← Back

GINKGO GEM

Year of Publication: 2008

Journal: JAMA

Link: https://doi.org/10.1001/jama.2008.683

PDF: https://jamanetwork.com/journals/jama/articlepdf/182854

Clinical Question

Does Ginkgo biloba 120 mg twice daily prevent incident all-cause dementia or Alzheimer disease in elderly adults with normal cognition or MCI?

Bottom Line

In 3,069 community-dwelling adults ≥75 years (normal cognition n=2,587 or MCI n=482) randomized to Ginkgo biloba 120 mg bid or placebo and followed median 6.1 years, Ginkgo did NOT reduce incident all-cause dementia (HR 1.12, 95% CI 0.94-1.33; p=0.21) or AD (HR 1.16, 95% CI 0.97-1.39; p=0.11). Progression from MCI to dementia also unaffected (HR 1.13; p=0.39). Definitively refuted Ginkgo for dementia prevention; $249 million/year US market for this indication proved unfounded.

        Major Points
        Multicenter randomized double-blind placebo-controlled clinical trial at 5 US academic medical centers (GEM Study, DeKosky JAMA 2008)
N=3,069 community-dwelling adults ≥75 years with normal cognition (n=2,587) or mild cognitive impairment (n=482)
1:1 randomization to Ginkgo biloba EGb 761 extract 120 mg bid vs matching placebo
Primary endpoint: incident all-cause dementia by expert panel consensus (DSM-IV + NINCDS-ADRDA)
Median follow-up 6.1 years — largest and longest Ginkgo dementia prevention trial
No reduction in all-cause dementia: HR 1.12 (95% CI 0.94-1.33); p=0.21
No reduction in Alzheimer disease specifically: HR 1.16 (95% CI 0.97-1.39); p=0.11
No effect on progression from MCI to dementia: HR 1.13 (95% CI 0.85-1.50); p=0.39
No benefit in any prespecified subgroup (age, sex, APOE ε4 status, baseline cognition)
Safety balanced: no signal for bleeding, cardiovascular events, or hepatotoxicity
Low dropout (6.3%) and good adherence support validity of negative result
Definitively refuted Ginkgo biloba for dementia prevention; confirmed by parallel GuidAge 2012 (France)

      

Design

Study Type: Multicenter randomized double-blind placebo-controlled primary-prevention trial

Randomization: 1

Blinding: Double-blind

Follow-up Duration: Median 6.1 years

Sample Size: 3069

Analyzed: 3069

Analysis: Cox proportional hazards

Baseline Characteristics

Characteristic	Control	Active
N	1524	1545
Age mean	~79	~79
MMSE	~28	~28
MCI	~16%	~16%
APOE ε4 carrier	~20%	~20%

Arms

Field	Control	Ginkgo biloba
N	1524	1545
Intervention	Matching placebo twice daily	EGb 761 Ginkgo biloba extract 120 mg twice daily
Duration	Median 6.1 years	Median 6.1 years

Outcomes

Outcome	Type	Control	Intervention	HR / OR / RR	P-value
Time to incident all-cause dementia (expert panel consensus)	Primary	246 events (2.9/100 person-years)	277 events (3.3/100 person-years)		p=0.21 (not significant)
Incident Alzheimer disease	Secondary	Reference	Slightly higher rate	HR 1.16 (95% CI 0.97-1.39)	p=0.11
Progression from MCI to dementia	Secondary	Reference	Similar	HR 1.13 (95% CI 0.85-1.50)	p=0.39
Incident vascular dementia	Secondary	Reference	Similar	Not significantly different	NS
Cognitive decline on 3MS (secondary analysis)	Secondary	Reference	No benefit		NS
Mortality	Secondary	Similar	Similar	No difference	NS
Any AE	Adverse	Similar	Similar		No significant difference
Bleeding	Adverse	Baseline rate	Not significantly increased		No bleeding signal
Cardiovascular events	Adverse	Baseline rate	Not significantly increased		No CV signal
GI symptoms	Adverse	Low	Low		Similar
Hepatotoxicity	Adverse	Rare	Rare		No signal
Serious AE	Adverse	Similar overall rate	Similar overall rate		No imbalance
Dropout/LTFU	Adverse	6.3% (pooled)	6.3% (pooled)		Low and balanced

Subgroup Analysis

No evidence of benefit in any prespecified subgroup: normal cognition at baseline, MCI at baseline, APOE ε4 carriers or noncarriers, age, sex. The HR consistently hovered around or slightly above 1.0 for all subgroups, firmly ruling out a meaningful preventive effect. A parallel trial in France (GuidAge, 2012) using the same extract in older adults with memory complaints also found no preventive effect — confirming the GEM finding.

Criticisms

Ginkgo 120 mg bid (240 mg/day) was the commonly marketed dose — higher doses not tested (dose-response possibility)
EGb 761 is one specific Ginkgo preparation; other formulations or plant parts could theoretically differ
Average 6.1-year follow-up may still be too short to capture preventive effect on pathology accumulating over decades
Patients already ≥75 may be too late for primary prevention; preclinical or midlife intervention not tested
Incident dementia diagnosis required symptom progression — could miss pre-symptomatic biomarker changes
Expert consensus diagnosis without universal MRI/amyloid biomarkers — less rigorous than current biomarker-based AD criteria

Funding

National Center for Complementary and Alternative Medicine (NCCAM), National Institute on Aging, Office of Dietary Supplements; EGb 761 donated by Schwabe Pharmaceuticals

Based on: GINKGO GEM (JAMA, 2008)

Content summarized and formatted by NeuroTrials.ai.

GINKGO GEM

(2008)

Objective

Ginkgo biloba extract 120 mg twice daily — to test whether long-term use prevents dementia (primarily Alzheimer disease) in elderly adults with normal cognition or mild cognitive impairment.

Study Summary

• Largest dementia prevention trial of Ginkgo biloba: 3,069 elderly adults, median 6.1 years follow-up.
• No reduction in all-cause dementia with Ginkgo 120 mg bid: HR 1.12 (95% CI 0.94-1.33; p=0.21).
• No reduction in Alzheimer disease specifically: HR 1.16 (95% CI 0.97-1.39; p=0.11).
• No effect on progression from MCI to dementia: HR 1.13 (95% CI 0.85-1.50; p=0.39).
• Adverse-event profile similar between groups; dropout rates low (6.3%).
• Definitive negative result refuted the $249M/year US market for Ginkgo in memory prevention.

Intervention

Ginkgo biloba extract 120 mg twice daily (EGb 761, Schwabe Pharmaceuticals) or identical placebo in blister packs, over median 6.1 years.

Inclusion Criteria

Community-dwelling adults aged ≥75 years with either normal cognition or mild cognitive impairment (Petersen criteria); willing to take twice-daily oral medication; no dementia at baseline.

Study Design

Arms: Ginkgo biloba 120 mg bid vs Placebo

Patients per Arm: Ginkgo 1,545; Placebo 1,524 (N=3,069)

Outcome

• Primary: All-cause dementia — 3.3/100 person-yr (Ginkgo) vs 2.9/100 py (placebo); HR 1.12 (95% CI 0.94-1.33); p=0.21 — NOT significant
• Alzheimer disease: HR 1.16 (95% CI 0.97-1.39); p=0.11
• Progression from MCI to dementia: HR 1.13 (95% CI 0.85-1.50); p=0.39
• AE profile similar between arms; no safety signal for bleeding or cardiovascular events
• Dropout/LTFU 6.3%; adherence good; 92% of incident dementia cases were AD or AD + vascular

Bottom Line

Major Points

Multicenter randomized double-blind placebo-controlled clinical trial at 5 US academic medical centers (GEM Study, DeKosky JAMA 2008)
N=3,069 community-dwelling adults ≥75 years with normal cognition (n=2,587) or mild cognitive impairment (n=482)
1:1 randomization to Ginkgo biloba EGb 761 extract 120 mg bid vs matching placebo
Primary endpoint: incident all-cause dementia by expert panel consensus (DSM-IV + NINCDS-ADRDA)
Median follow-up 6.1 years — largest and longest Ginkgo dementia prevention trial
No reduction in all-cause dementia: HR 1.12 (95% CI 0.94-1.33); p=0.21
No reduction in Alzheimer disease specifically: HR 1.16 (95% CI 0.97-1.39); p=0.11
No effect on progression from MCI to dementia: HR 1.13 (95% CI 0.85-1.50); p=0.39
No benefit in any prespecified subgroup (age, sex, APOE ε4 status, baseline cognition)
Safety balanced: no signal for bleeding, cardiovascular events, or hepatotoxicity
Low dropout (6.3%) and good adherence support validity of negative result
Definitively refuted Ginkgo biloba for dementia prevention; confirmed by parallel GuidAge 2012 (France)

Study Design

Study Type: Multicenter randomized double-blind placebo-controlled primary-prevention trial
Randomization: Yes
Blinding: Double-blind
Sample Size: 3069
Follow-up: Median 6.1 years

Primary Outcome

Definition: Time to incident all-cause dementia (expert panel consensus)

Control	Intervention	HR/OR	P-value
246 events (2.9/100 person-years)	277 events (3.3/100 person-years)	- (0.94 to 1.33)	p=0.21 (not significant)

Limitations & Criticisms

Ginkgo 120 mg bid (240 mg/day) was the commonly marketed dose — higher doses not tested (dose-response possibility)
EGb 761 is one specific Ginkgo preparation; other formulations or plant parts could theoretically differ
Average 6.1-year follow-up may still be too short to capture preventive effect on pathology accumulating over decades
Patients already ≥75 may be too late for primary prevention; preclinical or midlife intervention not tested
Incident dementia diagnosis required symptom progression — could miss pre-symptomatic biomarker changes
Expert consensus diagnosis without universal MRI/amyloid biomarkers — less rigorous than current biomarker-based AD criteria

Citation

View Original Publication →

GINKGO GEM

Clinical Question

Bottom Line

Major Points

Design

Baseline Characteristics

Arms

Outcomes

Subgroup Analysis

Criticisms

Funding

Ask about GINKGO GEM