Cell Type-Specific Aging Predicts Disease Onset

A new study has used aging trajectories of various cell types to predict diseases such as Alzheimer’s and lung cancer [1]. This expands on previous research into organ-specific aging.

Age is more than one number

Gone are the days when aging was assumed to be happening uniformly across every tissue at once. Today, we know that certain organs and systems in the body often exhibit either accelerated aging or, conversely, unusual resilience, and that can influence morbidity and mortality. This idea is associated in particular with the group led by Tony Wyss-Coray of Stanford University and their 2023 seminal paper on organ-specific aging [2]. The diagnostic and therapeutic potential of this area of research is hard to overstate.

In a new study published in Nature Medicine, Wyss-Coray and colleagues took their idea further, making a conceptual jump from organs to cell types. The researchers “developed machine learning models to estimate the biological age of over 40 cell types,” per the abstract, and the results are informative and, at times, surprising.

Clocks and predictions

Using single-cell RNA sequencing from the Human Protein Atlas, the authors classified a gene as “cell-type-enriched” if it was expressed much more strongly in one cell type than in any other and then linked those genes to their plasma protein products. For each cell type, the researchers trained a machine-learning model to estimate a person’s chronological age from how those associated proteins rise, fall, or hold steady across the lifespan. In total, data from about 60,000 people across three cohorts was analyzed to test whether these cell-type aging signatures track disease and death.

Notably, the clocks’ predictive power varied widely; for instance, predictions from liver cells (hepatocytes) were more robust than predictions from excitatory neurons. These cell type-specific clocks revealed that 35% of people had no extreme age gaps in any cell type, 24% had extreme aging in exactly one cell type, and 1.5% exhibited extreme aging across ten or more cell types. The rest showed extreme aging in anywhere from two to nine cell types.

Apparently, at least some of these clocks can actually predict diseases. For instance, extreme aging of skeletal muscle cells (myocytes) strongly predicted incident amyotrophic lateral sclerosis (ALS), a devastating neurodegenerative disease, even for cases diagnosed more than three years after the blood draw. Extreme aging of neuron-supporting brain cells (astrocytes) predicted incident Alzheimer’s disease. Other diseases the researchers were able to predict, although less robustly, included lung cancer, lymphoma, type 2 diabetes, COPD, and stroke.

“Twenty years ago, we started to explore the idea that immune and other signaling proteins in the circulation could provide insights into Alzheimer’s disease,” said Wyss-Coray to Lifespan News. “Taking advantage of technical innovations that allow us to quantify thousands of proteins in a drop of blood, we pushed this concept further to show that circulating proteins derived from specific cell types in the brain and other tissues allow us to assess the physiological state and the risk to develop disease at an individual level. We were most surprised about strong links between accelerated aging of astrocytes and muscle cells with Alzheimer’s disease and ALS, respectively.”

Interactions with genetic risk factors

Cellular aging interacted with the APOE genotype, the strongest genetic predictor of Alzheimer’s, in a peculiar way. People homozygous for APOE4 – the most Alzheimer’s-prone genotype – were almost three times as likely to develop the disease if they also had rapidly aging astrocytes. This result may eventually lead to earlier and more robust prediction of Alzheimer’s risk and new therapeutical options aimed at keeping astrocytes young.

Interestingly, APOE4 carriers showed older astrocytes, but younger macrophages, and the carriers of the protective allele APOE2 showed the opposite. The authors suggest this might be a case of antagonistic pleiotropy – when the same trait is beneficial in one respect but costly in another [3]. APOE4-enhanced immune vigilance may have been an asset in our pathogen-laden past, so it was favored by selection despite an accompanying cost of faster brain aging. The latter only became consequential once more people began surviving into the advanced age where Alzheimer’s strikes.

Another famous risk factor, smoking, was also mitigated by younger cells: in this case, alveolar type 2 cells and respiratory epithelial cells. Smokers whose cells stayed young were much less susceptible to lung cancer. These cellular clocks could also predict mortality. For all-cause death, skeletal myocyte aging carried the strongest signal, followed by neurons, fibroblasts, alveolar type 2, and myeloid cells.

Literature

[1] Ding, D. Y., Bot, V. A., Chen, K. L., Groves, J. W., Pálovics, R., Masuda, D., … & Wyss-Coray, T. (2026). Plasma proteomic signatures of cellular aging predict human disease. Nature Medicine, 1-13.

[2] Oh, H. S. H., Rutledge, J., Nachun, D., Pálovics, R., Abiose, O., Moran-Losada, P., … & Wyss-Coray, T. (2023). Organ aging signatures in the plasma proteome track health and disease. Nature, 624(7990), 164-172.

[3] Austad, S. N., & Hoffman, J. M. (2018). Is antagonistic pleiotropy ubiquitous in aging biology? Evolution, medicine, and public health, 2018(1), 287-294.