Objective: To summarise effects of intravenous bisphosphonates (IVBP) in patients with symptomatic knee osteoarthritis (OA) and bone marrow lesions (BMLs), using a meta-analysis of randomized controlled trials (RCTs).

Methods: Literature databases were searched for placebo-controlled RCTs of IVBPs for knee OA from inception, and included validated pain and function scales, BML size, and incidence of adverse events. Efficacy was compared using standardized mean differences (SMD) and risk ratios (RR) with fixed-effect or random-effects models. Methodological quality was assessed using the Cochrane risk of bias tool, heterogeneity was assessed by I² statistics.

Results: We included 428 patients in 4 RCTs of 2–24 months duration; most patients (84%) received zoledronic acid (ZA). Risk of bias was low-moderate. IVBP had large effect sizes on pain within 3 months (SMD= -2.33 (95% confidence interval= -3.02, -1.65)) mainly driven by neridronate (resulting in substantial heterogeneity, I²=92%) with no effect for ZA alone. Differences in knee function were statistically significant at 3 months (SMD=-0.22 (-0.43, -0.01), I²=0.2%). Effect sizes for pain did not reach statistical significance at any other time point. IVBPs improved a semi-quantitative measure of BML size within 6 months (SMD= -0.52 (-0.89, -0.14), I²=0%) but not at 12 months or two years. Adverse events (RR = 1.19 (1.00, 1.41) I²=52%), occurred more frequently with IVBP.

Conclusion: ZA has no effect on knee pain, possibly a short-term effect on BML size and higher rates of adverse events. Neridronate may improve pain in the short term, but this is based on a single trial.