https://orcid.org/0000-0003-1893-8307
Genome-wide analyses to identify targeted therapy for calcific aortic valve stenosis.
This editorial refers to ‘Dyslipidemia, inflammation, calcification, and adiposity in aortic stenosis: a genome-wide study’, by H.Y. Chen et al., https://doi.org/10.1093/eurheartj/ehad142.
Calcific aortic valve stenosis (CAVS) is a frequent condition associated with significant morbidity. In Western countries, severe CAVS is present in ∼3–5% of individuals >70 years old.1 The number of individuals with an indication for aortic valve replacement is expected to double by 2050, leading to a significant health and economic burden on society.1 There is currently no effective medical therapy to prevent this condition, in part because of our incomplete understanding of the disease process. Despite evidence for genetic susceptibility from family studies, only a handful of genes have been identified as directly associated with disease risk. Genome-wide association studies (GWAS) for CAVS performed thus far have included up to 5115 cases, leading to the identification of seven loci, namely 6q25.3-q26-LPA, 1p21.2-PALMD, 7p15.3-IL6, 1p36.12-ALPL, 1q32.1-NAV1, 2q22.3-TEX41, and 11q12.2-FADS1/2 (Graphical Abstract).2–5 Genetic studies in complex diseases taught us that the sample size at which the discovery of statistically significant loci emerges from GWAS varies by disease. For coronary artery disease, the number of associated loci has increased steadily at a rate of approximately one new locus per thousand cases included, up to ∼60 000 cases. The subsequent addition of large biobanks (e.g. the UK Biobank) has led to the identification of 241 loci from 181 522 cases in individuals of European ancestry in the latest study.6 For CAVS, we can decisively suggest that the sample size in GWAS above which the rate of locus discovery accelerates has not been reached yet. Progress in elucidating the polygenic inheritance of this disease is thus expected from studies including a larger sample size.
The study published by Chen et al.7 in this issue of the European Heart Journal brings us one step forward in our quest to find susceptibility genes for CAVS. The authors report the largest meta-analysis of GWAS to date for CAVS, including 13 765 cases of European ancestry. They identified 10 novel genetic loci associated with CAVS, including eight that were replicated in an independent cohort (n = 7111 cases). The novel replicated loci were tagged by non-coding variants of small effect sizes with odds ratios in the meta-analysis of the discovery and replication cohorts ranging from 1.07 to 1.16 per risk allele (Graphical Abstract). Among the new loci, the authors report a signal near CELSR2-SORT1; the risk allele at the lead single nucleotide polymorphism (SNP) is associated with lower hepatic expression of SORT1, higher circulating LDL-cholesterol and apolipoprotein B, as well as increased risk of coronary artery disease. Sortilin 1, the protein encoded by SORT1, has also recently been implicated in fibrosis and calcification of human valve interstitial cell cultures, garnering further interest in this locus in CAVS.8 Whether sortilin 1 acts through modulation of circulating lipoproteins or an aortic valve-specific mechanism remains to be elucidated. The authors also report a novel association near NLRP6, a gene involved in the formation of inflammasome complexes. Of note, a genetic variant influencing the expression in blood of NLRP3, a gene in the same family, has previously been associated with cardiovascular mortality.9 Some of the variants at the NLRP6 locus may impact gene function according to functional annotations. The lead variant was also associated with aortic valve calcification measured by computed tomography, which suggests its implication in calcification and the early disease process. Two other novel lead variants, located near PRRX1 and ACTR2, were associated with aortic valve calcification. Interestingly, genetically predicted expression of both genes in the aorta was positively associated with CAVS risk. While several of the lead SNPs were also associated with coronary artery disease and common cardiovascular risk factors, others seemed to be more specific to CAVS. Considering the complexity of the aortic valve tissue and its crucial role in the cardiac cycle, we can expect to discover genetic variations that affect tissue and vascular architecture, promote calcification, or impact haemodynamics. In this sense, further exploration of the mechanisms behind the associations at NLRP6, PRRX1, and ACTR2, among others, could lead to further insights on CAVS pathophysiology. The biological role of the other novel loci at 3q25.2-ARHGEF26, 9q31.1-SMC2, 16q23.1-TMEM170A, and 18q11.2 is currently unknown.
The authors also report the first genetic risk scores (GRSs) derived from CAVS GWAS. They developed three scores with weights obtained from a GWAS meta-analysis excluding the UK Biobank: one including the 18 lead variants, a second including 559 variants with P < 1 × 10−4, and a third using LDpred2, a Bayesian method that accounts for linkage disequilibrium between variants. All three were associated with prevalent CAVS in the UK Biobank (n = 3091–3410 cases), with odds ratios of 1.37, 1.55, and 1.45 per standard deviation, respectively, which is of the same order of magnitude as previous studies for coronary artery disease.10 All three scores modestly improved discrimination capacity when added to age, sex, and clinical risk factors. One cannot directly compare the three scores since the population used for validation for the first two only included individuals aged 55 years or older as opposed to all individuals for the LDpred2 score. Nevertheless, scores including a large number of variants (usually >1 million) with genome-wide representation, such as those obtained with LDpred2, tend to perform better for the prediction of complex traits.11 GRSs for CAVS have the potential to become important tools for clinical research and disease management. They could be used to identify individuals at high genetic risk of developing CAVS to be recruited for clinical trials (Graphical Abstract). They could also help to inform about the optimal use of echocardiographic screening and guide the clinical follow-up of patients with asymptomatic CAVS. Ultimately, prospective studies are needed to determine if such GRSs are associated with the rate of disease progression.
Finally, the genetic correlation and Mendelian randomization (MR) analyses of Chen et al. 7 point to a causal role for circulating lipids and obesity in CAVS risk, corroborating previous findings12,13 while using more robust data for genetic susceptibility to CAVS. The discovery of several risk loci associated with an increased plasma level of apolipoprotein B along with the MR results support an important role for circulating atherogenic lipoproteins. It is interesting to note that several risk loci were also associated with blood pressure, another actionable target for preventative treatments. Although not assessed in this study, previous MR analyses have suggested a causal role for systolic blood pressure in CAVS.14 The accumulation of evidence for the causal implication of circulating lipids, obesity, and blood pressure in CAVS supports further investigation into the potential of pharmacological agents targeting these risk factors for disease prevention and treatment (Graphical Abstract).
The study by Chen et al.7 has several strengths. It represents the largest GWAS for CAVS to date, including 10 cohorts with replication in an independent population. The development of a first GRS specific to CAVS independently associated with the disease offers some insights about the potential clinical utility of the new genetic knowledge derived from GWAS. MR analyses confirmed a causal role for circulating blood lipids and obesity in CAVS using a more robust instrument for genetic susceptibility to CAVS. Among the limitations, we note the low number of participants of non-European ancestry, which limits generalizability of the findings, as acknowledged by the authors. It is also unclear how many individuals with a bicuspid aortic valve (BAV) were included in the studies. BAV is frequent in patients undergoing aortic valve replacement, has a very high heritability, and specific genetic determinants have been identified.15 Finally, in silico post-GWAS analyses performed with public datasets are a necessary passage to interpret the biology underpinning GWAS loci. However, in this study, orthogonal omics data in the disease-relevant tissue (i.e. the aortic valve) are missing. The regulatory landscape is known to be context dependent as well as cell type and tissue specific; the lack of quantitative trait loci data in diseased and non-diseased valve tissues therefore limits the interpretation in this case.
Overall, this study may have reached the effective sample size in CAVS where the rate of locus discovery is ramping up. While the results point to potentially actionable pathways such as blood lipids, obesity, and inflammation, further molecular and functional studies in relevant cells, tissues, and living organisms are needed to fully characterize the molecular mechanisms at play. The development of GRSs in addition to evidence from MR nevertheless provide valuable information to guide the design of future randomized clinical trials. This study is an important step in the journey towards the identification of molecular targets for medical management of CAVS and may represent, with international collaborations, the tipping point of a new era in genetics of CAVS.
Acknowledgements
The graphical abstract was created with BioRender.com. The authors would like to thank Hasanga Manikpurage, Ursula Houessou, and Pardis Zamani for their assistance in creating the graphical abstract.
Data availability
No new data were generated or analysed in support of this research.
Funding
All authors declare no funding for this contribution.
References
Author notes
The opinions expressed in this article are not necessarily those of the Editors of the European Heart Journal or of the European Society of Cardiology.
Conflict of interest None declared.
