Purchase this article with an account.
Henrietta Ho, Peng Guan Ong, Charumathi Sabanayagam, Paul Mitchell, E Shyong Tai, Tien Yin Wong, Carol Yim-lui Cheung; Retinal Microvascular Parameters Improve The Prediction and Reclassification of Cardiovascular Disease Risk in Asian Diabetics. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):2016.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Retinal microvascular changes like retinal arteriolar narrowing, are measures of subclinical vascular disease. The aim of our study is to develop a cardiovascular diseases (CVD) prediction algorithm, and investigate the added prognostic value of retinal microvascular changes on traditional risk factors, in an Asian cohort with diabetes.
We recruited subjects from a prospective, population-based cohort, the Singapore Malay Eye Study (SiMES). The main outcome measure was incident CVD event, defined as myocardial infarction, stroke and CVD mortality, documented after baseline examination. Retinal arteriolar and venular calibers, retinal vascular fractal dimension and the presence of retinopathy were assessed from fundus photography; these were added to a basic risk prediction model derived from Cox regression analysis that included standard CVD risk factors. We validated the risk-prediction model with 10-fold cross validation internally, then externally in another independent prospective Asian cohort, the Singapore Prospective Study (SP2). The C statistics and net reclassification index (NRI) were calculated.
709 diabetics without CVD at baseline in SiMES were included and followed-up for 6.05±6.33 years. A total of 86 cases (12.13%) of CVD occurred. The best fitting basic model included age, gender, systolic blood pressure, lipids, diabetic control (HbA1c), renal function (creatinine) and inflammatory biomarker (high-sensitivity C-reactive protein). Addition of retinal microvascular parameters improved discrimination (C-statistics 0.71 to 0.75, p=0.011) and overall reclassification, mainly of subjects with no CVD event from a higher to a lower risk category (NRI=11.1%, p<0.02). Internal validation of the new model revealed better performance (C-statistic 0.69 to 0.73, p=0.02) and overall discrimination (NRI=12.9%, p=0.04). Among the 454 subjects with diabetes from SP2, 32 cases (7.0%) had an incident CVD event. In this external cohort, the new model performed consistently (C-statistics=0.74 to 0.79, p=0.042), and the total NRI was quantitatively similar to SiMES, and significant (NRI=18.1%, p=0.03).
Retinal microvascular changes measured from retinal photographs have incremental value in reclassifying CVD risk beyond traditional risk factors in diabetics. Future studies are needed to validate our findings in the wider population.
This PDF is available to Subscribers Only