Rapamycin and Dietary Restriction Extend Average Lifespan but Widen the Spread of When Animals Die
The ideal of modern aging research has a name: squaring the survival curve. The phrase describes a future where most people live close to maximum lifespan in good health and then decline steeply and briefly at the end - a curve that looks like a square on a graph rather than a long diagonal slope. It implies not just longer lives but more predictable and equitable ones, where the gap between the earliest and latest deaths narrows.
A re-analysis published by the University of Sydney suggests the three most widely studied interventions in longevity science - dietary restriction, rapamycin, and metformin - move in the opposite direction. All three increased variance in age at death in vertebrate animal studies, even in cases where they also raised average lifespan. The survival curve, under these interventions, gets wider rather than squarer.
The re-analysis and what it found
Dr. Tahlia Fulton and Associate Professor Alistair Senior of the University of Sydney's School of Life and Environmental Sciences drew on a recent meta-analysis of vertebrate longevity data to examine not just mean lifespan but its standard deviation. Their focus was on three interventions with strong evidence bases: caloric restriction, which reduces food intake without malnutrition; rapamycin, which inhibits the mTOR pathway involved in cellular growth and senescence; and metformin, a diabetes drug with apparent anti-aging properties in animal models.
Two of the three - dietary restriction and rapamycin - raised average lifespan as expected from the existing literature. Metformin's effect on mean lifespan in this dataset was less clear. But critically, all three interventions also expanded variance in age at death. The standard deviation around the mean grew proportionally with any gains in average longevity. Some animals lived substantially longer; others did not benefit comparably.
The result is what Fulton described as a biological lottery: "These approaches can make animals live longer, but the benefits aren't shared equally. Without more information, the outcome looks like a biological lottery. We're working to understand why, so future longevity science helps everyone."
Why variance matters as much as mean
Average lifespan extension sounds straightforwardly good. But if the same intervention that adds five years to the average also means that some individuals gain fifteen years while others gain none - or die earlier than untreated controls - the social and medical implications are more complicated. In human terms, interventions with high variance would produce a population where life expectancy statistics look improved but where predicting individual outcomes becomes harder and where the benefits concentrate in some individuals while others see little change.
The squaring of the survival curve matters precisely because it implies that compression of morbidity - not just lifespan extension - is achievable. If people are going to live longer, the goal is to keep them healthier and functional for more of that extended period, not simply to push the distribution wider. An intervention that extends the maximum but also extends the period of decline for some individuals may not represent progress toward that goal.
Limitations of the analysis
This work is a re-analysis of existing meta-analytic data rather than a primary experimental study. The vertebrate data it draws on come largely from rodent models - primarily mice and rats - and translating lifespan findings from rodents to humans has proven difficult historically. The mechanisms by which rapamycin, metformin, and dietary restriction affect human aging are likely to differ from those in short-lived rodents.
The study also cannot resolve why variance increases alongside mean longevity. Whether it reflects genuine biological heterogeneity in response to these interventions, variation in how consistently interventions are applied across studies, or some interaction between the intervention and genetic background of the animals studied remains unknown. Understanding the drivers of that variance is explicitly flagged as the team's next research priority.
Clinical trials of rapamycin and related compounds in humans are ongoing, including the VITAL-H trial at the University of Texas San Antonio, which will study rapamycin alongside two other drugs in healthy 60 to 65-year-old adults. Human data will ultimately provide the most relevant test of whether the variance problem seen in animals applies to people.