Calcific aortic stenosis: omics-based target discovery and therapy development

Abstract

Calcific aortic valve disease (CAVD) resulting in aortic stenosis (AS) is the most common form of valvular heart disease, affecting 2% of those over age 65. Those who develop symptomatic severe AS have an average further lifespan of <2 years without valve replacement, and three-quarters of these patients will develop heart failure, undergo valve replacement, or die within 5 years. There are no approved pharmaceutical therapies for AS, due primarily to a limited understanding of the molecular mechanisms that direct CAVD progression in the complex haemodynamic environment. Here, advances in efforts to understand the pathogenesis of CAVD and to identify putative drug targets derived from recent multi-omics studies [including (epi)genomics, transcriptomics, proteomics, and metabolomics] of blood and valvular tissues are reviewed. The recent explosion of single-cell omics-based studies in CAVD and the pathobiological and potential drug discovery insights gained from the application of omics to this disease area are a primary focus. Lastly, the translation of knowledge gained in valvular pathobiology into clinical therapies is addressed, with a particular emphasis on treatment regimens that consider sex-specific, renal, and lipid-mediated contributors to CAVD, and ongoing Phase I/II/III trials aimed at the prevention/treatment of AS are described.

Graphical abstract

Calcific aortic valve disease (CAVD) results from a complex physiological microenvironment and multifaceted interplay amongst a number of putative initiators and drivers of disease. Multi-omics enables a holistic assessment of molecular drivers of CAVD via quantitation of the (epi)genome, transcriptome, proteome, and metabolome at both a bulk and single-cell level. Integration of multiple omics modalities facilitates the identification of novel, subtle, or cross-layer molecules and interactions. Target prioritization via systems biology precedes drug development and eventual pharmaceutical alternatives to surgical aortic valve replacement or transcatheter aortic valve replacement.

Open in new tabDownload slide

Introduction

Calcific aortic valve disease (CAVD) resulting in aortic stenosis (AS) is a significant driver of cardiovascular morbidity and mortality and encompasses a spectrum of early leaflet fibrosis and thickening [aortic valve (AV) sclerosis, present in ∼25% of seniors] to highly fibrocalcific and symptomatic AS affecting nearly 3% of those over age 75.¹ Beyond surgical aortic valve replacement (AVR, SAVR) or transcatheter AVR (TAVR), there are no approved pharmaceutical therapies for AS. As a result, some ∼400 000 AVR procedures are performed annually worldwide.² These interventions are specialized, resource-intensive (>$10 billion/year in the USA alone³), invasive, and not without risk.⁴ In addition, structural valve degeneration of implanted bioprosthetic valves limits their lifespan to roughly a decade after implantation.⁵ As AVRs expand into younger patient cohorts who are projected to require multiple valve re-interventions and replacements in their lives due to repeated bioprosthetic valve failure, these technological limitations have sparked discussion and implementation of lifetime management strategies to the point where preemptive aortic root enlargement, early pulmonary autograft replacements, and valve-in-valve-in-valve procedures are contemplated and enacted.^6,7 If left untreated after the onset of symptoms, most patients with severe AS die within 2 years,⁸ and the vast majority will require invasive AVR or suffer from heart failure or death within half a decade.⁹ This grim prognosis—worse than that of many cancers—only serves to place in stark relief the limited therapeutic options available for patients suffering from AS. There are also important implications for global health equity: concurrent epidemics of aging, obesity, diabetes, and chronic kidney disease are predicted to lead to a rapid 2.5-fold increase in global AS incidence in the next 15 years which will be particularly concentrated in low- and middle-income countries, developing countries and the global south.^10–12 In turn, these regions have a critical lack of capacity for cardiac surgery or interventional cardiology,^13–15 and will be unable to leverage the otherwise-revolutionary benefits to access and equity of care provided by TAVR. Even within high-income and developed countries, massive racial, ethnic, and socioeconomic disparities exist in risk and/or diagnosis of AS,^16,17 access to TAVR,¹⁸ and outcomes after AVR.^19,20 In the face of these many challenges, there is a clear and urgent unmet need to unravel CAVD mechanisms, discover molecular targets, and drive the development of novel pharmacotherapies.²¹

The anatomy of the AV is intricate, with a tri-layered architecture ∼1 mm thick:²² the collagen-rich fibrosa faces the ascending aorta, the elastin-rich ventricularis is positioned next to the left ventricle, and the proteoglycan-rich spongiosa rests in between.²³ In normal tissues, two main mechanosensitive cell types dominate the valve: quiescent, fibroblastic valvular interstitial cells (VICs) are found throughout the three leaflet layers,²⁴ while a monolayer of valvular endothelial cells (VECs) lines the blood-contacting surfaces of the leaflet.²⁵ Disease occurs preferentially in the fibrosa,²⁶ where tissue and blood biomechanics, lipid-mediated inflammation, extracellular vesicles (EVs), platelet activation, myofibroblastic/osteoblastic/chondrogenic differentiation, immune cell infiltration, endothelial-to-mesenchymal transition (EndMT), aberrant extracellular matrix (ECM) turnover, fibrosis, impaired calcium and phosphate metabolism, cellular senescence, and other as-yet undiscovered mechanisms all conspire, to one degree or another, to actively drive tissue fibrosis, leaflet thickening, and mineralization.^24,27–36 This leads inexorably towards impaired leaflet opening, left ventricular remodelling, heart failure, and eventual death (Figure 1).

Much of the failure to bring effective CAVD treatments to bear is the result of still-limited understanding of the molecular mechanisms that govern pathogenesis in the AV. Initiation and progression of CAVD is highly complex and multi-factorial, existing in vitro and in vivo models of disease remain imperfect,^37,38 and the standard of care-driven absence of excised human tissue samples from normal, sclerotic, or mild/moderate AS has potentiated disease stage-specific segmentation of end-stage specimens.^30,39–41 When paired with cutting-edge systems biology approaches, multi-omics of the valvular (epi)genome, transcriptome, proteome, and metabolome hold significant promise to enable greater holistic comprehension of the molecular fingerprint of CAVD via improved identification of risk genes, circulating biomarkers, or disease-driving proteins, lipids, and metabolites.^42,43 These strategies may enable the rigorous definition of healthy baselines, building new prediction models of disease progression, the benchmarking of in vitro, in vivo, or in silico disease model performance against CAVD in humans, or improved characterization of valve cell heterogeneity to allow cell- or tissue-specific drug targeting. Most importantly, these technologies permit the bench-to-bedside prioritization of novel signalling pathways and individual molecules as promising targets befitting of investment in validation, drug development, and clinical trials (Figure 2). While pharmacotherapy development is challenged by the multifaceted nature of AS initiation and progression, multi-omics, systems biology, and network medicine approaches may allow the hierarchical characterization of disease to enable the targeting of mechanisms that are highly upstream or downstream of calcification, and the quantitative prioritization of molecular pathways with an outsized contribution to disease progression. Due to the rising popularity and near-ubiquity of this approach, contemporary (multi-)omics studies are an important and useful lens through which to observe the progress of basic and translational science. Prior multi-omics-driven advances in CAVD have been reviewed elsewhere,^42,43 and we will focus the bulk of this review on recent progress in the study of pathogenic mechanisms of CAVD, and position integrative physiology as one means by which new technologies may deliver pharmacotherapeutic relief of this disease via identification of novel drug targets and treatments.⁴⁴

Assessing (epi-)genetic regulation and risk in aortic stenosis

After early landmark studies revealed LPA as the first genetic driver of AS risk,⁴⁵ follow-on work in larger cohorts continued to confirm this association and identified additional risk alleles in PALMD, TEX41, IL6, and ALPL.^46–48 More recently, an explosion of high-quality, large-scale GWAS (genome-wide association studies) has revolutionized our understanding of the depth and breadth of genetic risk in AS. These meta-analysis studies leverage large cohorts (nearly 15 000 cases and 1 000 000 controls) to both replicate prior findings and add tens of novel risk genes to the pool of AS contributors, including the replicated loci CEP85L, FTO, SLMAP, CELSR2-SORT1, MECOM, CDAN1, NLRP6, and SMC2.^49–51 Small et al.⁴⁹ performed a multi-ancestry GWAS analysis that identified ancestry-specific SNPs and identified shared and differential genomic drivers of AS and atherosclerosis, Larsson et al.⁵² found via Mendelian randomization (MR) that genetically predicted body mass index and adiposity were most strongly associated with AS vs. other cardiovascular diseases, Chen et al.⁵⁰ generated cross-ancestry analyses, genetic risk scores, and linked causal roles for dyslipidemia and adiposity to AS, while Thériault et al.⁵¹ performed RNA-sequencing in a cohort of 500 human aortic valves for confirmatory genome-wide eQTL (expression quantitative trait loci) analyses and TWAS (transcriptome-wide association studies). In all cases, leveraging of sophisticated phenome-wide association studies, MR, colocalization analyses with tissue expression data, TWAS, etc. have become powerful and standard tools for interpretation and translation in genomic studies of AS. One notable challenge resulting from this dramatic expansion of cohort size, study power, and genome-wide significant risk alleles is the need to accurately annotate/map tens or hundreds of significant SNPs to onto protein-coding genes; this is particularly non-trivial as the complexity of genomic regulation continues to be unravelled. As additional risk alleles are identified, polygenic risk scores (PRSs) in AS have now been generated that reveal a weak correlation between PRS of AS vs. coronary artery disease (CAD), consistent with reports of unique whole blood transcriptional signatures between these conditions.^53,54 While the AS PRS performs comparably to the vast majority of clinical risk factors, they do not yet exceed these traditional clinical approaches and further development is required.⁵³

Bicuspid aortic valve (BAV) is the most common congenital heart defect and drives the early and accelerated onset of AS through mechanisms which remain largely unelucidated.⁵⁵ One hypothetical aetiology arises from the colocalization of BAV calcification with well-described abnormal haemodynamic forces and altered intra-leaflet mechanical stress concentrations found in BAVs,^56,57 though a causal role has yet to be established and separation of mechanical and genetic drivers in vivo is challenging. Genotype–phenotype linkages in BAV are complicated, the heritability of BAV is highly polygenic, and BAVs are found in only ∼2% of humans.⁵⁸ This context complicates the identification of genetic causes of BAV and BAV-associated AS. Initial GWAS in 466 cases and 4660 controls found a non-coding loss-of-function variant of the cardiac-specific transcription factor GATA4,⁵⁹ while recent studies in larger cohorts detected BAV risk loci in MUC4 (deleterious), PALMD (loss-of-function), TEX41, and MIB1 (loss-of-function).^58,60 Confirmatory experiments revealed that disruption of Muc4, a gene involved in epithelial-to-mesenchymal transformation, impaired AV development in zebrafish and linked genomic risk drivers to aberrant endothelial transcriptomic profiles. This finding, along with evidence from CRISPR-iPSC lines showing GATA4 reductions inhibited EndMT, raise the importance of this pathway in AV development and disease.^58,61 In turn, the E3-ubiquitin ligase MIB1 was shown to be essential for the activation of NOTCH signalling during AV development and implicated in ∼2% of non-syndromic cases of BAV.⁶⁰ Other studies have leveraged targeted DNA sequencing of HOXA1 and ADAMTS family genes, phenotype screening in zebrafish, and RNA-sequencing in knockout mice to implicate loss-of-function of these well-described regulators of cardiovascular development in human BAV incidence.^62,63 Beyond early transcriptomics work,^64–66 there is a notable absence of multi-omics studies aimed at differentiating mechanisms of BAV-driven AS from those of prototypical tricuspid aortic valve (TAV)-associated AS. In practice, it is challenging to conduct GWAS studies of AS exclusively on individuals with TAV and most GWAS studies of AS make no distinction between valvular morphotypes.

Germline point mutations are not the only DNA-associated drivers of CAVD, as evidenced by prior seminal works in AS epigenetics that found (i) hypomethylation drove expression of the lncRNA H19, impaired NOTCH signalling, and VIC calcification⁶⁷ and (ii) methylation of the intronic enhancer of PLPP3 (phospholipid phosphatase-3) attenuated the degradation of lipoprotein(a) [Lp(a)] metabolites and drove pro-osteogenic VIC responses.⁶⁸ In a particularly elegant study, Han and colleagues⁶⁹ revealed a putative pathogenic mechanism for the GWAS-identified CAVD risk gene and m6A methyltransferase FTO.^49–51 By sequencing non-coding PIWI-interacting RNAs (piRNA) in calcified human AVs, they identified elevated hsa-piR-25624, or AVCAPIR (AV calcification-associated piRNA), which induced VIC calcification in vitro. Using mass spectrometric proteomics, they identified FTO as an AVCAPIR binding partner, then employed methylated RNA IP-seq (m6A-seq) and RNA-seq to show this binding blocks FTO-mediated demethylation of CD36 mRNA. Highly methylated CD36 mRNA is recognized by the m6a reader IGF2BP1, which enhances the stability and abundance of CD36 mRNA and, thus, protein. Elevated CD36 protein levels directly bind to, stabilize, and promote the abundance of pro-calcific PCSK9 protein, leading to progressive CAVD through pleiotropic (non-lipid) downstream mechanisms.⁶⁹ Clonal haematopoiesis of indeterminate potential (CHIP)-driving mutations have recently emerged as a key and novel cause of atherosclerosis⁷⁰ and targeted amplicon sequencing has revealed that approximately a third of patients undergoing AVR carry these somatic mutations and suffer from elevated mortality after TAVR.⁷¹ Mechanistic insight into this phenomenon came from bulk transcriptomics (RNA-seq) of human valve tissues from SAVR recipients who were CHIP- and non-CHIP-mutation carriers; these experiments revealed broad inflammatory cell infiltration and inflammatory signalling in CHIP-carrier AVs with a particular exacerbation of activated B cells.⁷² Somatic alterations may also contribute to well-described sex-based differences in CAVD pathogenesis, whereby in both TAVs and BAVs, female AS patients have lower calcification levels but elevated leaflet fibrosis while incurring similar haemodynamic severity of disease.^73–75 Mosaic loss of chromosome Y (LOY) in circulating blood cells is the most common acquired mutation and is strongly associated with major adverse cardiovascular events (MACE) and heart failure in men.^76,77 After identifying exacerbated LOY and correlations with increased mortality in male TAVR recipients, Mas-Peiro et al.⁷⁸ performed single-cell RNA-seq (scRNA-seq) on monocytes with LOY to identify a TGFβ-mediated pro-fibrotic transcriptional programme in these patients, though the relative contributions of LOY to cardiac vs. valvular fibrosis remains to be elucidated. On the other hand, genes that escape X-chromosome inactivation also appear to regulate myofibrogenic VIC differentiation and resultant CAVD: transcriptomics of male- and female-derived porcine VICs in hydrogel culture systems found that elevated Rho-associated protein kinase (ROCK) signalling enabled the genes BMX and STS to avoid inactivation and elevate female VIC myofibrogenesis.⁷⁹ As a whole, these findings are consistent with several prior transcriptomics studies which have demonstrated elevated pro-fibrotic pathways in female AS leaflet tissues.^80,81

Multi-omics and integrative approaches to pathobiological insight

Mechanisms of genetic risk

Beyond the genome, efforts to capture global, omics-based insights into valvular phenotypes have been successfully applied in a widespread variety of contexts, and fall under two primary avenues: first, the identification of new mechanisms of homeostasis or pathogenesis through non-biased screening. Second, to confirm clinical findings or validate targets generated by hypothesis-driven research. In some cases, omics approaches can assist in understanding AV-targeted drug side effects: transcriptome arrays have revealed that downregulation of discoidin domain receptor 2 (DDR2), the target of the chronic myeloid leukaemia treatment Nilotinib, is present in calcified human AVs. In vitro experiments found that VICs treated with Nilotinib underwent robust osteogenesis and mineralization, suggesting a potential contribution to CAVD-associated cardiotoxicity.⁸² In other ways, multi-omics enables the construction of mechanistic linkages between genotypic drivers of CAVD and disease phenotypes. The paralemmin protein Palmdelphin (PALMD) has also long been connected by genomic evidence to AS onset in humans, though only recently have the downstream mechanisms been clarified. By mining murine scRNA-seq datasets, Sáinz-Jaspeado et al.⁴¹ found Palmd was highly expressed in the endothelium and controlled endothelial actin cap formation and nuclear alignment in response to fluid flow, then went on to show how PALMD exerted RAN GTPase-dependent control of nucleocytoplasmic shuttling via XPO1. Human CAVD tissue transcriptomics found disease stage-specific reductions in PALMD and suggested that nuclear fragility caused by loss of PALMD-dependent nucleocytoplasmic shuttling underpins genetic risk associated with SNPs in this gene. Importantly, the association of PALMD with bicuspid AS^47,58 could be explained by this shear-sensitive mechanosensitive nuclear resilience mechanism, as well as potential control over EndMT.⁸³ Palmdelphin appears to exert pleiotropic effects on cellular actin stress fibre dynamics: Chignon et al.⁸⁴ employed 3D genetic mapping, CRISPR activation screens, and DNA-binding assays to demonstrate that an AS risk variant at 1p21.2 controlled PALMD expression, and that the risk variant altered DNA shape in such a way as to lower binding of the transcription factor NFATC2 and resultant PALMD expression. The authors went on to examine the PALMD protein–protein interactome by mass spectrometry and identified a role for PALMD in regulating actin polymerization and exerted control of VIC fibrogenesis via myocardin-related transcription factor. Contemporary AS GWAS studies have also identified the CELSR2–SORT1 risk locus.^49–51 The SORT1 protein Sortilin has been extensively studied in the context of EV-mediated vascular calcification,⁸⁵ though its role in CAVD was unclear until Iqbal and colleagues showed Sortilin-deficient mice with wire-injured AVs were protected against aortic valve fibrocalcification. Proteomics and scRNA-seq of cultured human VICs revealed that Sortilin loss-of-function prevented calcification, and that Sortilin signalled via p38 MAPK to drive expansion of a complex inflammatory-myofibroblastic-osteogenic VIC subtype in vitro.⁸⁶

Inflammatory drivers of calcific aortic valve disease

Some of the earliest indications of a potential role for inflammatory drivers in AS stem from those associated with infection. Indeed, the incidence of infective endocarditis in patients with AS is a function of AS severity,⁸⁷ multiple bacterial species are present in the majority of stenotic aortic valves that undergo replacement,⁸⁸ inoculation of the rabbit aortic valve with Corynebacterium matruchotti and/or Streptococcus sanguis II is sufficient to induce large macrocalcifications,⁸⁹ and proteomics has recently implicated autoimmunity-associated inflammatory responses in the progression of rheumatic aortic valve disease.⁹⁰ Among the genetically identified targets (Table 1), the pivotal incrimination of the IL6 gene (encoding the pro-inflammatory cytokine interleukin 6) as a driver of genetic AS risk⁴⁸ emphasized the role of inflammatory processes in CAVD. Toll-like receptors (TLRs), an evolutionarily conserved family of pattern recognition receptors, are typically activated by pathogen detection and endogenous alarm signals from injured cells. Activation of TLR2 and TLR4 on VICs is directly linked to procalcification pathways.⁹⁴ In addition, recent evidence identified TLR3 activation in aortic valves leading to osteogenic VIC differentiation through type 1 interferon (IFN) signalling.⁹⁵ Systematic screening identified the proteoglycan Biglycan (BGN) as a novel TLR3 ligand, requiring a xylosyltransferase-mediated maturation of BGN’s glycosaminoglycan side chains. The functional impact of TLR3 signalling in AS has received support from the protective effects against CAVD of in vivo models with a genetically targeted BGN–TLR3–IFN pathway.^95,96 In contrast to TLR2, 3, and 4, the TLR family member TLR7 co-localized with anti-inflammatory M2 macrophages in stenotic human valves, and TLR7 ligand administration drove the secretion of anti-inflammatory cytokines.⁹⁷ Taken together, aortic valve TLRs may represent a starting point of aortic valve inflammation triggered by cellular damage and ECM components, and are coupled to immunomodulatory signalling.

Another genetic indication for the importance of immune activation and suppression in CAVD is represented by SNPs in the fatty acid desaturase FADS1/2 loci which were associated with plasma phospholipid fatty acid levels and, specifically, AS incidence.⁹⁸ Indeed, transcriptomics and lipidomics of stenotic human valves revealed that the FADS genotype associates with fatty acid desaturase (FADS) mRNA expression, tissue fatty acid content/metabolism, and leaflet calcification.⁹² Other work has further reinforced genetic associations between AS, FADS1, and pro-inflammatory omega-6 polyunsaturated fatty acids (n-6 PUFAs).⁹¹ Lipid-driven regulation of valvular inflammation is also fundamentally important to CAVD initiation and progression. Classical pro-inflammatory mediators of inflammation from the 5-lipoxygenase and cyclooxygenase enzymatic pathways have been associated with AS severity.⁹⁹ Furthermore, there is lipidomics-based evidence that the inflammation-resolving, omega-3 polyunsaturated fatty acid (n-3 PUFA)-derived lipid mediator resolvin E1 is dysregulated in calcified AV areas.¹⁰⁰ Mice with endogenous synthesis of n-3 PUFA presented with reduced calcification, improved valvular haemodynamics, and elevated M2 macrophage burden in a manner dependent on the resolvin E1 receptor ChemR23. These findings suggest that as in the vasculature,¹⁰¹ valvular calcification may result from improper resolution of inflammatory processes.

Bioactive extracellular vesicle cargoes

EVs are nanometer-sized membrane-bound particles actively released by all eukaryotic cell types and loaded with bioactive cargoes including non-coding RNA, proteins, lipids, and metabolites.¹⁰² They are well known to play key roles in cell–cell communication,¹⁰³ and miRNA transcriptomics of circulating EV cargoes has identified that shuttling of the proapoptotic EV-incorporated miR-122-5p into cardiomyocytes is a driver of reduced expression of the antiapoptotic gene BCL2 and responsible for reduced cardiac function after TAVR.¹⁰⁴ In cardiovascular tissues, EVs act as the biophysical building blocks of microcalcification,¹⁰⁵ and osteogenic culture conditions induce alterations to EV proteomic cargoes, extracellular EV aggregation, and formation of microcalcifications in the valve and vasculature.^{29,85,106,107} Notably, proteomics of VIC and vascular smooth muscle cell (SMC)-derived EVs from calcifying culture conditions found elevated abundances of the tethering protein ANXA1 (annexin A1).²⁹ ANXA1 was localized to collagen–EV contacts in calcified human cardiovascular tissues, while loss-of-function experiments rescued VIC and SMC calcification in vitro, and abrogated EV-driven microcalcification in 3D collagen hydrogels. Trailblazing efforts to quantitatively and systematically examine the cargoes of EVs entrapped in calcifying vascular and valvular tissue were motivated by the first comparative global proteomic study of disease progression in human atherosclerotic plaques and calcified aortic valves,⁴⁰ which identified unique protein drivers of CAVD and atherosclerosis and implicated EVs in both conditions. Enrichment of EVs directly from fibrocalcific cardiovascular tissues enabled vesiculomics assessments of tissue-entrapped EV protein and miR cargoes, and unbiased network medicine approaches integrated and prioritized 6 vesicle-borne targets for in vitro validation: FGFR2, PPP2CA, ADAM17, APC, APP, and WNT5A. Strikingly, knockdown of all six targets significantly modulated VIC and SMC calcification in vitro, including cell type-specific effects of FGFR2 (a pro-osteogenic fibroblast growth factor receptor) and WNT5A (control of osteogenesis via noncanonical PKC/JNK signalling and β-catenin antagonization), which mark them as high-priority candidates for future translational efforts aimed at targeted and specific treatment of vascular vs. valvular calcification.⁴⁰ Vesiculomics techniques have also been applied to tissue-engineered, biomechanically relevant, multi-layered, 3D-bioprinted, VIC-laden hydrogel valve organoids as a means of benchmarking model performance against native human tissues. Notably, network analyses of organoid-derived VIC and secreted EV proteomics determined that organoid culture mimicked 94% of the valve tissue proteome, while 2D monolayer culture only mirrored 70%.^108,109

Target discovery at the single-cell level

Early single-cell omics studies of CAVD were focused on transcriptomics atlas development in mouse and human tissues, and identified the presence of substantial cellular heterogeneity in both VIC/VEC and immune/inflammatory cell types.^110,111 Further studies have dissected and compared basal differences in cellular landscapes across all four cardiac valves¹¹² When applied to hyperlipidemic mouse models of CAVD, scRNA-seq was suggestive of lipid-driven recruitment of pro-inflammatory macrophages to the valve and protective activation of peroxisome proliferator-activated receptor-gamma (PPARγ) signalling in VECs, and further reinforced the notion that (non-)resident immune cells contribute to CAVD progression.¹¹³ Beyond atlas development, scRNA-seq has begun to support CAVD drug target identification: study of cells enzymatically isolated from young non-diseased, aged non-diseased, and CAVD human valves found the proteoglycan Lumican (LUM) marked a pathological pro-osteogenic VIC subpopulation. Multi-omics assessments revealed that LUM drove lactylation of histone H3 and activation of prototypical osteogenic transcriptional programmes.¹¹⁴ Studies of cultured VICs or VECs have been utilized to successfully identify key pathogenic subpopulations: scRNA-seq has identified sortilin-driven differentiation of an inflammatory-myofibroblastic-osteogenic VIC subtype in response to calcifying media conditions,⁸⁶ and single-cell multi-omics (scRNA-seq and single-cell mass cytometry) of VIC cultures isolated a disease-driving CD44^highCD29⁺CD59⁺CD73⁺CD45^low population with exacerbated osteogenic potential. Temporal bulk proteomics of this disease-driving VIC subpopulation and loss-of-function experiments identified monoamine oxidase (MAOA) as a potential therapeutic target.²⁴ As we noted above, the AV has a particularly elaborate anatomy, and spatial segregation of disease may hold the key to effective therapeutics. Early efforts at spatial omics in CAVD employed laser-capture microdissection of tissue sections and prototypical downstream bulk transcriptomics to examine layer-specific gene expression in diseased human valves.³⁹ Subsequent studies have leveraged 10× Genomics slide-based molecular profiling techniques to examine embryonic development of the murine AV and associate defective retinol-binding protein 1 (RBP1) signalling with BAV formation, though the application of high-resolution spatial omics approaches to CAVD remain to be performed.¹¹⁵

Clinical translational towards pharmacotherapy for aortic stenosis

Lipoprotein-lowering strategies

AS pharmacotherapeutics represent a global and rapidly growing potential market with the opportunity to cover the gap in care caused by inequalities in access to TAVR/SAVR throughout much of the world. Unfortunately, the cost and challenging nature of cardiovascular outcomes trials in general, and for AS in particular, has limited industry interest in this area to the point where any successful therapy would enjoy a first-to-market advantage. Despite a total lack of success to date (reviewed in Kraler et al.²¹ and Lindman et al.¹¹⁶), there are a number of ongoing clinical trials aimed at testing potential medical therapies for AS (Table 2). Thus far, these programmes have focused on attempts to repurpose a wide variety of drugs initially approved or developed for other indications; few if any first-in-class drugs have been developed against AS (Figure 3). Despite many shared risk factors and certain gross histological similarities, anti-atherosclerosis drugs (e.g. statins) which successfully reduce pro-inflammatory lipid accumulation have thus far failed to provide any protective effect against CAVD progression. Though low-density lipoprotein-cholesterol (LDL-C) is a strong and well-described clinical risk factor of AS,⁹ statin-mediated LDL-C lowering has continually failed in a number of prospective trials (e.g. SEAS,¹¹⁷ ASTRONOMER,¹¹⁸ SALTIRE¹¹⁹). While post-hoc analyses have evaluated the influence of baseline AS disease severity and LDL-C levels,^120,121 the window in which LDL-C lowering by statins may have therapeutic benefits remains to be established. Another contributor to such therapeutic benefits against AS may related to the degree of LDL-C lowering, which can be obtained to a greater extent by inhibitors of the LDL receptor (LDL-R)-degrading proprotein convertase subtilisin/kexin type 9 (PCSK9). In the FOURIER trial, which was one of the studies showing additional beneficial effects of PCSK9 inhibition in addition to statin therapy in atherosclerotic cardiovascular disease prevention, an exploratory analysis pointed to potential protective effects of the PCSK9 inhibitor Evolocumab on AS measures when compared with placebo beyond 1 year of treatment.¹²² The EPISODE trial is evaluating three such PCSK9 inhibitors (Evolocumab, Alirocumab, and Tafolecimab) in patients with mild, moderate, or asymptomatic severe AS.^123,124 In addition to systemic LDL-C lowering, PCSK9 inhibitors may directly affect local valve inflammation and calcification processes, as demonstrated by the presence of PCSK9 secretion by VICs and in vitro reductions in VIC calcification upon PCSK9 inhibition (Figure 3).¹²⁵

Other lipoproteins, particularly Lp(a), confer significant and substantial AS risk.⁹ Over a decade ago, a pioneering GWAS and MR study implicated the first genetic drivers of AS and demonstrated that a SNP within the LPA locus drove genetically determined alterations in circulating Lp(a) levels that were significantly and causally associated with AS (Table 1).⁴⁵ In the ∼7% of AS cases accounted for by Lp(a), these effects appear to be mediated by the oxidized phospholipids and the ectonucleotide pyrophosphatase/phosphodiesterase autotaxin carried by Lp(a), which directly drive VIC osteogenesis.^32,33 Until recently, Lp(a) was challenging to target therapeutically—statins do not lower Lp(a) levels, and niacin lowers circulating concentrations by ∼20%–30% alongside an increased incidence of adverse events. As a result, recruitment to the EAVaLL trial was slow and eventually withdrawn.¹²⁶ Because LDL-R internalizes both LDL and Lp(a), PCSK9 inhibitors reduce circulating levels of both lipoproteins—though MR studies suggest the expected Lp(a)-lowering capability of existing PCSK9 inhibitors (similar to that of niacin) may be insufficient for hope of a therapeutic benefit.¹²⁷ In a more specific approach, the Lp(a)FRONTIERS CAVS trial is currently recruiting individuals with mild or moderate AS to test the efficacy of Lp(a) lowering using the anti-LPA antisense oligonucleotide Pelacarsen.¹²⁸ Outside of LDL-R, a specific valvular receptor responsible for internalization of Lp(a) remained undiscovered until Rogers and colleagues employed unbiased ligand–receptor capture mass spectrometry to identify major facilitator superfamily domain containing 5 (MFSD5) as a putative Lp(a) receptor/cofactor.¹²⁹In vitro, chemical or genetic reduction of MFSD5 expression impaired Lp(a) uptake and suppressed calcification in both VICs and VECs, and targeted assessment of genetic MFSD5 variants was significantly associated with AS. Together, these data suggest MFSD5 as a high-priority drug target for inhibition of Lp(a)-specific internalization. Beyond Lp(a), targeted proteomics of apolipoproteins has unearthed enrichment of Apolipoprotein C-III (apo-CIII) in calcified valve regions, and shown that apo-CIII induced calcification in primary human VIC cultures via a mitochondrial dysfunction/inflammation-mediated pathway.³⁰ These findings raise the intriguing notion of a potentially novel class of repurposed AS treatments through direct therapeutic reduction of apo-CIII (e.g. via antisense approaches such as Olezarsen¹³⁰), or by targeting apo-CIII-modulated hypertriglyceridemia through fibrates or icosapent ethyl.^131,132

Therapeutic disruption of pathogenic valvular cell mechanisms

Indeed, due to their active roles in driving disease most remaining drug development endeavours have focused on directly modulating pathogenic disease drivers in VICs, VECs, macrophages, and other valve-resident cell types. After notable reductions of cardiovascular events in patients with chronic CAD,¹³³ interest grew in repurposing the anti-inflammatory alkaloid colchicine towards other cardiovascular indications. The CHIANTI and COPAS-Pilot trials are testing the impact of low-dose colchicine therapy on mild to moderate AS progression.^134,135 In pre-clinical studies, scRNA-seq studies found pro-inflammatory valvular macrophages were increased in hyperlipidemic murine models of AS along with activation of PPARγ signalling in VECs.¹¹³ Notably, PPARγ agonists reduced valvular inflammation, monocyte-derived macrophage content, and calcification in mice and prospective trials of pioglitazone in mild-to-moderate AS are underway.^136,137 Transcriptomics has also revealed that oxidative stress and resultant endothelial dysfunction appear to drive elevated VEC expression of the exopeptidase dipeptidyl peptidase-4 (DPP4), which contributes to recruitment of inflammatory macrophages and inhibits homeostatic insulin-like growth factor 1 (IGF-1) signalling in VICs, driving osteogenesis and calcification.¹³⁸ Inhibition of DPP4 by Sitagliptin in a rabbit AS model drove significant valvular calcium and haemodynamic improvements; the EVOID-AS and DIP-CAVD trials are testing the DPP4 inhibitor Evogliptin in AS outcomes.^139,140 Though DIP-CAVD recently reported negative outcomes on aortic valve calcification burden, Evogliptin drove significant improvement in some secondary endpoints (early reductions in calcification as measured by more sensitive radiotracer-based imaging and suppression of increases in calcification burden at later timepoints).¹⁴¹ Many DPP4 inhibitors also prevent catabolism of incretins [e.g. glucagon-like peptide 1 (GLP1)]. In obese individuals without diabetes, the incretin-mimic semaglutide improved cardiovascular outcomes in a manner that may be at least partially independent of body weight reductions and suggested direct physiological effects.¹⁴² GLP1 levels are reduced in human AS blood and valvular tissue, and antagonized pro-osteogenic signalling pathways and VIC calcification in vitro.¹⁴³ In hyperlipidemic mice, liraglutide administration attenuated AV calcification, and leaflet transcriptomics implicated reduced VIC myofibro/osteogenic differentiation and lower inflammatory macrophage accumulation in incretin mimetic-treated mice, suggestive of potential therapeutic benefits in human AS.¹⁴⁴ Downstream of endothelial dysfunction and nitric oxide¹⁴⁵ or natriuretic peptide signalling, valvular cyclic guanosine monophosphate levels contribute to embryonic leaflet development along with adult VIC/VEC mechanosensation and homeostasis.^55,146 The AS trial is current testing the nitric oxide-independent soluble guanylate cyclase (sGC) activator Ataciguat in patients with moderate AS; early readouts in mice and humans were suggestive of positive effects on VIC osteogenesis and progression of calcification.^147,148

Extracellular targets

Other efforts have aimed at directly targeting extracellular or ECM-associated targets, with a particular focus on mineral metabolism. After some retrospective evidence that extra-skeletal effects of anti-osteoporosis drugs might slow AV calcification progression, the SALTIRE II trial randomized patients with mild or moderate AS to the RANKL inhibitor denosumab or the bisphosphonate alendronate but failed to alter AV calcium burden nor haemodynamic measures.^149,150 Other modes appear more promising: SNF-472 is the hexasodium salt of myo-inositol hexaphosphate (phytate) that inhibits ectopic calcification by selectively binding to the surface of hydroxyapatite crystals, thereby blocking ion binding and further crystal growth. Pre-clinical studies found that SNF-472 inhibited VIC calcification in vitro, abrogated calcium accumulation in an ex vivo porcine AV leaflet culture model, and reduced cardiovascular calcification levels in uraemic rodents.^151–153 In patients with end-stage renal disease and on haemodialysis, intravenous delivery of SNF-472 in the CaLIPSO trial significantly slowed imaging-measured progression of both coronary artery and AV calcification.^154,155 The closely related compound INS-3001 is an inositol phosphate derivatized with ethylene glycol oligomers, which stabilizes calciprotein particle growth and offers improved pharmacokinetic profiles. INS-3001 inhibited calcification of cultured SMCs, explanted human femoral arteries, and AVs, and reduced vascular calcification in vitamin D-induced murine models without changes in trabecular bone area.¹⁵⁶ First-in-human studies of INS-3001 in patients with moderate AS are underway.¹⁵⁷ Other direct protein modulators of mineralization include the master calcification inhibitor matrix Gla protein (MGP), a vitamin K-dependent calcium-binding protein that negatively regulates cultured VIC calcification,¹⁵⁸ while circulating levels are reduced in AS patients.¹⁵⁹ In a trio of larger ongoing trials (BASIK2, AVADEC, DECAV-K2), vitamin K2 supplementation aimed at driving carboxylation and activation of MGP is actively being assessed in patients with both prototypical tricuspid and congenital bicuspid AS.^160–162 ECM components have also been directly targeted: covalent binding of the chelator diethylenetriaminepentaacetic acid onto albumin-based nanoparticles conjugated to anti-elastin targeting antibodies arrested and reversed ex vivo mineralization of intact porcine AV leaflets.¹⁶³ In the valve, misfolded extracellular protein aggregates (amyloids) may act as a strong pro-calcifying substrate. Cardiomyopathic transthyretin amyloidosis (ATTR-CM) is an under-recognized form of heart failure that is often comorbid with AS. Multi-omics has shown amyloid proteins are highly and specifically elevated in human CAVD tissues, carried by valvular tissue-entrapped EVs,^40,164 and amyloid plaque co-localizes with valvular nodule formation.¹⁶⁵ Recent successes in small molecule (e.g. Tafamidis), RNAi, CRISPR–Cas9, and antibody-based therapies for ATTR-CM are promising future areas for therapeutic discovery and drug repurposing.¹⁶⁶

Myocardial mediators of heart failure

In recent years, therapeutic inhibition of sodium–glucose cotransporter 2 (SGLT2) has expanded beyond the treatment of Type 2 diabetes after unexpected reductions in heart failure hospitalization rates. Follow-up studies found broad effects on cellular metabolic processes, inflammation, and fibrosis, though the exact mechanisms of these beneficial effects remain unclear. Regardless, cardiac SGLT2 abundances are elevated in low-flow, low-gradient AS, and in preliminary studies, the SGLT2 inhibitor Empagliflozin improved heart failure phenotypes in supra-aortic-banded rats.¹⁶⁷ While inhibition of angiotensin-converting enzyme (ACE) has shown variable and conflicting effects on AS outcomes in a number of trials, retrospective assessments of anti-hypertensive blockade of the renin–angiotensin system in AS via angiotensin II receptor blockers (ARBs) suggests this drug class may leverage its cardioprotective benefits against AS class with a modicum of success.¹⁶⁸ The ongoing ARBAS and ALFA trials will assess the efficacy of angiotensin II receptor type 1 (AT1R) inhibition on AS progression, with a focus on valvular haemodynamic parameters and cardiac function.^169,170 Notably, outside of impacts on the heart, AT1R is expressed in VICs^21,171 with elevated levels found in stenotic AVs¹⁷² and may also directly regulate pathological valvular macrophage and VIC phenotypes (Figure 3).^173,174 Interestingly, such an effect may be sex-specific and in alignment with the female fibrotic AS phenotype:⁷³ a recent retrospective assessment found that ARB intake was only associated with reduced valvular leaflet fibrosis in female patients, not males.¹⁷⁵

Conclusions

As we enter into the age of testing the first pharmaceutical products designed against early genomics-derived anti-AS drug targets [e.g. Lp(a)⁴⁵], so too is the field now positioned to expand this set of risk genes with large-cohort genomics meta-analyses and MR approaches.^49–51 The fusion of large-cohort databases with robust clinical and molecular datasets, or clinomics, promises to improve our understanding and treatment of AS. By integrating deep clinical phenotyping (clinical factors, imaging, events, outcomes, etc.) of large longitudinal cohorts with genomics and unbiased profiling of circulating or tissue molecular profiles (often via affinity- or aptamer-based proteomics methods^176,177), MR-based inference may enable precise identification of causal targets, subpopulations with greater potential therapeutic efficacy, anticipation of safety signals, and personalized prediction of AS progression as has been done for CAD.^52,178,179 Rapid maturation of both the technical capabilities of multi-omics methods and downstream systems analyses has also begun to spur pre-clinical translation of the first non-genomic targets towards drug development.^24,69,86,129 Single-cell technologies have recently expanded beyond the transcriptome and into epigenomics, immune repertoire profiling, and cell surface protein quantitation; these modalities promise to facilitate the specific targeting of crucial cellular subpopulations that contribute to CAVD pathogenesis.^{110,112–114} Single-cell techniques may also enable differentiation of BAV vs. TAV-associated disease-driving mechanisms, while cusp-specific studies could shed light on how shear-stress or flow-related modulate cellular subpopulations in the valve. More broadly, single-cell studies across various stages of disease (e.g. early initiation, mid-stage fibrosis, end-stage calcification) will help to better understand the complex continuum of pathobiology present in CAVD. In turn, spatial transcriptomics may unravel the complex anatomical and layer-specific cellular regulatory programmes that underpin the disease-prone nature of the fibrosa. Lastly, proteomics techniques have now advanced to the point where circulating biomarkers of AS, AS risk, and AS progression can now be reliably and economically assayed in large cohorts.^176,177

These important advances are also tinged by a growing clarity and comprehension of the pitfalls that will be faced along the path towards efficacious AS pharmacotherapies. As easily detected targets with the largest effect sizes are discovered, larger cohorts and costlier/deeper sequencing and phenotyping will be needed to identify significant hits and threaten to provide ever-diminishing returns. Multi-omics promises to ease these concerns by leveraging cross-modality interactions to identify more subtle hits in existing sample sets. The cost, complexity, and length of rigorous clinical trials in AS will also remain a major impediment to translation. Cost-efficient trials will require advances in early detection and diagnosis, improved prediction models of sclerosis-to-stenosis progression, careful selection of trial enrolment populations, and non-traditional clinical endpoints (reviewed elsewhere^21,116). Development of targeted therapeutic strategies may be assisted by trial stratification and potentiated via a clearer differentiation of sex-based differences in AS pathogenesis,^{73–75,79–81,180} of disease-initiating and disease-driving factors in CKD- and non-CKD-associated CAVD,¹⁸¹ between pathogenic mechanisms and markers of CAVD vs. CAD,^40,53,182 or via insights into disease stages and vulnerable phases of CAVD development potentially associated with intra-leaflet haemorrhage.^183–185 Systematic multi-omics assessments of concordant and paradoxical mechanisms between valvular calcification and skeletal bone formation remain unstudied, and could shed new light on AS pathogenesis and aid in drug repurposing strategies.^186,187

In one under-appreciated way, the field will soon be a victim of its own success: in the TAVR era, SAVR volumes have dropped precipitously—some estimates place this reduction at almost 45% over the past decade.¹⁸⁸ The resultant scarcity of SAVR-derived diseased human tissue samples is becoming an impediment to basic science efforts and has prompted renewed attempts to develop new animal models (reviewed in Sider et al.³⁷ and Zhu et al.¹⁸⁹) or engineer organoids and other suitable 3D models of CAVD.^{108,109,190–192} Appropriate model selection, where mode(s) of model AS induction match putative target-driven disease mechanisms, is a critical aspect of both the accurate study of pathogenic mechanisms and rigorous pre-clinical target validation.^38,193 Robust and multi-factorial benchmarking of model recapitulation of native human CAVD pathobiology via multi-omics will be of fundamental importance to these initiatives.¹⁰⁹ Should these surgical trends continue, there may be a narrow window of time within which the field must work to mitigate the spectre of bioprosthetic valve degeneration and gather the resources needed to support successful pharmaceutical translation in the future.

Supplementary data

Supplementary data are not available at European Heart Journal online.

Declarations

Disclosure of Interest

M.C.B. is a consultant for BioMarin Pharmaceuticals, Inc. M.B.’s institution has received speaker and consultant fees from Amarin, Amgen, Heel, Novartis, and Fresenius Kabi. T.F.L. has received educational and/or research grants from Abbot, Amgen, AstraZeneca, BAYER Health Care, Boehringer Ingelheim, Daichi-Sankyo, Eli Lilly, Novartis, Novo Nordisk, Sanofi, and Vifor, and honoraria or consulting fees from Milestone Pharmaceuticals and Novo Nordisk. E.A. is a member of the scientific board of Elastrin Therapeutics, Inc.

Data Availability

No data were generated or analysed for this manuscript.

Funding

Swedish Research Council (grant number 2023-02652) to M.B., Swedish Heart and Lung Foundation (grant number 20210560) to M.B.; and National Institutes of Health grant number R01HL141917 to E.A.

References