|Home | About | Journals | Submit | Contact Us | Français|
European Regulation 1924/2006 states that all health claims made on foods need to be substantiated scientifically.
To apply the PASSCLAIM criteria for the scientific substantiation of health claims on foods to herbal supplements containing Ginkgo biloba. Evaluation of three selected claimed health effects for G. biloba (improvement of blood circulation, improvement of symptoms of old age, and improvement of memory) was achieved through review of publicly available scientific data. A total of 35 human intervention studies were evaluated. Commercially available products claimed to contain mainly G. biloba (N=29) were randomly sampled in the Netherlands and analyzed for their content on ginkgo extract. Also, a toxicological risk assessment was performed.
The three selected health claims investigated could not be substantiated. This was mainly because of a lack of data from studies in healthy volunteers. In most studies results performed with a 24% standardized G. biloba extract were described. However, our chemical analysis showed that 25 of the 29 sampled products did not contain the required minimum 24% standardized extract. Moreover, in most preparations the content of substances typical for G. biloba did not conform to what was declared on the label. Since toxicity data for G. biloba are very limited, a safety limit could not be established.
Evidence is lacking for three health claims of herbal products with G. biloba. Neither safety nor efficacy can be guaranteed at the recommended daily dose. The multidisciplinary approach described in this paper provides good insight into issues that are relevant for the evaluation of health claims for herbal food supplements.
The European Regulation 1924/2006 on nutrition and health claims made on foods states that health claims made on foods, including functional foods and dietary supplements, have to be scientifically substantiated (1). Articles 5.1.a and 6 state that generally accepted scientific evidence is needed for this. An overview of the new European Regulation for health claims has recently been published (2–4). In short, reduction of disease risk claims and claims referring to children's development and health are accounted for in Articles 14 and 13.5, and require to be reviewed separately by the European Food Safety Authority (EFSA). A complete dossier containing details of all studies pertinent to the proposed claim is required for an application under Article 14 of the Health Claim Regulation and should be supplied to EFSA via one of the EC member states. EFSA will assess these health claims on a case-by-case basis, leading to an authorization by the European Commission (2, 4, 5). Health claims other than Article 14 and 13.5 claims, the so-called ‘generic claims’ are accounted for in Article 13.1. These claims are based on ‘generally accepted scientific evidence.’ For these ‘generic claims,’ no complete dossier per claim is required (6), and similar applications can be evaluated in batches. The European Commission aimed to adopt a Community list of permitted claims by 31 January 2010, but this deadline was not achieved owing to the very high number of proposals for health claims submitted (2).
To help manufacturers with their preparation and submission of Article 14 health claims (reduction of disease risk claims and claims referring to children's development and health) and 13.5 health claims (based on new scientific developments), in July 2007, EFSA published a guidance document ‘Scientific and technical guidance for the preparation and presentations of the application for authorization of a health claim’ (2, 7). The document specifies what information manufacturers have to include in their application, in particular, with regard to the scientific data that is needed (7). Criteria for the scientific substantiation of health claims on foods were developed from 2001 to 2005 in the EU-sponsored PASSCLAIM project (8, 9) and these were taken into account in setting down the EFSA guidance. According to these criteria, a comprehensive review of human studies on the relationship of the food or nutrient to the claimed health effect is systematically and transparently undertaken before a decision can be made on the claim. Human studies are considered pertinent to the claim. Data of animal or in vitro studies can be used as supporting evidence. The totality of scientific data needs to be included and then the evidence needs to be weighed. Blinded placebo-controlled trials in healthy human samples carry most weight.
The project described in this paper applies the PASSCLAIM criteria for scientific substantiation of health claims on foods to herbal supplements containing Ginkgo biloba, which were used as a model. This paper is largely based on a RIVM report in Dutch from 2007 which was posted on the RIVM website only (http://www.rivm.nl/bibliotheek/rapporten/320106001.html). Upon many requests to have this report also available in the English scientific literature, we have written the current version. We have not changed the conclusions of the original report, but we have updated the literature and developments where applicable.
G. biloba originates from the G. biloba tree. For herbal supplements, usually a G. Biloba leaf extract is used, although the seeds can also be used in dietary supplements. However, in the scientific literature, only effects of the standardized leaf extract are supported (10). A standardized leaf extract (GB-STE; extract of dry G. biloba leaves with acetone and water) contains 22–27% of flavone glycosides and 5–7% of terpene lactones, with 2.8–3.4% of ginkgolide A, B, and C, and 2.6–3.2% of bilobalide (Table 1). The concentration of ginkgolic acid (GAS) needs to be below 5 ppm in a GB-STE (11). The two most considered GB-STEs in studies are EGb761 (12) and LI 1370. A WHO monograph indicates the use of G. biloba extract for symptomatic treatment of mild to moderate cerebrovascular insufficiency, improvement of pain-free walking distance in patients with peripheral arterial occlusive disease (PAOD), and treatment of inner ear problems like tinnitus (11). These are not health claims but medical claims, which are prohibited on foods and food products (1). In Germany, Commission E has approved the use of G. biloba for the same three indications (13).
A random check on the existence and quality of files for substantiation of claims on herbal supplements of the Dutch Food and Consumer Product Safety Authority in 2003 showed that the claim substantiation process was unsatisfactory for herbal supplements in general (14). The outcome of that study resulted in this evaluation study. The aims of the current study were threefold: (a) analysis of herbal supplements containing G. biloba (N=29) randomly sampled in Dutch shops by the Dutch Food and Consumer Product Safety Authority (VWA); (b) scientific substantiation of three selected claimed health effects of G. biloba (including effective dose of substances typical for G. biloba, if available) through an overview of publicly available, scientific data; and (c) risk assessment of G. biloba and, if possible, derivation of a level for safe use.
In August and September 2005, 29 G. biloba food supplements were sampled in the Netherlands. An inventory was made of the indications for use as stated on the label. Claims for a positive influence on memory and/or concentration or improvement of blood circulation were most frequently mentioned. Only four products didn't have any claims on the label (see Table 2).
To determine the degree to which the content of the substances typical for G. biloba were analog to what was declared on the label, the main components of G. biloba, flavone glycosides (also known as flavonoids), and terpene lactones were analyzed. Flavone glycosides were analyzed by using high-pressure liquid chromatography (HPLC) according to the method of Mesbah et al. (15), but with an extra hydrolysis step added to the pretreatment step. Without this hydrolysis step, no flavonoids were detected in the samples. Terpene lactones were analyzed using HPLC with refractive index-detection (RI), also based on the method of Mesbah et al. (15), but with RI-detection instead of the UV-detection, because UV-detection leads to interfering peaks. Reference material included a standardized 24% extract and Tavonin® (16, 17). The Dutch guideline ‘Besluit Geneesmiddelen Wet’ (18) states that the amount of the active substance present in a pharmaceutical product has to be 90–110% of the declared amount. To prove the presence of a 24% standardized G. biloba extract in the products, the amount of bilobalide compared to the total amount of terpene lactones was calculated. A ratio of 39–59% was used as a reference, corresponding to the ratio used in the European Scientific Cooperative On Phytotherapy (ESCOP) monograph (19).
Three commonly encountered claimed health effects for G. biloba were selected: improved blood circulation, effects on symptoms of old age, and improved memory. Examples of such claims used are: ‘helps for cold hands and feet; improves memory; protects against symptoms of old age; natural memory booster.’ Per health effect a literature search was conducted in publicly available, objective scientific data:
An overview of the available literature per health effect can be found in Table 3, including not only studies in healthy volunteers (n=24), but also some important patient studies (n=11). The PASSCLAIM criteria were applied to assess if the health claims can be supported by the totality of the data and weighing of the evidence (8). The criteria are that the product needs to be characterized and administered in amounts consistent with its intended consumption. Studies should include representative human study groups, have appropriate controls, an adequate study duration, characterized background diets, and monitor compliance of the study group. The target variable should change in a statistically significant way and the change should be biologically meaningful for the target group. Dependent on the quality of the data, per health claim a conclusion was drawn as to the substantiation of the claim with reference to the strengths and weaknesses therein.
Information on toxicity of G. biloba was obtained by searching literature databases (Medline, PubMed, Toxline), international documentation (Commission E monograph, ESCOP monograph, Hager's Handbuch, European Medicines Agency – EMEA – information on animal drugs, CBG – College ter Beoordeling van Geneesmiddelen – [Dutch Medicines Evaluation Board] information on drugs, WHO, Herbalgram information, National Toxicology Program – NTP), and the internet. In the Commission E monograph, ESCOP monograph, and Hager's Handbuch, toxicity of the standardized extract EGb 761 is described. The toxicity evaluation in the current evaluation is based on those reports as well as recent NTP studies and recent literature, predominantly case reports, and reports on interaction with frequently/commonly used medication. In addition to G. biloba and the GB-STE, data on GAS were considered for evaluation. Because of their toxic properties, the presence of GAS is limited in standardized G. biloba preparations to maximum 5 ppm (11). However, in non-standardized G. biloba preparations, GAS can be present at higher levels.
For 16 products, both the detailed content and the recommended dose were declared on the label (see Table 2). For 13 products, the declaration was unclear, e.g. the recommended dose was stated, but the amount of flavonoids and terpene lactones in this dose was not stated in the declaration. Only two of the 29 products met the pharmaceutical guideline (the amount present in the product is 90–110% of the declared amount) for terpene lactones. For flavonoids, seven out of the 29 products met the guideline. Only one product met the guideline for both components. Only four products met the criteria of a ratio of 39–59% (the amount of bilobalide compared to the total amount of terpene lactones). Based on the daily dose and the concentration of terpene lactones and flavonoids found in the products, a dose range of substances typical for G. biloba was calculated; 16 products did not meet the dose range for terpene lactones and flavonoids for a 24% standardized extract.
In summary, when the components of herbal supplements were evaluated according to the guideline for pharmaceuticals, that require the presence of 90–110% of the declared amount, one out of 29 G. biloba products meets the guideline.1 The amount of terpene lactones and flavonoids found vary between 27–358% for terpene lactones and between 86–418% for flavonoids. The presence of a 24% standardized extract was detected in four products.
The (combined) PASSCLAIM criteria were checked per health effect, for the selected studies (see Table 3). The final step of the assessment, PASSCLAIM criterion 6, weighing of the evidence, is neither worked out in detail in PASSCLAIM (8), nor in the EFSA guidance document (7). In practice, this is done now by EFSA on a case-by-case basis (2). In our assessment we have used the WHO criteria (20), which categorizes data into convincing evidence (consistent results from adequately powered, randomized placebo-controlled trials of the claimed substance in human subjects representative of the normal/target population, under normal conditions of use and assessing endpoints relevant to the claim), probable evidence (consistent results from small randomized placebo-controlled trials of the claimed substance or active ingredients contained therein in sufficient numbers of human subjects representative of the normal/target population, under normal conditions of use and assessing endpoints relevant to the claim and/or human epidemiological studies), possible (or supporting) evidence (consistent results from animal or in vitro studies of the claimed substance or active ingredients contained therein and assessing endpoints relevant to the claim) and insufficient evidence (inconsistent results and/or studies which do not consider the substance which is the subject of the claim and/or the target population and/or endpoints relevant to the claim). Below, a summary of the available evidence per health effect will be given. A detailed description and comments on individual meta-analyses, systematic reviews, and individual studies are given in Table 3.
Treatment of the disease claudication intermittens is one of the best-known uses of G. biloba. There is a registered medicine on the market containing G. biloba, Tavonin®, which is used for patients that don't respond (enough) to walking exercise (16). In patients with claudicatio intermittens, a positive effect at a dose of 120 mg/day was seen in two meta-analyses (21, 22). However, a recent systematic review that included 739 patients from 14 trials (with different dosage and duration of treatment) didn't find a significant effect on walking distance in people with claudication intermittens (23). Such clinical studies, however, may not be considered pertinent to a claim for the general population.
‘Cerebrovascular insufficiency’ is an expression often used in literature concerning G. biloba. It is an inaccurate expression that is often used to describe a collection of symptoms associated with dementia (11) or the effects of reduced cerebral blood flow in the elderly (24). For the purpose of this evaluation, therefore, cerebrovascular insufficiency has been considered a disease. A systematic review by Kleijnen and Knipschild including patients with mild to moderate cerebrovascular insufficiency showed that treatment with 112–160 mg/day G. biloba for 4–6 weeks was effective for cerebrovascular insufficiency (24).
An overview of available studies in healthy subjects can be found in Table 3. Few human studies are available that have investigated specifically the microcirculation and improved blood circulation in healthy subjects consuming G. biloba. Santos et al. (25) did find a lower blood viscosity (determined with a rotational viscosimeter) in 48 older men that used 80 mg/day G. biloba extract for a period of 8 months. Another study by the same group also described a decreased blood viscosity (measured using ‘Wells-Brookfield Cone/Plate Viscometer’ DV-I) in 25 adult men taking 80 mg GB-STE a day (26). A randomized double blind placebo-controlled trial with 240 mg/day EGb761 for 3 weeks in 27 older subjects found a vasoregulation role of G. biloba (27). Unfortunately, the studies described above included different blood flow parameters, therefore, comparison of the studies is complicated.
We concluded there was insufficient evidence that GB-STE use results in improved blood circulation in healthy subjects.
To enable scientific evaluation, a claim must include in the wording some reference to the target population and the health effect. Current practice within EFSA holds that such a claim is too vague for evaluation. ‘Symptoms’ and ‘old age’ need to be defined and the wording of the claim altered to reflect relevant measurable endpoints. If the claim was submitted through Article 13.1, it would not be evaluated. If submitted through Article 13.5, the dossier could be returned to the applicants who in some cases would be afforded the opportunity to re-word the claim to enable evaluation subsequent to re-submission.
Studies on Alzheimer and dementia show inconsistent results for the effect of G. biloba (28–31). In the latest updated version of a systematic review, looking at cognitive impairment and dementia, it was concluded that the evidence for a clinically significant benefit of G. biloba is unreliable (32).
A systematic review on the effects of G. biloba on the memory of healthy subjects concluded that G. biloba didn't improve memory (33). The results of several trials using G. biloba have been contradictory.
Most studies of G. biloba and aging have considered cognitive endpoints in subjects with mild impairment (13) or Alzheimer and dementia patients (28–31) with inconsistent results for a beneficial effect of G. biloba. In the latest updated version of a systematic review, looking at cognitive impairment and dementia, it was concluded that the evidence for a clinically significant benefit of G. biloba is unreliable (32). Studies employing clinical samples, however, cannot be considered pertinent to a claim for the general population. Evidence for benefit in healthy subjects is mixed.
A large trial including 262 healthy subjects aged 60 years and over, observed improved performance on validated memory tests with GB-STE (180 mg/day, 6 weeks) (34). In other trials, G. biloba did not enhance performance on tests of memory or other measures of cognitive function (35, 36). A double blind, longitudinal, intervention study by Dodge et al. (37) found some evidence for improved memory function in healthy elderly aged 85 + years (N=118) treated with G. biloba (240 mg/day) compared to placebo after controlling for compliance. The observed effect, however, was weak and disappeared when other variables were controlled. That the subjects were also administered a multi-vitamin, may have confounded any effect for ginkgo. The authors do not appear to have declared the content of the placebo making it difficult to evaluate the report. The most recent trial, the GEM study, looking at cognitive decline in healthy older adults, did not find a difference in cognitive functioning between the G. biloba and the placebo group (38).
In a small (N=20) placebo-controlled study with GB-STE 240 or 360 mg/day, a positive effect was observed in young healthy volunteers in improvement in quality of memory and attention (39). However, this result could not be replicated in a similar study in the same study population (40). A double blind controlled study showed no effect of GB-STE 120 mg/day on the memory of healthy young males (N=104) (41).
In conclusion, there is insufficient evidence to substantiate the claim that G. biloba can improve memory in healthy subjects.
In the open literature there are no (animal) data on acute toxicity, carcinogenicity, reproduction toxicity, teratogenicity, neurotoxicity or immunotoxicity, but information on these toxicological endpoints are available in the Commission E monograph, ESCOP monograph, the EMEA report, and a recent NTP study (10, 19, 42). Reportedly, G. biloba is well absorbed upon oral intake in animals and human. However, this should be interpreted with care as G. biloba is an herbal preparation consisting of many components, which may show interactions. Indeed, most reports on kinetics focus on known ingredients of G. biloba such as ginkgolides A and B and bilobalide.
For GB-STE, low acute toxicity has been reported for rodents with LD50 values ranging from 1,100 (intravenous, rats and mice) to 7,725 mg/kg body weight (bw) (oral, mice). A 13-week gavage study in mice and rats with GB-STE (EGb761 dose levels 0, 65.5 [rats only], 125, 250, 500, 1,000 (rats and mice), and 2,000 (mice only) mg/kg bw per day) showed an increase in liver weight at all dose levels tested, and a dose-related increase in hepatocyte hypertrophy in male and female mice at and above 250 mg/kg bw and in male rats at all dose levels. In oral studies with rats and dogs during 6 months, a dose- and time-related mild temporary vasodilatation in cranial blood vessels was observed in dogs dosed at and above 100 mg GB-STE/kg bw from 35 days on (19).
Although GB-STE was reported not mutagenic in (among other tests) a Ames test by Commission E 1994 and ESCOP 2003, recently NTP (43) reported a positive Ames test. An in vivo micronucleus assay was negative for male mice, whereas for female mice the results were equivocal. ESCOP (19) reported no carcinogenic effects in a 2-year study in rats at dose levels of 4, 20, and 100 mg/kg bw. Oral administration of GB-STE to rats (up to 1.6 g/kg bw day) and rabbits (up to 0.9 g/kg bw day) did not show embryotoxic, teratogenic or reproduction toxicity effects (19).
GB-STE also contains GAS, to which immunotoxic, cytotoxic, mutagenic, and carcinogenic properties are ascribed (44–48). As a consequence, a maximum level for the presence of GAS in the GB-STE was set at 5 mg/kg (11). However, as not all G. biloba preparations included in the present study contain GB-STE, maximum levels of 5 mg/kg are not guaranteed, which may pose a health risk (49). Indeed, Chiu et al. (50) report higher levels (16–733 times) than 5 ppm in 13 of 14 G. biloba preparations.
Reportedly, G. biloba is generally well tolerated by human (19, 51), albeit that some mild side effects have been reported [gastrointestinal complaints, headache, allergic skin reactions, nausea, dizziness, restlessness, heart palpitation, and weakness (16, 22, 51, 52)]. In addition to these mild effects, in several case studies side effects have been reported related to blood platelets, hemorrhage, and blood coagulation, which is in agreement with the PAF-inhibiting properties of G. biloba (53–61). In these case reports, reference is made to ‘a ginkgo extract’ or ‘a ginkgo containing preparation,’ without further specifications of the G. biloba preparation used. The duration and dose of intake of G. biloba prior to the reported PAF-related side effects ranges from 2 weeks to 2 years, respectively, 80–160 mg/day, well within the ‘recommended’ dose range.
As for case studies on effects on bleeding, available data on interactions with drugs are also mainly associated with drugs involved in the treatment of blood coagulation and bleeding (53, 62–64). The ESCOP monograph (19) reports that there is no clear evidence for GB-STE on blood coagulation (alone or in combination with medicines). However, various case reports of interactions with drugs have been reported (16, 53, 54, 65–71). Interactions of herbal preparations with drugs are particularly relevant for drugs with a narrow therapeutic margin, it can result in more or stronger side effects or effect on the bioavailability of the drug (72, 73).
Although a direct causal relationship between G. biloba intake and observed effects in case studies is lacking (74), the absence of other risk factors, the PAF-inhibiting activity of G. biloba, and the disappearance of effects upon cessation of intake suggest such causality (59–61). The WHO has indicated to be careful with the use of G. biloba in combination with drugs that affect blood coagulation or platelet aggregation and to stop G. biloba use before a diagnostic treatment or an operation (49, 75).
Summarizing, the potential safety issues associated with G. biloba are severalfold. First, in most literature reports on G. biloba, the preparation used is not specified. There is no information on the levels of terpenoids, flavonoids, and GAS. Animal toxicity studies typically were performed with GB-STE, whereas in case studies no specification on the G. biloba preparation is given. Second, because of the immunotoxic, cytotoxic, carcinogenic, and genotoxic potential of GAS, the maximum level of GAS in GB-STE is limited to 5 ppm. However, if a preparation of G. biloba does not contain the GB-STE, a higher level of GAS cannot be excluded. Finally, G. biloba has PAF-antagonist activity, which may result in effects on platelet aggregation and blood coagulation. Indeed, there are various case studies reporting hemorrhages with intake of G. biloba preparations alone or in combination with drugs affecting platelet aggregation and blood coagulation. Moreover, the dose levels of case reports are within the range of ‘recommended’ dose levels. Nevertheless, exact dose levels are unknown. As a result of abovementioned available data, it is not possible to establish a safe level for intake of G. biloba.
There was insufficient evidence to substantiate the three selected health claims for G. biloba, when applying the PASSCLAIM criteria. However, there were some difficulties when applying the PASSCLAIM criteria. First, a clear definition of what sort of study sample constitutes a healthy target population is lacking. This relates to the representativeness of the study population for the target population. Asp and Bryngelsson (3) concluded that PASSCLAIM is useful for applicants for health claims and for the agencies that evaluate the evidence. Since herbal food supplements are intended for healthy persons, health claims for herbal food supplements have to be proven in studies performed with healthy persons who are representative of the target population. However, distinction between healthy, complaint and illness is not always clear (definition of ‘a healthy person’ is not clear). For this analysis we excluded patient studies, which left only a small number of available studies per health effect. The first opinions from EFSA seem to support this view (76).
Secondly, it is unclear to which degree a herbal supplement of an applicant should be identical to the product used in the human studies. Should they be exactly the same, or should they contain the same standardized extract? This study showed that most studies are performed reportedly using standardized G. biloba extract, but also that nearly every product found in the Dutch shops was not in conformity with the label declaration. When a claim can be substantiated using those studies, the product that sets the claim should resemble the standardized extract.
Thirdly, the PASSCLAIM criteria 4 and 5 require consideration of specific markers, or endpoints, depending on the type of claim made. Evaluation by an expert committee is recommended.
Finally, the procedure for weighing the evidence was not described in detail in the PASSCLAIM project. We used the WHO criteria to weigh the evidence. In contrast to the USA, where qualified health claims are allowed (77), the EU Regulation 1924/2006 was not designed for supporting qualified health claims. However, qualification of the evidence according to the WHO criteria has proved useful for scientific purposes. The value of such criteria in consumer communication has not yet been established. Indeed, in the USA experience, consumers do not make the anticipated distinction between substantiated and not substantiated (=qualified) health claims (2, 77–80). Hence the need for interdisciplinary groups of expert scientists to monitor and evaluate such claims.
There were some limitations of the present study. Firstly, there was no case with product-specific information, delivered by a manufacturer, to evaluate. Therefore, a search of publicly available literature was performed. It is possible that information is missing, e.g. information from negative or null trials or unpublished proprietary or confidential studies. In future, EFSA evaluations will endeavor to include all available information, including confidential data, for the substantiation of Article 14 and 13.5 claims. Secondly, regarding the G. biloba products that were included in this study, the sampling of these products was random from Dutch shops.
That the search was not exhaustive leaves the possibility that there were products available for purchase which were not analyzed and which contained at least the minimum amount of G. biloba extract. The finding that the standardized 24% of G. biloba extract is often not incorporated into the food supplements, nevertheless, raises doubts as to the efficacy of such products. This analysis has also implied that it is not possible to establish a safe level for intake of G. biloba. Review of the science provided insufficient evidence with which to substantiate the three claims for G. biloba. In conclusion, the claim G. biloba and improvement in the symptoms of old age could not be evaluated. A cause and effect relationship between G. biloba and improved circulation or memory has not been established.
The main conclusions in this study were:
Moreover, the PASSCLAIM criteria need to be developed in more detail when used for general application on health claim substantiation for herbal supplements and there is also a need for clarification of the study conditions that are required for claim substantiation, especially with respect to patient studies. A multidisciplinary approach is recommended to assess health claims: analyses of substances typical for G. biloba in the products and toxicological risk assessment are needed for a full assessment. This provides a good insight into the several aspects that are significant for the evaluation of health claims for herbal substances.
We would like to thank Marja Pronk, Ellen de Rooij, and Bert Jan Baars (RIVM) for their valuable input. We also thank Martijn Martena (VWA) for his input during the first phases of the project.
1The guideline for pharmaceuticals (18) was used because no alternative guidelines for herbal supplements were available.
The authors have not received any funding or benefits from industry to conduct this study.