Introduction: Research populations harbor innate and acquired characteristics that can introduce substantial treatment response heterogeneity. Weight loss interventions show this variability, likely due to behavioral and physiological characteristics. Genetic risk for obesity is a potential but largely unexplored factor in weight loss trials. We examined whether genetic risk for obesity modified the effect of a low-fat diet on weight change in a randomized controlled trial. Methods: A data from the Women’s Health Initiative Dietary Modification Trial, including 10,353 postmenopausal women aged 50-79 with genotype data [63% European American (EAs), 37% African American (AAs)] were randomized into control and low-fat diet groups. The outcome was visit-specific weight difference (kg) from baseline. The Polygenic Risk Score of body mass index (PRS BMI ) was computed using the PRS-CS method and an independent trans-ancestry genome-wide association study of BMI (N = ~ 2M; 79 % EAs and 4 % AAs). We operationalized PRS BMI as a binary variable, distinguishing individuals in the 95 th percentile from the remaining individuals. Association was tested using generalized estimated equations with a 3-way interaction term (PRS х randomization х time), adjusting for 10 principal components, age, baseline BMI, and follow-up status. Analyses were performed separately by race/ethnicity, because body weight is influenced by ancestral diversity, which affects both linkage disequilibrium, PRS effect sizes, and cultural factors (e.g., racism). Results: Baseline, we had 6,473 EAs (mean age: 65.4 ± 6.6 years, mean BMI: 29.5 ± 6.5 kg/m 2 ) and 3,880 AAs (mean age: 60.9 ± 6.7years, mean BMI: 32.1 ± 6.8 kg/m 2 ) and mean follow-up was 6.0 ± 1.8 years. PRS BMI explained 12% of the variance in the baseline BMI measure in EAs and 8% in AAs. During the first year of intensive dietary intervention (18 sessions/yr), the low-fat diet group experienced 2.1 kg and 1.1 kg weight loss in European and AAs respectively, with no differences in PRS BMI . During years 1-7 with reduced intervention intensity (4 sessions/yr), the rate of weight change in both race/ethnic groups were modified by PRS BMI , although in different directions. EAs with high PRS (95 th percentile) in the intervention group rebounded towards initial weight (0.35 kg/year, p<0.001) nearly twice as fast as those women with PRS < 95 th percentile (0.19 kg/year, p<0.001). Conversely, high PRS African American women in the intervention group maintained their weight loss without rebound (-0.10 kg/year, p=0.48), distinctly different from those with PRS < 95 th percentile, who gained 0.25 kg annually (p<0.001). Conclusions: Impact of long-term low-fat dietary intervention on weight change was modified by genetic risk and race/ethnicity. Genetic predisposition and other contextual factors may be considered in the design and analysis of relevant clinical trials.
Read full abstract