Slowing ageing itself

Every other driver on this site follows a disease. This one follows the attempt to act on the thing underneath them. The case for it is arithmetic: curing any single cause of death buys a population a few years, because the people saved remain old and die of something else soon afterwards. Acting on ageing, if it could be done, would move every cause at once. That is the claim, and it is the reason a field exists.

What exists to show for it is thin and improving. Senolytics have cleared small human trials on safety and biomarkers, not on outcomes. Metformin's effect on ageing rests largely on a retrospective comparison whose control group was in worse health; the trial designed to test it properly has spent years seeking funding. Rapamycin has the strongest animal evidence in the field and almost no human data at doses anyone would use. Partial reprogramming has restored function in mice and has not been given to a person. Caloric restriction changes biomarkers in humans and its effect on human survival is unknown.

The measurement half is in better shape and is where the near-term interest lies. Epigenetic, proteomic and phenotypic clocks give a number for how old a body looks against how old it is, and that number predicts death more sharply than most single risk factors. But clocks disagree with one another, they are trained on the outcomes they then predict, and no clock has yet been shown to move in response to an intervention and to carry a mortality benefit with it. Until one does, a clock is a measurement, not a target.

The field also carries more money and more noise than its evidence supports, and a good deal of what is published is mechanism rather than outcome. The rule this site applies — that a study must measure something in people — removes most of it.

Status

emerging

Direction

mixed

Would change our view

A randomised trial of any intervention aimed at ageing that reports a clinical endpoint rather than a biomarker. A clock that moves under an intervention and carries a measured survival difference with it. Agreement between two independently built clocks on the same cohort. A regulator accepting ageing as an indication.

Trackers

Gaps

Any human trial long enough to have a mortality endpoint. Clocks validated outside the cohorts they were trained on. Almost everything above age 85, where the intervention would matter most and the data are thinnest. And independent replication: much of the field's evidence comes from a small number of groups with an interest in the answer.