Obesity at 20 and Cancer Risk: 6.3 Million Korean Adults

In 6.3 million Korean adults aged 20 to 39, higher BMI was linked to a modestly higher chance of a cancer diagnosis before midlife — 1.84% over ten years at normal weight versus 2.62% with class II obesity — an absolute difference of less than one extra case per 100 people.

A nationwide Korean cohort followed 6,313,338 adults aged 20 to 39 for a mean of 11.2 years and recorded 151,893 cancers. Over ten years, cumulative cancer incidence was 1.84% in the normal-weight BMI category, 2.24% with class I obesity and 2.62% with class II obesity — absolute differences of 0.40 and 0.78 percentage points. That is a real gradient and a small one: fewer than one extra diagnosis per 100 people over a decade. The study is observational, the authors name unmeasured confounding including smoking, the cohort is screening attendees in Korea rather than a general population, and the BMI categories are Asian-specific.

What this means

In a Korean national screening cohort of 6.3 million adults aged 20 to 39, higher BMI went with a modestly higher ten-year chance of a cancer diagnosis: 1.84% at normal weight, 2.24% with class I obesity, 2.62% with class II obesity. This is an association in one population, not evidence that weight causes cancer and not evidence that changing weight would change anyone’s risk — the study did not measure weight change. The absolute gap at the top category is 0.78 percentage points, under one extra case per 100 people over ten years, and the authors list unmeasured confounding including smoking. Waist circumference sharpened the pattern, but nothing here amounts to a screening recommendation.

The cohort behind the numbers

The study drew on the Korean National Health Insurance Service database, linked to the national screening program, 2009 to 2021. It followed 6,313,338 adults aged 20 to 39 with no prior cancer and recorded 151,893 incident cancers over a mean 11.2 years. Body mass index was sorted into five categories, from underweight through class II obesity, using cutoffs developed for Asian populations; waist circumference was analyzed alongside it.

Two boundaries matter more than any other detail. Participants were people who attended a national health screening — a group the authors describe as not representative, since attendees differ systematically from those who do not. And the categories follow Asian-specific cutoffs, not the thresholds most US readers have in mind, so a label does not transfer cleanly between countries.

Only the abstract was available to us at press time, so the site-specific hazard ratios and the model’s covariates were not examined; PubMed’s record could only tell us that disclosure forms are available with the article online, so conflicts of interest are recorded in Sources as unknown, not as none.

What 0.78 percentage points looks like

Over ten years, cumulative cancer incidence was 1.84% in the normal-weight category (95% CI 1.83 to 1.86), 2.24% with class I obesity (2.22 to 2.27) and 2.62% with class II obesity (2.56 to 2.68). Against normal weight, the absolute differences were 0.40 percentage points (95% CI 0.39 to 0.41) and 0.78 percentage points (0.73 to 0.82). Restated per 10,000 adults over the same decade: about 184 diagnoses at normal weight, 224 with class I obesity, 262 with class II obesity.

Fewer than one extra diagnosis per 100 people across a decade is what the top category amounts to. The abstract reports absolute risks and no relative-risk estimate — and relative framing is where a difference of this size starts to sound large.

Why this is an association, not a cause

The design is a retrospective cohort built from administrative records: it tracks who was diagnosed with what, not why. The authors list unmeasured confounding, naming smoking burden. People in different BMI categories also differ in ways the dataset may not capture.

The study did not measure weight change. It compared people by the category they were in, at a point the abstract does not specify. Nothing in it tests whether losing weight would alter anyone’s risk, and these numbers cannot be read as evidence that it would. That is the whole distance between a cohort finding and a prevention claim.

What waist circumference added

Combining BMI with waist circumference separated the categories further. Class I obesity with abdominal obesity carried a ten-year incidence of 2.45% (2.41 to 2.50), against 2.13% (2.10 to 2.16) for class I obesity without it — and above class II obesity without abdominal obesity, at 2.20% (2.03 to 2.36). Where fat sits changed the estimate at a given BMI.

That is a sharper association in one cohort; it stops there. The abstract does not describe how waist was measured at screening, and none of it supports using waist circumference as a screening tool.

Where this sits

This is the largest of a cluster of studies from the same Korean national database. An earlier one followed 935,600 women aged 20 to 39 across two health exams three years apart and found 798 endometrial cancers, with sustained obesity at a hazard ratio of 2.08 (1.45 to 3.00) and sustained abdominal obesity at 4.39 (3.56 to 5.43). Another drew on 2,877,245 adults aged 20 to 39, of whom the study classified some with metabolic dysfunction-associated steatotic liver disease: 46,729 cancers over 22.8 million person-years, with the fatty-liver group at a hazard ratio of 1.19 (1.16 to 1.22) for cancer overall against adults without steatotic liver disease.

What the new paper adds is scale, a ten-year cumulative-incidence framing rather than hazard ratios alone, and the BMI-plus-waist comparison. It does not add a mechanism, and the sizes are modest in absolute terms. For how many adults meet an obesity definition at all, see how-many-people-have-clinical-obesity.

What it leaves open

Three things remain unresolved. The cancers with the highest cumulative incidence were thyroid, endometrial and ovarian — named, with no magnitudes given, so nothing here says how large those excesses were. The relative association was more pronounced in men while absolute differences were greater in women; explaining that asymmetry needs the full text. And obesity was not associated with a higher risk of breast cancer diagnosed by age 50, a null that cuts against any reading in which excess weight raises every early cancer.

Terms explained

Cumulative incidence: the share of a group diagnosed within a set period — here, ten years. It is a proportion of everyone at risk at the start.
Absolute vs relative risk: an absolute difference subtracts one risk from another (2.62% minus 1.84% leaves 0.78 percentage points). A relative figure divides one risk by the other and looks larger.
Hazard ratio: the rate of an outcome in one group divided by the rate in another over the same follow-up. A value of 1.00 means no difference.
Confounding: a difference between groups — smoking, for instance — that moves the outcome along with the exposure, so part of an apparent association may belong to it.
Asian-specific BMI cutoffs: the threshold values for underweight, overweight and obesity used in this study, defined for Asian populations and different from the cutoffs most US readers know.

Sources

Yun JS, Mun H, Jung JH, Park Y, Kim SY, Kim K, Ko SH, Ahn YB, Kwon HS, Han K, Son JW. “Obesity and Cancer Risk in Young Adults in Korea: A Nationwide Health Screening Cohort Study.” Annals of Internal Medicine 2026, ahead of print, published October 27, 2026. DOI: 10.7326/ANNALS-25-02612 · PMID: 42932883. Peer-reviewed retrospective cohort; abstract only.
Landscape — endometrial cancer: “Impact of Changes in Obesity and Abdominal Obesity on Endometrial Cancer Risk in Young Korean Women: A Nationwide Cohort Study.” Cancer Epidemiology, Biomarkers & Prevention 2025;34(10):1794–1800. DOI: 10.1158/1055-9965.EPI-25-0615 · PMID: 40694047. Abstract-level access.
Landscape — fatty liver disease: “Young Adults With Metabolic Dysfunction-Associated Steatotic Liver Disease Have an Increased Risk of Early-Onset Cancer: A Nationwide Cohort Study Differentiating the Risk of 23 Site-Specific Cancers.” Clinical Gastroenterology and Hepatology 2025;23(13):2540–2549.e8. DOI: 10.1016/j.cgh.2025.01.042 · PMID: 40348107. Abstract-level access.
Note on funding and conflicts: the primary paper’s abstract states “Primary Funding Source: None.” PubMed’s abstract record states “Disclosures: Disclosure forms are available with the article online”; the forms themselves were not reachable, so conflicts of interest are recorded here as unknown, not as none.