Background Currently, magnetic resonance imaging techniques have been developed for evaluation of cartilage damage in early-stage osteoarthritis (eOA). These imaging techniques have approximately 70% sensitivity and 90% specificity compared to reference diagnosis by arthroscopy. They require expensive instrumentation, time and facilities. In searching for biomarkers for clinical diagnosis we found that the proteolysis of damaged (oxidised, glycated and nitrated) proteins gives a unique pattern in the plasma of arthritic patients with the severity of the disease – trace level oxidised, glycated and nitrated amino acids. In this study we developed a two-step algorithm using these analytes as features.
Objectives The objective was to distinguish between the following four groups: healthy control, eOA, early stage rheumatoid arthritis (eRA) and other inflammatory joint disease (non-RA).
Methods Four algorithm types were tested for performance using random forests, multiclass logistic regression, multi-class sparse logistic regression and support vector machines. In all cases, the diagnostic algorithms were trained on the training data set, before being used to predict the disease class for each sample in the test data set. A two-stage approach was taken: (i) to distinguish between disease and healthy control; and (ii) to distinguish between eOA, eRA and non-RA. The area under the curve of the receiver operating characteristic plot (AUROC) statistic was used as measure of performance
Results Random forest was the best-performing method. Application of two algorithms consecutively gave the best diagnostic outcome. The AUROC (sensitivity/specificity) values for disease/health were: eOA, 0.99 (0.92/0.91); eRA, 0.96 (0.89/0.90) and non-RA, 0.77 (0.73/0.72) for the training set and test set validations. A random outcome is 0.50. For typing arthritis, eOA, eRA and non-RA, AUROC values were in the range 0.68–0.98, 0.77–0.93 and 0.62–0.91 for training set and test set cross-validations and test set validation, respectively. A random outcome is 0.33.
Conclusions A two-step algorithm approach based on trace level damaged amino acids gave better diagnostic performance that MRI for detecting and typing eOA. It is low cost and suitable for rollout as a clinical screening test.
Menashe, L., Hirko, K., Losina, E., Kloppenburg, M., Zhang, W., Li, L. and Hunter, D. J. (2012) The diagnostic performance of MRI in osteoarthritis: a systematic review and meta-analysis. Osteoarthritis and Cartilage. 20, 13–21.
Ahmed, U., Anwar, A., Savage, R.S., Thornalley, P.J. and Rabbani, N. (2016) Application of protein glycation, oxidation and nitration for early-stage diagnosis and severity of osteoarthritis and other arthritic disease. Arthritis Research & Therapy 18:250.
Disclosure of Interest None declared