The diagnosis of early-stage osteoarthritis (eOA) is important in disease management and outcomes. Herein we report the clinical validation of a blood test for the diagnosis of eOA in a large patient cohort using trace-level glycated and oxidized amino acid analytes. Subjects were recruited and enrolled in two study groups: subjects with eOA of the hip (n = 110) and asymptomatic controls (n = 120). Their plasma was analyzed for glycated and oxidized amino acids by quantitative liquid chromatography-tandem mass spectrometry. Algorithms were developed using plasma hydroxyproline and 12 glycated and oxidized amino acid analyte features to classify the subjects with eOA and asymptomatic controls. The accuracy was defined as the percentage of the subjects correctly classified in the test set validation. The minimum number of analyte features required for the optimum accuracy was five glycated amino acid analytes: Nω-carboxymethyl-arginine, hydroimidazolones derived from glyoxal, methylglyoxal and 3-deoxyglucosone, and glucosepane. The classification performance metrics included an accuracy of 95%, sensitivity of 96%, specificity of 94%, area under the curve of the receiver operating characteristic curve of 99%, and positive and negative predictive values of 94% and 97%. We concluded that an assay of five trace-level glycated amino acids present in plasma can provide a simple blood test for the screening of eOA. This is predicted to improve the case identification for expert referral 9-fold.
Keywords: glycation; machine learning; osteoarthritis; oxidative stress.