A Drop of Blood Reveals the Pace of Cellular Aging: Protein Clocks Map Risks of Dementia, ALS, and Cancer Years Earlier

Different cells in the same person do not necessarily age according to the same clock. A study published in *Nature Medicine* shows that protein signals circulating in the blood may allow researchers to estimate the biological age of different cell populations, including astrocytes, skeletal muscle cells, and alveolar cells, without directly accessing the skin or organs, and to identify groups at high risk of certain diseases before symptoms appear.

The research team analyzed data from 60,542 people across the Global Neurodegeneration Proteomics Consortium, UK Biobank, and the UK 1946 Birth Cohort Study. Using expression sources recorded in the Human Protein Atlas, the researchers mapped more than 7,000 plasma proteins to neural, immune, glial, epithelial, musculoskeletal, and other cells. Machine-learning models first learned the relationship between proteins and chronological age among healthy participants, then calculated whether each cell type was “older” or “younger” than the average for people of the same age. Models covering more than 40 cell types passed quality screening.

These clocks did not depict uniform aging throughout the body. Across different datasets, approximately 20% to 25% of participants showed extreme accelerated aging in only one cell type, while approximately 1% to 3% showed accelerated aging in at least ten cell types simultaneously. People who smoked and had obesity generally exhibited older cellular profiles, while those with healthier lifestyles tended to have younger ones. However, these observations remain statistical associations and do not establish that changing lifestyle habits can reverse the age of specific cells.

During up to 15 years of follow-up in UK Biobank, people with extremely aged skeletal muscle cell profiles had a 12.74-fold risk of subsequently developing amyotrophic lateral sclerosis compared with those whose profiles were relatively young. The association remained even when only cases diagnosed at least three years after blood collection were included. Extremely aged astrocytes corresponded to a 12.59-fold risk of Alzheimer’s disease. These figures represent relative risks between different aging groups; they do not mean the test can predict that a particular person will inevitably develop the disease, nor should the 15-year follow-up period be interpreted as meaning that every case was detected 15 years in advance.

The intersection of genetics and cellular state also provided a more detailed picture of risk. On average, APOE4 carriers showed older astrocytes and younger macrophages, while APOE2 showed the opposite pattern. The study also found that smokers who simultaneously had signals of extreme aging in airway epithelial cells and type II alveolar cells had a 58% higher risk of lung cancer than smokers overall. The team’s multicellular aging risk score also reproduced mortality risk stratification across different cohorts and two proteomics platforms.

However, plasma proteins are not cellular identity cards: the same protein may be produced by multiple cell types, and transcriptional expression does not necessarily correspond to protein concentrations in the blood. The models were limited to cells included in the Human Protein Atlas, and the study population consisted predominantly of older White individuals, so whether the findings apply to younger and more diverse populations remains to be validated. The analysis is currently for research use only. Before it can be used for screening, prospective clinical studies will still need to clarify its accuracy, the consequences of false-positive results, testing thresholds, and whether high-risk results can ultimately lead to effective preventive measures.