Serum proteomic panel validated for prediction of knee osteoarthritis progression
Virginia B. Kraus, Alexander Reed, Erik J. Soderblom, M. Arthur Moseley, M.-F. Hsueh, Mukundun G. Attur, Jonathan Samuels, Steven B. Abramson, Yi‐Ju Li
Abstract
Objective: To further validate a serum proteomics panel for predicting radiographic (structural) knee OA progression. Design: Serum peptides were targeted by multiple-reaction-monitoring mass spectrometry in the New York University cohort (n = 104). Knee OA progression was defined as joint space narrowing ≥1 in the tibiofemoral compartment of one knee per study participant over a 24-month follow-up. The discriminative ability of an 11-peptide panel was evaluated by multivariable logistic regression and area under the receiver operating characteristic curve (AUC), without and with demographic characteristics of age, sex, and body mass index. The association of each peptide with OA progression was assessed by odds ratios (OR) in multivariable logistic regression models adjusted for demographics. Results: The cohort included 46 (44%) knee OA progressors. The panel of 11 peptides alone yielded AUC = 0.66 (95% CI [0.55, 0.77]) for discriminating progressors from non-progressors; demographic traits alone yielded AUC = 0.66 (95% CI [0.55, 0.77]). Together the 11 peptides and demographics yielded AUC = 0.72 (95% CI [0.62, 0.83]). CRAC1 had the highest odds for predicting OA progression (OR 2.014, 95% CI [0.996, 4.296], p = 0.058). Conclusions: We evaluated a parsimonious serum proteomic panel and found it to be a good discriminator of knee radiographic OA progression from non-progression. Since these biomarkers are quantifiable in serum, they could be deployed relatively easily to provide a simple, cost-effective strategy for identifying and monitoring individuals at high risk of knee OA progression.