Obesity and overweight are major components of metabolic syndrome, significantly increasing the risk of various lifestyle-related diseases like type 2 diabetes. The global prevalence of obesity has tripled since 1975, highlighting the urgent need for effective prevention strategies. While lifestyle changes are widely recommended for obesity prevention, individual susceptibility is influenced by both lifestyle and genetic factors. Recent research indicates a significant heritability estimate for body mass index (BMI). Polygenic scores (PGSs) offer a quantitative measure of an individual's genetic risk for obesity. However, the extent to which lifestyle modifications can improve outcomes in individuals with a high PGS for obesity remains unclear. This study aimed to address this gap by analyzing a large cross-sectional dataset from a Japanese community cohort to determine the optimal model for calculating PGS for predicting obesity status and to assess the impact of lifestyle factors, specifically leisure-time physical activity and sodium intake, on obesity risk across various genetic risk categories. The findings contribute valuable insights into the integration of PGSs into preventive medicine and personalized approaches to weight management.

Numerous studies have established the strong link between obesity and an increased risk of various lifestyle-related diseases. The global burden of obesity necessitates effective preventive measures. While the role of genetics in obesity is well-recognized, the interaction between genetic predisposition and lifestyle choices is less understood. Previous research has explored the impact of lifestyle changes on individuals with varying genetic risks for specific diseases, such as cardiovascular disease, but comprehensive evidence across various phenotypes, particularly obesity, remains limited. This study builds upon this existing research by focusing specifically on the interaction between PGS for obesity and lifestyle factors to provide a more nuanced understanding of obesity risk and prevention.

The study utilized four datasets from the Tohoku Medical Megabank Community-Based Cohort (TMM CommCohort): TMM10K (n=9811, used for model selection), TMM67K (n=50195), TMM18K (n=11947), and TMM8K (n=6796) (all used for validation and meta-analysis). The TMM10K dataset was used to determine the optimal model for calculating the polygenic score (PGS) for obesity, using linkage disequilibrium (LD) pruning plus thresholding (P+T) and LDpred methods. The best model was selected based on the area under the curve (AUC) in predicting obesity (BMI ≥ 25 kg/m²). The remaining datasets were used to validate this model. Sodium intake was estimated from spot urine samples using Tanaka's formula, and physical activity was assessed using self-administered questionnaires, distinguishing between leisure-time exercise (LTE) and daily life activities (DLA). Statistical analyses included AUC calculations, meta-analysis (fixed-effect and random-effects models), and logistic regression to evaluate the association between PGS, lifestyle factors (LTE, DLA, sodium intake), and obesity risk across different PGS categories (low, intermediate, high). The study adhered to the Declaration of Helsinki, and ethical approval was obtained.

The LDpred model at p=0.03 showed the highest AUC (0.63) for predicting obesity in the TMM10K dataset and was selected for subsequent analyses. The PGS distributions across the validation datasets were similar. The prevalence of obesity increased significantly with increasing PGS percentiles. Meta-analysis revealed that the odds ratio (OR) for obesity was 2.27 (95% CI: 2.12–2.44) for the intermediate PGS group and 4.83 (95% CI: 4.45–5.25) for the high PGS group compared to the low PGS group. Increased leisure-time physical activity (LTE) was significantly associated with reduced obesity risk across all genetic risk categories (OR ≈ 0.9 for each incremental increase in LTE quintile). In contrast, daily life activities (DLA) showed no significant association with obesity risk. Higher sodium intake was positively associated with obesity risk across all PGS categories (OR ≈ 1.24 in the high PGS group for each incremental increase in sodium intake quintile). Sensitivity analyses using alternative obesity definitions (waist circumference, body fat percentage, visceral fat area) yielded consistent results. Stratified analyses by sex generally mirrored the main findings.

This study's key finding is the demonstration of a significant association between lifestyle factors, particularly leisure-time physical activity and sodium intake, and obesity risk across various genetic risk categories, as measured by PGS. The observation that increased LTE is associated with reduced obesity risk even in individuals with a high genetic predisposition underscores the potential of lifestyle interventions in mitigating obesity risk. The positive association between sodium intake and obesity highlights the importance of dietary modifications. While the study's cross-sectional design limits the establishment of causality, the findings strongly suggest that lifestyle interventions could significantly contribute to obesity prevention even in individuals with a high genetic risk. The results emphasize the potential of integrating PGS into preventive medicine for personalized interventions.

This study successfully developed a PGS for obesity risk prediction and demonstrated its associations with lifestyle factors like physical activity and sodium intake. These findings support the development of personalized and targeted interventions that address both genetic predispositions and modifiable lifestyle factors for more effective obesity prevention. Future longitudinal studies are needed to further confirm these findings and establish causal relationships.

The study's cross-sectional design limits the ability to establish causal relationships between PGS, lifestyle factors, and obesity. The focus on a specific Japanese community cohort may limit the generalizability of the findings to other populations. The use of self-reported data for physical activity and spot urine samples for sodium intake introduces potential measurement error. Finally, normalizing the PGS within each dataset might introduce bias if PGS distributions significantly differed between datasets, although the sensitivity analysis suggests this bias was limited.