38 organ aging clocks trained separately for men and women for the first time: the same acceleration in brain aging raises the risk of progression to Alzheimer’s disease by 125% in women and 74% in men
38 organ aging clocks trained separately for men and women for the first time: the same acceleration in brain aging raises the risk of progression to Alzheimer’s disease by 125% in women and 74% in men
On September 16, a Columbia University team that has been developing models of organ biological age since 2023 retrained all 38 models separately for men and women for the first time. The models had previously been trained on a combined sample. The sex-specific models were more accurate than the combined models, and a genome-wide analysis linked them to 359 DNA regions, many associated with aging in only one sex. In a cohort of people at risk of Alzheimer’s disease, the same measure of “brain age” predicted progression from mild cognitive impairment to Alzheimer’s disease differently in men and women.
“Aging clocks” are AI models that predict age from organ scans, blood proteins, or blood metabolites. The difference between predicted and chronological age indicates how far an organ’s aging has advanced beyond what would be expected. Since 2023, Junhao Wen’s laboratory at Columbia University has been developing a set of these clocks, which now includes 23 models based on data from nearly half a million UK Biobank participants. All were trained on a combined sample of men and women, as were almost all comparable models worldwide, even though women live longer and experience more severe Alzheimer’s disease, while men develop heart problems earlier. Wen and colleagues investigated whether a single fitted curve averages together two different trajectories with similar means.
The findings appeared in a Nature Medicine article. To determine whether sample size explained the difference, the authors compared sex-specific and combined models using groups matched for size and age. The sex-specific models remained more accurate and showed less overfitting. The team then investigated the biological basis of the difference.
A genome-wide analysis of the 38 models identified 359 associations between DNA regions and organ aging. Liver and metabolic aging were more heritable in women, while immune and skin aging were more heritable in men. Brain aging measured by MRI was also associated with different blood proteins: SLITRK1 in women and a set of eight other proteins in men.
The main clinical validation used the ADNI cohort, a long-term study following Americans from normal cognition through mild cognitive impairment to Alzheimer’s disease. Each standard deviation increase in how much “older” the brain appeared was associated with a 74% increase in the risk of progression from mild cognitive impairment to Alzheimer’s disease in men and a 125% increase in women. The same model was tested using data from a trial of solanezumab, an antibody targeting amyloid plaques in the brain, in healthy people at risk of Alzheimer’s disease. Among participants with accelerated brain aging, memory declined markedly over 240 weeks in both sexes. Among those with slower brain aging, women retained better cognitive performance than men at that time point. The combined version of seven MRI-based organ aging clocks had already identified a similar shift in these data, without separating it from the drug’s effects.
The sex-specific models also differed in their predictions of mortality. A combination of eight organ systems predicted mortality in women, while a broader set did so in men. Within that broader set, the liver and spleen associations ran in the opposite direction, with lower mortality risk. The authors qualify their conclusion:
“Sex-specific models should be used according to the context. Their performance does not justify treating them as universally superior or adopting them as a default approach.”
A reviewer of the paper independently described Wen’s team as follows:
“The team is among the first in the world to develop multi-organ, multi-omics aging clocks to model human aging and disease as a whole.”
Biological aging clocks are a tool researchers hope to use when testing drugs intended to slow aging, so they can assess potential effects on lifespan without waiting decades. If these clocks have spent years averaging male and female biology, their use in selecting trial participants and assessing outcomes may have produced differences between the sexes that vary by organ and disease.