Genetic analysis of over 1 million people identifies 535 new loci associated with blood pressure traits

Abstract

High blood pressure is a highly heritable and modifiable risk factor for cardiovascular disease. We report the largest genetic association study of blood pressure traits (systolic, diastolic and pulse pressure) to date in over 1 million people of European ancestry. We identify 535 novel blood pressure loci that not only offer new biological insights into blood pressure regulation but also highlight shared genetic architecture between blood pressure and lifestyle exposures. Our findings identify new biological pathways for blood pressure regulation with potential for improved cardiovascular disease prevention in the future.

Figure 1

Study design schematic for discovery and validation of loci. ICBP; International Consortium for Blood Pressure; N, sample size; QC, quality control; PCA, principal-component analysis; GWAS, Genome-wide Association Study; 1000G 1000 Genomes; HRC, Haplotype Reference Panel; BP: blood pressure; SNPs, single nucleotide polymorphisms; BMI, body mass index; LMM; linear mixed model; UKB, UK Biobank, MAF, minor allele frequency; HLA, Human Leukocyte Antigen; MVP, Million Veterans Program; EGCUT; Estonian Genome Center, University of Tartu; SBP, systolic blood pressure; DBP, diastolic blood pressure; PP, pulse pressure.

Figure 2

Manhattan plot showing the minimum P-value for the association across all blood pressure traits in the discovery stage excluding known and previously reported variants. Manhattan plot of the discovery genome-wide association meta-analysis in 757,601 individuals excluding variants in 274 known loci. The minimum P-value, computed using inverse variance fixed effects meta-analysis, across SBP, DBP and PP is presented. The y axis shows the –log₁₀ P values and the x axis shows their chromosomal positions. Horizontal red and blue line represents the thresholds of P = 5 x 10^-8 for genome-wide significance and P = 1 x 10^-6 for selecting SNPs for replication, respectively. SNPs in blue are in LD (r² > 0.8) with the 325 novel variants independently replicated from the 2-stage design whereas SNPs in red are in LD (r² > 0.8) with 210 SNPs identified through the 1-stage design with internal replication. Any loci in black or grey that exceed the significance thresholds were significant in the discovery meta-analysis, but did not meet the criteria of replication in the one- or two-stage designs.

Figure 3

Venn Diagrams of Novel Loci Results (a) “Comparison of 1-stage and 2-stage design analysis criteria”: For all 535 novel loci, we compare the results according to the association criteria used for the one-stage and the two-stage design. Two-hundred and ten loci exclusively met the one-stage analysis criteria (P <5x10^-9 in the discovery meta-analysis [N=757,601], P < 0.01 in UKB [N=458,577], P < 0.01 in ICBP [N=299,024] and concordant direction of effect between UKB and ICBP). The P-values for the discovery and the ICBP meta-analyses were calculated using inverse variance fixed effects meta-analysis. The P-values in UKB were derived from linear mixed modeling using BOLT-LMM. Of the 325 novel replicated loci from the 2-stage analysis (genome-wide significance in the combined meta-analysis, P < 0.01 in the replication meta-analysis and concordant direction of effect), 204 loci would also have met the one-stage criteria, whereas 121 were only identified by the two-stage analysis. (b) “Overlap of Associations across Blood Pressure Traits”. For all 535 novel loci, we show the number of loci associated with each blood pressure trait. We present the two-stage loci first, followed by the one-stage loci. SBP: systolic blood pressure; DBP: diastolic blood pressure; PP: pulse pressure; UKB: UK Biobank; ICBP: International Consortium of Blood Pressure.

Figure 4

Association of blood pressure loci with lifestyle traits. Plot shows unsupervised hierarchical clustering of BP loci based on associations with lifestyle-related factors. For the sentinel SNP at each BP locus (x-axis), we calculated the -log₁₀(P)*sign(β) (aligned to BP-raising allele) as retrieved from the Gene Atlas catalogue (http://geneatlas.roslin.ed.ac.uk). The P-values in Gene Atlas were calculated applying linear mixed models. BP loci and traits were clustered according to the Euclidean distance amongst -log₁₀(P)*sign(β). Red squares indicate direct associations with the trait of interest and blue squares inverse associations. Only SNPs with at least one association at P <10^-6 with at least one of the traits examined are annotated in the heat-map. All 901 loci are considered, both known and novel: novel loci are printed in bold font. SNPs: Single Nucleotide Polymorphisms; BP: Blood Pressure.

Figure 5

Association of blood pressure loci with other traits. Plot shows results from associations with other traits which were extracted from the GWAS catalog and PhenoScanner databases for the 535 novel sentinel SNPs including proxies in Linkage Disequilibrium (r² ≥ 0.8) with genome-wide significant associations. SBP: Systolic Blood Pressure; DBP: Diastolic Blood Pressure; PP: Pulse Pressure; HR: Heart Rate; ECG: Electrocardiographic traits; CAD: Coronary Artery Disease CHD; Coronary Heart Disease MI; Myocardial Infraction; T2D: Type II Diabetes.

Figure 6

Association of blood pressure loci with other traits. Plots (a) and (b) show overlap between variants associated to (a) traits and (b) diseases in the manually-curated version of the DisGeNET database, and all variants in LD r²>0.8 with the known (red bars) SNPs from the 274 published loci, and all (green bars) BP variants from all 901 loci. Numbers on top of the bars denote the number of SNPs included in DisGeNET for the specific trait or disease. Traits/diseases with an overlap of at least 5 variants in LD with all markers are shown. The Y axis shows the percentage of variants associated with the diseases that is covered by the overlap. For the sake of clarity, the DisGeNET terms for blood pressure and hypertension are not displayed, whereas the following diseases have been combined: coronary artery disease (CAD), coronary heart disease (CHD) and myocardial infarction (MI); prostate and breast carcinoma; Crohn's and inflammatory bowel diseases.

Figure 7

Relationship of deciles of the genetic risk score (GRS) based on all 901 loci with blood pressure, risk of hypertension and cardiovascular disease in UK Biobank. The plots show sex-adjusted (a) mean systolic blood pressure (SBP) and odds ratios of hypertension (HTN) (N=364,520) and (b) odds ratios of incident cardiovascular disease (CVD), myocardial infarction (MI) and stroke (N=392,092), comparing each of the upper nine GRS deciles with the lowest decile; dotted lines represent the upper 95% confidence intervals.

Figure 8

Known and novel BP associations in the TGFβ signalling pathway. Genes with known associations with BP are indicated in cyan. Genes with novel associations with BP reported in this study are indicated in red. TGFβ pathway was derived from an ingenuity canonical pathway. BP: Blood Pressure.