Large-scale exome sequence analysis identifies sex- and age-specific determinants of obesity

Abstract

Obesity contributes substantially to the global burden of disease and has a significant heritable component. Recent large-scale exome sequencing studies identified several genes in which rare, protein-coding variants have large effects on adult body mass index (BMI). Here we extended such work by performing sex-stratified associations in the UK Biobank study (N∼420,000). We identified genes in which rare heterozygous loss-of-function increases adult BMI in women (DIDO1, PTPRG, and SLC12A5) and in men (SLTM), with effect sizes up to ∼8 kg/m². This is complemented by analyses implicating rare variants in OBSCN and MADD for recalled childhood adiposity. The known functions of these genes, as well as findings of common variant genome-wide pathway enrichment analyses, suggest a role for neuron death, apoptosis, and DNA damage response mechanisms in the susceptibility to obesity across the life-course. These findings highlight the importance of considering sex-specific and life-course effects in the genetic regulation of obesity.

Keywords: DNA damage; GWAS; UK Biobank; exome-wide association study; obesity; rare variant; type 2 diabetes.

Graphical abstract

Figure 1

Gene burden associations of rare variants with adult BMI by sex (A) Miami plot showing significantly associated genes (Bonferroni corrected p < 7.76 × 10⁻⁷) separately in women (upper) and men (lower). (B) QQ plot of the same data. (C) Effect estimates and 95% confidence intervals for each identified gene. For further details, see Tables S2 and S3.

Figure 2

Distributions of adult BMI by sex (A) In all UK Biobank participants; (B) among carriers of rare variants (DMG, damaging; PTV, protein truncating) in genes associated with sex-stratified BMI. Mean and 95% CI for each group are indicated by horizontal bars and boxes. Summarized group data can be found in Table S22.

Figure 3

Adult and childhood obesity risk in carriers of rare damaging variants in the exome-identified genes (A) Comparative size at age 10; “Thinner,” “Average,” or “Plumper” was treated as an ordered categorical outcome to indicate childhood obesity. Adult BMI was similarly split into three categories: <20, >20 but <30, and >30. (B) These two categorical outcomes were tested in cumulative link models against carrier status for qualifying rare exome variants. Displayed log(OR) with 95% CIs and underlying data can be found in Table S7.

Figure 4

Gene burden associations of rare variants with comparative size at age 10 (A) Manhattan plot showing significantly associated genes (Bonferroni corrected p < 1.47 × 10⁻⁶). (B) QQ plot of the same data. (C) Effect estimates and 95% confidence intervals for each identified gene. For further details, see Tables S2 and S3.

Figure 5

Exome associations between the functional domains of POMC, MC4R and SAC10 in the UK Biobank Included variants in the POMC (A) and MC4R (B) genes from our discovery analyses had a minor allele frequency (MAF) smaller than 0.1% and were annotated to be either high-confidence protein truncating variants or missense variants with a high CADD score (≥25). Each variant is presented as an individual line extending to its association p value (−log10), in the direction indicating the direction of effect on SAC10 in carriers of the alternate allele, while the point size indicates the comparative number of carriers of each variant (i.e., allele count), as indicated in the figure legend. Domain-level association statistics can be found in Table S9.

Figure 6

Comparison of rare variant gene-level effects on adult BMI and comparative size at age 10 For each identified exome gene, the adjusted R² for carrier status of qualifying rare exome variants against residual variance in the outcome phenotype after adjusting for covariates. For each gene, the “discovery” trait-sex combinations are shown. Underlying data can be found in Table S13.