A person’s genetic risk for obesity might feel like a fixed sentence, but a new randomized controlled trial suggests it does not determine how well diet-based weight loss works. The GENEROOS study, published in Nature Communications, found that adults in the top 5% of the BMI polygenic score distribution lost just as much weight through dietary coaching as those in the bottom 5%.
The finding challenges a plausible intuition: that people carrying hundreds of obesity-associated gene variants face a steeper biological hill when trying to lose weight. The data say otherwise.
The trial
GENEROOS (Genetic and Environmental Response to Obesity Intervention Study) recontacted 38,621 genotyped adults from the FinnGen biobank in Finland. From these, 223 participants were selected from the extreme tails of the BMI polygenic score distribution, the top 5% (highest genetic risk) and the bottom 5% (lowest genetic risk). They were randomized 1:1 to either 6 months of structured dietary coaching or usual care.
The dietary intervention prescribed a 500 kcal/day deficit with personalized meal plans delivered through a mobile app with coach chat support. The control group received no nutritional coaching.
The results
After 6 months, the diet intervention group lost an average of 1.51 kg/m² in BMI (95% CI: -1.72 to -1.31), equivalent to about 4.7% of body weight. The control group showed essentially no change (-0.01 kg/m², 95% CI: -0.22 to 0.19). The difference between groups was -1.50 kg/m² (P < 0.001).
The critical test was the interaction between polygenic score group and intervention effect. The interaction term was 0.06 (95% CI: -1.33 to 1.44, P = 0.94), unequivocally non-significant. People in the highest genetic risk category lost weight at the same rate as those with the lowest genetic risk.
Participants in the high-genetic-risk group did start with a higher baseline BMI (about 3.0 kg/m² higher), confirming that polygenic scores capture real biological differences. But once the intervention began, the trajectories converged.
Why this matters
The idea that genetic predisposition determines weight loss success has gained traction in popular discourse, often fueling a sense of futility. The GENEROOS trial directly tested this assumption and found no evidence for it.
“Current genome-wide polygenic scores for BMI do not appear useful for stratifying individuals for short-term dietary weight loss interventions,” the authors concluded.
The study is not definitive on its own. The investigators acknowledge several limitations. The trial was powered to detect medium-to-large interaction effects, meaning smaller genetic effects could still exist. The duration was only 6 months, weight loss maintenance and long-term differential responses remain unknown. The sample was drawn from a single Finnish population, and the results may not generalize to other ancestries. The cohort was also predominantly female (74.9%).
Nevertheless, the message for clinical practice is pragmatic: genetic risk should not be used to deny or discourage dietary interventions, nor to predict who will benefit most.
The polygenic score
The BMI polygenic score used in GENEROOS incorporated 1,097,993 SNPs derived from a meta-analysis of Yengo et al.’s GWAS (~700,000 individuals) and the FinnGen GWAS (315,000 individuals). It explained 12.7% of BMI variance in the Finnish Clinical Biobank Tampere cohort, a substantial fraction by polygenic score standards.
Funding and competing interests
The study was funded by the European Research Council under Horizon 2020 grant 945733 (Andrea Ganna). Ganna is a founder of Real World Genetics Oy, Finland. Co-author Ettore Brenna is an employee of Eli Lilly and Company.
Sources
Rodosthenous RS, Viiri LE, Carson AM, et al. “Dietary intervention and BMI reduction in individuals at the extremes of genetic predisposition to higher BMI: a randomized controlled trial.” Nature Communications (2026). DOI: 10.1038/s41467-026-75224-0