June 2013
Volume 54, Issue 15
Free
ARVO Annual Meeting Abstract  |   June 2013
Novel and Traditional Biomarkers of Diabetic Retinopathy Severity: Multi-category Classifications Modeling
Author Affiliations & Notes
  • Wan Ling Wong
    Department of Ophthalmology, National University Health System, Singapore, Singapore
    Department of Ophthalmology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore
  • Jialiang Li
    Department of Statistics and Applied Probability, National University of Singapore, Singapore, Singapore
  • Xiang LI
    Singapore Eye Research Institute, Singapore National Eye Centre, Singapore, Singapore
    Department of Statistics and Applied Probability, National University of Singapore, Singapore, Singapore
  • Ecosse Lamoureux
    Singapore Eye Research Institute, Singapore National Eye Centre, Singapore, Singapore
    Centre for Eye Research Australia, Royal Victorian Eye and Ear Hospital, University of Melbourne, Melbourne, VIC, Australia
  • Carol Cheung
    Singapore Eye Research Institute, Singapore National Eye Centre, Singapore, Singapore
  • Tien Wong
    Department of Ophthalmology, National University Health System, Singapore, Singapore
    Singapore Eye Research Institute, Singapore National Eye Centre, Singapore, Singapore
  • Footnotes
    Commercial Relationships Wan Ling Wong, None; Jialiang Li, None; Xiang LI, None; Ecosse Lamoureux, None; Carol Cheung, None; Tien Wong, Allergan (C), Bayer (C), Novartis (C), Pfizer (C), GSK (F), Roche (F)
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science June 2013, Vol.54, 1533. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to Subscribers Only
      Sign In or Create an Account ×
    • Get Citation

      Wan Ling Wong, Jialiang Li, Xiang LI, Ecosse Lamoureux, Carol Cheung, Tien Wong; Novel and Traditional Biomarkers of Diabetic Retinopathy Severity: Multi-category Classifications Modeling. Invest. Ophthalmol. Vis. Sci. 2013;54(15):1533.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose: Most previous studies have examined biomarkers of diabetic retinopathy (DR) based on absence or presence of disease. We evaluated if novel biomarkers (serum creatinine, C-reactive protein (CRP) and retinal vascular imaging parameters) add discriminative value beyond traditional risk factors for identifying increasing severity levels of DR.

Methods: We used data from the Singapore Malay Eye Study, a population-based, survey of 3,280 (78.7% response) Malays aged 40 to 80 years. DR was graded from fundus photographs using the modified Airline House classification system and categorized as none/minimal (L1), mild/moderate (L2), and severe/vision threatening (L3). Blood samples collected were measured for serum creatinine and CRP. Retinal vascular parameters were measured quantitatively using a semi-automated computer-based program. Support Vector Machines (SVMs) is a machine learning method that performs classification tasks by constructing hyper-planes in a multi-dimensional space to separate outcome categories. Hyper-volume under the receiver operating characteristic curve manifold was used as inputs to non-linear SVMs to assess the accuracy and discrimination ability of novel biomarkers versus traditional risk factors (age, gender, body mass index, systolic blood pressure, hemoglobin A1c, low-density lipoprotein cholesterol and diabetes duration) for classifying DR severity.

Results: 740 diabetic participants were analyzed (582 L1; 87 L2; 71 L3). The discrimination ability, which is the probability of correctly classifying three random subjects from the population, each from one of the three stages of DR severity using traditional risk factors was 62%. The addition of novel biomarkers (serum creatinine, CRP, retinal venular tortuosity, fractal dimension, retinal arteriolar and venular caliber) increased the overall discriminating power of DR severity by 23% to 85%. Compared to traditional risk factors at specific DR severity levels, serum creatinine and CRP together have greater relative contribution (24%) to the increase in discriminatory ability of L3 DR while retinal arteriolar diameter and fractal dimension have greater relative contribution (22% and 10% respectively) to L2 DR.

Conclusions: The addition of novel biomarkers of serum creatinine, CRP and retinal vascular imaging parameters improves the discrimination and classification accuracy of DR severity levels.

Keywords: 459 clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology  
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×