A clinical model including protein biomarkers predicts radiographic knee osteoarthritis: a prospective study using data from the Osteoarthritis Initiative

Abstract

Objective: We aimed to provide a model to predict the prospective development of radiographic KOA (rKOA).

Method: Baseline sera from 333 non-radiographic KOA subjects belonging to OA Initiative (OAI) who developed or not, rKOA during a follow-up period of 96 months were used in this study. The exploratory cohort included 200 subjects, whereas the replication cohort included 133. The levels of inter-alpha trypsin inhibitor heavy chain 1 (ITIH1), complement C3 (C3) and calcyclin (S100A6), identified in previous large proteomic analysis, were analyzed by using sandwich immunoassays on suspension bead arrays. The association of protein levels and clinical covariates with rKOA incidence was assessed by combining logistic regression analysis, Receiver Operating Characteristic (ROC) analysis, Integrated Discrimination Improvement (IDI) analysis and Kaplan-Meier curves.

Results: Levels of ITIH1, C3 and S100A6 were significantly associated with the prospective development of rKOA, showing an area under the curve (AUC) of 0.713 (0.624-0.802), 0.708 (0.618-0.799) and 0.654 (0.559-0.749), respectively to predict rKOA in the replication cohort. The inclusion of ITIH1 in the clinical model (age, gender, BMI, previous knee injury and WOMAC pain) improved the predictive capacity of the clinical covariates (AUC = 0.754 [0.670-0.838]) producing the model with the highest AUC (0.786 [0.705-0.867]) and the highest IDI index (9%). High levels of ITIH1 were also associated with an earlier onset of the disease.

Conclusion: A clinical model including protein biomarkers that predicts incident rKOA has been developed. Among the tested biomarkers, ITIH1 showed potential to improve the capacity to predict rKOA incidence in clinical practice.

Keywords: ITIH1; Incidence; Knee osteoarthritis; Predictive model.