Diabetic retinopathy and other retinal vascular measures with the risk of cardiovascular events: a cohort study of type 2 diabetes in Singapore

Abstract

Background. While cardiovascular disease risk assessment is needed to inform clinical decisions, predicting individual risk in diabetics remains challenging. The evidence base regarding the effect of diabetic retinopathy on cardiovascular disease risk in a multi-ethnic Asian diabetic population is lacking.

Methods. This was a retrospective cohort study examining two cohorts of patients (n=17,705 and n=10,280) from the Singapore Diabetes Registry and Singapore Integrated Diabetic Retinopathy Programme. Given the nonavailability of cause-of-death data, the primary outcome (time to first major adverse cardiovascular event) was defined as a composite of non-fatal stroke, non-fatal myocardial infarction and all-cause mortality, with a secondary definition treating all-cause mortality as a competing risk to non-fatal stroke and myocardial infarction was also used. Multivariable Cox and Fine-Gray models were used to estimate hazard ratios (HRs) for the primary outcome associated with diabetic retinopathy, diabetic macular edema, and retinal vascular calibres among individuals with no prior history of cardiovascular disease at baseline.

Results. During an average follow-up of 4.4 years in our primary cohort, 790 (4.46%) and 223 (1.26%) individuals experienced composite MACE and non-fatal MACE respectively. After adjustment for significant covariates and established risk factors, the presence of clinically referable diabetic retinopathy was associated with composite MACE (HR 1.64, 95% CI 1.18-2.30), non-fatal MACE (2.66, 1.47-4.81), composite CVD (1.69, 1.37-2.10) and non-fatal CVD (1.87, 1.48-2.37). Retinal arteriolar calibre at the extended zone was observed to be associated with non-fatal MACE (1.01, 1.00-1.02). In the primary cohort, inclusion of clinically referable diabetic retinopathy into prognostic models produced significant improvements in measures of risk discrimination, with an increase from 0.7321 to 0.7402 and 0.6892 to 0.7106 in time-dependent AUC for Framingham risk models predicting composite and non-fatal MACE respectively at the 4-year mark.

Conclusion. The presence of clinically referable diabetic retinopathy significantly increases the risk of incident cardiovascular disease in Asian individuals with type 2 diabetes and shows potential to improve clinical risk assessment in diabetics. With the increasing diabetic population worldwide, further efforts to incorporate diabetic retinopathy into diabetes-specific cardiovascular disease risk assessment is recommended.