GWAS of allometric body-shape indices in UK Biobank identifies loci suggesting associations with morphogenesis, organogenesis, adrenal cell renewal and cancer

Abstract

Genetic studies have examined body-shape measures adjusted for body mass index (BMI), while allometric indices are additionally adjusted for height. We performed the first genome-wide association study of A Body Shape Index (ABSI), Hip Index (HI) and the new Waist-to-Hip Index and compared these with traditional indices, using data from the UK Biobank Resource for 219,872 women and 186,825 men with white British ancestry and Bayesian linear mixed-models (BOLT-LMM). One to two thirds of the loci identified for allometric body-shape indices were novel. Most prominent was rs72959041 variant in RSPO3 gene, expressed in visceral adipose tissue and regulating adrenal cell renewal. Highly ranked were genes related to morphogenesis and organogenesis, previously additionally linked to cancer development and progression. Genetic associations were fewer in men compared to women. Prominent region-specific associations showed variants in loci VEGFA and HMGA1 for ABSI and KLF14 for HI in women, and C5orf67 and HOXC4/5 for ABSI and RSPO3, VEGFA and SLC30A10 for HI in men. Although more variants were associated with waist and hip circumference adjusted for BMI compared to ABSI and HI, associations with height had previously been reported for many of the additional variants, illustrating the importance of adjusting correctly for height.

Figure 1

Miami plots of candidate SNPs identified for allometric body-shape indices in women and men. (a) GWAS of waist-to-hip index calibrated for UK Biobank participants (WHI_UKB); (b) GWAS of a body shape index calibrated for UK Biobank participants (ABSI_UKB); (c) GWAS of hip index calibrated for UK Biobank participants (HI_UKB); P-P-values were derived from BOLT-LMM infinitesimal models; SNP-single nucleotide polymorphism; horizontal lines correspond to the genome-wide significance cut-off P = 5*10^–8. Genomic risk loci with their corresponding locus lead SNPs were identified with FUMA v1.3.6a. All candidate SNPs are shown in grey, locus lead SNPs are colour-coded as follows: (grey circle) candidate SNPs; (dark red circle) novel genomic risk locus identified in the current study, with no previously reported candidate SNPs; (orange circle) genomic risk locus with a previously reported locus lead SNP; (yellow circle) genomic risk locus with a previously reported SNP in strong linkage disequilibrium (LD) with the locus lead SNP at r² ≥ 0.6; (cyan circle) genomic risk locus with previously reported other candidate SNP. SNPs identified for allometric body-shape indices in the current study were matched against SNPs reported in the NHGRI-EBI GWAS Catalog (https://www.ebi.ac.uk/gwas/home, accessed on 07/04/2021) in association with the corresponding traditional body-shape index (with or without adjustment for body mass index, BMI), i.e. the waist-to-hip ratio for WHI_UKB (catalogue sets EFO_0004343, EFO_0007788, EFO_0004302); waist circumference for ABSI_UKB (EFO_0004342, EFO_0007789, EFO_0004302); hip circumference for HI_UKB (EFO_0005093, EFO_0008039, EFO_0004302).

Figure 2

Deleteriousness (CADD) score of locus lead SNPs identified for allometric body-shape indices. (a) waist-to-hip index (WHI_UKB) calibrated for UK Biobank women (n = 282 genomic risk loci with the corresponding locus lead SNPs); (b) a body shape index (ABSI_UKB) calibrated for UK Biobank women (n = 200); (c) hip index (HI_UKB) calibrated for UK Biobank women (n = 171); (d) WHI_UKB for UK Biobank men (n = 97); (e) ABSI_UKB for UK Biobank men (n = 65); (f) HI_UKB for UK Biobank men (n = 75). CADD-Combined Annotation Dependent Depletion; LD-linkage disequilibrium; SNP-single nucleotide polymorphism; red vertical lines-recommended cut-off 12.37 for CADD (the higher the more deleterious); horizontal line-genome-wide significance cut-off (P = 5*10^–8); left-hand side-CADD for the locus lead SNP of each genomic risk locus, with the proportion above the cut-off; right-hand side-candidate SNPs in strong LD with the locus lead SNP (colour-coded according to r² ≥ 0.6), showing the maximum CADD within the LD block, plotted with the corresponding significance on the y-axis; P_{WHI / ABSI / HI}-P-values for body-shape indices, derived from BOLT-LMM infinitesimal models; (black circle)-marks the locus lead SNP when this is showing the maximum CADD within the corresponding LD block; percentages (top corners)-percentage above the cut-offs for both, CADD and genome-wide significance (all differences between left-hand side and right-hand side proportions were significant at P < 0.0001 when compared with Fisher’s exact test, except for ABSI_UKB in men (P = 0.0008).

Figure 3

Miami plots of significant genes identified for allometric body-shape indices in women and men. (a) waist-to-hip index calibrated for UK Biobank participants (WHI_UKB); (b) a body shape index calibrated for UK Biobank participants (ABSI_UKB); (c) hip index calibrated for UK Biobank participants (HI_UKB); P-P-values were derived from MAGMA v1.08 employed in FUMA v1.3.6a and were adjusted with Bonferroni correction for 19,088 protein-coding genes; SNP-single nucleotide polymorphism; horizontal lines-correspond to P = 0.05 after Bonferroni correction. Significant genes within 250 kb window were consolidated in genomic risk regions represented by a lead gene. All significant genes are shown in grey, lead genes are colour-coded as follows: (grey circle)-significant genes; (dark red circle)-novel genomic risk region identified in the current study, with no previously reported significant gene; (orange circle)-genomic risk region with a previously reported lead gene; (cyan circle)-genomic risk region including a previously reported significant gene (not the lead gene). Genes identified for allometric body-shape indices in the current study were matched against genes reported in the NHGRI-EBI GWAS Catalog (https://www.ebi.ac.uk/gwas/home, accessed on 07/04/2021) in association with the corresponding traditional body-shape index (with or without adjustment for body mass index, BMI), i.e. the waist-to-hip ratio for WHI_UKB (catalogue sets EFO_0004343, EFO_0007788, EFO_0004302); waist circumference for ABSI_UKB (EFO_0004342, EFO_0007789, EFO_0004302); hip circumference for HI_UKB (EFO_0005093, EFO_0008039, EFO_0004302).

Figure 4

Gene sets associated with allometric body-shape indices. ABSI-a body shape index calibrated for UK Biobank participants (ABSI_UKB); Currated-curated gene set; GO-BP-gene ontology biological process; GO-CC-gene ontology cellular component; GO-MM-gene ontology molecular function; HI-hip index calibrated for UK Biobank participants (HI_UKB); WHI-waist-to-hip index calibrated for UK Biobank participants (WHI_UKB); #-reactome runx1 regulates transcription of genes involved in wnt signaling. The figure includes gene sets identified using competitive gene-set analysis with MAGMA v1.08 employed in FUMA v1.3.6a as significant for at least one of the allometric indices in women or men with adjusted P_adj < 0.05, incorporating Bonferroni correction for multiple comparisons for 15,485 gene sets. Gene sets with P_adj < 0.05 for WHI_UKB in women were sorted in descending order of P_adj, for WHI_UKB in women, then the remaining gene sets with P_adj < 0.05 for WHI_UKB in men were sorted in descending order of P_adj for WHI_UKB in men, then the remaining gene sets with P_adj < 0.05 for HI_UKB in women were sorted in descending order of P_adj for HI_UKB in women and last, the remaining gene sets with P_adj < 0.05 for HI_UKB in men were sorted in descending order of P_adj for HI_UKB in men, leaving a single gene set associated only with ABSI_UKB in women. The horizontal lines mark the start of the next sorting criterion. White cells are gene sets with P_adj ≥ 0.05 for the corresponding body-shape index.

Figure 5

eQTLs associated with allometric body-shape indices. ABSI-a body shape index calibrated for UK Biobank participants (ABSI_UKB); eQTLs-expression Quantitative Trait Loci; HI-hip index calibrated for UK Biobank participants (HI_UKB); WHI-waist-to-hip index calibrated for UK Biobank participants (WHI_UKB). The figure includes specific tissues from the Genotype-Tissue Expression (GTEx) v8.0 database, queried for associations with eQTLs with gene-property analysis, which were identified by MAGMA v1.08 employed in FUMA v1.3.6a as significant for at least one of the allometric indices in women or men, i.e. with adjusted P_adj < 0.05, incorporating Bonferroni correction for multiple comparisons for 54 tissues. White cells are tissues with P_adj ≥ 0.05 for the corresponding body-shape index.

Figure 6

Significance ranking of independent significant SNPs: comparison between pairs of traditional and allometric body-shape indices. (a) independent significant SNPs (at r² < 0.6 within 1 Mb window) for WHI_UKB (n = 1,988) or WHR_adjBMI (n = 1,959) in women; (b) independent significant SNPs for ABSI_UKB (n = 1,273) or WC_adjBMI (n = 1,586) in women; (c) independent significant SNPs for HI_UKB (n = 814) or HC_adjBMI (n = 2,008) in women; (d) independent significant SNPs for WHI_UKB (n = 617) or WHR_adjBMI (n = 571) in men; (e) independent significant SNPs for ABSI_UKB (n = 266) or WC_adjBMI (n = 842) in men; (f) independent significant SNPs for HI_UKB (n = 222) or HC_adjBMI (1,400) in men. Association statistics p-values were derived from BOLT-LMM infinitesimal models. ABSI_UKB-a body shape index calibrated for UK Biobank participants; BMI-body mass index; HC_adjBMI-hip circumference adjusted for BMI; HI_UKB-hip index calibrated for UK Biobank participants; SNP-single nucleotide polymorphism; WC_adjBMI-waist circumference adjusted for BMI; WHR_adjBMI-waist-to-hip ratio adjusted for BMI; WHI_UKB-waist-to-hip index calibrated for UK Biobank participants. Colour scale-colour-marked were only SNPs reported as associated with height in the NHGRI-EBI GWAS Catalog (https://www.ebi.ac.uk/gwas/home, accessed on 07/04/2021), i.e. included in catalogue sets EFO_0004339 or EFO_0004302; (navy circle)-independent significant SNP reported in association with height; (cyan circle)-independent significant SNP in strong LD (at r² ≥ 0.6) with a SNP reported in association with height.