Predictive Value of the Systemic Immune-Inflammation Index for the Clinical Efficacy of Acupuncture and Exercise Rehabilitation in Knee Osteoarthritis: A Retrospective Study.

objectiveTo evaluate the predictive value of the systemic immune-inflammation index (SII) for clinical efficacy of acupuncture combined with exercise rehabilitation in patients with knee osteoarthritis (KOA).

methodsIn this retrospective observational study, 151 patients with radiographically confirmed Kellgren-Lawrence Grade II or III KOA who completed an 8-week standardized protocol of acupuncture and exercise rehabilitation were enrolled. Clinical efficacy was defined as the attainment of minimal clinically important improvement (MCII) in both the 11-point Numeric Rating Scale (NRS) for pain (≥ 2-point reduction) and the 68-point WOMAC function subscale (≥ 6-point improvement, Likert version 3.1). Baseline demographic, clinical, and hematological data were collected, and SII was subsequently calculated from baseline hematological parameters. Uni- and multivariate logistic regression analyses were used to identify independent predictors of treatment response. Receiver operating characteristic (ROC) curves were constructed to assess the discriminative ability of SII, body mass index (BMI), and lymphocyte count, individually and in combination.

resultsOf the 151 patients, 73 were classified as responders and 78 as nonresponders. Baseline BMI and SII were significantly lower, while lymphocyte count was higher in the effective group (all p < 0.001). Multivariate logistic regression identified BMI (OR = 0.596, 95% CI: 0.472-0.753), lymphocyte count (OR = 34.597, 95% CI: 3.234-370.131), and SII (OR = 0.912, 95% CI: 0.873-0.953) as independent predictors of treatment response. ROC analysis showed that SII demonstrated only moderate predictive power (AUC = 0.749, 95% CI: 0.670-0.828), with improved but still moderate accuracy when combined with BMI and lymphocyte count (AUC = 0.858, 95% CI: 0.799-0.917).

conclusionAlthough SII is an independent and accessible biomarker for predicting clinical efficacy, its discriminative ability is moderate. A combined model incorporating SII, BMI, and lymphocyte count may provide better, though not strong, predictive performance and could support clinical decision-making and individualized treatment planning. However, the retrospective single-center design may introduce selection bias and limit generalizability. Residual confounding from unmeasured variables cannot be excluded, and the absence of longitudinal SII measurements limits evaluation of temporal changes. Future multicenter prospective studies are warranted to validate these findings.