Review Strategy: General Considerations

For the current review neither the principles of criterion-based quality assessment, nor the rules of statistical pooling were applied. Instead a middle course was followed between the narrative approach and the criterion-based quality assessment approach. Through several tables, organized according to the type of health problem of interest, relevant information will be presented on crucial design aspects as well as on the numerical and statistical results for each Ginkgo biloba intervention trial which fulfilled the selection criteria mentioned in the introduction of this chapter. Each table will be accompanied by an overall judgement of the meaningfulness of the studies enclosed. For some trials it was difficult to assess to what extent the criteria for inclusion in the review were actually met, due to a lack of specific information in the text of the publication. Those studies were given the benefit of the doubt and are presented in the overview. The considerations which motivated the current approach will be explained below.

Completeness and transparancy of study reports

Many studies published in the remote past suffer from a rather poor reporting style (non-transparant structure, essential information lacking). This impedes a critical reviewer to distinguish between a poor design and conduct of the trial at the one hand, giving rise to invalid results, and a poor description of valid results at the other hand. One should not blame the responsible authors too much for this, as the criteria for reporting clinical trials in general and trials in the field of cognitive functioning and dementia in particular, have evolved gradually during the last few decades. Nevertheless, this circumstance seriously hampers any attempt to perform a meaningful criterion-based methodological quality assessment.

Provided that sufficient information is forwarded through a well-structured publication that meets the current reporting standards, the methodological assessment of most of the design aspects, such as random allocation, placebo-intervention, and blinded outcome measurement, is rather straightforward and not that complicated. For Ginkgo biloba trials the trouble starts as soon as the patient definition (diagnostic criteria used for the selection of study subjects) and the outcome measurement (evaluation criteria used for identifying a treatment effect) have to be assessed. These problems are more pronounced for clinical Ginkgo studies focussing on cerebral insufficiency than, for instance, peripheral arterial disease.

Diagnostic uncertainties

Before the methodological problems at the outcome measurement level are discussed, it is necessary to elaborate a bit more on the diagnostic difficulties and uncertainties which surround the identification of the target conditions for Ginkgo treatment. Most of the Ginkgo trials in the past, mainly from Germany and France, were conducted in patients suffering from cerebral insufficiency. Cerebral insufficiency and related concepts, such as cerebrovascular insufficiency and brain organic psychosyndrome, have been regarded as a disease entity for quite a long time. The use of these terms and concepts reflected the contemporary etiologic hypothesis that stenotic vascular changes, related to the ageing process, may cause a progressive decrease in cerebral blood flow, leading to a decline in mental and physical functioning. The clinical manifestations of this pathophysiological process were thought to encompass impairment of memory and other cognitive functions, affective symptoms such as anxiety and depression, and physical complaints such as tinnitus, vertigo, and headache. During recent years this concept of cerebral insufficiency has become more or less obsolete, especially after one came to recognize that neural degeneration as in Alzheimer’s disease is a more frequent cause of cognitive impairment in elderly patients than vascular deficiency. The clinical features of these central nervous system disorders correspond to the syndrome of dementia, a pattern of disturbance in which several higher mental functions are affected simultaneously, with impairment of memory, abstract thinking, and psychomotor functions like speech as crucial symptoms, and irreversible changes in mood, social functioning, and personality as accompanying phenomena. The criteria for the differential diagnosis of dementia, dementia subtypes — Alzheimer’s disease (AD) due to an idiopathic degeneration of the central nervous system, vascular dementia (VD); previously called multi-infarction dementia (MID)) due to changes in the vascular system, mixed forms of Alzheimer’s disease and VD, dementia due to other causes, such as AIDS, brain tumours, etc. — and related diseases have evolved steadily during the past 15 years .

To satisfy the need for disease classification within the domain adjacent to dementia and obviously not covered by the diagnostic criteria for dementia, the concept of age-associated memory impairment (AAMI) has been proposed. This concept refers to a pre-dementia state of impairment of memory function and other cognitive functions (language, orientation, constructional abilities, abstract thinking, problem solving, praxis), established both by means of psychometric testing and subjective reporting, and not yet accompanied by impairment of social functions (occupational or social performance). Although AAMI and comparable concepts — like cognitive impairment no dementia (CIND) and age-associated cognitive decline (AACD) — have always remained controversial as meaningful diagnostic entities, formal criteria to identify cases affected by this syndrome have been suggested at least. Eventually the concept of age-associated cognitive decline has acquired an autonomous position in the DSM-IV classification system.

Given these nosological developments, it is not surprising that nowadays a more specific and focussed diagnosis is required for those disorders that were once classified under the heading “cerebral insufficiency”. Unfortunately, the majority of the Ginkgo trial reports that can be retrieved from the literature, do not allow for such a differentiation. Again, it would not be fair to blame the investigators, as most of the trials have been completed or initiated at a time that the new criteria, guidelines and screening tools for such a differential diagnosis were still awaiting approval and publication. Even nowadays the diagnosis of cognitive neurological disorders is surrounded by much uncertainty and dissent. In a recent review, which compared the designs and the results of 25 Ginkgo trials in cerebral insufficiency with those of two other active substance classes (tacrine, nimodipine), Letzel et al. (1996) concluded that only a few recent Ginkgo trials fulfilled the strictest criteria which nowadays apply to the diagnosis of dementia. At the same time the authors gave several arguments that seem to justify a rather indulgent stand against the design choices made in Ginkgo trials conducted in the past. Nevertheless, both a criterion-based systematic review and a meta-analysis based on statistical pooling would be seriously hampered by these diagnostic problems.

Outcome measurement uncertainties

Analogous problems to those complicating the diagnosis of dementia and related disorders have to be coped with in the outcome measurement domain. To evaluate the efficacy of Ginkgo treatment most of the trials from the past have relied on a wealth of different outcome measures, covering various levels of functioning (general well-being, cognitive functioning, daily life activities, psychopathology, psychometric capacities, clinical signs and symptoms, etc.). Most of the trials have applied all these outcome measures without a predefined hierarchy. Again, the former researchers can hardly be criticized for this. Only quite recently stricter requirements have been formulated for the evaluation of nootropics and anti-dementia drugs, as far as the outcome measurement and evaluation of efficacy of these drugs are concerned. These regulations and guidelines, both at the national and the international level, are revised and adapted periodically. They state, for instance, that the outcome measures and the success criteria derived from these outcome measures should be chosen in advance of the trial onset, i.e. should be included in the trial protocol. Success criteria should not only be formulated in terms of statistical significance (hypothesis testing), but also in terms of clinical relevance (practical meaning of absolute changes and differences in parameter values). The assessment of the ability of interventions to slow down, stabilize, or improve cognitive impairment should be investigated at least at 3 complementary levels: 1. psychopathology (presence of clinical signs and symptoms); 2. cognitive functioning (psychometric testing); 3. behaviour (reflection of cognitive functioning in daily life activities, social functioning, etc.). Moreover, it is required to make a distinction between primary or main, and secondary outcome parameters, and, connected with this, between a confirmatory and a descriptive analysis of the data collected during a trial. A responder analysis should be considered as well. Just as for the diagnosis of dementia, the review of Letzel et al. (1996) demonstrates also for the outcome evaluation that only a few recent trials have been able to keep pace with the evolving standards and requirements. Again, this circumstance seriously hampers the conduct of a systematic review.

The structure of the tables which were composed to summarize the design and the results of the intervention trials retrieved for the current review, will be explained in this section. The information contained by each of the seven tables was organized under 8 different headings.

Column 1

Column 1 shows the names of the author(s) and the year of publication of the trial report. Within each table the trials are arranged in a chronological sequence (year of publication).

Column 2

Column 2 shows the key information regarding the patient population in each trial: diagnostic criteria, and a summary of the main disease characteristics (stage, duration, treatment history, etc.). Some additional information on the composition of the study population may be listed in column 8, under the heading “Remarks”.

Column 3

Column 3 shows the total number of randomized study subjects (first row), and the numbers allocated to each of the intervention subgroups (second row). For crossover trials the second row remains empty, as each study participant has been exposed to both the experimental drug and the reference drug. If the numbers of subjects included in the analysis differed from the number randomized, e.g. due to dropping out, the total and treatment-specific numbers of analysed persons are mentioned on the third and fourth rows of the same field, provided that this information could be retrieved from the trial report. The analysis may have been based on either the “intention-to-treat” (I-to-T) or the “per-protocol” (= “valid-cases”) principle. For some studies additional information was entered into column 8 of the table. Some trials adhered to a multi-comparison design, e.g. contained both a placebo control group and a reference drug control group, in addition to the Ginkgo group. In such cases only the information concerning the verum-placebo part of the trial will be presented. In this column as well as in other columns the notation “n.d.” (=not described) is used to indicate that the relevant information could not be retrieved from the published data.

Column 4

Column 4 provides some elementary information regarding the age and gender composition of the study population. The age range of the persons admitted to the trial and/or the mean (or median) age for the undivided study population are specified in the first row. The sex ratio is shown in the second row, preferably for the randomized population, and otherwise for the analysed population. Occasionally, information regarding the distribution according to some other very important group characteristic, e.g. Alzheimer’s dementia vs. vascular dementia, is entered under this heading as well.

Column 5

Column 5 deals with the study treatment: product name (brand name and/or special extract code name), route of administration, daily dosage and duration of the intervention.

Ginkgo biloba extract can be administered in various ways. Most of the times it is supplied per os, either in a solid form (film-coated tablets, dragees, capsules), or as a solution (drops). However, intravenous or intramuscular injection, and intravenous infusion, are alternative options, especially in clinical settings. Tablets and solutions may contain various amounts of the extract that encompasses the active principles. Eventually, the effect is assumed to be dependent on the total daily dose and its time-related distribution (number and timing of intakes). Regular dosage schedules for Tebonin tablets — Tebonin is one of the most popular Ginkgo brand names — are, for instance: 2 or 3 film-coated tablets a day (1 tablet=40, or 80, or 120mg of active substance (dry extract; input: output ratio=50:l)). In the past a total daily amount of 80-160mg special extract, usually corresponding with 1-4 tablets, was regarded to be sufficient, but recently a tendency can be noticed to elevate this dose, especially within the context of clinical research.

Preferably, the quantity supplied on a daily base, expressed in mg of active substance, is given in the table, to indicate the level of exposure to the study medication. If this figure was not available, the next best information is given (e.g. ml of solution per day or number of drops per day). The dosage scheme is not further specified (e.g. time-related distribution of the daily dose). The last row shows the duration of the intervention, which covers the period from onset (baseline-measurement, after randomization) until final, blinded outcome measurement. In case of a crossover trial the period of Ginkgo treatment is presented. The length of treatment is expressed in either hours (h.; so, registration of an acute intervention effect), days (d.), weeks (w.), or months (m.). The intervention contrast — GBSE vs. either placebo, or a reference drug, or no treatment at all — is specified in column 8. In a few studies Ginkgo was prescribed as an adjuvans therapy to support another therapy. If this was the case the standard therapy is mentioned as well.

Column 6

Column 6 contains information on the outcome measures in each study. A wealth of different endpoints have been chosen to evaluate Ginkgo effects. In particular this is true for the trials dealing with cerebral insufficiency. These outcome measures cover several levels and domains of achievement: global assessment of (change in) health and cognitive state; assessment of clinical signs and symptoms (presence, severity); assessment of functional signs and symptoms; cognitive rating scales; measurement of functional capacity under laboratory conditions (psychometric and physical testing, e.g. aspects of memory function, attention, reaction time, specific parameters of information processing in the brain, concentration, coping with interference, visuo-spatial orientation); neurophysiological and neuroanatomical testing (e.g. EEG-patterns, electronystagmography, craniocorpography); measurement of functional capacity in daily life; and behaviour.

Only a minority of the study reports communicated an explicit hierarchy of endpoints. If so, this structure is conserved in the tables, for instance by distinguishing primary and secondary outcome measures (printed in italics). Otherwise, only the most relevant data are selected for presentation, according to the following priority setting: global assessment; general well-being, quality-of-life assessment; behaviour, daily life activities and functioning; psychometric and neuropsychological test results (functional capacity under laboratory conditions); clinical signs and symptoms; neurophysiological test results; physical signs and symptoms; and physiological and biochemical laboratory values.

With regard to the questionnaires, clinical rating scales and psychometric tests, the tables provide information on both the generic typing and the specific format (“brand name”) (e.g.: depression, Hamilton Depression Scale). In order to facilitate the interpretation of the outcomes recorded, for the most frequently used outcome measures some information is added on scale or test characteristics, such as dimensions and subscales, and range of scores (underlining of one extreme value indicates the most favourable result). For subjective outcome measures the source of information is specified (patient, physician, caregiver, relative, investigator).

In most of the trials, especially the recent ones, much information on safety and tolerance aspects was collected, in addition to efficacy. This was often done in a systematic and extensive fashion (monitoring of adverse events and serious adverse events). The same holds for information on compliance and adherence to the treatment protocol, withdrawal, missing values of outcome parameters, etc. This type of information is hot given in the tables. In general, one agrees that Ginkgo is very safe and tolerable. Almost all studies failed to demonstrate a relationship between study treatment and the occurrence of adverse events, and, if such a relationship could not be excluded fully, the main concern was with fairly minor health consequences.

Column 7

Column 7 summarizes the reported results on the outcome measures mentioned in the corresponding rows of column 6. Unless otherwise stated, only information related to the final outcome measurement time-point will be presented. Invariably “vs” is inserted to contrast the results registered for the Ginkgo group (mentioned first) and the reference group (either placebo group, or alternative drug group). In the case of a continuous outcome measure the numerical results may refer to the mean (or median) status scores at the end of the intervention period. They may also refer to the mean (or median) changes in status score from beginning to end. When change in status over time is the outcome parameter of primary concern, the numerical value is preceded by either “+” or “-” to indicate improvement and deterioration, respectively, for the relevant comparison group. In the case of a discrete (categorical) outcome measure, most of the times ” % ” is used to indicate the relative size of each category. Sometimes, especially for small-sized trials, absolute numbers were entered to avoid sham precision. For some trials numerical results could only be generated via extrapolation of the information stored in a graphical format. This may have resulted in slightly inaccurate outcome estimates.

If retrievable from the trial report the results of formal hypothesis testing (P-values) were added to the numerical outcome values, in order to show the statistical significance of each effect. This information was just copied, without further information on the test statistic used and without checking the appropriateness of the statistical data processing.

Column 8

The final column provides some additional, miscellaneous information on the trial design and the trial results. First of all this column contains information on the study design and the main intervention contrast. The third row throws some light on the source of recruitment of the patient population: out-patient versus hospitalized, type of medical practice, etc. The rest of the information is optional and not systematically entered for each trial (dependent on space, relevance, etc.).

Dementia

The literature search yielded 9 RCTs which suggested that they evaluated the effect of Ginkgo in patients diagnosed with dementia. A more detailed analysis reveals, however, that only five of these alleged anti-dementia trials were designed and conducted in agreement with the gradually evolved and currently accepted scientific guidelines and prescriptions for the diagnosis of dementia and the evaluation of therapy-induced changes in the state of dementia. According to these guidelines and prescriptions, the patient selection should be based on the DSMTII-R/DSMTV, the ICD-10, or the NINCDS-ADRD criteria for dementia. And the outcome assessment should be based on three different levels of achievement: clinical psychopathology (both global and specific), objective psychometric performance, and (cognition-related) daily life activities. It is not surprising that only trials completed during the 1990s fulfil these criteria.

It was decided to deal under the dementia heading also with other trials which claim that they have included patients with dementia, although the appropriateness of this claim cannot be verified from the documentation available. Moreover, one should recognize that many of the trials allocated to some of the other headings (e.g. cognitive impairment) probably have harboured a considerable number or even a majority of demented patients as well, who have been overlooked due to the deployment of insufficient diagnostic equipment.

Hofferberth performed a RCT in 40 hospitalized patients with incipient dementia of the Alzheimer type. All patients (age range: 50-75 years) were treated during 3 months with either Tebonin forte (tablets, 80mg/day) or placebo. The Syndrom Kurz Test (SKT), indicating the level of memory and attention functioning (scale 0-27; the underlined extreme represents the most favourable score), was elected as the primary outcome measure in this trial. Ginkgo treatment turned out to be beneficial, according to the observed mean change in SKT-score over 3 months (+5 vs. -2 points; p<0.001). Fifty percent of the Ginkgo users and none of the placebo users showed an improvement that exceeded 5 SKT-points. A global assessment by the physician, the Sandoz Clinical Assessment-Geriatric Scale (SCAG; measuring the degree of clinical psychopathology), the Vienna Determination Test, a saccade test, and various EEG characteristics were applied as secondary outcome parameters in this study. The patients in the verum group showed more favourable scores on these parameters than those in the placebo group.

Kanowski et al. (1996) conducted a placebo-controlled, parallel trial in 216 outpatients with either Alzheimer’s dementia or vascular dementia, in a mild to moderate state. The patients were recruited from 41 centres (practices of general practitioners, neurologists, psychiatrists, internists). After a 4 week run-in phase, the participants were randomly allocated to a 24-week treatment with either Ginkgo Biloba Extract 761 (240mg/day) or placebo capsules. 205 Patients were involved in the intention-to-treat analysis. However, an analysis based on the 156 participants who completed the trial, was preferred as the confirmatory analysis of the trial results. The clinical global impression of change by the physician (CGI-2, range of scores 1-7), the SKT (range of scores 0-27), and the Nuremberg Geriatric Observation Scale (NAI-NAB) — a rating scale measuring behaviour indicating the level of coping with everyday tasks and the level of independency of care (rated by relatives of the patient) — were applied as the primary outcome measures. Thirty-two percent of the Ginkgo users and 17% of the placebo users were found to respond according to the CGI-2, when the categories “much improved” and “very much improved” were used as the success criterion. The mean change in the SKT sum score was +2.2 for the Ginkgo group vs. +0.8 for the placebo group. With a decrease of at least 4 points on the SKT-scale acting as the criterion for success, 38% of the Ginkgo users vs. 18% of the placebo users turned out to be responders (p<0.05). The mean changes on the NAI-NAB were +0.9 vs. +0.6 points for the Ginkgo group and the placebo group, respectively, with 33% vs. 23% responders (p<0.1; response criterion: >2 NAI-NAB points). An overall responder analysis — criterion: response for at least 2 out of the 3 primary outcome measures — yielded 28% vs. 10% (p=0.005) responders among the Ginkgo users and the placebo users, respectively.

Haase et al. (1996) reported on the results of a parallel, placebo-controlled trial in 40 out-patients of a neuro-psychiatric practice. The study subjects received intravenous treatment, which contained 200mg EGb per day for the Ginkgo group and which was continued over a 4 week period. The daily dose was administered by infusion, 4 days a week. Included in the trial were patients with a moderate state of dementia, either AD, or VD, or a mixed form. The Nuremberg Geriatric Observation Scale (NAI-NAB), the clinical global impression of change by the physician (CGI-2), and the Kurztest fur Allgemeine Intelligenz (KAI) were the primary outcome measures in this trial. With respect to the NAI-NAB 17 Ginkgo users vs. 3 placebo users improved (mean change in score: +3.6 vs. +0.3 points (p<0.01)). With regard to the CGI-2 17 vs. 7 participants improved (p<0.0001). The KAI-test for information processing capacity resulted in improvement for 7 vs. 1 of the participants (p<0.03).

Maurer and colleagues studied the effect of treatment with Tebonin forte (tablets, 240mg/day, intervention stretching over 3 months) in 20 out-patients suffering from Alzheimer’s dementia. A CT-scan revealed either diffuse atrophy or no pathological signs at all. Patients excluded from participation encompassed subjects with an advanced stage of dementia, and patients with dementia of known etiology. Eighteen subjects were included in the analysis. The Syndrom Kurz Test was chosen as the outcome measure of primary interest. The changes in SKT sum score were +2.89 vs. -0.78 (p=0.013), departing from baseline scores amounting to 19.67 vs. 18.11 (scale range: 0-27). Positive, though not statistically significant effects were also reported for the secondary outcome parameters: the number-connection test (NAI-ZVT-G), the ADAS cognitive and non-cognitive subscales, the clinical global impression of change (CGI), and several electrophysiological parameters (AEP-300, EEG-topography). Eight out of 9 Ginkgo users and 5 out of 9 placebo users stated that they would like to continue the treatment.

The first USA-based clinical trial on the efficacy of Ginkgo biloba special extract was reported by Le Bars et al. (1997). Included were 327 persons (aged 45-90 years) from 6 out-patient centres, with a mild to moderately severe state of dementia according to the DSM-III-R and ICD-10 criteria, either Alzheimer’s disease or VD. After a 2-week run-in period the patients were randomly allocated to either Ginkgo (tablets, daily dose of 120mg) or placebo, for a period of one year. 78 Patients in the Ginkgo group and 59 in the placebo group completed the trial. 309 Subjects could be entered in the intention-to-treat analysis, whereas 202 patients met the criteria for the per-protocol analysis, with a deviation of less than 20% from the protocolized intake based on pill counts being one of the criteria. For this trial three primary outcome measures were defined. Firstly, the cognitive subscale of the Alzheimer Disease Assessment Scale at the cognitive function level (ADAS-Cog; performance-based (memory, language, praxis, orientation); 11 items, yielding a 0-70 sum score range). Secondly, the Geriatric Evaluation by Relative’s Rating Instrument at the daily living and social behaviour level (GERRI; rated by the care-giver after 14 days of observation; 49 items, divided over 3 subdomains: cognition, social functioning, and mood; item-, subscale- and total scale score range: 1-5). And, thirdly, the Clinical Global Impression of Change (CGIC; scored by the clinician; scale: 1-7). The mean change in ADAS-Cog score showed less impairment for patients allocated to the Ginkgo group than for patients in the placebo group: -0.1 vs. -1.5 (p=0.04). Although this difference is statistically significant, its clinical meaningfulness is debatable. The improvement amounted to >2 points for 50% of Ginkgo users and 29% of the placebo users. The mean change of the GERRI-overall score was also more favourable for the Ginkgo treatment group: +0.06 vs. -0.08 (p=0.004). Improvement was registered for 37% and 23% of the Ginkgo and placebo groups, respectively. The treatment groups did not differ with respect to the CGI-scores: 4.2 vs. 4.2 (p=0.77). The analysis did not reveal effect modification by AD/ VD-subgroup.

The four remaining trials belonging to this category, which used more relaxed criteria to identify patients with dementia, were all designed as parallel, placebo-controlled trials. The experimental treatment consisted of 120-150mg of active substance per day, for about 3 months. Weitbrecht and Jansen reported a beneficial effect of Ginkgo treatment in patients with a mild to moderate form of primary degenerative dementia according to various clinical rating scales (hetero-assessment of general health state, Crichton Geriatric Scale, SCAG) and various psychometric tests (figure-symbol test, number-repeat test, flicker fusion frequency, reaction time). Halama evaluated the efficacy of Ginkgo treatment in 50 subjects with cerebral insufficiency classified as either degenerative dementia (Alzheimer’s disease) or vascular dementia. The Ginkgo cohort showed better results for the Syndrom Kurz Test (SKT; +4.15 vs. +1.32 points), the global assessment of treatment efficacy, and the number-connection test. Significantly more Ginkgo users than placebo users improved for 7 out of 11 clinical symptoms of cerebral insufficiency. Hartmann and Frick (1991) studied the effect of Ginkgo biloba in 52 patients with vascular dementia. Only patients with at least one symptom out of four symptom categories that are typical for the so-called organic brain psycho syndrome were considered eligible for the trial. A statistically significant effect was recorded only with the Griinberg Verbal Memory test, not with the other outcome parameters (global assessment of efficacy, 12 clinical symptoms, trail making test). In a trial with 90 subjects Vesper and Hansgen (1994) reported a positive effect in favour of the Ginkgo group on a scale of subjective cerebral insufficiency troubles (Cl-scale), both during the first part and the second part of the intervention trial. The rating of changes in each of 8 mental and physical functions on a 1-7 point transition scale (4 indicating “no change”, 7 “very much improvement”) pointed also to a beneficial effect: the scores on the 8 functions varied from about 4.7 to 5.0 in the Ginkgo group and from about 3.9 to 4.2 in the placebo group. A series of computer-aided psychometric tests also suggested a positive effect of Ginkgo treatment. No difference was found for the Short Test of General Intelligence (KAI).

As a conclusion it can be stated that the published randomized controlled trials which have been designed to prove the efficacy of Ginkgo biloba special extract in demented patients, show remarkably consistent results. All trials show positive, often statistically significant effects in patients subjected to Ginkgo treatment compared with patients under placebo treatment, for most of the outcome parameters assessed. One may argue, however, that the reported effect sizes are less impressive and convincing from a clinical point of view. How should one value, for instance, a mean gain of 1.4 points for Ginkgo over placebo on the SKT-scale of memory and attention (maximum improvement: 27 points), or on the ADAS-cog scale (maximum improvement: 70 points) in the sense of the amount of clinical benefit for the patients involved?

Cognitive impairment

The second category consists of 8 RCTs that focus on patients with cerebral insufficiency, manifested by cognitive impairment probably not being dementia. Four studies were conducted in France, two in the U.K., one in Germany, and one in Austria. Again, it should be emphasized that the reported diagnostic criteria do not always preclude the presence of dementia. Several terms and criteria have been proposed to indicate a type of cognitive decline that might be distinguishable from dementia as a separate pathological concept and entity, such as cognitive decline no dementia (CIND), age-associated cognitive decline (AACD), and especially age-associated memory impairment (AAMI). Only one trial strictly adheres to the AAMI-criteria as far as the patient recruitment is concerned.

In addition to seven parallel trials — six of them placebo-controlled, and one with nicergolin (NCG) as the reference substance — one cross-over trial was identified. One of the placebo-controlled trials was based on a factorial design, with memory training as the second intervention factor of interest. The most frequently occurring daily dose was 120mg/day, and the duration of treatment varied from 2-12 months. The cross-over trial aimed at demonstrating the acute effect of a large single dose of Ginkgo Biloba Extract 761.

The outcome measures used in these trials enclosed the global assessment of the intervention effect (4 studies); various clinical rating scales (e.g. MMSE, Global Deterioration Scale); quality-of-life assessment; depression; self-assessesment of well-being and behaviour; hetero-assessment of functional capacity; memory and attention; level of drug prescription; and finally psychophysiological parameters (e.g. EEG). Except for two studies, a clear hierarchy of the outcome measures was not given.

Chartres et al. (1987) found no difference in global effect between Ginkgo and placebo, a small gain in the level of cognitive functioning (MMSE, Reisberg’s GDS, Plutchik GRS) for the Ginkgo users, no difference in the development of several clinical signs, but a remarkably larger decline in the prescription of (other) psychotropic drugs for the persons in the Ginkgo group. Israel et al. (1987) observed an extra effect of Ginkgo, on top of the memory training effect. Within the subgroup that did not receive memory training the Ginkgo users felt more satisfied than the placebo users. Part of the results of the memory battery (long-term memory, fluency test) was in favour of Ginkgo as well. Wesnes et al. (1987) failed to detect a significant overall Ginkgo effect. They also found no difference in actual daily activities between the Ginkgo and placebo users. However, the Ginkgo group performed better on most of the psychometric tests, which were the outcome measures at the forefront in this study. In the intervention trial of Rai et al. (1991) (computerized) psychometric testing was strongly emphasized as well. When compared with the placebo group, the Ginkgo group performed better on most of the tests. A positive effect, be it small, was also found for the MMSE. In the trial of Franco et al. (1991) Ginkgo surpassed nicergolin on almost all the outcome parameters. The cross-over trial conducted by Allain et al. (1993) evaluated the acute effect of two different doses of Ginkgo and placebo through a Latin square design. Series of drawings and words were presented, with varying presentation times. The number of correctly recalled objects was registered. It is a well-known fact that drawings can be remembered more easily than words. The trial results demonstrated that the breakpoint and the “dual coding” point marking the onset of a significant difference between word recall and drawing recall, had been shifted to a shorter presentation time under influence of Ginkgo, irrespective of the treatment dose. This suggests an increase of the relative speed of information processing due to Ginkgo treatment. Semlitsch et al. (1995) assessed the influence of Ginkgo biloba special extract in AAMI-patients by means of psychometric tests, EEG brain-mapping, a checklist expressing subjective thymopsychical well-being, and neurophysiological examinations indicating the quality of cognitive information processing. An acute (3 hours after first dose, week 0), chronic (before the last dose, week 8 of treatment), and superimposed (3 hours after first dose, week 8 of treatment) effect of Ginkobene on P300 latency was registered. The shortened latency may reflect a decrease of the stimulus-evaluation time, evoked by Ginkgo treatment. Oswald et al. (1997) evaluated the effect of Ginkgo biloba within three domains of performance: fluid cognitive performance (objective psychometric testing), subjective well-being (self-rating), and functional performance (hetero-assessment). Only at the fluid cognition level a significant effect of Ginkgo was found. A responder analysis that covered all three domains of performance did not yield a significant result.

It can be concluded that the results of the trials discussed under this heading point to a beneficial effect of Ginkgo treatment, although not unequivocally and not for all outcome parameters.

Non-cognitive signs and symptoms

Twelve randomized clinical trials were identified which had their main focus on the treatment of “physical”, non-cognitive signs and symptoms of cerebral insufficiency, such as vertigo (dizziness), tinnitus (“ear noise”), hearing loss, and ataxia.

These studies were reported from France (6), Germany (5), and Sweden (1). A cross-over design, which contrasted Ginkgo with cinnarizine (CIN) and placebo, respectively, was employed in two trials. The other studies were designed as parallel trials. Six of them contrasted Ginkgo biloba with placebo, the remaining four compared Ginkgo with another active compound (nicergolin (NCG), almitrine-raubasine (ARB), naftidrofuryl). Four studies focussed on vertigo/dizziness as the leading symptom, also four on tinnitus, two on cochlear deafness or sudden hearing loss, and two included a mixture of patients with vertigo, tinnitus and hearing loss. The number of patients varied from 20 to 80 in most of the trials, and exceeded 100 in only two trials. The mean age of the study population was close to 50 in the majority of the studies. The “average” intervention scheme consisted of a daily Ginkgo dosage of 120-160mg/ day, administered as an oral solution during 1-3 months. In one study Ginkgo extract was given as a supplement to a physical vestibular training programme.

The following endpoints were used in these trials: global assessment of the treatment effect; assessment of functional signs and symptoms through a questionnaire, VAS-scale, or clinical rating scale; and various forms of paraclinical otoneurological examination, e.g. vocal and tonal audiometry, electronystagmography, craniocorpography (CCG) and Romberg’s test (posturographic body sway).

The trials which contrasted Ginkgo with a reference drug may be considered less informative with regard to the potential efficacy of Ginkgo. Nevertheless, it is remarkable that in most of these trials Ginkgo surpassed the reference drug for nearly all the parameters evaluated. In the trial reported by Schwerdtfeger (1981) the overall conclusion, based on a questionnaire, electronystagmographic evaluation, and several other neuro(oto)logical tests, pointed to a beneficial effect of Ginkgo. Claussen found a positive effect on both parameters evaluated: subjective dizziness symptoms and lateral body sway. Hamann (1985) reported a positive effect on the degree of posturographic body sway, whereas the change in subjective dizziness complaints was almost equal for the Ginkgo and placebo groups. Meyer (1986a) reported beneficial effects of Ginkgo, which were (borderline) statistically significant for all the relevant endpoints: global assessment of change in clinical signs, change of intensity of tinnitus complaints, change of level of discomfort due to tinnitus, time needed to reach significant improvement or disappearance of complaints. Statistically significant results for all outcomes measured were also communicated by Haguenauer et al. (1986). Morgenstern and Biermann (1997) found a larger regression of tinnitus loudness for the Ginkgo group than for the placebo group. However, this result was not accompanied by a positive effect of Ginkgo on the patient’s subjective impression of the change in tinnitus complaints. In contrast with the studies mentioned before, the placebo-controlled cross-over trial published by Holgers et al. (1994) did not reveal a positive effect of Ginkgo in patients suffering from tinnitus.

A rather new approach in the treatment of tinnitus is the application of Ginkgo biloba extract in combination with soft-laser therapy. One randomized trial, based on a factorial design, could be identified in this field. It was excluded from the review table, since the emphasis was on soft-laser irradiation and random allocation of the study subjects was not stated explicitly.

It can be concluded that almost all trials in this category suggest that Ginkgo is beneficial for patients with vertigo, tinnitus and related manifestations of cerebrovascular insufficiency, with only one trial clearly out of tune.

Subjective cognitive and non-cognitive symptoms

A common feature of these trials is that the investigators applied a symptom checklist to assess the presence and severity of 11 or 12 clinical symptoms, both for patient selection and for outcome evaluation. Memory loss, forgetfulness, lack of concentration, dizziness, tinnitus (ear noise), fatigue, headache, loss of performance, lack of motivation, depression, anxiety, and sometimes confusion, made up the fixed combination of symptoms, which were scored on an ordinal response scale, expressing the degree of nuisance caused by each symptom (absent, mild, moderate, severe). It should be noted that in one study a slightly different symptom checklist was used , and that in another study the primary concern was with psychometric and neurophysiological testing. Furthermore, one should be aware that the same symptom checklist played a prominent role in a few other trials, which were discussed previously under the “dementia” heading.

All studies in this category were placebo-controlled trials. The size of the study population varied from 50 to more than 300 participants, the daily dose of Ginkgo biloba special extract from 120 to 160mg/day, and the duration of the intervention from 4 to 12 weeks. As Table 5 demonstrates, most of the symptoms were found to react (significantly) better to Ginkgo treatment than to placebo treatment. Global assessment of the treatment effect in addition to the administration of the symptom checklist, either by the physician or by the patient, always proved to be in favour of Ginkgo. Hofferberth (1991) reported a positive effect of Ginkgo for the psychometric and neurophysiological parameters as well.

Depression as the leading symptom of cerebral insufficiency

Three RCTs were retrieved which concentrated on depressive mood as the major cerebral insufficiency symptom of interest.

In each of these trials treatment with tablets of Tebonin forte (120-240mg/day, during 8-12 weeks) was compared with placebo treatment. In two studies patients were included who had resisted previous therapy with (other) antidepressants. In both trials Ginkgo Biloba Extract 761 was administered in combination with other antidepressants.

A beneficial effect of Ginkgo was reported in all three trials, viz. 5.3 points improvement vs. 5.2 points deterioration on Zung’s Self-Rating Depression Scale, 6.0 vs. 4.6 points improvement on the Hamilton Depression Scale (HAMD), and 9.5 vs. 1.0 points improvement, also on the Hamilton Depression Scale (p<0.01), respectively. In the last study 75% of the Ginkgo users vs. 30% of the placebo users showed >5 points decrease on the HAMD-scale. In the study of Halama (1990) a positive effect of Ginkgo was also reported for an additional primary outcome measure (late acoustic-evoked potentials) and for a series of secondary outcome measures, e.g. the Kurztest fiir Allgemeine Intelligenz (KAI), and the subjective assessment of depression and concentration. No significant differences were registered for the secondary outcome measures in both other depression trials, except for a better result on the KAI for the subgroup under Ginkgo treatment and a larger decrease in the number of patients with moderate or severe symptoms of depression and concentration, based on subjective assessment by the physician, in the study of Schubert and Halama (1993).

It can be concluded that the trials in this category consistently suggest a favourable effect of Ginkgo on depressive mood.

Cerebral insufficiency, not clearly specified

A rest category was reserved for 18 randomized trials in the field of cerebral insufficiency which were published without sufficient diagnostic information to allow for allocation to one of the preceding categories. Most of these trials date from the 70s and the first half of the 80s. Nine were conducted in Germany, 6 in France, and 3 in Italy.

14 trials have been designed as a placebo-controlled parallel trial, 2 as parallel trials contrasting Ginkgo treatment with a (vasoactive) reference drug (vincamine, dihydroergotoxine), and 2 as placebo-controlled cross-over trials. The number of participants amounted to 30 and 90 in the crossover trials, and varied from 14 to 189 in the parallel trials. The average age of the participants approached 70 years. In the majority of the trials a daily amount of 120mg of active substance was administered. A higher exposure was chosen in only 5 trials. The most common mode of administration of the study medication was per os — in 8 trials as a solution, in 5 trials in tablet form — while in one study the active substance was administered intravenously for 15 days. The reports of 4 other studies lack a clear description of the mode of administration. In most trials the treatment was continued during 1-3 months, in two studies 6 months, and in one study even 12 months.

A large variety of outcome measures was applied to evaluate the Ginkgo effects. In all eight studies in which a global assessment of the treatment effect was made — either by the patient, or by the physician/caregiver, or by both of them — Ginkgo showed the best result. Statistical significance of this global effect was reported in four studies. Clinical rating scales were used in eight trials as well. The application of the Crichton Geriatric Scale in its original or in a modified format yielded a significantly positive effect for Ginkgo in one trial, no effect in another trial, while in a third study no results for this rating scale were given. Application of the Plutchic Geriatric Rating Scale yielded a beneficial Ginkgo effect in two trials. In the study by Halama et al. (1988) both the Sandoz Clinical Assessment Geriatric Scale (SCAG) — being the primary outcome measure — and the Syndrom Kurz Test (SKT) revealed a (statistically) significant effect of Ginkgo. Together with the SGRS, another clinical rating scale, the SCAG was also assessed in the trial reported by Pidoux et al. (1983), again with a positive result. A geriatric clinical evaluation scale (EACG), comparable with the SCAG, showed positive results, just like a self-rating clinical symptoms questionnaire. In the trial of Halama et al. (1988) clinical symptom rating gave mixed results.

In ten trials ample attention was paid to psychometric testing of any Ginkgo biloba effect, through a wealth of different tests and test batteries. The influence of Ginkgo treatment on daily life activities was assessed in three trials, each time with a positive result. Neurophysiological characteristics, the presence and severity of physical signs and functional symptoms, state of anxiety, and clinical and laboratory examination results were used as other endpoints to quantify the treatment effect.

All trials but one failed to explain the relative importance of the outcome measures used. The consequences of this neglect are not that far-reaching, as almost all authors could report beneficial effects of Ginkgo treatment for (almost) all endpoints considered. Israel et al. (1977) reported a positive Ginkgo effect on three out of four cognitive functions, GeEner et al. (1985) reported a positive effect for global evaluation and for two reaction time tests, but not for several other psycho(physio)logical tests, and the positive results reported by Halama et al. (1988) did not extend to the Crichton scale and the craniocorpographic measurements. When compared to a reference drug Ginkgo did slightly better than vincamine, but slightly worse than dihydroergotoxine.

Reflection on the Study Results

Fifty-five randomized clinical trials of Ginkgo biloba and cerebral insufficiency were identified, which were placed a little bit artificially under six different headings, according to the target health problem addressed. Most of the studies were reported from Germany and France: 31 and 16 trials, respectively. Occasionally, patients were recruited in other countries: Italy (3 trials), UK (2 trials), USA, Sweden, and Austria (1 trial each). Almost all trials in which Ginkgo was contrasted with placebo showed a statistically significant effect or a positive trend in favour of Ginkgo. Summarizing and pooling of these results appeared to be a harsh endeavour, not only because of the heterogeneity of study design aspects, especially outcome measurements and definition of success criteria, but also because of the large variety in reporting style. For instance, many reports communicated only part of the outcome results. Statistical test results were not recorded in a systematical way either. The contents of the summary tables reflect these problems. A serious shortcoming of many studies is that hardly any attention is paid to the hierarchy of the outcome measures and to the clinical meaning of the recorded effects. Actually, this is the reason why the study results are presented in a rather detailed fashion and were not reduced to a more aggregate and summary level. Major methodological weaknesses of many trials were, in addition to the outcome definition, the choice and description of the patient definition and eligibility criteria, and the small sample size. Generally speaking, the evidence contained by the more recent trials, including most of the trials which focussed on demented patients, should be valued considerably higher than what can be learned from earlier trials.