Quantitative retinal traits and their association with cardiovascular disease and cardio-metabolic genetic variants in people with type 2 diabetes

Abstract

Introduction: Type 2 diabetes (T2D) is one of the most prevalent noncommunicable diseases in the world and its cardiovascular complications present a huge socio-economic burden. In 2015, in the UK alone, 3.8 million people have been diagnosed with T2D and cardiovascular disease accounts for almost 1.7 million episodes throughout the country. Early diagnosis of cardiovascular disease in people with T2D thus becomes critical. The retina gives a unique opportunity to study the human microcirculation, which can then offer insights into the pathophysiology of cardiovascular disease. By using semi-automatic software, retinal images can provide quantitative traits derived from the microvasculature. Previous research has found that arteriolar and venular calibres are associated with cardiovascular outcomes such as hypertension and stroke. Moreover, retinal vascular tortuosity, a novel quantitative biomarker which measures the degree to which blood vessels visible in the retina twist and turn, has been associated with traditional cardiovascular risk factors in the general population. However, this area needs to be further explored, especially in the population with T2D and in prospective analyses. Aims: To determine whether quantitative retinal traits such as vessel widths, vessel tortuosity and multifractal dimensions are associated with the subsequent development of major cardiovascular events such as ischaemic heart disease and stroke in people with T2D. Also, to use a genome-wide association approach to investigate if these quantitative retinal traits are associated with cardio-metabolic genetic variants, which could help identify novel biomarkers of cardiovascular disease for future research. Methods: Analyses used the Edinburgh Type 2 Diabetes Study, a prospective cohort of 1066 men and women with T2D aged 60-75 years at baseline with eight years of follow-up for cardiovascular events. A total of 1028 retinal images from baseline were analysed using the semi-automatic retinal software VAMPIRE (Vascular Assessment and Measurement Platform for Images of the Retina). Cross-sectional analyses including ANOVA and Chi-square test were performed along with prospective analysis using Cox regression. Additionally, a genome-wide association study was performed to explore the association of 12 quantitative retinal traits with cardio-metabolic genetic variants. Imputation of variants included in the MetaboChip array was used. Results: In an unadjusted model, there was a significant association between arteriolar tortuosity and incident stroke (Hazard Ratio (HR) 1.26; 95% CI 1.02, 1.57; p=0.03). This association remained significant after full adjustment for age, sex, cardiovascular risk factors (body mass index, HbA1c, total cholesterol, duration of diabetes, renal dysfunction) and previous cardiovascular events (HR 1.26; 95% CI 1.01, 1.58; p=0.04). Multifractal dimensions, a novel retinal biomarker which provides an insight into vascular geometry, was inversely associated with incident stroke (unadjusted HR 0.73; 95% CI 0.57, 0.94; p=0.01). This association also remained significant after adjustment for age, sex, cardiovascular risk factors and previous cardiovascular event (HR 0.73; 95% CI 0.56, 0.94 p=0.02). Associations between other retinal traits and stroke, and between traits and ischaemic heart disease, tended not to be statistically significant, especially after multivariable adjustment. The genome-wide association analysis of arteriovenous ratio (ratio of arteriolar to venular vessel width) revealed a genome wide significant locus, rs73198094 (p = 5.27 x 10-8), an intergenic variant located between ASAH1 and LOC101929066 genes in chromosome 8. The ASAH1 gene has been associated with atrial fibrillation. Although no further single nucleotide polymorphisms reached genome-wide significance, some additional promising findings emerged. Analysis for retinal arteriolar width revealed a genome-wide suggestive intronic locus, rs4944903 (p= 8.5x10-7), of the gene POLD3 in chromosome 11. Identified loci for minimum arteriolar tortuosity, rs7991332 (p=1.54 x 10-6) and rs2172724 (p=2.46 x10-6), are located in the COL4A2 gene in chromosome 13 and another identified variant, rs7319323 (p=3.53 x 10-6), is located in an intron of the neighbouring COL4A1 gene. Previous studies showed the relevance of these genes including an association with stroke and intracerebral haemorrhage. These two genes encode collagen protein chains, which are major components of the vascular basement membrane. Another promising variant identified was the rs34013641 locus, associated with minimum venular tortuosity (p=2.81 x 10- 6), which is located in the MYH11 gene in chromosome 2. This gene encodes smooth muscle myosin heavy chain protein, which is highly expressed in human arteries. Finally, identified loci for the multifractal dimension D0, rs10963694 (p=8.53 x 10-7) and rs4977506 (p=4.95 x 10-6), located on the ADAMTSL-1 gene in chromosome 9, are also strong candidates in the pathophysiology of vascular disorders. Conclusions: In older people with T2D, arteriolar tortuosity and multifractal dimensions were significantly and independently associated with incident stroke. GWAS findings for these and other quantitative retinal traits offer insight into pathophysiological changes of the vasculature, which may result in cardiovascular disease. These findings, in the context of further research, could potentially be used to reveal biological mechanisms related to major cardiovascular complications of T2D and to guide efforts on prevention and early interventions.