Multitrait analyses identify genetic variants associated with aortic valve function and aortic stenosis risk

Figures

• 

  Fig. 1. Study overview. a, Deep learning was used to create segmentation masks of the ascending aorta in the aortic flow image series in the UK Biobank. b, The masks were used to create velocity maps that were used to construct three cMRI-derived phenotypes (AVA, mean gradient and peak velocity). The orange color denotes female participants and the blue color denotes male participants. c, The traits were then used to conduct GWAS of common variants. The summary statistics were further used to create MTAG of AVA, mean gradient and peak velocity with a disease-based aortic stenosis GWAS. d, Gene-set enrichment and analyses for tissue enrichment, as well as cell-specific analysis for the thoracic aorta and left ventricle, were performed. e, Further analyses were performed using Mendelian randomization and polygenic scores of lipid-based traits and other common risk factors with the cMRI-derived traits. Additionally, polygenic risk scores were constructed for each phenotype. Polygenic scores from GWAS before MTAG were used to predict incident aortic stenosis in the external FinnGen cohort. f, MTAG-adjusted scores were applied to the external All of Us cohort and the MGB Biobank. cMRI images…

• 

  Fig. 2. Manhattan plot of GWAS results after MTAG. Genetic associations for AVA, peak velocity and mean gradient after MTAG. Loci with P < 5 × 10−8 are colored red and labeled with the name of the gene in closest proximity to the strongest associated variant. MTAG uses a generalization of IVW meta-analysis and provides updated effect sizes, effective sample sizes and P values for the summary statistics for each trait. For visualization purposes, the Manhattan plot is truncated at a −log10(P) = 30. X axis, chromosomal position; y axis, −log10(P).

• 

  Fig. 3. GWAS for aortic stenosis after MTAG. Genetic associations for aortic stenosis after MTAG. The color is blue when the locus was associated at P < 5 × 10−8. The color is red when the locus was associated at P < 5 × 10−8 with aortic stenosis but not aortic valve measurements after MTAG; the loci are annotated with the name of the gene in closest proximity to the strongest associated variant. MTAG uses a generalization of IVW meta-analysis and provides updated effect sizes, effective sample sizes and P values for the summary statistics. For visualization purposes, the Manhattan plot is truncated at a −log10(P) = 30. X axis, chromosomal position; y axis, −log10(P).

• 

  Fig. 4. Variant effect alignment at MTAG loci. Depiction of the loci significant for at least one trait in MTAG analysis. Effect estimates and P values are taken from the original (non-MTAG) analysis. Effect direction is depicted with respect to the reference allele. Variants with a risk-decreasing reference allele for aortic stenosis are depicted in the left panel. Those with a risk-increasing reference allele are depicted in the right panel. Within each panel, variants are sorted by effect size for aortic stenosis. Effect sizes have units of s.d. for AVA, peak velocity, mean gradient and log odds for aortic stenosis. Variants are represented by the nearest gene name and the variant identifier. Multiple variants at one locus are listed when the lead variant differs across phenotypes.

• 

  Fig. 5. Cumulative incidence of aortic stenosis in All of Us stratified by an MTAG polygenic score. Kaplan–Meier cumulative incidence of aortic stenosis in 496 cases and 243,954 controls and 95% CI in All of Us (496 cases and 243,954 controls, left) and MGB Biobank (680 cases and 42,328 controls, right). Red indicates the top 5% for the aortic stenosis polygenic score. Blue indicates all other participants. The plot is terminated at the 95th percentile of follow-up time (5.0 and 9.0 years). For All of Us, the HR for the top 5% of participants was 3.32 (P = 8.8 × 10−22) compared to the remaining 95% of participants. For MGB Biobank, the HR for the top 5% of participants was 2.76 (P = 7.8 × 10−15) compared to the remaining 95% of participants.