Higher Teen BMI Could Speed Up Biological Aging Decades Later

At age 9, the children in a long-running Finnish health study had an average body mass index, or BMI, of 16.6. By 18, it had risen to 21.4. For most, that is the ordinary arithmetic of growing up. But when researchers returned years later to blood samples collected in adulthood, those early measurements appeared to have left a molecular trace.

The people who had entered adolescence with a higher BMI tended to show a faster pace of biological aging decades afterward, according to a study of the Young Finns cohort. The relationship was clearest in a DNA methylation measure called DunedinPACE, and it was entangled with an inherited tendency toward a higher BMI. Understanding the impact of teen BMI can provide insights into long-term health outcomes.

The study does not show that a teenager’s BMI locks in their health or their lifespan. It asks a narrower question: whether genetic susceptibility to higher BMI, BMI measured from ages 9 to 18 and later DNA-based measures of aging fit together in a pattern that looks partly causal. The answer was suggestive, but not uniform across every genetic score or every clock the team tested.

A Blood Sample Turned Into a Pace

Biological aging is not a second birthday. It is a collection of measurements intended to capture whether the body’s systems are changing faster or more slowly than its calendar age would suggest. In this study, the researchers read patterns of DNA methylation, chemical tags that can influence how genes are used without altering the DNA sequence itself.

They applied two algorithms to the blood data. PC-GrimAge produces an age estimate, from which the team calculated age acceleration. DunedinPACE estimates a rate: the number of biological years accumulated for each calendar year. The latter was developed by tracking change in 19 measures of organ-system integrity over two decades, then distilling that longitudinal signal into a single blood test. Its developers found that it was associated with later morbidity, disability and mortality in several cohorts, but it remains a research biomarker, not a diagnosis for an individual patient.

For the present study, the team used blood samples collected in 1986, 2011 and 2018. The earliest samples came from 310 participants aged 15 to 24. The later samples came from 1,711 participants aged 34 to 49 and 1,296 aged 41 to 56. In total, 2,045 people had at least one usable biological-aging measurement.

Genes Raised the Starting Level

The study began with 3,596 participants in the Cardiovascular Risk in Young Finns Study, who were 3 to 18 years old in 1980 and were followed through 2018 to 2020. Rather than treating their adolescent BMI as one isolated number, the researchers modeled trajectories from measurements at ages 9, 12, 15 and 18.

They then calculated two polygenic risk scores. One gathered the effects of 941 genetic variants associated with adult BMI. The other used 286 available variants associated with recalled childhood body size. Neither is a genetic verdict. Each is a weighted summary of many common variants, and in this cohort the adult BMI score explained only about 3.5 to 4.3 percent of the variation in adolescent BMI.

Still, the pattern was visible. A higher adult BMI polygenic score predicted a higher BMI level during adolescence. In turn, that higher level, rather than the rate at which BMI rose through adolescence, was associated with a faster DunedinPACE result in 2011 and 2018. The same pattern was weaker and statistically non-significant for PC-GrimAge.

“Our findings suggest that genetic predisposition to high body mass index is associated with accelerated biological aging,” lead researcher Anni Pitkänen, of the University of Jyväskylä, said in a statement. “Part of this association may be explained by the fact that the inherited predisposition is reflected in higher adolescent BMI.”

In other words, the signal was not that a particularly rapid adolescent gain in BMI explained later aging. It was that people who began adolescence at a higher BMI tended, on average, to have a faster molecular pace of aging later on.

The Causal Test Did Not Give One Answer

To press beyond correlation, the researchers used Mendelian randomization. This approach uses genetic variants linked to an exposure as instruments for testing whether the exposure is plausibly on a causal path to an outcome. It can reduce confounding and reverse causation, but it only works if demanding assumptions about the genetic instruments hold. In particular, the variants must be associated with the exposure, not share a confounding pathway with the outcome and affect the outcome only through the exposure.

Using the adult BMI polygenic score as the instrument, the study found a positive effect of genetically predicted adolescent BMI on DunedinPACE. The estimates were 0.020 in 2011, with a 95 percent confidence interval from 0.008 to 0.031, and 0.019 in 2018, with an interval from 0.003 to 0.035. Put plainly, the instrumental-variable model estimated that a 1 kg/m² increase in genetically predicted adolescent BMI was associated with a 0.02 biological-year-per-calendar-year faster pace of aging.

But the causal inference has an awkward seam. When the researchers used the childhood body-size score instead, they found no clear causal effect, although the estimates for DunedinPACE remained positive. They also found no strong evidence of directional pleiotropy in their sensitivity analyses, meaning that the tested variants did not show a detectable tendency to influence the outcome through unrelated pathways. That does not eliminate the possibility of such bias, and it does mean the study’s strongest causal language belongs to the adult-BMI genetic score, not to every available genetic instrument.

The authors offer one possible explanation: variants selected for adult BMI may capture genetic influences that operate across the life course, whereas a score based on recalled childhood body size may describe something more specific. That distinction matters because genetic results reflect long-term differences in an exposure. They do not tell us that changing BMI by one unit at a particular age will produce the same numerical shift in an epigenetic clock.

Adult Weight Complicates the Story

The study’s most consequential limitation appears when adult BMI enters the model. When the researchers adjusted the biological-aging measures for BMI measured at the same adult follow-up, the indirect pathways from genetic susceptibility to later aging through adolescent BMI became smaller and no longer statistically significant.

That does not settle the issue. Adult BMI may be a confounder, a continuation of adolescent BMI or part of the pathway through which adolescent BMI affects later health. Adjusting for it can therefore erase a real life-course effect as well as reveal one that had been overstated. What it does establish is that this study cannot cleanly separate adolescence from the years that followed.

That broader trajectory is already a public-health concern. A 2025 systematic review and meta-analysis of 38 studies found that higher BMI from ages 2 to 22 was associated with higher risks of adult cardiovascular disease, coronary heart disease and heart failure. Such evidence gives the Finnish findings context, but it does not turn a DNA methylation score into proof that a particular disease will occur.

In the Young Finns data, only about 7 to 10 percent of participants met international BMI cutoffs for overweight during adolescence, and about 1 percent met the cutoff for obesity. By the 2011 and 2018 adult follow-ups, 40 percent had overweight and 21 to 28 percent had obesity. The cohort was also limited to people of European ancestry, a restriction that matters for both the generalizability of polygenic scores and the clocks’ interpretation.

The enduring image from the study is not a molecular timer set in adolescence. It is a line that begins with a measured BMI at 9, carries through adulthood and reaches a later vial of blood. The researchers can see that line in aggregate. They cannot yet say where along it an intervention would have its greatest effect, or whether changing the line would change the clock.

FAQ