universal. They urge caution “to avoid overinterpretation”.
• The exercise finding is weak. The main test of exercise was not significant; the within-group pattern is suggestive at most. Exercise did not lower the score.
• Who was studied. UK Biobank participants are mostly of European descent and better off than the UK average. In the older MEDEX group, the score correlated much less strongly with age.
• Not a test you can order. This is a research tool. The team shares its code for other researchers.

Why it matters

Trials of drugs or lifestyle changes aimed at senescent cells need a way to see whether they work without waiting years for illness or death. A blood score that reflects the overall load of senescent cells, and that can be calculated from data gathered with different lab methods, could make such trials cheaper and faster. This study shows the score predicts health outcomes in a large population; the next step, as the authors say, is to confirm it in other groups and see whether treatments move it.