Is There a Risk of Bleeding Associated with Standardized Ginkgo biloba Extract Therapy? A Systematic Review and Meta-analysis

Study Objective. To determine the effect of standardized Ginkgo biloba leaf extracts (GBE) on outcome parameters of hemostasis associated with risk of bleeding. Design. Meta-analysis of 18 randomized controlled trials. Subjects. A total of 1985 adults were enrolled to receive either GBE or placebo; 87% were patients with dementia, peripheral artery disease, or diabetes mellitus; 13% were healthy volunteers. Measurements and Main Results. The MEDLINE, EMBASE, Cochrane Library, and SciSearch databases were searched from inception through 2009. The following outcome parameters of hemostasis were assessed: blood flow, blood viscosity, adenosine 5'-diphosphate (ADP)-induced platelet aggregation, fibrinogen concentration, activated partial thromboplastin time (aPTT), and prothrombin time (PT). Reference parameter values were taken into account when assessing clinical relevance of statistically significant treatment effects. The quality and risk of bias of each study were evaluated by using the Jadad score and Cochrane Collaboration tool, respectively. According to the continuous data, the following statistical methods were used: weighted mean difference (WMD), standardized mean difference (SMD), and generic inverse variance (GIV), with 95% confidence intervals (CIs). Random-effects models of effects on baseline change or mean difference showed a positive effect of GBE on blood perfusion, as shown by a significant reduction in blood viscosity (WMD -1.03 mPa degrees sec, 95% CI -1.29 to -0.78 mPa degrees sec), but no evidence of any significant effect on ADP-induced platelet aggregation (WMD -0.35%, 95% CI -15.16-14.46%), fibrinogen concentration (GIV -2.45 mg/dl, 95% CI -11.59-6.70 mg/dl), aPTT (GIV -0.42 sec, 95% CI -0.97-0.12 sec), and PT (SMD 0.00, 95% CI -0.09-0.09). Subgroup analyses revealed a statistically significant reduction in aPTT for subgroups receiving high-dose GBE of 240 mg/day or more (GIV -0.47 sec, 95% CI -0.88 to -0.05 sec) and for studies including only patients, not healthy volunteers (GIV -0.61 sec, 95% CI -0.95 to -0.27 sec); however, both findings were not clinically relevant. Conclusion. Based on meta-analysis of hemostasis outcomes, comparison of mean difference or baseline change between treatment and placebo groups did not indicate a higher bleeding risk associated with standardized GBE. This finding ultimately contributes to an informed evaluation of GBE use, including patient self-medication.