Retinopathy Signs Improved Prediction and Reclassification of Cardiovascular Disease Risk in Diabetes: A prospective cohort study

• Published in: Scientific reports Vol. 7; no. 1; p. 41492
• Main Authors: Ho, Henrietta, Cheung, Carol Y., Sabanayagam, Charumathi, Yip, Wanfen, Ikram, Mohammad Kamran, Ong, Peng Guan, Mitchell, Paul, Chow, Khuan Yew, Cheng, Ching Yu, Tai, E. Shyong, Wong, Tien Yin
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 02.02.2017; Nature Publishing Group
• Subjects: 692/163/2743/137; 692/53/2423; 692/699/75; Biochemical markers; Biomarkers; Biomarkers - blood; C-reactive protein; Cardiovascular diseases; Cardiovascular Diseases - blood; Cardiovascular Diseases - epidemiology; Cohort analysis; Diabetes; Diabetes mellitus; Diabetic Retinopathy - blood; Diabetic Retinopathy - complications; Epidermal growth factor receptors; Eye; Female; Glomerular filtration rate; Health risk assessment; Health risks; Humanities and Social Sciences; Humans; Male; Microvasculature; Microvessels - pathology; Middle Aged; multidisciplinary; Prediction models; Prospective Studies; Reclassification; Regression analysis; Reproducibility of Results; Retina; Retinal Vessels - pathology; Retinopathy; Risk Factors; Science; Science (multidisciplinary)
• ISSN: 2045-2322; 2045-2322
• DOI: 10.1038/srep41492

CVD risk prediction in diabetics is imperfect, as risk models are derived mainly from the general population. We investigate whether the addition of retinopathy and retinal vascular caliber improve CVD prediction beyond established risk factors in persons with diabetes. We recruited participants from the Singapore Malay Eye Study (SiMES, 2004–2006) and Singapore Prospective Study Program (SP2, 2004–2007), diagnosed with diabetes but no known history of CVD at baseline. Retinopathy and retinal vascular (arteriolar and venular) caliber measurements were added to risk prediction models derived from Cox regression model that included established CVD risk factors and serum biomarkers in SiMES, and validated this internally and externally in SP2. We found that the addition of retinal parameters improved discrimination compared to the addition of biochemical markers of estimated glomerular filtration rate (eGFR) and high-sensitivity C-reactive protein (hsCRP). This was even better when the retinal parameters and biomarkers were used in combination (C statistic 0.721 to 0.774, p = 0.013), showing improved discrimination, and overall reclassification (NRI = 17.0%, p = 0.004). External validation was consistent (C-statistics from 0.763 to 0.813, p = 0.045; NRI = 19.11%, p = 0.036). Our findings show that in persons with diabetes, retinopathy and retinal microvascular parameters add significant incremental value in reclassifying CVD risk, beyond established risk factors.