From polygenic risk scores to integrative epigenomics: the dawn of a new era for cardiovascular precision medicine

Commentary on ‘Regulatory genomic circuitry of human disease loci by integrative epigenomics’ by C. Boix et al. Nature, 2021, 590, 300–307.

In the last two decades, genome-wide association studies (GWASs) took advantage of advances in genome-wide genotyping technology, and large population and patient datasets, to explore the role of common variants on phenotypic traits and disease susceptibility. GWAS have the major advantage of being hypothesis free, thus allowing investigators to uncover associations between cardiovascular traits and genetic polymorphisms, which can help point to unexpected pathways involved in cardiovascular pathophysiology such as the important role of inflammatory-related genes in coronary artery disease (CAD)¹ or as the identification of the PITX2 transcription factor in atrial fibrillation.² According to the GWAS catalogue database (release of 25 March 2021), 1329 polymorphism-cardiovascular trait associations have been so far described. This growing catalogue of genome-wide (P < 5 × 10⁻⁸) and nominally (P < 0.05) significant variants has also opened the door to creating polygenic risk scores (PRS) and other personalized medicine approaches, by identifying individuals at risk of developing specific cardiovascular diseases or sub-groups of patients with a more severe prognosis.

But despite their remarkable success, GWAS also have their limitations. Most notably, the core output of GWAS, being long lists of disease-associated genes and single-nucleotide polymorphisms (SNPs), provide no additional insights on potential mechanisms of action. As a result of this, in parallel with the mounting success of GWAS, has been the growing list of disease-associated genes and SNPs for which we have little or no idea about how they might cause disease. This issue is compounded by the fact that the large majority of genetic signals (90%) are located in non-coding regions of the genome, likely corresponding to transcriptional regulatory elements.³ Thus, a major challenge that has arisen, has been understanding the impact of variants in such regulatory elements, and the added complexity of needing to consider not only gene regulation but also the tissue pleiotropy of an enhancer. That is, a given enhancer or transcriptional regulatory element might have a profound effect in a specific tissue (i.e. in liver), but have no effect or even the oppositive effect in another tissue (i.e. the heart). At a minimum, any attempt at a mechanistic molecular understanding of these factors requires the integration of: (i) GWAS data comprising hundreds of disease-associated SNPs and genes; (ii) epigenetic data, obtained from specific human tissues since the location and role of the regulatory elements (promoter or enhancer) are tissue- and species-dependent; (iii) transcriptomic (RNA-seq) data sets to assess human tissue gene expression levels (or even gene isoforms) that can be correlated with genetic variants (eQTL: expression Quantitative Trait Loci) and which can then point to causal variants or target genes thus controlled or modulated.

In a recent publication in Nature, Boix et al. in their study ‘Regulatory genomic circuitry of human disease loci by integrative epigenomics’, undertook a tour de force where they tackle these exact challenges.⁴ They leverage unprecedented network data and tools to disclose, among other things: (i) the effect of individual variants on multiple traits through effects in different tissues; (ii) the combined effect of multiple variants on the expression pattern of individual genes in multiple different tissues; and (iii) the combined effect of multiple genes in one or multiple tissues. They then developed a repository called Epimap (http://compbio.mit.edu/epimap) based on the uniform processing of 833 biosamples (comprising 33 tissue categories) for 18 epigenomics marks resulting in 2850 observed and 14 510 imputed tracks. This comprehensive annotation revealed more than 2 million enhancers, with an average of 13 enhancers modulating each gene and 1.5 genes controlled by each enhancer. Boix et al. also demonstrated enhancer co-regulation and described 300 enhancer modules including 290 tissue-specific modules. These different layers of data integration appear to be essential for understanding the genetic architecture underlying the broad phenotypic traits encountered in common and complex cardiovascular diseases such as CAD (Figure 1). As an example, Boix et al. described 56 ‘unifactorial’ traits enriched in only one tissue such as the QT interval in heart (where the heart is the only tissue that might contribute to controlling the QT interval), whereas 192 ‘multifactorial’ traits find enrichment on average in five different tissues such as waist-to-hip ratio in adipose, muscle, kidney, and digestive tissues, and finally, 26 ‘polyfactorial’ traits in 14 tissue categories on average such as CAD that was associated with enrichment in 19 tissues including liver, heart, adipose, muscle, and endocrine samples.

Figure 1.

Pleotropic impact at the gene- and tissue-level of variants identified in GWAS. The integration of different layers of data (epigenomic and transcriptional datasets) is essential to understand the genetic architecture underlying the broad phenotypic traits encountered in common and complex cardiovascular diseases such as CAD. Organ's cartoons have been taken from https://smart.servier.com.

In addition, their study provides an unprecedented integrative comprehension at the gene and tissue level of the pleotropic impact of variants identified in GWAS. Common variants identified by GWAS (P < 5 × 10⁻⁸) are mostly considered as modest contributors to disease inheritance. However, this study supports the emerging concept of the likely larger contribution of variants exhibiting nominally significant P-values (P < 0.05 but >5 × 10⁻⁸), as also suggested by studies examining either gene-regulatory networks⁵ or polygenic risk scores^6,7 for CAD. Furthermore, their study supports the concept that common variant effects may be potentiated by combined genetic factors impacting the same gene-pathway, in combination with other (rare or low frequency) variants, drugs, or environments.^8,9

The study by Boix et al. is at the same time a rich scientific resource, but also a lesson regarding the profound and magnificent complexity of the human genome and the causal basis of common diseases like CAD. Their enhancer modules and trait-trait networks open the door to understanding fine gene regulation on one hand and broad distance gene regulation pleiotropy orchestrated by enhancers of multiple genes and multiple tissues. The study also underscores the importance of developing global tissue research programs to decipher the broad phenotypic spectrum of complex diseases, while also highlighting the complex and multi-layered analyses that are required to meaningfully interpret these datasets. More newsworthy than ever, multidisciplinary research that integrates complex multi-layered datasets from various sources appears to be key for deciphering the full complexity of cardiovascular diseases.

Precision medicine is being increasingly applied to manage patients with Mendelian disorders, where genotype can directly inform phenotype and management decisions can be guided by patient-specific information. But given their remarkably complex genetic and epigenetic basis, common diseases like CAD, obesity and hypertension lag behind Mendelian disorders in terms of being able to apply precision medicine. However, by continuing to advance our understanding of why individual patients develop common complex diseases, we foresee that advanced approaches such as integrative epigenomics, in combination with other techniques, will eventually help to guide patient management decisions. For example, while PRS have so far been marginal in their ability to predict disease or disease severity, integrative epigenomics will likely facilitate the fine-tuning of PRS by re-assigning weights to differing variants according to their specific pleiotropic effects. So, while we are not there yet, through integrative epigenomics and other advances it seems that we are inching towards the long-held goal that precision medicine will one day inform the delivery of the right treatment, to the right patient, at the right time, for every patient.

Conflict of interest: none declared.

Funding

J.B. is supported by the research program Etoiles montantes des Pays de la Loire REGIOCARD RPH081-U1087-REG-PDL, by the ANR LEARN (R21006NN) and by the ANR EJPRD LQTS-NEXT. J.C.K. acknowledges research support from the National Institutes of Health (R01HL130423, R01HL135093, R01HL148167-01A1) and New South Wales health grant RG194194.

Authors

Biography: Julien Barc' s group leads genetics and epigenetics research programs on the inherited cardiac arrhythmia at l institut du thorax in Nantes, France, where he obtained his PhD in 2009 in genetics of ventricular cardiac arrhythmia. Julien spent then 5 years at the department of Experimental Cardiology of Amsterdam, the Netherlands, a world leader group in genetics and functional studies on cardiac arrhythmias. He dedicated the early part of his carrier to identify new syndromes, genes and mutations associated with inherited cardiac arrhythmia at risk of sudden death. More recently he developed research programs on patient population to investigate the role of common variants in cardiac disorders. This conducts Julien to explore the role of regulatory region by initiating epigenomics projects and develop models based on genome editing in induced pluripotent stem cell-derived cardiomyocytes.

Biography: Professor Jason Kovacic graduated from The University of Melbourne Medical School in 1994, and then undertook cardiology training at St Vincent s Hospital in Sydney including a PhD in cardiovascular medicine at the Victor Chang Cardiac Research Institute. In 2007, he relocated to the USA, initially to do a postdoc at the National Heart, Lung and Blood Institute (NHLBI) at the National Institutes of Health (NIH) in Bethesda, Maryland. After this postdoc he then moved to The Icahn School of Medicine at Mount Sinai in New York. In early 2020 Jason was appointed as the Executive Director of the Victor Chang Cardiac Research Institute in Sydney, Australia. He continues to practice as a clinical cardiologist and runs a research program focused on atherosclerosis, fibromuscular dysplasia and other vascular diseases.