A new multi-ancestry polygenic score (PGS) for body mass index (BMI), developed using genetic data from over 5.1 million people, substantially improves the prediction of obesity across the life course compared to previous scores.¹ The research, which drew data from the GIANT consortium and 23andMe, demonstrates particular value in predicting obesity risk from early childhood and in understanding weight change dynamics in response to lifestyle interventions.

Researchers developed and validated ancestry-specific and multi-ancestry PGSs using genome-wide association study (GWAS) summary statistics from a cohort of up to 5.1 million individuals of diverse ancestries (71.1% European, 14.4% American, 8.4% East Asian, 4.6% African, and 1.5% South Asian). The best-performing multi-ancestry PGS, a linear combination of five ancestry-specific scores (PGSLC), was then tested in several independent populations.

The score’s predictive ability was assessed across different life stages: for adult BMI and obesity in cohorts including the UK Biobank (UKBB) and the Million Veteran Program (MVP); for childhood BMI trajectories in the Avon Longitudinal Study of Parents and Children (ALSPAC); for adult weight gain in the PLCO Cancer Screening Trial; and for response to intensive lifestyle interventions (ILIs) in the Diabetes Prevention Program (DPP) and Look AHEAD trials.

In UKBB participants of European ancestry, the new PGSLC explained 17.6% of BMI variation, more than doubling the predictive power of a widely used previous score.² Performance varied across other populations, ranging from 16% in East Asian-Americans to 2.2% in rural Ugandans, highlighting a persistent performance gap in populations of African ancestry.

In the ALSPAC cohort, children with higher polygenic scores showed accelerated BMI gain from 2.5 years of age. When added to clinical predictors available at birth, the PGS nearly doubled the explained variance for BMI from age 5 onwards (from 11% to 21% at age 8). Furthermore, adding the PGS to BMI measurements taken at age 5 increased the explained variance for BMI at age 18 from 22% to 35%.

In two ILI trials, individuals with a higher PGS lost modestly more weight during the first year of intervention (0.55 kg per standard deviation increase in PGS) but were also more likely to regain it in the following years.

This improved PGS offers a more powerful tool for identifying individuals at high genetic risk of obesity. Its utility is most pronounced in early life, where it can significantly enhance risk prediction beyond what is possible with early-life BMI or birth characteristics alone. The findings from the lifestyle intervention trials suggest that a high genetic predisposition is not deterministic; rather, these individuals may be more responsive to lifestyle changes, underscoring the importance of sustained support to prevent weight regain. The study authors noted, "Overall, these data show that PGSs have the potential to improve obesity prediction, particularly when implemented early in life."¹

Further research is needed to improve the performance of PGSs in underrepresented populations, particularly those of African ancestry, to ensure their equitable application in clinical practice. Future studies may also investigate the utility of these scores in predicting response to the growing number of pharmacotherapies for weight loss.

References

1. Smit RAJ, Wade KH, Hui Q, et al. Polygenic prediction of body mass index and obesity through the life course and across ancestries. _Nat Med_ 2025. https://doi.org/10.1038/s41591-025-03827-z

2. Khera AV, Chaffin M, Wade KH, et al. Polygenic prediction of weight and obesity trajectories from birth to adulthood. _Cell_ 2019;177:587–596. https://doi.org/10.1016/j.cell.2019.03.028

Disclaimer

Disclaimer: The information presented in this article is for educational purposes only. Any quotes included reflect the opinions of the individual quoted, and do not necessarily reflect the views of the publisher. The publisher does not guarantee the accuracy or completeness of the content and accepts no responsibility for any errors, or any consequences arising from its use.