New genetic research shows a simple genetic test can predict who’s most at risk for obesity, offering hope for early prevention, but also raises tough questions about genetic fairness and healthcare access.
Study: Polygenic prediction of body mass index and obesity through the life course and across ancestries. Image credit: Andrii Yalanskyi/Shutterstock.com
A recent study in Nature Medicine developed ancestry-specific and multi-ancestry polygenic scores (PGSs) for early prevention and targeted treatment of obesity. Researchers highlighted that PGSs could be implemented early in life to prevent obesity. However, performance of PGSs can vary significantly across different populations, and careful implementation is necessary to avoid widening health disparities.
Obesity - a major public health threat
Obesity is a chronic medical condition characterized by the accumulation of excessive body fat. It is considered a significant public health concern because it enhances the risk of developing many chronic diseases, which may reduce life expectancy.
According to a recent prediction, more than half of the global population would become overweight or obese by 2035. Although scientists have developed multiple strategies to combat obesity, including intensive lifestyle interventions (ILIs), weight loss medications, and bariatric surgery, the associated risks of these approaches and inaccessibility to most people have limited their widespread implementation. Therefore, it is essential to develop an effective strategy that could help prevent obesity.
Many children develop obesity, which may persist into their adulthood. Therefore, early predictors, such as genetic variants, could be extremely valuable in preventing obesity. Previous studies have highlighted the potential of PGSs in disease risk prediction and population screening. This predictive quality is based on their ability to capture an individual’s inherited polygenic susceptibility to a trait or disease. It is essential to examine the circumstances in which PGSs for obesity may be helpful in risk prediction.
A previous study demonstrated the use of PGSs for obesity based on a genome-wide association study (GWAS) of BMI in over 339,000 people of predominantly European ancestry. However, a PGS based on one ancestry population may not accurately reflect other ancestry populations. Using PGSs developed primarily in one ancestry group may result in lower prediction accuracy and worsen health inequities if not carefully addressed.
About the study
The current study exploited the findings of GWAS meta-analyses to develop a PGS for BMI. GWAS meta-analysis included summary statistics for BMI from over 200 studies from the GIANT consortium and 23andMe.
The GWAS summary statistics encompassed over 5.1 million individuals from diverse populations. This diverse population included 71.1% of participants with European ancestry, 14.4% of Hispanic ethnicity with typically admixed ancestries, 8.4% of East Asian ancestry, 4.6% of African origin, and 1.5% with South Asian ancestry.
Participants with closely aligned genetic relationships were grouped and referred to as having European-like ancestry (EUR), African-like ancestry (AFR), East Asian-like ancestry (EAS), American-like ancestry (AMR), and South Asian-like ancestry (SAS). It is worth noting that the authors acknowledged these groupings oversimplify the actual genetic diversity among participants.
PRS-CS(x), a standard method for generating cross-population polygenetic risk scores, was used to develop ancestry-specific and multi-ancestry PGSs leveraging up to 1.3 million common variants.
Study findings
The current study identified the optimal genome-wide shrinkage parameter and linear combination weights for PRS-CS(x) that demonstrated highest explained variance for BMI in six ancestry subpopulations of the UK Biobank (UKBB), including individuals of Middle Eastern-like ancestry (MID). A random subset of 20,000 unrelated individuals was used for the EUR-tuning population.
A multi-ancestry PGS that comprised five ancestry-specific PGSs (PGSLC) generated best prediction scores. Compared to PGSs trained only with GWAS summary statistics, multi-ancestry PGS exhibited higher explained variance for BMI, ranging between 7.2% (AFR) and 17.5% (EUR), with a median of 14.0%.
Except for the East Asian-like and European-like ancestry, performance of a PGS containing genome-wide significant variants was generally intermediate to that of the ancestry-matched and multi-ancestry PGSs.
The prediction accuracy of PGSLC for BMI and obesity was also assessed in independent validation populations of 482,135 participants, from the UKBB, the Million Veteran Program (MVP), the Uganda General Population Cohort (GPC-UGR), and the BioMe Biobank.
The current study highlighted that prevalence of obesity varies significantly across populations and cohorts. It must be noted that the mean BMI ranged between 22.2 kg m2 and 30.6 kg m2. The performance of the PGSLC was highest in participants with EUR ancestry from the UKBB, with an explained variance of 17.6%. In contrast, a lower PGSLC performance was found for African-like ancestry with explained variance of 6.3%, 5.1% in African American populations, and 2.2% in the GPC-UGR population from rural southwestern Uganda.
Within EUR from the UKBB participants, the explained variance was marginally higher in males than in females. It was also found to be higher in younger participants than in those belonging to advanced age groups. Within the EUR population, the PGSLC demonstrated improved performance in differentiating between participants with and without obesity.
The area under the receiver operating characteristic curve (AUC) increased with severity of obesity. The AUC for PGSLC was significantly larger on its own. BMI in children with higher genetic predisposition (PGS ≥10th percentile) increased faster than in those with lower genetic predisposition. The added value of the PGS for predicting BMI was greatest at a very young age, particularly up to age five, before BMI becomes a strong predictor of later obesity. In older children, measured BMI provides much of the predictive information, and the incremental value of PGS is smaller.
Children's higher mean PGS is a well-established predictor of future obesity risk. To predict BMI in early adulthood, PGS in the first few years after birth was found to be a more reliable obesity predictor. PGS was also much more predictive of BMI than other body composition traits, such as body fat percentage or waist-to-hip ratio.
The current study indicated that individuals with a higher PGSLC underwent greater weight loss during the first year in response to the ILI than the control group. However, these individuals were also more likely to regain weight after the first year, highlighting the importance of ongoing support for weight maintenance among those at higher genetic risk.
Importantly, the authors stress that a higher genetic risk, measured by PGS, does not mean that obesity is inevitable. Individuals with a higher PGS may be particularly responsive to changes in environment and lifestyle interventions, and preventative strategies can be effective. The researchers caution that implementation of PGS-based risk tools must account for differences in predictive performance across populations, especially to avoid worsening health inequities among underrepresented groups such as those of African ancestry. There is future potential for PGSs to guide lifestyle interventions and new weight loss drug therapies, though more research is needed.
Conclusions
The current study demonstrates the potential of BMI PGSs as a tool for predicting adult obesity throughout life, particularly in early life. This tool can identify individuals at high risk of obesity, enabling the timely and effective implementation of preventive strategies.
However, the use of PGS in clinical or public health practice must be accompanied by careful attention to population differences and ethical considerations of genetic risk prediction.
Download your PDF copy now!
Study uncovers a specialized gut-brain pathway for glucose sensing