Ultimately, the goal of defining the clinical and genetic factors underlying disease is to improve identification of at-risk individuals, implement targeted prevention strategies, and develop effective therapies. For example, the identification and treatment of epidemiological risk factors such as smoking, hypertension, and cholesterol levels have been associated with marked declines in coronary heart disease mortality. More recently, genomic studies have provided important insights into our understanding of complex cardiovascular diseases.1–3 More than 2500 genome-wide association studies have been published, uncovering thousands of unique single nucleotide polymorphism-trait associations,4 and genes identified in genome-wide association studies are nearly 3 times more likely to be targetable than other genomic regions.5 However, the underlying basis of most diseases remains poorly understood, and therapeutic interventions are often targeted toward disease manifestations rather than their mechanistic causes.
Such is the case for varicose veins. These enlarged, tortuous vessels of the superficial venous system are often written off as an unsightly nuisance, and their clinical implications have largely been underappreciated. However, varicose veins affect nearly a quarter of US adults and negatively impact quality of life.6,7 For a significant number of patients, varicose veins and venous insufficiency may be complicated by superficial thrombophlebitis or progress to stasis dermatitis and ulceration.8 Moreover, patients with varicose veins appear to have increased risk for other vascular diseases, including deep venous thrombosis.9
Despite their commonness and associated morbidity, varicose veins have not been the focus of extensive investigation. Studies seeking to define clinical risk factors have been relatively small and have often provided inconsistent results.10–12 Genetic studies of varicose veins are similarly limited. A small number of genes have been implicated in the context of congenital disorders,13 and the only genome-wide association study to date, including ≈10 000 individuals (2269 cases), identified just 2 loci with genome-wide significance.14 Hence, the clinical and genetic epidemiology of varicose veins remains incompletely defined, and overall the tempo of mechanistic discovery and therapeutic innovation has been slow.
In this context, Fukaya et al15 present the largest and most comprehensive epidemiological and genetic study of varicose veins to date in a cohort of ≈500 000 subjects from the UK Biobank. One innovative aspect of the study was the use of unsupervised machine learning for agnostic interrogation of >2700 diverse clinical variables to identify those associated with incident varicose veins. This approach identified established risk factors such as increasing age, obesity, and history of deep venous thrombosis, as well as a number of novel candidate risk factors, including increased height and leg bioimpedance, both of which remained independently associated with varicose vein risk when evaluated using conventional statistical models. This analysis also provided insight into clinical features whose associations with varicose veins were ambiguous because of conflicting data from prior reports (eg, identifying systolic blood pressure as a likely risk factor but detecting no increased risk with oral contraceptive use).
The investigators also conducted multiple analyses to investigate the genetic architecture of varicose veins. A genome-wide association study in 300 000 subjects, including 10 000 cases, showed varicose veins to be a highly polygenic disease and dramatically expanded the list of genetic associations, identifying 30 new loci. The most strongly associated loci were near genes associated with blood pressure and vascular mechanosensing. Many regions also contained single nucleotide polymorphisms previously associated with traits such as height and waist-to-hip ratio as well as expression quantitative trait loci for biologically plausible genes. It is interesting to note that genes in associated loci showed enrichment in pathways such as vascular development, endothelial cell differentiation, and vascular endothelial growth factor signaling. To further probe the relationship between clinical risk factors and varicose veins, investigators assessed their genetic correlations, finding a 16% genetic overlap between varicose veins and height, a 10% correlation with body mass index, and a striking 36% with deep venous thrombosis. These analyses deepen our understanding of the relationship between each risk factor and varicose veins and provide a direct assessment of the shared genetic and, by extension, biological drivers of disease risk. Finally, because height was associated with varicose veins in both observational and genetic analyses, the authors sought to further evaluate this association using Mendelian randomization, which strongly implicated increased height as a causal risk factor. Although a connection between higher lower extremity venous hydrostatic pressure in taller individuals seems intuitive, it has been inconsistently identified in prior reports, and a major finding of the current study is the establishment of height as directly related to varicose vein risk.10–12
Several aspects of the study are especially notable, including the scale (>60 time larger than prior investigations) and utilization of diverse, complementary analytic methodologies. The application of machine learning algorithms to a large clinical cohort represents a new paradigm for exploring the epidemiology of varicose veins. These methods, in combination with contemporary genomic approaches, provide unprecedented insight into the clinical epidemiology and genetic architecture of varicose veins and represent a significant advance in the field.
The authors should be complemented on this important contribution. However, in some respects, it is only the first step. Future epidemiological efforts are needed to determine whether the observed clinical and genetic associations can be replicated, especially in populations outside the middle-age British subjects included in this study. The finding of shared genetic risk between varicose veins and clinical risk factors, especially deep venous thrombosis, is intriguing and suggests the possibility of shared pathobiology that remains to be elucidated. Similarly, it remains to be seen whether implicated biological pathways, although plausible, can be therapeutically targeted. Like most genome-wide analyses, the majority of associated loci do not contain annotated genes and have unknown underlying biology. These investigators have made many interesting observations, and the field must now begin the difficult tasks of uncovering the biological basis of these associations, understanding their significance, and translating them to improved patient care.
More generally, the current study represents a realization of the power of large-scale analyses to uncover the clinical and genetic determinants of complex disease. However, it also highlights an important inequity. Recent years have seen rapid growth in the amount and diversity of data available to biomedical scientists. Genetic data are increasingly abundant and linked to large, phenotypically rich data derived from electronic health records, clinical trials, and population-based studies. In parallel, there has been a rapid evolution of analytic methods for their interrogation. However, most large resources, such as the 1 used in this study, contain predominantly subjects of European ancestry. This is especially true for genetic studies. Because patterns of clinical and genetic disease associations are complex and may vary by factors such as ancestry, socioeconomic status, and geography, the scientific community must continue striving to increase diversity of these important resources so that all groups benefit from the discoveries they yield.
Quinn S. WellsMD, PharmD, MScDepartment of Medicine, Division of Cardiovascular Medicine, Vanderbilt University Medical Center, Nashville, TN.
https://www.ahajournals.org/doi/10.1161/CIRCULATIONAHA.118.037219.
Gracias.