Stroke Genetics, Genomics, and Precision Medicine

Increasingly large cross-ancestry GWAS meta-analyses have discovered over 100 independent genetic risk loci harboring common single nucleotide variants associated with stroke risk. The largest stroke GWAS to date comprised 110 000 patients with stroke, with additional 90 000 patients with stroke to follow-up significant findings. Most associations were observed for any stroke and ischemic stroke,^3,4 with so far few genome-wide significant associations for intracerebral hemorrhage, due to a smaller sample size.^6,7 Extensive efforts have been made to identify genetic associations for individual ischemic stroke subtypes,^4,8 but numbers remain considerably smaller (<15 000 for small vessel, large artery, or cardioembolic stroke). Moreover, thanks to worldwide international collaborations, the contribution of non-European data sets to stroke GWAS has considerably increased in recent years, representing about a third of patients (most of whom are of East-Asian ancestry). This has enabled a notable gain in power and made it possible to compare risk variants and effect sizes of genetic associations across ancestries. Seventeen genetic risk loci have been identified for intracranial aneurysms, the leading cause of subarachnoid hemorrhage.⁹ In parallel, over 60 loci were found to be associated with magnetic resonance imaging-based endophenotypes of stroke, especially magnetic resonance imaging markers of cerebral small vessel disease.^10,11 Recent efforts also revealed genetic variants modulating acute¹² or long-term outcomes after stroke.¹³ Although substantial shared genetic variation with vascular risk factors (particularly blood pressure) was observed, many stroke risk loci point to novel pathways.⁴

Whole exome sequencing¹⁴ and whole genome sequencing^15,16 studies of stroke, enabling the detection of rare stroke risk variants, respectively, within genes and at the much broader genome-wide level (enabling to capture variants that regulate gene expression), are still scarce and underpowered. These will likely provide important novel insight once larger samples become available, especially if rare deleterious or protective variants with large effect sizes can be identified. A few studies also explored other types of genetic variants, such as structural variants (copy number variants, mitochondrial DNA copy number, etc)^17,18 in relation with stroke, but these remain largely unexplored. Increasingly scalable long-read sequencing technologies will open new perspectives for deciphering the impact of such variants with greater accuracy.¹⁹

Developments in next-generation sequencing technologies have also enabled the study of rare genetic variants underpinning Mendelian (monogenic) strokes on a scale never possible before, yielding important insights. First, rare genetic variants hitherto believed to cause Mendelian strokes in a quasi-deterministic fashion are in fact much more common than anticipated on a population scale.²³ A study of 5 monogenic cerebral small vessel disease genes in 200 000 persons reported a prevalence of putative pathogenic variants of 1:200.²⁴ Across 18 genes causing Mendelian stroke the aggregate prevalence of pathogenic variants varied between 0.7% and 2.8% depending on genetic ancestry,²⁵ suggesting variable penetrance and the need for additional factors (including common genetic variants) for strokes to manifest clinically. The identification of genes (eg, COL4A1/2, HTRA1) harboring both common variants associated with complex stroke and rare mutations causing monogenic stroke further adds to the blurring distinction between monogenic and multifactorial stroke.^3,11 Another major discovery is the recognition that somatic mutations can lead to vascular malformations causing stroke, such as arterio-venous malformations,²⁶ paving the way for novel therapeutics.²⁷

Translating genetic loci to therapies requires precise knowledge of both causal genes and cell-specific directionality of expression effects. Combining genomics with other omics technologies (epigenomics, transcriptomics, proteomics, etc) that are increasingly affordable creates an opportunity to make rapid progress toward defining molecular mechanisms and causal pathways underlying genetic associations.²⁹ Single-cell transcriptomics and its spatial profiling can enhance this potential by enabling mechanisms to be mapped to specific cell (sub)types, and have for instance shown enrichment of genetic risk loci for perivascular space burden, an imaging marker of cerebral small vessel disease, in brain vascular endothelial cells based on a human single-cell atlas of fetal gene expression.¹⁰ Recent work also revealed that many genes differentially expressed in vascular brain cells of Alzheimer disease patients are directly linked to Alzheimer disease-associated GWAS loci, highlighting the strong interconnection of vascular and neurodegenerative processes.³⁰ Various bioinformatics approaches facilitate the integration of genomic with other omic data (transcriptome- and proteome-wide association studies, colocalization analyses, Mendelian randomization, etc), leveraging increasingly rich resources of tissue- and cell-specific multiomics.^4,29 These are beginning to yield major results for a variety of complex diseases, such as a cerebrospinal fluid-based proteogenomic study that identified 42 putative causal proteins for Alzheimer disease, a third of which point to drug repositioning opportunities, also shedding new light on how independent Alzheimer disease GWAS loci are connected to each other.³¹ Artificial intelligence-based approaches are also growingly proposed for multidimensional integration of omics,³² although this field is still at a preliminary stage for brain diseases, with a growing interest in expanding that integration further to the combination of omics and imaging dimensions, with even greater complexity, as outlined for the Alzheimer precision medicine initiative.³³

As sequencing becomes more affordable and rare mutations causing Mendelian forms of stroke are more common than previously suspected, genetic testing may eventually be more frequently included in the clinical work-up of patients with stroke. However, the added value of systematic mutation screening in sporadic patients with stroke is still unknown and requires additional research. PGS also have the potential to improve disease prevention,³⁴ and to enable predictive and prognostic enrichment of clinical trials, by selecting individuals who are most likely to develop the disease and respond to treatment.³⁵ The clinical utility of PGS in the specific setting of stroke prevention remains to be determined, for example, to discriminate individuals at high risk of ischemic versus hemorrhagic stroke to guide antithrombotic drug prescription in situations of clinical equipoise. Implementing PGS will require addressing ethical and societal challenges and gaining acceptance by physicians.³⁶ Whether PGS will ultimately have broad applications in the clinic,³⁷ or whether their use will mostly remain restricted to a research setting is still unclear. Recent pharmacogenomic studies also show that clinically relevant adverse drug reactions can be significantly reduced by testing for genetic variants modifying drug metabolism, demonstrating feasibility across diverse health care systems.³⁸ However, the implementation of pharmacogenetic testing in clinical practice is slow, due to some conflicting findings, a lack of stroke-specific comparative effectiveness trials, and practical aspects including cost-effectiveness. As an example, despite encouraging results supporting the use of pharmacogenetics to guide antiplatelet (clopidogrel) and anticoagulant (warfarin) treatment, clinical implementation remains sporadic.³⁹ Overall, such developments will require genomics savvy health care workforce and public health specialists,⁴⁰ with an urgent need for curricula in precision medicine and public health.

Providing genetic evidence for drug effects was retrospectively estimated to increase the success rate of clinical trials by >2-fold.⁴¹ Currently available genetic evidence supported 63% of drugs approved by the Food and Drug Administration in the past decade.⁴² However, genetics drove drug discovery only in a small fraction of these cases.⁴³ Several lipid-lowering therapies derived from genetic findings, such as PCSK9 (proprotein convertase subtilisin/kexin type 9) and ANGPTL3 (angiopoietin like 3) inhibitors and gene-editing/-silencing therapies in advanced development are of direct relevance for stroke prevention.⁴³ The trajectory from discovery of genetic associations to drug development is long (25 years on average) and complex. Robust evidence for causal genes and downstream effectors is required. Gene-editing/-silencing experiments in cellular and animal models are typically used, but remain challenging.⁴⁴ Resource-intensive complementary approaches such as massively parallel reporter assays or Clustered Regularly Interspaced Short Palindromic Repeats-mediated gene-editing of induced pluripotent stem cell lines (including in 3-dimensional models such as organoids), may eventually play a crucial role in larger scale screening of putative causal variants and genes at genetic risk loci for stroke and related disorders, as was recently successfully implemented for autism spectrum disorder.⁴⁵ In combination with drug databases, GWAS can also provide guidance for repositioning of drugs for stroke that are already commercialized or tested in clinical trials for other indications. Such evidence was recently shown for F11 inhibitors and PROC (protein C) agonists, currently in phase II/III trials for stroke, and drugs targeting LAMC1 (laminin subunit gamma 1) and GP1BA (glycoprotein Ib platelet subunit alpha).⁴ Finally, the convergence of genomic breakthroughs with the advent of next-generation programable therapies (RNA based and other) will likely accelerate genomics-driven drug discovery.

The past 10 years have been transformative to the field of stroke genetics, yet the full promise of precision medicine has not reached clinical practices. Translation of genetic findings into improved care is arguably the most important challenge for the next 10 years. Many of the genetic associations already identified likely hold the key to new avenues for prevention and treatment of stroke, yet the investment necessary to translate these findings is considerable. As rapid progress continues to be made in stroke genetics, the urgency of translational opportunities only becomes more pressing.