Ginkgo Biloba EGb 761: Why the Dementia Evidence Hasn’t Translated

← Back to articles

Ginkgo biloba is one of the top-selling and most-studied herbal extracts in the world, so the question is not whether it has been tested but what the testing has actually shown. The pattern is consistent and telling: the largest, best-designed prevention trials are flatly negative, while the more favorable signals come from smaller treatment trials of uneven quality.

The GEM trial and dementia prevention

The Ginkgo Evaluation of Memory (GEM) study — the largest trial of ginkgo ever conducted — randomized 3,069 community-dwelling adults aged 75 or older to EGb 761 120 mg twice daily (240 mg/day) or placebo, with a median follow-up of 6.1 years (DeKosky and colleagues, 2008). The primary endpoint was incident dementia. The result was unambiguously negative: 277 dementia cases on ginkgo versus 246 on placebo, a hazard ratio of 1.12 (95% CI 0.94–1.33), with no benefit for Alzheimer's disease specifically and no effect on progression in the subgroup that already had mild cognitive impairment. Ginkgo simply did not prevent dementia.

The European counterpart reached the same conclusion. The GuidAge trial (Vellas and colleagues, 2012) randomized 2,854 adults aged 70 or older with memory complaints to EGb 761 240 mg/day or placebo and followed them for five years. Conversion to probable Alzheimer's disease was 1.2 cases per 100 person-years on ginkgo versus 1.4 on placebo (hazard ratio 0.84, 95% CI 0.60–1.18) — not significant. Two large, long, well-conducted trials with consistent null results are about as definitive as prevention evidence gets: ginkgo does not lower dementia risk in older adults.

Mild cognitive impairment and symptomatic dementia

For people who already have mild cognitive impairment or symptomatic dementia, the picture is more favorable but considerably weaker in quality. A 2015 meta-analysis (Tan and colleagues) pooled nine trials of 22–26 weeks duration (2,561 patients) and found that EGb 761, mainly at 240 mg/day, produced statistically significant improvements in cognition, activities of daily living and clinician-rated global change versus placebo. A separate 2016 meta-analysis (Yang and colleagues, 21 trials, 2,608 patients) reported similar benefits but graded the underlying trials as moderate-to-poor in methodological quality and found results inconsistent across individual studies. The older Cochrane review (Birks and Grimley Evans, 2009) was more cautious still, judging the evidence for clinically meaningful benefit "inconsistent and unreliable." In short, there may be a small symptomatic effect, but it is smaller than that of acetylcholinesterase inhibitors and rests on a shaky evidence base.

Tinnitus and peripheral vascular disease

Ginkgo has been tested extensively in tinnitus, with meta-analyses producing mixed or negative results; it is not a reliable treatment. For intermittent claudication (peripheral arterial disease), a 2013 Cochrane review (Nicolai and colleagues) found that ginkgo extended pain-free walking distance only slightly compared with placebo — a difference too small to be clinically meaningful, and far less than supervised exercise therapy delivers.

The EGb 761 preparation is specific

Almost all of the positive evidence refers specifically to EGb 761 (sold as Tanakan and Tebonin) — a proprietary extract standardized to about 24% flavonoid glycosides and 6% terpene lactones, with ginkgolic acids kept below 5 ppm. Generic ginkgo products vary enormously in specification and contamination, and cannot be assumed to reproduce the trial results. This is a recurring theme in botanical research: the evidence attaches to a defined extract, not to the plant in general.

Safety

Across the large trials, ginkgo's overall adverse-event and mortality profile was similar to placebo — in GEM and GuidAge, rates of death, stroke and serious bleeding did not differ significantly between groups. The clinically important concern is its antiplatelet activity: case reports document post-surgical and intracranial bleeding, and interactions are plausible with warfarin, direct oral anticoagulants, and aspirin. It is prudent to stop ginkgo about two weeks before surgery. Caution is also warranted with anticonvulsants (a theoretical lowering of seizure threshold). At standard doses in otherwise healthy users, ginkgo is generally well tolerated; the drug interactions matter more than any intrinsic toxicity.

Bottom line

Ginkgo's reputation rests on a body of research that, read carefully, does not support its biggest marketing claim. Two large randomized trials show it does not prevent dementia or Alzheimer's disease. For people who already have cognitive impairment, meta-analyses suggest a small symptomatic benefit, but the trials are of modest quality and the effect is smaller than standard drugs. For tinnitus and claudication the evidence is weak. If used, it should be a standardized EGb 761 product, and not by anyone on antiplatelet or anticoagulant therapy or approaching surgery.