Background: Obesity is a global public health crisis associated with high morbidity and mortality rates. Previous genome-wide association studies (GWAS) investigating body mass index (BMI) have primarily relied on imputed data from European individuals. Consequently, most BMI risk variants identified to date are common and in primarily European ancestry populations exhibiting small effect sizes, leaving low and rare variants with potentially large effects largely unexplored.
Methods: This study leveraged whole-genome sequencing (WGS) data from 88,873 participants from the Trans-Omics for Precision Medicine (TOPMed) Program, of which 51% belonged to non-European population groups. We performed GWAS of BMI adjusting for age, age2, sex, study, population group, principal components, sequencing center, sequencing phase, and project (TOPMed vs. Centers for Common Disease Genomics). We subsequently conducted replication analyses, stepwise conditional analyses, rare variant (MAF ≤ 1%) aggregate association analyses, and fine-mapping.
Results: We discovered a total of 18 BMI-associated signals (P < 5 × 10-9), including a novel low-frequency single nucleotide polymorphism (SNP), rs111490516, in MTMR3 that was common in individuals of African descent (minor allele frequency [MAF] = 13% in African and Barbadian population groups). We successfully replicated this novel SNP and observed directionally consistent associations across replication studies of similar background. In the meta-analysis of 198,621 individuals from both discovery and replication studies, the estimated effect of rs111490516 was 0.037 kg/m2 with a SE of 0.006 (P = 4.19 × 10-9). Using our diverse study population, we additionally identified two potentially novel secondary signals in known BMI loci (rs2206277 in TFAP2B and rs3838785 in BDNF) in our conditional analyses and pinpointed two likely causal variants in the POC5 (rs2307111, posterior probability [PP] = 0.99) and DMD loci (rs1379871, PP = 1.00) in our fine-mapping analyses. Finally, we successfully replicated previous gene-based associations with the well-known MC4R gene (P = 8.47 × 10-8) by aggregating 111 alleles across 37 sites within coding regions, enhancers, and promoters.
Conclusion: Our work shows the benefits of linking WGS in diverse cohorts for discovering new variants and genes that confer risk for obesity. Ultimately, our study brings us one step closer to understanding the complex genetic underpinnings of obesity, translating these leads into mechanistic insights, and developing targeted preventions and interventions to address this global public health challenge.