Epidemiology and Genetics of Venous Thromboembolism and Chronic Venous Disease

Abstract

Venous Disease is a term that broadly covers both venous thromboembolic (VTE) disease and chronic venous disease (CVD). The basic pathophysiology of VTE and CVD differ as VTE results from an imbalance of hemostasis and thrombosis while CVD occurs in the setting of tissue damage due to prolonged venous hypertension. Both diseases are common and account for significant mortality and morbidity, respectively, and collectively make up a large health care burden. Despite both diseases having well characterized environmental components, it has been known for decades that family history is an important risk factor, implicating a genetic element to a patient’s risk. Our understanding of the pathogenesis of these diseases has greatly benefited from an expansion of population genetic studies from pioneering familial studies to large genome-wide association studies; we now have multiple risk loci for each venous disease. In this review, we will highlight the current state of knowledge on the epidemiology and genetics of VTE and CVD and directions for future research.

Introduction to Venous Thromboembolism and Chronic Venous Disease

Venous disease is a general term that applies broadly to venous pathology of both venous thromboembolism (VTE) and chronic venous disease (CVD). Although there is some overlap, the pathogenesis of these diseases is quite different. VTE primarily manifests as either deep vein thrombosis (DVT) and/or pulmonary embolism (PE) driven by dysregulated coagulation and the diagnosis is usually event driven, however the burden of subclinical VTE is likely substantial. On the other hand, CVD is thought to result from tissue damage due to prolonged venous hypertension, which results from chronic venous insufficiency (CVI) due to primary venous reflux or other causes of increased central venous pressures including obesity, prolonged standing, pregnancy and other situations causing vascular compression. CVD develops over a longer period of time (months to decades) and has a range of clinical manifestations including varicose veins, swelling, skin changes, and venous leg ulcers, which are often associated with significant discomfort¹. The interface between VTE and CVD can present as post-thrombotic syndrome (PTS), which can occur when the residual clot burden of a DVT leads to venous stenosis, obstruction, or valvular damage resulting in prolonged increases in venous pressures and eventually CVI². PTS is estimated to affect 23–60% of individuals with DVT^3–5. Additionally, an epidemiologic study using the UK Biobank found that a history of DVT increased the incidence of developing varicose vein with a hazard ratio of 2.6⁶. The inverse can also occur when primary CVI, which results in venous stasis or turbulent flow, predisposes to thrombotic events including superficial venous thrombosis (SVT) or DVT in the calf veins⁷. Indeed, a retrospective cohort study investigating more than 400,000 patients in Taiwan’s Health Insurance program found that varicose veins dramatically increased the incidence of DVT with a hazard ratio of 5.3⁸, a finding that has been supported by multiple other epidemiologic studies^9–14.

Venous Disease Epidemiology

VTE Demographics:

VTE was reported in over 1.2 million people in the United States in 2016 of which 60% were diagnosed as DVT alone while 40% presented as PE with or without DVT¹⁵. Interestingly, there has been a steady increase in VTE events over the last two decades, however it is unclear if this is due to increased disease prevalence or increased detection through improved diagnostics. Mortality from PE at one year is 19.6%, which unfortunately has been unchanged since 1999, and there is a high rate of VTE recurrence with 30% of patients having a recurrence in 10 years¹⁶. Further, 30% of DVT are complicated by PTS, which often results in chronic venous dysfunction. Indeed, severe PTS can cause recurrent venous leg ulcers (VLU) and debilitating leg symptoms including chronic pain and severe swelling, which often leads to impaired quality of life and is poorly managed with existing treatment options. It is estimated that VTE and its complications account for 7–10 billion US dollars annually¹⁷.

VTE Risk Factors:

VTE is often multifactorial and therefore, when approaching a patient with VTE, it is critical to inventory both transient and persistent risk factors for each patient, which is essential to accurately prognosticate the risk of recurrence and to inform management. Traditionally, VTE was dichotomized as provoked or unprovoked but it is now appreciated that this dichotomy can be misleading and rather VTE should be thought of as a continuum of disease¹⁸. Recurrence rates for VTE are as high as 30–50% which depends largely on the existing risk factors. For example, an underlying cancer diagnosis can increase risk by as much as 20–30%^{19, 20}. Factors associated with thromboembolic risk are classically described using Virchow’s triad (endothelial injury, hypercoagulability, and stasis), which continues to be a useful pneumonic for the pathophysiology of the disease. However, as described in the genetics section several of the novel risk loci suggests mechanisms outside of the classic triad.

The traditional risk factors for VTE are advanced age, obesity, family history, previous DVT, recent surgery/trauma/fracture, hospitalizations, prolonged immobility^{21, 22}, disorders that induce a hypercoagulable state like cancer^{19, 20}, thrombophilic disorders, chronic kidney disease, autoimmune disorders^{23, 24}, sickle cell anemia²⁵, and infection including with viruses like SARS-CoV-2²⁶. For this later risk factor, a meta-analysis of 42 studies that included 8,271 patients with Covid-19 found the rate of VTE to be 5% in non-ICU patients and 31% for patients treated in the ICU²⁷. The increased risk appears to be multifactorial including immobility due to critical illness, increased circulating prothrombotic factors²⁸, and endothelial cell injury²⁹. Race also appears to impact risk with the highest rates of VTE in the Black population followed by the White, Hispanic, and Asian populations, in that order^30–33. This is certainly multifactorial but may have a genetic component, however further studies with greater diversity will be cirtical to understand these differences. There are also several female-specific risk factors including the use of estrogen-based oral contraceptive therapy, hormonal therapy, in vitro fertilization treatments, pregnancy³⁴, and the peripartum period^{35, 36}. Given hormonal therapy may increase VTE risk more than 2–6-fold³⁷ and is a modifiable risk factor, several studies have attempted to identify which therapies result in the greatest risk. Several comprehensive reviews have concluded that all combined oral contraceptives are associated with increased risk of venous thrombosis³⁸, but the effect size depends on which progestogen is used and the dose of ethinylestradiol³⁹. Progestogens at higher doses were associated with an increased risk of venous thrombosis but progestin (a synthetic form of progestogen) in the doses used for progestin-only contraceptives or emergency contraception has not been shown to increase the risk of DVT or PE³⁸.

CVD Demographics:

CVD is thought to affect more than 25 million adults in the United States with 6–7 million having advanced venous disease⁴⁰, which is characterized by formation of VLU and accounts for this condition’s significant morbidity with >40% of these patients suffering from frequent recurrences⁴¹. Although it is well known that CVD is common in the general population, the true disease burden is difficult to estimate given its spectrum of phenotypes, which are often misclassified. These phenotypes are delineated in the CEAP classification system [Clinical (C), Etiological (E), Anatomical (A), and Pathophysiological (P)] ⁴² (Figure 1B), however this system is not widely implemented outside of vascular specialties and therefore phenotypic classification including the diagnosis codes that describe a patient’s CVD are often inaccurate. For example, it is common to see patients with extensive venous reflux and symptomatic CVD with swelling and skin changes (CEAP 4) or patients that develop venous hypertension due to PTS or increased central pressures due to morbid obesity causing swelling (CEAP 3), inaccurately codified as “varicose veins” (CEAP 2). Both are common scenarios and obfuscate the true depiction of CVD demographics. VLUs (CEAP 5 or 6 disease) are more easily identifiable phenotypes of CVD but account for only a small percentage of all CVD patients. Most epidemiology studies reporting incidence of CVD report on “varicose vein”, “venous reflux”, or “venous ulcer” phenotypes, but inaccurate clinical classification likely lead to significant underestimation of the true CVD prevalence. Indeed, one study that included CEAP 1 through 6 estimated the global prevalence of CVD to be as high as 83.6%⁴³.

Figure 1: Chronic Venous Disease.

(A) Normal venous return occurs through a system of superficial, perforating, and deep (intermuscular) veins propelled by the pressure gradient derived from muscle contraction and a network of bicuspid venous valves to prevent retrograde blood flow. During proper functioning, the pressure within the thin-walled veins remains relatively low (~20–30 mmHg) (B) Chronic venous disease occurs in the setting of prolonged venous hypertension arising from multiple factors including degenerate venous valves, weakened muscle contraction, and proximal obstruction. Rather than the standard flow of blood from the superficial to the deep veins, incompetent valves allow for blood to flow back into the superficial veins or pool in the deep vein increasing their local pressure. Overtime, the increased volume and pressures in the veins incites endothelial dysfunction including loss of the protective glycocalyx and venous wall inflammation resulting in adverse extracellular matrix remodeling. This is further exacerbated by RBC extravasation and leukocyte adhesion. Further support for these mechanistic changes within the venous wall have been derived from gene expression studies comparing healthy and diseased veins. (C) CEAP [Clinical-Eitology-Anatomy-Pathphysiology] classification system for patients with chronic venous disorders. (D) Summary table of genetic abnormalities (and when appropriate their associated genetic syndrome) that have been associated with varicose vein formation. These associations are nicely reviewed by Anwar et al¹¹⁹.

CVD and VLU significantly impact our patients’ quality of life due to the heavy burden of wound care requiring frequent medical appointments and significant time spent tending to wounds at home. This often limits their ability to participate in social activities, their employment options, and takes a large emotional toll on patients. The average VLU can take over 3 months to heal⁴⁴ making it an enormous burden to the United States healthcare system with over 1 billion dollars spent annually on treatment of VLU alone⁴⁵.

CVD Risk Factors:

There are multiple established risk factors for CVD including older age, female sex, family history, prolonged standing, and obesity. The correlation between increased age and disease prevalence is thought to be due to vessel wall deterioration over time coupled with an increased venous pressure resulting from weakened calf muscles⁴⁰. Female sex is also a risk factor for varicose veins with pregnancy proving to be a major contributing factor due to the dramatic physiologic changes including increased blood volume, weight gain, elevated intra-abdominal pressures, and decreased venous return⁴⁶. Interestingly, increased production of relaxin, a hormone secreted by the corpus luteum to relax the pelvic ligaments and a potent vasodilator, may contribute to the increased venous pressures in the lower extremities^{47, 48}. A recent study published in Circulation used a machine learning approach and confirmed many of the known risk factors but also identified several new strong predictors, including leg bioimpedance and height⁴⁹. Height had been identified as a potential risk factor in an early epidemiological study several decades ago⁵⁰ but had been inconsistently reported since^50–52. The other newly identified risk factor, bioimpedance, defined as the ability of tissue to impede electric current, reflects the amount of fluid accumulation in body tissue⁵³. These novel predictors likely point to the link between high-volume venous reflux, increased hydrostatic pressure, and resulting venous hypertension.

Venous Disease Genetics

VTE genetics:

The heritable nature of VTE has been documented for over a century when it was shown that certain families were afflicted at unusually high rates⁵⁴. Family studies have estimated the genetic factors contribute nearly 50% to risk.⁵⁵ Further, a patient with an affected sibling has a familial standardized incidence ratio of 2.5, which jumps to 51.8 if the patient has 2 or more affected siblings.⁵⁶ Numerous approaches including familial studies, candidate gene analyses, GWAS, and now deep sequencing have been applied to decipher the genetic factors that influence a patient’s propensity to develop VTE. Most of the work presented below addresses predictors of incident VTE; the literature on the genetic predictors of recurrent VTE has been limited, primarily due to poor statistical power due to small sample sizes.^57–59

Familial Studies of VTE Genetics:

Early studies investigating the etiology of VTE were conducted in families and linked inherited characteristics with VTE. Often the presenting phenotype was a deficiency in a natural anti-coagulant, such was antithrombin, protein C, or protein S, which are strong predictors of VTE.^60–63 Family studies are useful when the condition has a high penetrance and when the VTE is the result of a monogenetic trait with an intermediate phenotype that has a large association with the clinical outcome, such as the familial mutations in the natural anticoagulant genes. It must be mentioned that one of the earliest studies identifying the genetic associations with VTE was an international cooperative study investigating VTE risk associated with oral contraceptive use using traditional case-control epidemiologic methods.⁶⁴ Investigators observed that women with blood group O (determined by ABO) had a decreased risk of VTE compared with other blood groups.

Candidate-Gene Approach to Genetic Discovery:

Starting in the early 1990s, basic biology discoveries and insights led investigators to target candidate genes and their proteins that were suspected of regulating the risk of VTE. These investigations have often been conducted in key proteins of the coagulation/anti-coagulation and fibrinolysis/anti-fibrinolysis pathways. The realization that genetics may influence VTE risk started with the recognition of activated protein C resistance in 1993⁶⁵ and then characterization by two separate groups in 1994 of the causal factor V Q506 non-synonymous amino acid substitution resulting from single nucleotide polymorphism (variant) in the F5 gene.^{63, 66, 67} This variant, F5 1691A (rs6025), prevalent in 2–5% of European-ancestry (EA) individuals, is known as the FV Leiden variant and increases the risk of VTE 2 to 5-fold in carries of risk allele.⁶⁸ Other candidate-gene discoveries include the prothrombin mutation, F2 20210A (rs1799963), prevalent in <2% of EA individuals and associated with a greater than 2 to 3-fold increase in risk of VTE.^{68, 69} The prevalence of the F5 and F2 variants in non-EA individuals is markedly lower—or not present. Other candidate gene studies led to discoveries of genetic variation in the protein C receptor (PROCR) and in 1 of the 3 fibrinogen genes (FGG), relatively common variants that initially reported increases in risk of 2-fold magnitude but are now thought to be smaller.^{68, 70–72} Indeed, the candidate approach has been deemphasized in the modern genomics era given challenges in reproducing findings, which are thought to arise from publication bias and often failing to correct for multiple comparisons.

With the advent of genotyping arrays in the early 21^st century and better characterization of genetic variation in key cardiovascular genes, investigators were able to simultaneously interrogate multiple variants at reasonable costs. The 2007 JAMA publication investigated 172 variants in 24 clotting-related genes for association with an incident VTE among 349 cases and 1680 controls.⁷³ The investigation identified tissue factor pathway inhibitor (TFPI) with global (gene-wide) association with VTE risk and also identified five variants associated with VTE in F5, F11, and protein C (PROC) including three novel associations.

Genome-wide Approach to Genetic Discovery:

Beginning at the end of the first decade of the 21^st century, a series of genome-wide association studies (GWAS) were published with consecutively larger samples sizes (larger populations) and larger number of variants investigated. In 2008, the first genome-wide interrogation of genetic variation for VTE was conducted using 19,682 gene-centric (mainly-non-synonymous amino acid substitutions) variants in 3 case control studies of first DVT.⁷⁴ The Dutch investigators identified three SNPs that were strongly associated with DVT in CYP4V2 (later found to be in linkage disequilibrium with F11),⁷⁵ antithrombin (SERPINC1), and GP6. In 2009, French investigators analyzed 317,139 variants in 453 cases and 1327 controls.⁷⁶ This was the first agnostic investigation of the human genome for association with VTE. The team identified 3 variants in 2 genes (F5 and ABO) known to be associated with VTE. Additional work by the French investigators using a larger genotyping array that covered over 551,000 variants, investigated association with VTE in 1,542 cases and 1,110 controls.⁷⁷ Agnostic variant-based analyses identified associations with 4 known VTE loci (ABO, F5, F11, and FGG) while haplotype analyses provided some evidence of association of 3 genes (HIVEP1, STAB2, and PROCR) with VTE. Next, in 2012 a Mayo Clinic team published a GWAS that interrogated 2.5 million imputed variants in 1503 cases and VTE associated loci using Mayo Clinic patients and a separate replication among Olmsted County data with 1,407 cases and 1,418 controls.⁷⁸ The team identified associations at the F5 (including NME7) and ABO loci. A team of investigators from the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) Consortium meta-analyzed GWAS data from 6 studies that included 1,618 VTE cases from 44,499 participants plus a replication sample of 3,231 VTE cases and 3,536 controls from 3 case-control studies.⁷⁹ Nearly 2.5 million variants were interrogated and identified associations in F5 and ABO; the discovery and replication dataset were combined and association in the F11 and FGG loci reached genome-wide significance. In 2019, an exome-wide array investigation of VTE identified variants in known VTE loci but was underpowered to identify novel rare variation throughout the exome.⁸⁰ More recently exome array findings have been published on a smaller set of VTE cases and results suggested new candidate loci.⁸¹

Although important, these early GWAS were unable to identify new candidate genes associated with VTE. It was not until the International Network Against Venous Thrombosis (INVENT) Consortium was established, a collaboration that meta-analyzed previously published GWAS, that novel loci were identified.⁸² In 2015 INVENT investigators performed a meta-analysis of 6,751,884 variants from VTE GWAS that included 7,507 cases and 52,632 controls and then replicated using 3 case control studies with 3,009 cases and 2,586 controls. In addition to identifying 6 previously known loci (ABO, F2, F5, F11, FGG, and PROCR), investigators identified 3 novel loci with variants reaching genome-wide significance in ZFPM2, TSPAN15, and SLC44A2. The TSPAN15 and SLC44A2 loci replicated whereas ZFPM2 would be replicated later in a UK Biobank publication (3,290 cases and 116,868 controls) in 2016 that interrogated nearly 9 million variants.⁸³ The 2019 publications in Blood and Nature Genetics further expanded the number of novel loci associated with VTE.^{72, 84} The INVENT Blood manuscript published a trans-ancestry meta-analysis of data from 18 VTE GWAS studies with replication in the Department of Veterans Affairs Million Veteran Program (MVP).⁷² Discovery meta-analysis that included 30,234 cases and 172,122 controls and interrogated nearly ~13 million variants identified 34 independent signals across 24 genetic loci. Among these, 14 were newly reported associations (11 newly associated loci, 3 new independent signals in known genes). Further discovery using a transcriptome-wide association study using blood and liver expression data identified an addition 5 loci associated with VTE. Although the biologic role of the new loci has not yet been characterized, several loci have been associated with red blood cell and platelet traits suggesting an important role of these phenotypes in the development of VTE. The MVP investigators published their discovery GWAS in Nature Genetics that tested ~13 million variants (26,066 cases and 624,053 controls) with replication in INVENT.⁸⁴ In addition to replicating 11 previously described loci, they identified 28 novel candidate loci, 22 of which were replicated in INVENT.

Whole Exome and Whole Genome Sequencing Approach:

Only recently has a whole-exome investigation been published.⁸⁵ In case-control study of 393 idiopathic VTE cases, a novel VTE association was found for STAB2, previously described to associate with plasma levels of von Willebrand factor and factor VIII.^{86, 87} Whole-genome sequencing has been completed for over 3,700 VTE cases in the National Heart, Lung, and Blood Transcript-Omics in Precision Medicine (TOPMed) program and a publication of the deeply sequenced participants is expected to appear in press in 2021.

Summary of VTE Genetic Discovery:

Over the past 50 years, different approaches have been used to identify genetic determinants of VTE risk (summarized in Figure 2 and Table 1). It has only been in the past decade that these studies have expanded our biologic knowledge base of VTE beyond the well-characterized coagulation/anti-coagulation and fibrinolysis/anti-fibrinolysis pathways. Of the 2 earliest novel, well-replicated discoveries for VTE, TSPAN15, and SLC44A2, we are only beginning to understand how genetic variation in the choline transporter Slc44a2 gene (SLC44A2) controls platelet activation and thrombosis through the regulation of mitochondrial function and still know little about how the tetraspanin 15 protein (TSPAN15) modulates thrombotic risk.^{82, 88} A diverse trove of other genes have shown novel association with VTE, some of which point to the role of immune function, platelets, and red blood cells in modifying VTE risk.^{72, 84}

Figure 2: Placing the genetics of venous thromboembolism into context.

Many of the early discoveries implicated genes coding for protein components of the coagulation/ anticoagulation pathways. Those factors that have been identified in previous studies to play a role in VTE risk have been bolded in the coagulation cascade or included at the bottom of the box if not neatly fitting into the cascade. As our understanding of the genetic influence on VTE has expanded, genes that appear to have a functional role outside of classical coagulation have been identified and appear to implicate multiple hematopoietic cell lineages including immune cells, platelets, and red blood cells. The functional role of many of the risk loci has yet to be determined. This figure was adapted from figure 2 of Zoller et al (2020)¹²⁰ and was created using BioRender.com.

Polygenic Risk Score for VTE:

The clinical application of the genetic discoveries to date remains limited. Several groups have created polygenetic risk scores (PRSs) over the past decade to help characterize a more global burden of genetic risk and its association with incident and recurrent VTE.^{59, 72, 84, 89} In the INVENT Consortium 2019 publication, the PRS was based exclusively on UK Biobank data and the 37-variant risk score identified an associated 50% reduction in risk for those at or below the 5^th percentile of the risk-score distribution (OR 0.51, 95% confidence interval 0.43–0.61) compared with half the population in the middle range and a 3-fold increase in risk for those at or above the 95^th percentile (OR 3.2; 95% CI: 2.9–3.5).⁷² In the MVP 2019 publication, similar estimates were made using a score generated from 297 variants.⁸⁴ The utility of PRS was recently demonstrated in a study of high risk cardiovascular disease patients using the 297 variant PRS to predict VTE events among patients in three Thrombolysis in Myocardial Infarction (TIMI) trials. Over a median follow up of 2.4 years, there were 174 VTE events. Those patients in the upper tertile of PRS had a 2.7-fold higher risk of VTE with each standard deviation increase leading to a 47% higher risk.⁹⁰

CVD genetics:

Family history is a well-known risk factor for CVD however the true heritability of CVD has been difficult to ascertain given its multifactorial nature where prominent environmental risk factors interact with genetic variation that dictate a patient’s predisposition to the disease. There have been numerous epidemiologic studies that have established the heritable nature of varicose veins. A study in 1969 determined the prevalence of varicose veins in the women who worked in cotton mills in England and Egypt⁹¹. In addition to implicating pregnancy, prolonged standing, and habitual corset wearing, the study found a family history to be a common finding among those afflicted. Later, a French study of 134 families found that the risk of developing varicose veins was 90% when both parents were affected by the disease⁹². Another study using the Swedish Hospital Discharge Register found that the standardized incidence ratio was 2.39 when patients had one affected parent and 5.52 in patients with two affected parents. Interestingly, they also observed an increased risk among spouses of affected patients (SIR ~1.7), reiterating a significant lifestyle component to the disease⁹³. Finally, a German study examining over 4,000 families found the narrow sense heritability of CVD (CEAP 2–6) to be 17.3%⁹⁴.

The genetics of CVD have received less attention than VTE but the progress has been made using similar strategies including candidate gene approaches and GWAS. Additional progress has also been made by comparing the gene expression in healthy versus diseased tissue. As described earlier, some of the challenge in elucidating the genetic component has been the variability in disease classification, thus the majority of studies have focused on the varicose vein phenotype.

The gene expression approach

Unlike VTE, where the pathogenesis is thought to arise largely intravascularly, CVD provides the opportunity to compare transcriptomic changes in the degenerate venous tissue that defines the disease with healthy venous tissue to gain insights into key pathogenic changes. One study using an ex vivo model of venous hypertension found that inferior vena cava segments from rats loaded with tension for 24 hours resulted in increased matrix metalloproteinase (MMP) expression and decreased stimuli-induced contraction⁹⁵. This has been supported by multiple human studies of varicose veins showing similar changes in MMPs and TIMPs, their natural inhibitors, suggesting a critical role for matrix remodeling in CVD^96–99. A transcriptomic analysis published in 2018 compared paired samples of varicose and normal venous tissue from the eight patients and found numerous differentially expressed genes found to be enriched in pathways dictating extracellular matrix organization and vascular morphogenesis¹⁰⁰. Although this approach has been useful in understanding the changes in gene expression that occur within diseased vessels, this approach is unable to distinguish whether changes in gene expression are the cause or a consequence of the disease process.

Deriving genetic insights from congenital diseases

Significant insights into CVD genetics have been made by studying the pathologic mutations underlying congenital conditions where CVD is a prominent feature of its symptomatology. This was the case for FOXC2, which was postulated to be associated with primary varicose veins^{101, 102} because of its role in lymphedema-distichiasis syndrome, a rare condition characterized by dysfunctional lymphatic vessels and extra eyelashes. In a study of 18 families with 74 affected individuals, there was a 49% prevalence of varicose veins (age of onset 7–28) and 72 of the 74 patients with harbored mutations in the FOXC2 gene, which were primarily small insertions and deletions making it the first study to identify FOXC2 as a candidate gene for the development of varicose veins¹⁰¹. Another example is cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), a rare congenital condition characterized by early onset varicose vein formation that is caused by mutations in the Notch3 gene, raising interest in the possible role of Notch3 in varicose vein pathogenesis¹⁰³. Additional examples are included in Figure 1D including Klippel-Trenaunay syndrome and Ehlers-Danlos syndrome¹⁰⁴ as well as other genetic mutations that have been associated with varicose vein development. The candidate gene approach is limited in that it has to be hypothesis driven and requires sufficient knowledge of disease pathophysiology for the hypothesis to be synthesized. Candidate gene approaches can reveal fundamental pathways that may inform broader hypotheses as to the general pathophysiology of the disease, however the vast majority of varicose veins occur in patients without a known genetic disorder and therefore knowledge of these particular mutations will only explain a small subset of the population.

The GWAS approach

Previously, GWAS studies for CVD have been limited by insufficient cases and controls needed to attain sufficient statistical power to identify associated loci but with larger biorepository data becoming publicly available, GWAS studies have become more feasible. A persistent challenge remains in ensuring accurate CVD phenotyping reported in the database. Indeed, the definition of the cases can result in over- or undercounting and have a tremendous impact on the resulting genetic analyses. However, despite these challenges there has been significant progress using GWAS to identify risk loci for varicose veins (summarized in Table 2).

In 2017, the first CVD GWAS was published comparing 323 unrelated cases with 4,619 controls in a German cohort and identified three novel CVD susceptibility loci, including EFEMP1 (fibulin-3, an extracellular matrix glycoprotein), KCNH8 (member of the human Elk voltage gated potassium channel) and SKAP2 (adaptor protein involved in Src signaling)¹⁰⁵.

This was followed a year later by a Circulation paper featuring a GWAS of varicose veins using ICD9 and −10 codes from the UK Biobank with expression quantitative trait loci (eQTL) and pathway analyses among 337,536 individuals (9,577 cases). This revealed 30 new genome-wide significant loci⁴⁹, however investigators were unable to perform a replication analysis in a separate cohort. The strongest association was located on chromosome 1 in the CASZ1 gene, a locus known to influence blood pressure (rs11121615; p=3.71×10⁻⁶⁵). Other lead variants were located in the 16q24 region, in or near vascular mechanosensory channel PIEZO1 and galactosamine-sulfatase enzyme GALNS (rs2911461, P=4.81×10⁻²⁹; rs8053350, p=8.93×10⁻¹⁸). eQTL analyses identified several genes significantly regulated by the varicose vein SNPs which impact vascular development and skeletal abnormalities. These analyses identified the lead variants rs2911463 (PIEZO1, p=1.50×10⁻⁵; GALNS, p=1.50×10⁻⁵) and rs3101725 (FBN2, p=9.50×10⁻⁷), which have been linked to a hereditary form of skeletal dysplasia and the Marfan-like syndrome congenital contractural arachnodactyly, respectively. FUMA analysis revealed SNPs were associated with varicose vein predictors including waist hip ratio (p=2.79×10⁻⁴) and height (p=2.30×10⁻³). Interestingly, Mendelian randomization analysis suggested that increased height was causally related to varicose veins (inverse-variance weighted odds ratio, 1.26; p=2.07×10⁻¹⁶). It is also noteworthy that Mendelian randomization has also shown height to be a risk factor for VTE in a meta-analysis¹⁰⁶ although prothrombotic genotypes did not appear to increase VTE risk in subjects with increased height.¹⁰⁷

The most recent CVD GWAS, a preprint on BioRxiv¹⁰⁸, used ICD10, procedure codes, and self-reported disease from the UK Biobank to assemble a 400,000 patient cohort with more than 22,000 cases and performed replication of their findings with self-reported cases from 23andMe. Their analyses revealed 46 susceptibility loci for varicose veins, including 28 novel loci. Interestingly, using linkage disequilibrium score regression, they replicated the association with height and weight but also discovered a novel genetic correlation of varicose veins with systemic lupus erythematosus. Their data implicate dysregulated ECM production, chronic inflammation, and angiogenesis as processes that may contribute to pathogenesis and represent putative therapeutic targets.

The differences between the Circulation and BioRxiv papers despite using the same patient dataset are fascinating and potentially shed additional light on the challenge of accurately defining CVD. The Circulation study reported varicose veins defined only by ICD codes while the preprint identified varicose vein patients through ICD codes (13363 cases - primary and secondary codes), operational codes (12195 cases - primary codes), and self-report of varicose vein surgery (20115 cases) with their total varicose vein patient subset being 22473 cases when overlapping cases were excluded. Given that patients with varicose vein operational codes should have ICD codes for this diagnosis, it seems unlikely that the inclusion of operational codes in the data set drove the differences. Indeed, most of the additional cases in of varicose vein cases appear to be coming from self-report codes, which introduces the possibility of over-counting as subjects may be counting extraneous procedures in the leg (e.g., sclerotherapy for superficial reticular veins or a stripping procedure for venous reflux in the absence of varicose veins) as a surgery for the colloquially well-known condition of varicose veins. However, it also true that those cases which have a true “self-report of varicose vein surgery” are undercounted in the dataset used in the Circulation study. Importantly, a study published shortly after replicated the findings in the Circulation study with the same UK Biobank dataset using ICD codes to define varicose veins and prioritized the same candidate genes.¹⁰⁹ Taken together, both approaches are reasonable and subject to their own set of limitations, but the findings are largely complementary, and both greatly enhance our understanding of the genetics underlying CVD heritability. As more data from additional biorepositories with information on CVD become available and GWAS is performed on multiple subsets of people, it will become easier to filter out the noise and the true signal will likely become stronger.

VLU genetics

Venous ulcer development is usually multifactorial but the general mechanism of ulcer formation is thought to occur with the following steps: (1) necrosis of the extracellular matrix due to TGFβ signaling and matrix metalloproteinase (MMP) activity^{110, 111}, (2) destruction of collagen microfibrils and proteoglycans^{110, 111}, and (3) increased apoptosis of keratinocytes and increased caspase-2 leading to dermal-epidermal detachment¹¹². Interestingly, recovery from VLU has been shown to be compromised in subjects with altered iron metabolisms¹¹³. Iron deposition, which occurs secondary to skin complications of chronic venous insufficiency, generates free radicals, drives proteolytic hyperactivity of MMPs, and downregulates tissue inhibitors of MMPs. Mutations in the hemochromatosis gene (HFE) including HFE C282Y and HFE H63D, which cause a deficiency of iron metabolism, also leads to a 5-fold increase in the risk of developing venous ulcers in subjects with venous disease¹¹³. Another related gene, SLC40A1, which codes for ferroportin (FPN1), a transmembrane protein involved in the export of intracellular iron, has a SNP in the promoter region (FPN1–8CG) that appears to affect gene expression. Carriers of this SNP have a 5-fold increase in venous ulcer susceptibility¹¹⁴ further implicating iron metabolism in the formation of VLU. Other genes that have been associated with VLU and poor healing include MTHFR^{115, 116}, MMP12¹¹⁴, FGFR2 ¹¹⁷ and FXIII¹¹⁸.

Conclusion

Venous disease, which includes VTE and CVD, are prevalent conditions that account for significant morbidity and mortality. Despite decades of research, both diseases lack adequate therapeutic options and remain a large burden on the US health care system. With the emergence of modern population genetics, we have started to decode the genetic factors mediating the heritable contribution to these diseases. We now have numerous candidate loci for both diseases but still have significant unexplained heritability and often a rudimentary understanding of the functional role of these loci on venous disease. We should celebrate the progress that’s been made, appreciate the remaining unexplained heritability, and armed with our growing understanding of the pathogenesis seek to discover novel safe and effective therapeutic strategies.

Non-standard Abbreviations and Acronyms:

CADASIL

Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy

CEAP

clinical, etiological, anatomical, pathophysiological

CHARGE

Cohorts for Heart and Aging Research in Genomic Epidemiology

CI

confidence interval

CVD

chronic venous disease

CVI

chronic venous insufficiency

DVT

deep venous thrombosis

ECM

extracellular matrix

eQTL

expresstion quantitative trait loci

FUMA

Functional Mapping and Annotation

GWAS

genome wide assocation study

ICD

international statistical classification of diseases

ICU

intensive care unit

INVENT

International Network of VENous Thromboembolism

MMP

matrix metalloproteinase

MVP

Million Veteran Program

OR

odds ratio

PE

pulmonary embolism

PRS

polygenic risk score

PTS

post thrombotic syndrome

SIR

standardized incidence ratio

SNP

single nuceoltide polymorphism

SVT

superficial venous thrombosis

TIMI

Thrombolysis in Myocardial Infarction

TIMP

Tissue inhibitor of matrix metalloproteinase

TOPMed

Trans-Omics for Precision Medicine

VLU

venous leg ulcer

VTE

venous thromboemolism