Eight years of follow-up in 335 women showed that molecular aging follows different paths among women of the same age
Eight years of follow-up in 335 women showed that molecular aging follows different paths among women of the same age
Researchers at King's College London repeatedly measured gene activity and blood composition in 335 women from the TwinsUK twin cohort over up to eight years. Among participants aged 32 to 80, the activity of more than five thousand genes and the levels of dozens of substances in the blood changed over time, but the speed and direction of these changes differed among women of the same age. The study was published in the journal Science.
Most studies of molecular aging take a single snapshot, measuring blood samples once in people of different ages. This approach conflates age-related differences across a population with changes within an individual. The authors of the paper, published in Science on September 3, took a different approach. Within TwinsUK, the UK's largest adult twin registry, maintained by King's College London since 1992 and now numbering more than 16 000 participants, they established a longitudinal subgroup called MultiMuTHER. Participants attended at least three visits, each at least a year apart. Blood was collected at every visit for RNA sequencing to measure gene activity and for analysis of 1197 metabolic measures.
Over a median follow-up of six years, more than five thousand genes and 45 substances in the blood showed consistent changes. Activity declined in the tumor suppressor gene TP53 and in genes associated with cardiovascular and neurodegenerative diseases, while activity increased in genes involved in mitochondrial function. Levels of the “forever chemicals” PFOA and PFOS fell sharply as restrictions on their use were introduced in the UK and worldwide. At the first visit, their levels were associated with the activity of more than 200 genes; by the final visit, these associations had disappeared entirely.
The group averages also concealed individual patterns. Some genes and substances showed no shared direction of change: levels rose in some participants and fell by similar amounts in others. These opposing changes canceled out in the group average, even though each woman followed a distinct trajectory. One example was CXCL9, an immune signaling protein. In 2021, the iAge model, trained on blood data from more than a thousand people, identified it as the largest contributor to an “inflammatory clock” that predicts frailty and cardiovascular aging.
A separate analysis of blood cell types showed that the immune system changed unevenly. In most participants, gene activity declined over time in T-cells, which recognize pathogens, and increased in NK-cells, which provide a rapid innate immune response. The authors describe this as a reorganization in which one part of the immune system weakens while another strengthens. Genes whose activity declined were substantially longer than those whose activity increased, possibly because cells find it harder to transcribe long stretches of DNA as they age.
The time of day and season of blood collection also affected gene activity. Approximately a quarter of genes showed fluctuations over the course of the day, and another quarter showed seasonal fluctuations. Activity of the clock genes PER1, PER2, PER3 and the stress-related gene FKBP5 declined during the day alongside cortisol levels. Immune gene activity was higher in winter, while mitochondrial gene activity was higher in summer. Without accounting for these fluctuations, a single measurement can easily be mistaken for a persistent age-related trend.
“Human molecular aging is a dynamic, context-dependent process that cannot be described as a uniform process governed by a single mechanism,”
the authors write. For a marker such as CXCL9, this means that a single blood measurement captures one moment in time. Repeated measurements are needed to determine whether an individual's risk is increasing or decreasing. The study is observational and includes only women. The next step is to examine these changes at the level of individual cells and test whether diverging trajectories, such as those seen for CXCL9, predict disease before symptoms appear.