59 - Predictors of Clinical Response Following Genicular Artery Embolization for Knee Osteoarthritis

Purpose: Patient-level predictors of treatment response after genicular artery embolization (GAE) for knee osteoarthritis (OA) are poorly understood. We evaluated candidate clinical, serum, and imaging biomarkers for their ability to predict achievement of the minimal clinically important difference (MCID) in WOMAC pain and stiffness scores following GAE.

Materials and Methods: Data from a prospective single-arm clinical trial of 25 patients who underwent GAE for symptomatic knee OA was retrospectively analyzed (1). Candidate predictors, selected a priori, included sex, age, BMI, contralateral Kellgren-Lawrence score, and baseline values for WOMAC pain, serum IL-1Ra, serum VEGF, and total bone marrow edema score on MRI using MOAKS methodology. The primary outcomes were frequency of achieving the MCID in WOMAC pain and WOMAC stiffness at 1, 3, and 12 months, modeled as an ordinal outcome (0–3). Ordinal logistic regression models were constructed. Variance inflation factors (VIFs) were assessed to detect multicollinearity, and leave-one-out cross-validation was performed to evaluate model robustness.

Results: All candidate predictors were successfully incorporated into regression models, with no evidence of multicollinearity by VIF analysis (1.157 ≤ VIF ≤ 2.035). Lower contralateral KL scores (OR: 0.087 [0.012-0.618], p=0.015) and higher BMI (OR :1.383 [1.001-1.910], p=0.049) were significantly associated with achievement of the MCID for WOMAC pain. Lower baseline serum IL-1Ra levels (OR: 0.122 [0.018-0.816], p=0.030) was significantly associated with achievement of the MCID for WOMAC stiffness. The remaining demographic, clinical, serum, and imaging biomarkers were not significantly associated with MCID achievement. In the cross-validation, contralateral KL score, serum IL-1Ra level, and BMI were significant in 96%, 88%, and 48% of the 25 replications, respectively, and the average and median p-values of these predictors in the nonsignificant replications were 0.062 and 0.058.

Conclusion: In this clinical trial cohort, specific baseline clinical and serum factors were associated with achievement of clinically meaningful improvements in pain and stiffness. Analysis of larger cohorts will help clarify ideal demographic-, biomarker- and imaging-based patient selection strategies which can improve prediction of treatment response and guide clinical decision-making in GAE for knee OA.