Statistical Correction of the Area Under the ROC Curve in the Presence of Random Measurement Error and Applications to Biomarkers of Oxidative Stress

Abstract

Despite the elegant patho-physiological mechanisms that have been identified linking lipid peroxidation to the development of atherosclerosis, the clinical evidence linking markers of oxidative stress to coronary heart disease is still controversial (1-4) in particular for thybarbituric acid reactive substances (TBARS). The reasons for this lack of consistency in research findings may be many and include: 1) the use of small and selected samples; 2) differences in methods used to quantify TBARS; and 3) lack of adjustment for potential confounders. The inconsistency in findings together with issues related to the lack of specificity of TBARS as a measurement of lipid peroxidation have prompted the questioning of TBARS measurement as a marker of lipid peroxidation (5). The use of TBARS as a summarizing value of total circulating oxidative stress in individuals is popular in laboratory research; however, its use is still controversial because of the lack of a uniform methodology between researchers to estimate TBARS levels.