Background
The standardized Ginkgo biloba extract EGb 761
® is one of the most widely used herbal remedies for dementia and cognitive impairment and remains one of the best evaluated and characterized extracts [
1]. Since 2000, according to the current ATC-classification, Ginkgo biloba special extract is listed in the group of anti-dementia drugs together with cholinesterase inhibitors and memantine. However, most efficacy and effectiveness studies are small, suffer from methodological limitations and are subject to considerable controversy.
Recently, the discussion about the benefits of Ginkgo biloba extracts for different indications has again been revitalized after the publication of two major trials. One study [
2] with 176 participants could not demonstrate any evidence of effectiveness of 120 mg Ginkgo biloba in mild to moderate dementia. In addition, the Ginkgo Evaluation of Memory (GEM) study [
3] found no favorable effects of 240 mg EGb 761
® for the prevention of dementia onset in older people without or with only mild cognitive impairment. Whereas the GEM trial was large enough to prove the robust result of a lacking effect of Ginkgo biloba in dementia prevention and overall mortality in people without dementia [
4], the randomized controlled trial (RCT) of McCarney et al. has been criticized due to methodological problems and insufficient sample size [
5,
6].
Most previous reviews have shown inconsistent results and fail to draw firm conclusions whether Ginkgo biloba has patient-relevant benefits in people with a diagnosis of dementia [
7,
8]. In the meantime, new studies have been published. A major limitation of the available Cochrane Review on the effectiveness of Ginkgo biloba is the combined evaluation of cognitive decline and dementia. No subgroup analyses were performed [
7]. As no valid definition of cognitive impairment has been used, and given the finding of differential effects of other anti-dementia drugs on dementia and cognitive impairment [
9], it appears no longer reasonable to pool these two indications. Furthermore, the uncertainty regarding mild cognitive impairment (MCI) as a clinical entity raises the question as to the scientific validity of MCI trials. In addition, the German Health Technology Assessment Institute IQWiG (Institute for Quality and Efficiency in Health Care) published a favorable report on the effectiveness of Ginkgo biloba, which was, however, limited to Alzheimer's disease and contradicted the Cochrane review [
10]. Taking into account the overlap between different types of dementia [
11] and the limited significance of dementia subtyping in routine use of anti-dementia drugs, there is no clear evidence for substantial differences in the effectiveness of those drugs in vascular or Alzheimer's dementia. As most anti-dementia drugs including ginkgo are prescribed without comprehensive assessment of the dementia subtype, and given continuing high prescription rates of ginkgo to people with an established dementia diagnosis in some countries, we felt that further decision support is necessary for this kind of routine use of the substance. Therefore, we performed a systematic review on the effects of Ginkgo biloba in Alzheimer's disease as well as vascular and mixed dementia covering a variety of outcome domains.
Methods
Data sources
We used a sensitive search strategy based on Cochrane and other systematic reviews in the field. We searched the following databases: MEDLINE (January 1, 1966, to August 2008), EMBASE (January 1, 1980 to August 2008), PsycINFO (January 1, 1982, to August 2008), CINAHL (Cumulative Index to Nursing and Allied Health Literature), the Cochrane Database of Systematic Reviews, and the Cochrane Controlled Trials Register (until Issue 3, 2008). The following search terms were used for dementia (with * characterizing a wildcard, and the items AND and OR being used as boolean functions): dementia; (senile* AND dement*); Alzheimer*; ((cognit* OR memory* OR mental*) AND (decline OR impair* OR los* OR deteriorat* OR diminish* OR insufficien* OR degenerat*)). The following search terms were used for Ginkgo biloba: ginkgo; ginko; gingko; bilobalid*; tebonin; egb 761; li 1370. To identify clinical trials, we used the following search terms: (clinical AND trial*); random*; placebo*; (controlled AND trial*); (multicent* AND stud*); (comparative AND stud*); follow-up; (research AND design). All search terms within the indication (dementia), ginkgo biloba and clinical trials section were searched individually in each database and combined together using the "OR" boolean. The results of each of the three sections were then combined by utilizing the "AND" boolean. Furthermore, the reference sections of systematic and narrative reviews were screened for primary publications. In addition, the manufacturer of EGb 761®, Dr. Willmar Schwabe GmbH & Co. KG, Karlsruhe, Germany, was queried and information was requested as needed.
Selection and study characteristics
We selected controlled clinical trials, with or without randomization, assessing the effects of treating people with a diagnosis of Alzheimer's disease, vascular or mixed dementia according to internationally valid diagnostic criteria, with a standardized Ginkgo biloba extract. Further study inclusion criteria were the use of an internationally accepted diagnostic for the dementia diagnosis such as the International Classification of Diseases (ICD) [
12], the Diagnostic and Statistical Manual of Mental Disorders (DSM) [
13], the NINCDS-ADRDA (National Institute of Neurological and Communicative Disorders and Stroke, Alzheimer's Disease and related Disorders Association) [
14], or the NINDS-AIREN criteria (National Institute for Neurological Disorders and Stroke and Association Internationale pour la Recherche et l'Enseignement en Neurosciences) [
15]; a minimum treatment duration of 12 weeks; a minimum number of participants of ten per group; and the availability of a full-text publication. Exclusion criteria were studies with a majority of people with specific types of non-vascular and non-Alzheimer's dementia, such as Lewy-body dementia or dementia due to Parkinson's disease; and a publication language other than English, German, French, Italian or Spanish.
Data extraction and critical appraisal of studies
Study selection and appraisal of studies were performed independently by two researchers (SW and SR). Disagreement was resolved by consensus. Information was extracted via a standardized checklist included study design, duration of the study, comparability of study groups with respect to sex, age, baseline scores of cognitive, psychopathological and activities of daily living scales and clinical outcomes on different rating scales in the following domains: cognition, activities of daily living (ADLs), neuropsychiatric symptoms, patients' quality of life, response according to the studies' definitions and drop out rates due to side effects. The study and publication quality was assessed on the basis of whether the following quality criteria had been adequately fulfilled: randomization and allocation concealment, blinding of patients and investigators, sample size estimation, handling and reporting of study discontinuations, application of the intent-to-treat principle, relevant data inconsistencies, and report of study funding. All clinical outcomes were assessed in the intent-to-treat samples. An intent-to-treat analysis was accepted when a last-observation-carried forward analysis was performed, or when no drop-outs occurred and all patients were included in the evaluation.
The following rating scales were accepted for clinical outcomes: (1) cognition: Alzheimer's Disease Assessment Scale, cognitive subscale (ADAS-cog) [
16], Syndrom-Kurztest (SKT) [
17]; (2) ADLs: GERRI [
18], Nürnberger Alters-Alltagsaktivitäten-Skala (NAA) and Nürnberger Alters-Beobachtungsskala (NAB) [
19], GBS-ADL
(Gottries-Bråne-Steen - Activities of daily living scale) [
20], ADL-IS (Alzheimer's Disease Activities-of-Daily-Living International Scale) [
21], (3) neuropsychiatric symptoms: Geriatric Depression Scale (GDS) [
22], Hamilton Depression Rating Scale (HAMD) [
23], Montgomery Asberg Depression Rating Scale (MADRS) [
24] and Neuropsychiatric Inventory (NPI) [
25]; and (4) quality of life: Quality of Life in Alzheimer's Disease (QOL-AD) [
26], DEMQOL-Proxy (Quality of Life Questionnaire for people with dementia, rated by caregivers) [
27], PDS (Progressive Deterioration Scale) [
28]. In addition, we intended to pool the response rates with response definitions as used in the studies.
Statistical analysis
If feasible and meaningful, data were pooled by means of meta-analysis. Effect measures presented in this publication were reported as Mantel Haenszel risk ratios (RRs) including 95% confidence intervals (CIs) for binary data. Effects on rating scales were expressed as standardized mean differences (SMDs) with the 95% CIs, N and p values. A random-effects model was used to calculate a pooled effect estimate, because we expected considerable heterogeneity. When different dose levels were used in different study arms, the study arm with the higher dosage was chosen, and a sensitivity analysis was performed. Further sensitivity analyses were performed for type of dementia (Alzheimer's dementia versus vascular or mixed dementia) and exclusion of studies with a high number of participants without dementia.
Statistical significance was assumed in case of p < 0.05. Heterogeneity of effect sizes was evaluated by the I2 statistic. An alpha error p < 0.05 and/or I2 of at least 50% were taken as indicators of substantial heterogeneity of outcomes. In this case, we examined the effect of removing single studies with special study designs on the results and on the heterogeneity. If meta-analyses were not possible, the results of individual studies are presented. Meta-analyses were performed using Review Manager, version 5 (The Cochrane Collaboration, Oxford, England) for all calculations.
Role of funding source
Funding for this review was provided via an unrestricted grant from Dr. Willmar Schwabe GmbH & Co. KG, Karlsruhe, Germany.
Discussion
This review was performed because the efficacy and effectiveness of Ginkgo biloba in dementia treatment has been an issue of controversy against a background of an increasing individual and societal burden due to these disorders, and given only moderate effects of cholinesterase inhibitors and memantine [
38,
39]. With only symptomatic treatments available, the clinical relevance of all anti-dementia drugs would be to improve cognitive and neuropsychological symptoms and activities of daily living and/or to delay deterioration.
We found a statistically significant advantage of Ginkgo biloba compared to placebo in improving cognition for the whole group of patients with Alzheimer's disease, vascular or mixed dementia. Regarding activities of daily living, there was no significant difference for the whole dementia group. However, in the subgroup of patients with Alzheimer's disease, the advantage of Ginkgo biloba compared to placebo was statistically significant. The benefit of Ginkgo biloba in overall response rates was quite robust when all response definitions of the studies were accepted. However, the available response definitions differed considerably. Therefore, we did not pool the results.
There was no consistent evidence of a benefit of Ginkgo biloba in the treatment of neuropsychiatric symptoms. However, the presence of psychological or behavioral symptoms in dementia may be an effect modifier. Two studies that included patients with neuropsychiatric features yielded a clinically relevant effect in this outcome domain. Based on our results, no solid conclusions can be given for quality of life. Subgroup analyses showed that a dosage of 240 mg of ginkgo might be necessary to yield clinically relevant effects.
According to the evaluated studies, Ginkgo biloba extract seems to be well tolerated with rates of adverse effects and study withdrawals not being different between medication and placebo. Nevertheless, it has to be taken into account that randomized controlled trials are not suitable to evaluate rare events and medication interactions. The included studies were undertaken with the standardized extract EGb 761
®. However, other ginkgo extracts and ginkgo-containing products may be associated with different side effects. During intake of EGb 761
®, mild gastro-intestinal symptoms, headache, dizziness or allergic skin reactions have occasionally been reported. Single cases of bleeding, e.g. intracranial haemorrhage, have been reported in the context of an intake of Ginkgo biloba preparations. However, some of these products concerned were of unknown origin and quality, some were multi-ingredient products, and in most instances anti-platelet agents or anticoagulants were taken in addition. A review of controlled studies indicated that the ginkgo extract EGb 761
® does neither influence blood clotting nor bleeding time nor significantly potentiates the effects of anticoagulant or anti-platelet drugs [
40]. In addition, a systematic review of case reports concluded that the clinical evidence for Ginkgo biloba causing bleeding is far from compelling [
41,
42].
We included efficacy as well as effectiveness studies in a variety of settings and regions where different additional psychosocial treatments were offered. In addition, in recent years the availability of cholinesterase inhibitors and memory clinics may have influenced the way dementia is seen and treated. With cholinesterase inhibitors being established treatments, it has become more difficult to run placebo- or actively controlled ginkgo trials in Western countries. The larger effect sizes in the Eastern European studies [[
32], Ihl R, Bachinskaya N, Korczyn AD, Tribanek M, Hoerr R, Napryeyenko O: Efficacy and Safety of a Once-Daily Formulation of Ginkgo biloba Extract EGb 761
® in Dementia with Neuropsychiatric Features. A Randomized Controlled Trial, submitted] may indicate a higher differential efficacy of Ginkgo biloba compared to placebo in a setting with less specialized dementia care and less availability or reimbursement of anti-dementia drugs. These two studies were the ones with the highest effect sizes for cognition and activities of daily living outcomes and had extremely low drop-out rates. Therefore, the higher effect sizes in these studies might be a consequence of lower drop-out rates and less influence of statistical assumptions necessary to calculate last-observation-carried-forward analyses. In addition, treatment can only work optimally in compliant patients staying in the study for the whole follow-up period. When more people drop out of the treatment arm, this would lead to an underestimation of the effect size of the intervention.
In a situation where the clinical significance of the effects of anti-dementia drugs is increasingly questioned, and neither empirical data nor a consensus on minimally clinically important differences in outcome parameters is available [
43], it is difficult to assess if the moderate effects of ginkgo on cognition and ADLs make a difference for the patients in the long run. The validity and reliability of clinical endpoints remain unclear with some early trials using outcomes that would not be accepted for modern studies. Furthermore, there was only one study with a 52-week-follow-up which showed a small but statistically significant advantage in cognition and daily activities. The effects did not differ in magnitude from the 26-week results. On the other hand, even a short follow-up time of 12 weeks was sufficient to separate ginkgo from placebo in one study [
31]. This indicates that the duration of the study and the setting as well as methodological factors may be stronger outcome modifiers than the effects of the medication itself.
With Ginkgo biloba being prescribed for a variety of indications, external validity and generalisability of study results to the target population is of utmost importance. Using criteria for external validity assessment as proposed by Bornhöft et al. [
44], we realized that none of the studies considered all those criteria. Some studies focused on the efficacy of ginkgo and tried to assure high internal validity with low selection, performance, detection and attrition bias [[
33], Ihl R, Bachinskaya N, Korczyn AD, Tribanek M, Hoerr R, Napryeyenko O: Efficacy and Safety of a Once-Daily Formulation of Ginkgo biloba Extract EGb 761
® in Dementia with Neuropsychiatric Features. A Randomized Controlled Trial, submitted]. In these studies, patients with somatic or psychiatric comorbidity were excluded, and other medications were not allowed. This may limit the generalisability of the study results, although in most of the studies included in this review, the setting reflected the everyday conditions with most patients being treated by outpatient clinics or practice-based physicians. One study was performed only with persons living at old people's homes in the Netherlands and had an over-representation of the very old [
34,
45]. This may have contributed to the lack of ginkgo effectiveness versus placebo. Other studies were community-based pragmatic randomized trials with substantial clinical relevance. For example, in the McCarney et al. study [
2], patients were included by general practitioners. The very pragmatic design, together with insufficient statistical power, however, may limit internal validity as many of the included patients were treated concurrently with cholinesterase inhibitors, which were allowed to be continued in both study arms. This may have led to a ceiling effect and the dilution of a significant effect [
5]. A high external validity may therefore compromise internal validity, particularly in those cases where study designs reflect everyday conditions but fail to control other influences on the outcomes. As Bornhöft et al. note [
46], in most ginkgo studies external validity has not sufficiently been addressed. With evidence of a considerable influence of contextual factors also in dementia treatment, it is probable that trial participation itself improves outcomes [
47] and modifies the effects of medication. Therefore, it seems important to keep study conditions as close to the clinical reality.
Our results are consistent with the Cochrane data indicating a small advantage for ginkgo compared with placebo at 12 and 24 weeks in dementia [
7]. However, we did not include trials where patients did not receive a validated dementia diagnosis. Due to our inclusion criteria, the overall methodological quality of studies was higher than in the Cochrane Review, which included some older studies without validated dementia diagnoses, less rigorous randomization and allocation schemes and, consequently, a higher risk of bias. In addition, we identified and included three recently performed trials. Two of these trials showed a considerable superiority of ginkgo in mild to moderate dementia. This may have contributed to the differences between ours and previous reviews. Contrary to previous reviews, we pooled the values of different outcomes scales of the same domain for meta-analysis (e.g. ADAS-cog and SKT for cognition). Given the paucity of studies using the same outcome scales, we felt that this was justified in order to increase the power of the meta-analysis.
Although there is some evidence of a Ginkgo biloba mode of action through mitochondrial stabilization and improvement of cerebral energy mechanism [
48,
49] in addition to hemodynamic effects, there is no consistent picture how this substance may alleviate dementia symptoms. However, this is also true for cholinesterase inhibitors.
Despite a considerable heterogeneity not fully explained by dementia type or ginkgo dosage, we think that ginkgo may be of benefit for a certain but unknown proportion of dementia patients. Against this background and in analogy with dementia prevention [
3,
50], it may be advisable to run a major multicenter study for dementia treatment to definitively answer the question of ginkgo effectiveness for different dementia subgroups in advanced health care systems. Treatment and setting should reflect everyday conditions as much as possible without compromising internal validity. We think that the hitherto gained results justify both symptomatic treatment of dementia and further research, as differential effects for Alzheimer's and vascular dementia are obvious and ginkgo is carried on being used as a complementary therapy in these disorders.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
SW, CV and SNW were primarily involved in conception and design of the study. SW, CS and SR were responsible for acquisition of data, analysis and interpretation of data, SW, SR, CS have been involved in drafting the manuscript, all authors had revised it critically and gave final approval of the version to be published.