A transcriptome-wide association study identifies PALMD as a susceptibility gene for calcific aortic valve stenosis

Abstract

Calcific aortic valve stenosis (CAVS) is a common and life-threatening heart disease and the current treatment options cannot stop or delay its progression. A GWAS on 1009 cases and 1017 ethnically matched controls was combined with a large-scale eQTL mapping study of human aortic valve tissues (n = 233) to identify susceptibility genes for CAVS. Replication was performed in the UK Biobank, including 1391 cases and 352,195 controls. A transcriptome-wide association study (TWAS) reveals PALMD (palmdelphin) as significantly associated with CAVS. The CAVS risk alleles and increasing disease severity are both associated with decreased mRNA expression levels of PALMD in valve tissues. The top variant identified shows a similar effect and strong association with CAVS (P = 1.53 × 10-10) in UK Biobank. The identification of PALMD as a susceptibility gene for CAVS provides insights into the genetic nature of this disease, opens avenues to investigate its etiology and to develop much-needed therapeutic options.

Conflict of interest statement

The authors declare no competing interests.

Figures

Fig. 1

Manhattan plots showing the GWAS in QUEBEC-CAVS and TWAS results. a Genetic associations with CAVS observed in 1009 cases and 1017 controls. The y-axis represents P value in −log10 scale. The horizontal blue line indicates a P value of 1 × 10−5. The magenta dot indicates rs6702619. b Transcriptome-wide association in valve tissue with CAVS. P values for gene expression-CAVS associations are on the y-axis in −log10 scale. The blue horizontal line represents PTWAS of 0.05. The magenta horizontal line represents the genome-wide significant threshold used in this study (PTWAS < 0.0001). Annotations for the top significant probes are indicated

Fig. 2

PALMD is the candidate causal gene on the 1p21.2 CAVS susceptibility locus. a GWAS and valve eQTL results surrounding PALMD on chromosome 1p21. The upper panel shows the genetic associations with CAVS. The bottom panel shows the valve eQTL statistics for the PALMD gene. The extent of linkage disequilibrium (LD; r2 values) for all SNPs with rs6702619 is indicated by colors. The location of genes in this locus is illustrated at the bottom. bPALMD gene expression levels in aortic valves by genotyping groups for SNP rs6702619. The y-axis shows PALMD mRNA expression levels. The x-axis denotes the three genotype groups. The number of individuals is indicated in parentheses. The CAVS risk allele is illustrated in red. Boxplot boundaries represent the first and third quartiles, whiskers are the most extreme data point, which is no more than 1.5 times the interquartile range, and the center mark represents the median. The black bar and the right y-axis indicate the variance in PALMD gene expression explained by rs6702619. c Scatterplot of the 1p21.2 susceptibility locus showing SNP associations with CAVS and PALMD gene expression in aortic valve tissues. The y-axis represents variant association with CAVS (Z score). The x-axis shows association with PALMD gene expression (t statistic). Variants are colored based on the degree of LD (r2) with the top CAVS-associated variant rs6702619. The blue line is the regression slope with 95% confidence interval (red lines)

Fig. 3

Mendelian randomization analysis of the association between PALMD gene expression and CAVS risk. Each circle represents 1 of 12 SNPs located within 200 kb of PALMD selected for association with PALMD gene expression (P < 0.05) using stepwise regression. The blue line is the regression slope using the Wald method. The magenta dashed lines represent 95% confidence intervals from bootstrap. a Effect on CAVS risk from QUEBEC-CAVS cohort against effect on PALMD gene expression (P = 0.0036). b Effect on CAVS risk from UK Biobank against effect on PALMD gene expression (P = 1.18 × 10−5)

Fig. 4

Relationship between PALMD expression levels and CAVS severity. PALMD-normalized, age- and sex-adjusted mRNA expression levels in 239 aortic valve tissues according to CAVS disease severity assessed by a aortic valve area (P = 0.0027), b mean (P = 0.0001), and c peak transvalvular gradients (P = 8.13 × 10−5). The blue lines represent the slopes obtained by linear regression with 95% confidence interval (magenta lines)

Fig. 5

Replication in UK Biobank. a Forest plot of overall effect size for rs6702619 in the discovery (QUEBEC-CAVS) and validation (UK Biobank) cohorts. The blue filled squares represent the odds ratio (OR) for each cohort with 95% confidence intervals illustrated by horizontal lines. The gray vertical line represents an OR of 1.0 and the dashed magenta line is the OR of the meta-analysis. b Manhattan plot showing the GWAS in UK Biobank comparing 1391 CAVS cases and 352,195 controls. The horizontal blue and magenta lines indicate PGWAS values of 1 × 10−5 and 5 × 10−8, respectively. Three SNPs were significant at the PGWAS threshold of 5 × 10−8: rs74617384 and rs10455872 at the LPA locus and rs7543130 (in perfect LD with rs6702619) at the PALMD locus