Ginkgo biloba and EGb 761: PAF, rheology and cerebral blood flow

Ginkgo between vascular biology and clinical outcomes

Ginkgo leaf extracts have long been studied for their potential effects on vascular function, cellular resistance to injury and symptoms of cognitive disorders. EGb 761 is particularly useful for understanding this research: it is a specific standardized extract with a reproducible composition. Its name identifies a pharmaceutical substance, rather than the entire family of Ginkgo biloba products. Leaf tea, powder, another extract and isolated ginkgolide B are not interchangeable research interventions.

Interest in ginkgo developed at the intersection of two aims: delivering oxygen to tissue and reducing cellular injury when delivery is impaired. An improved laboratory measure, preservation of mitochondrial potential and slowing of dementia are nevertheless different outcomes. The last requires long-term human studies and cannot be inferred from the first two. Ginkgo mechanisms therefore deserve sufficient detail to reveal their possibilities, while the dose, model and outcome of each clinical trial remain part of the interpretation.

Why capillary blood flow cannot be described by viscosity alone

Brain tissue is sensitive to interruption of oxygen and energy-substrate delivery. Between flow in a large supplying artery and the supply to an individual neuron lies a network of arterioles, capillaries and venules. Inlet pressure, the distribution of resistance, blood transit time and exchange across the vascular wall determine how much oxygen actually reaches tissue. Faster flow in one vessel does not by itself demonstrate better exchange.

A capillary lumen may be similar in size to a red blood cell, or smaller. A disc-shaped cell approximately 7–8 μm in diameter passes through a narrow segment by changing shape, not by losing haemoglobin. Capillary dimensions vary across the network; 4–7 μm is a useful illustration of a narrow vessel, rather than a universal reference range for the cortex. A vessel 100–500 μm in diameter must not be drawn as a carotid artery: that scale belongs to much smaller arterial branches.

For an ideal cylindrical tube, steady laminar flow of a Newtonian fluid is described by Poiseuille’s equation:

Q = ΔP·π·r⁴ / (8·η·L), R = 8·η·L / (π·r⁴), Q = ΔP / R.

Q is fluid volume per unit time, ΔP the pressure difference, r the internal radius, L segment length, η dynamic viscosity and R hydraulic resistance. With metres, pascals and seconds, Q is expressed in m³/s and η in Pa·s. The r⁴ term explains why small changes in radius strongly affect resistance in this model. Blood is a cellular suspension, however; vessels branch and change diameter, and capillary flow often consists of individual red cells passing in sequence. Substituting a single “blood viscosity” into the formula to calculate an extract’s effect would therefore be inappropriate.

In thin-tube experiments, apparent blood viscosity depends on diameter and haematocrit. Axial migration of red cells and a relatively cell-poor layer near the wall contribute to the Fåhræus–Lindqvist effect: within a certain range, decreasing diameter reduces apparent viscosity. As diameter approaches cell size, resistance to cell passage becomes important again. This relationship does not establish one threshold at which flow necessarily becomes “abruptly blocked”; cell properties, channel geometry and measurement conditions matter.

Cerebral microcirculation is regulated by the neurovascular unit. It includes endothelium, basement membrane, pericytes, astrocytic endfeet and neurons; nearby arteriolar smooth muscle provides an important share of vascular-resistance changes. Neuronal activity is coupled to local blood delivery, while endothelial junctions contribute to the blood–brain barrier. Ischaemia and inflammation can disrupt several components together, rather than reduce to a single NO deficiency or PAF excess. Endothelin-1 (ET-1) and thromboxane A₂ can contribute to vasoconstrictor signalling, whereas NO and prostacyclin restrain some contractile and platelet responses. Imbalance among them is one possible component of vascular dysfunction, not a single explanation for every memory complaint.

A capillary lacks the continuous smooth-muscle layer characteristic of an arteriole. “Endothelial capillary spasm” therefore conflates vessel-wall contraction, endothelial shape changes, oedema, cellular adhesion and mechanical obstruction. This distinction matters for ginkgo: finding vasorelaxation in a vascular preparation does not mean that all causes of capillary hypoperfusion have been eliminated.

What EGb 761 contains

Standardization describes composition but does not guarantee a clinical benefit for every symptom. EGb 761 is obtained from leaves through aqueous-acetone extraction and subsequent enrichment of selected fractions. For the corresponding quantified dry extract, the European monograph specifies a drug-to-extract ratio of 35–67:1 and 60% acetone as extraction solvent. These manufacturing characteristics must not be assigned to every product labelled “ginkgo”.

The flavonoid fraction includes glycosides of quercetin, kaempferol and isorhamnetin. Quercetin has several hydroxyl groups; kaempferol lacks one of the B-ring hydroxyls, and isorhamnetin is a methylated quercetin derivative. Sugar residues alter solubility and subsequent metabolism. Circulating metabolites are also present after oral intake, so a diagram showing unchanged free flavonol directly inserting into every membrane in the body oversimplifies pharmacokinetics.

In chemical and cellular systems, flavonoids can interact with radicals and transition-metal ions. Hydroxyl-group positions and stabilization of the resulting radical contribute to these properties. The Fenton reaction illustrates one source of oxidative injury:

Fe²⁺ + H₂O₂ → Fe³⁺ + OH⁻ + •OH.

The hydroxyl radical •OH is extremely reactive; preventing its formation is fundamentally different from trying to intercept it afterwards. Iron chelation may alter the reaction, but the direction and magnitude depend on the complex, medium and concentration. Test-tube antioxidant activity does not establish suppression of the Fenton reaction in a patient’s brain by a clinical extract dose.

Ginkgolides are diterpenes with a complex six-ring framework, three lactone rings and a tert-butyl group. These are features of molecular structure, not a “hexacyclic crystal lattice”. Bilobalide has a smaller carbon skeleton: its formula is C₁₅H₁₈O₈, whereas ginkgolides contain 20 carbons. These differences explain why components of one terpene fraction cannot all be assigned the same target.

Ginkgolic acids and the meaning of impurity limits

Ginkgolic acids are salicylic-acid derivatives with a long hydrocarbon side chain. C₁₃:₀, C₁₅:₁ and C₁₇:₁ denote side-chain length and its number of double bonds. C₁₅:₁ contains one double bond; the fully saturated fragment −(CH₂)₁₄−CH₃ is C₁₅:₀. This apparently minor structural detail must be retained in the illustration, or a different compound is depicted.

Ginkgolic-acid cytotoxicity has been described in cellular experiments, including hepatocyte cultures. Sensitization potential also explains the need for purification. Such experiments cannot establish inevitable intestinal toxicity or DNA injury in everyone consuming an unpurified leaf. The practical conclusion is different: a homemade tincture or tea is not controlled by the same parameters as the studied extract. Universal impurity concentrations for homemade products are unknown, and meeting 5 mg/kg reduces one risk without removing drug interactions or individual intolerance.

PAF: synthesis of a lipid signal and the cellular response

Platelet-activating factor, PAF, is the lipid mediator 1-O-alkyl-2-acetyl-sn-glycero-3-phosphocholine. Its name identifies the ether-linked alkyl chain at sn-1 and the short acetyl group at sn-2. Ordinary diacyl phosphatidylcholine must not be drawn as the direct equivalent of this precursor. PAF participates in inflammatory signalling, vascular permeability and cell activation, with its importance depending on tissue and circumstances of formation.

During inflammatory activation, remodelling of an existing ether phospholipid is important. Phospholipase A₂ first removes the acyl residue at sn-2; lyso-PAF acetyltransferase then transfers an acetyl group from acetyl-CoA. LPCAT2 can catalyse this acetyltransferase reaction and is important in inflammatory synthesis. This sequence is not the de novo pathway, which begins with other precursors.

1-O-alkyl-2-acyl-sn-glycero-3-phosphocholine + H₂O → lyso-PAF + fatty acid [phospholipase A₂].

Lyso-PAF + acetyl-CoA → PAF + CoA-SH [lyso-PAF acetyltransferase; LPCAT2].

PAF + H₂O → lyso-PAF + acetate [PAF acetylhydrolase].

The last reaction removes the short acetyl group and limits the mediator’s duration of action. Synthesis, degradation and receptor blockade are fundamentally different processes: a PAFR antagonist does not necessarily reduce PAF itself. It can attenuate the cellular response to mediator already present.

PAFR is a seven-transmembrane G-protein-coupled receptor. Gq can activate phospholipase Cβ, which cleaves membrane PIP₂ into IP₃ and DAG. IP₃ promotes Ca²⁺ release from the endoplasmic reticulum, while DAG and calcium signalling contribute to protein kinase C activation. Gi can inhibit adenylyl cyclase and reduce cAMP formation. The branches are not expressed equally in every cell type.

In platelets, calcium signalling helps change shape, release dense-granule contents and activate integrin αIIbβ₃. ADP and serotonin are in dense granules; P-selectin is associated with α-granule membranes and appears on the cell surface during secretion. α-granules also contain fibrinogen and platelet factor 4. Calcium–calmodulin regulation of myosin light-chain kinase, MLCK, contributes to changes in the contractile apparatus; this must be distinguished from claiming that direct phosphorylation of MLCK itself initiates the whole response. Thromboxane A₂ is synthesized from arachidonic acid after activation, rather than stored ready-made in a dense granule. Activated αIIbβ₃ binds fibrinogen, allowing bridges between neighbouring platelets, but aggregate stability depends on continued signalling and additional haemostatic mechanisms.

In endothelial and immune cells, PAF can increase permeability and inflammatory cell interactions. Adhesion molecules, including selectins and ICAM-1, matter here, but the platelet cascade cannot be transferred unchanged to an endothelial cell or neuron. PAF is also studied as a participant in synaptic transmission. Excess exposure altered long-term potentiation in hippocampal slices; this illustrates a context-dependent signal, not a universal clinical “PAF deficiency” or “PAF excess”.

What ginkgolide antagonism actually means

Ginkgolide B, also called BN 52021, antagonizes PAF effects in experimental systems. Molecular structure matters for this activity, but details of receptor binding cannot be reconstructed from superficial molecular resemblance. The published crystal structure of human PAFR with synthetic antagonists is not a complex with ginkgolide B. Specific hydrogen bonds to His188 and Tyr293 and one invariant binding constant for all systems are therefore not asserted here.

Egon Koch’s experiment on human platelets is particularly informative. It compares ginkgolides within one assay while demonstrating the difficulty of translating the result to oral extract intake.

IC₅₀ is the concentration at which a selected response in a particular assay is inhibited by half. It is not automatically Kᵢ, the target-interaction constant. Species, PAF concentration, medium and endpoint influence the result. Effective ginkgolide concentrations in this study were more than two orders of magnitude above concentrations measured after usual oral intake of the corresponding extracts. This limits explanations of bleeding based on strong direct platelet PAFR blockade, although it does not exclude other interactions or individual risk.

Neuronal effects likewise do not fit “PAF blockade — less glutamate — neuroprotection”. In a separate experiment on rat hippocampal nerve terminals, ginkgolide B increased evoked glutamate release through mechanisms involving calcium entry and PKA. These results do not establish harm at a clinical dose, but demonstrate why an isolated receptor mechanism cannot predict the behaviour of the entire nervous system.

Bilobalide and mitochondrial resilience

Bilobalide is studied mainly in relation to bioenergetics and cellular resistance to injury. The inner mitochondrial membrane maintains the electrochemical gradient used by ATP synthase. Oxygen shortage prevents normal respiratory-chain operation; changes in membrane potential ΔΨₘ and ATP content reflect this state, but each measure requires separate assessment.

An experiment using bilobalide in rat mitochondria showed protection of complexes I and III under the studied conditions. Protection of IV and V was not found in the same study. Later studies of the complete EGb 761 extract in aged mice observed favourable changes in complexes I, IV and V. These involve different substances, models and exposure regimens; they cannot be combined into selective complex-IV protection by bilobalide. Nor can a person be promised fixed preservation of “more than 40% ATP” without a model, dose and measurement method.

Cell-injury discussions must distinguish increased inner-membrane permeability from cytochrome c release across the outer membrane. mPTP opening may contribute to potential loss and swelling, whereas Bax/Bak-dependent outer-membrane permeabilization is associated with the apoptotic cascade. Cytosolic cytochrome c participates in apoptosome formation and caspase activation. Bcl-2, Bax and caspase-3 are informative participants, but changing one protein or its expression does not demonstrate preserved tissue function. Studies of EGb 761 and its components examine particular links; a universal sequence from capsule intake to prevention of neuronal death in patients is not established.

Terpene trilactones also interact with GABA_A and glycine receptors in experimental preparations. Attenuating these responses is not unconditionally beneficial “prevention of depolarizing inhibition”. Chloride-current direction depends on the ionic gradient, and suppressing inhibitory transmission can increase excitability. For specialists, this supports cautious neuropharmacological interpretation, particularly in seizure disorders, rather than choosing ginkgo as an anticonvulsant or neural stimulant.

Red cells, plasma and cellular obstacles to flow

Red-cell deformability depends on more than the lipid membrane. Connections to the spectrin–actin cytoskeleton, band 3 and ankyrin, the surface-area-to-volume ratio and intracellular viscosity are important. Oxidative lipid and protein damage may impair these properties. Cholesterol-to-phospholipid balance and membrane-protein sulfhydryl groups also matter, but their changes are not identical across diseases. Na⁺/K⁺-ATPase and Ca²⁺-ATPase contribute to ion homeostasis; an antioxidant marker does not automatically establish preservation of these pumps. Bilayer fluidity, whole-cell deformability and passage through a microchannel are related, but different measurements.

Calcium homeostasis adds another layer. Increased intracellular Ca²⁺ can activate the Gardos channel, molecularly identified as KCNN4. K⁺ efflux with subsequent Cl⁻ and water loss dehydrates the red cell and changes mechanical properties. Every affected cell does not necessarily become an echinocyte: shape depends on several factors. Pumps need energy, so “an antioxidant always restores plasticity” fails to encompass the full process.

Red-cell experiments, including samples from patients with inflammatory disease, have shown extract effects on oxidative-injury markers in vitro. Direct addition to cells does not reproduce metabolism after oral administration. Such results cannot establish restoration of damaged spectrin, recovery of every ATPase or elimination of capillary stasis by EGb 761 in patients. These claims require corresponding human outcomes.

Cell aggregation and apparent viscosity

At low shear, red cells can form reversible aggregates, including rouleaux. Plasma proteins, particularly fibrinogen, contribute, but literal bridges between sialic acids are not the only explanation: macromolecular depletion in the intercellular space is also important. Increasing shear disperses aggregates, making viscosity results incomplete without shear rate and haematocrit.

Shear rate is denoted γ̇ and measured in s⁻¹. It describes velocity differences between neighbouring fluid layers, not blood velocity along a vessel. The 0.2 and 128 s⁻¹ values encountered in rheological experiments specify different measurement conditions, rather than clinical targets or reference limits. Plasma viscosity, whole-blood viscosity and the elastic component of viscoelasticity are also not synonyms.

Historical ginkgo research included people with increased blood viscoelasticity. Witte assessed rheological changes with different administration routes. Santos combined cognitive assessment, SPECT and viscosity in older adults without dementia. These studies matter to the history of the hypothesis but do not establish viscosity reduction as the cause of clinical improvement. Different products, small samples and multiple outcomes limit translation. Universal reduction in fibrinogen or “plasma surface tension” must not be described as an established therapeutic mechanism of complete EGb 761.

Leukocytes can also impede microcirculation: activation, endothelial interaction and low deformability may increase resistance to passage through narrow segments. PAF contributes to some responses, motivating antagonist studies in inflammation and ischaemia. The existence of this pathway does not establish that oral ginkgo keeps brain capillaries open whenever perfusion pressure falls. Platelet aggregation, leukocyte adhesion and mechanical cell trapping require separate assessment.

NO, prostacyclin and the vascular wall

Endothelial NO synthase, eNOS, uses L-arginine to form NO and L-citrulline. Its function depends on cofactors and cell conditions. Akt-mediated eNOS phosphorylation, including human Ser1177, is an established regulatory mechanism. EGb 761 affected endothelial NO production in experiments, supporting a vascular hypothesis without demonstrating identical endothelial restoration in every patient with cognitive complaints.

Reduced tetrahydrobiopterin BH₄ availability and other adverse conditions can promote eNOS uncoupling, increasing superoxide relative to normal NO synthesis. This differs from simply reducing enzyme quantity. Superoxide reacts with NO to form peroxynitrite:

•NO + O₂•⁻ → ONOO⁻.

One classic direct measurement found a rate constant of (6.7 ± 0.9) × 10⁹ L·mol⁻¹·s⁻¹. For the bimolecular reaction, v = k·[NO]·[O₂•⁻]. Brackets represent reactant concentrations; with mol/L, the rate is mol·L⁻¹·s⁻¹. This number characterizes the chemical reaction under measurement conditions, not a ginkgo effect coefficient or rate of vascular recovery.

NO diffuses into smooth muscle and activates soluble guanylyl cyclase, sGC. This enzyme converts GTP to cGMP, which contributes to PKG activation and regulation of calcium signalling and contractile machinery. Reduced Ca²⁺ sensitivity and altered myosin phosphorylation support wall relaxation. Cytosolic calcium removal and myosin light-chain phosphatase, MLCP, also participate, with relationships depending on the cell and signalling conditions. eNOS phosphorylation nevertheless does not guarantee more available NO: formation, destruction and enzyme state all matter.

Prostacyclin PGI₂ and thromboxane TXA₂ share the prostanoid precursor PGH₂ but are formed by different terminal enzymes. In typical vascular and platelet contexts, PGI₂ favours relaxation and restrains aggregation, whereas TXA₂ acts oppositely. PGI₂ does not activate adenylyl cyclase by simply diffusing into the cell: it binds the Gs-coupled IP receptor, increasing cAMP formation from ATP. TXA₂ signalling involves the TP receptor.

These pathways explain vascular-wall physiology. They do not demonstrate that EGb 761 necessarily stimulates COX-1, increases prostacyclin synthase or normalizes a numerical PGI₂/TXA₂ ratio in humans. This would require particular metabolite and functional measurements in the relevant clinical setting. “Eutonic vasodilation” is not itself a measured outcome.

Brain, eye and inner ear

Autoregulation, neurovascular coupling and supplying-vessel condition are central to brain perfusion. Small perfusion studies, including a pilot MRI study, can inform discussion of flow changes but do not establish selective effects in watershed zones or absence of vascular steal in every situation. Perfusion imaging likewise does not demonstrate stroke prevention, neuronal preservation or greater everyday independence.

The eye has distinct retinal, choroidal and optic-nerve vascular supplies with different control mechanisms. Limited autonomic innervation of retinal vessels cannot be generalized to all ocular microcirculation. A short-term visual-field change is not equivalent to slowing ganglion-cell death. This is particularly important in normal-tension glaucoma, where a vascular-component hypothesis does not remove the need for established ophthalmological treatment.

The inner-ear stria vascularis supports the ionic environment and positive endocochlear potential. Its dysfunction can affect hearing, but tinnitus has many mechanisms, including changes in central auditory processing. “Cochlear ischaemia” cannot be diagnosed from tinnitus or subjective response to ginkgo. Pulsatile tinnitus, sudden hearing loss and neurological symptoms require their own diagnostic pathway rather than an empirical extract course.

What human studies found

Results are best compared by clinical question, not presumed molecular target. Preventing dementia in someone without the diagnosis, symptomatic support in established dementia, treating primary tinnitus and changing capillary perfusion are four different aims. Combining them into one “strength of microcirculatory action” scale obscures evidence limits.

Dementia prevention: GEM and GuidAge

GEM included 3069 adults aged 75 or older without baseline dementia; some had mild cognitive impairment. Participants received 120 mg extract twice daily or placebo, with median follow-up of 6.1 years. Dementia incidence did not decrease: hazard ratio 1.12, 95% confidence interval 0.94–1.33. A further cognitive-trajectory analysis did not find slower decline in the measures studied. The confidence interval includes no difference; the result does not support preventive benefit, but also does not show that every biological ginkgo hypothesis is false.

The French GuidAge trial studied 2854 adults aged 70 or older with memory complaints, using 240 mg extract daily and planning five-year follow-up. The primary endpoint showed no convincing reduction in progression to Alzheimer’s disease: hazard ratio 0.84, 95% confidence interval 0.60–1.18. Signals discussed after longer observation or in subgroups do not change the main result. In particular, they do not establish prevention beginning after four years of use.

Multiple comparisons, time-varying risk, attrition and subgroup rules matter in such analyses. Selecting participants according to duration or adherence may disrupt randomized group comparability when those characteristics are determined after randomization. Even prespecified subgroup analyses require caution, especially after a negative primary endpoint. “At least four years are needed for neuroprotection” is therefore not a justified clinical conclusion from GuidAge.

Established dementia and neuropsychiatric symptoms

With diagnosed dementia, the question changes: can an extract improve symptoms and daily functioning over a limited period? In one placebo-controlled trial, 410 patients with mild or moderate dementia and neuropsychiatric features received 240 mg EGb 761 daily for 24 weeks. Cognitive and neuropsychiatric differences favoured the extract. The findings concern a selected clinical population and a specific product.

Systematic reviews discuss positive signals on cognitive scales, activities of daily living and neuropsychiatric symptoms. These do not promise equal benefit in every type or stage of cognitive disorder. Study heterogeneity, scale choice, randomization quality and effect magnitude matter. Some studies and reviews have links to the extract manufacturer, requiring attention to design and independent replication without automatically invalidating results.

A lower SKT score, ADAS-Cog change and reduced NPI symptoms measure different dimensions. ADL describes everyday actions and is closer to a patient’s lived outcome, but even improved ADL over months does not establish halted neurodegeneration. Cross-study comparisons also cannot establish non-inferiority to donepezil: that requires a direct design with a prespecified non-inferiority margin. MCI cannot automatically inherit results from dementia trials.

Primary tinnitus

A large placebo-controlled trial of 1121 people using another extract, LI 1370, did not confirm benefit when tinnitus was the primary symptom. This is not an EGb 761 trial, but contradicts claims that every ginkgo preparation treats tinnitus. Conversely, some EGb 761 studies assessed tinnitus as a concomitant symptom in other populations; those results cannot simply be transferred to primary chronic tinnitus either.

Review disagreements reflect trial selection and populations. They do not establish a boundary whereby “vascular tinnitus responds, nonvascular tinnitus does not”. That conclusion needs a clear vascular definition and reproducible evidence of an interaction between disease subtype and treatment effect. Subjective intensity, hearing thresholds and quality of life should be considered separately; eliminating stria vascularis ischaemia cannot be declared the mechanism without appropriate measurements.

Raynaud phenomenon, peripheral arteries and glaucoma

Raynaud attacks involve changes in finger colour and temperature in response to cold or stress. Primary Raynaud phenomenon must first be distinguished from secondary disease, including systemic illness. Fewer attacks do not equal restored capillary architecture or prevented tissue ischaemia.

A small early trial reported an approximately 56% reduction in attack frequency versus 27% with placebo. It studied Seredrin, however, not EGb 761, and did not show substantial advantages in attack duration, severity or haemorrheological measures. A later EGb 761 trial in 41 patients found reductions of about 33% versus 31%, without a significant difference. These data do not position the extract as standard treatment for severe Raynaud phenomenon.

In peripheral artery disease, a 62-participant trial did not establish convincing improvement in walking time under the studied conditions. A viscosity hypothesis cannot replace a clinical result for intermittent claudication. Vascular-risk control, a walking programme and decisions about specialist treatment need separate assessment.

In normal-tension glaucoma, an early small crossover study found visual-field changes after a short course. Another randomized crossover trial found no significant improvement in visual field or contrast sensitivity. Small samples, short duration and possible learning from repeated perimetry limit inference. Preserved ganglion cells or slower glaucoma progression cannot be claimed from these studies.

Cognitive complaints and vascular surrogates

Zhang and Xue studied addition of ginkgo tablets to usual treatment in 80 patients without dementia, including MoCA and transcranial Doppler assessments. They reported favourable changes, but the preparation and regimen must not automatically be equated with EGb 761. Doppler velocity is not volumetric tissue perfusion and is influenced by vessel diameter. The study supports further investigation of vascular cognitive impairment, rather than universal ginkgo prescribing for memory complaints.

Safety: assess signal, mechanism and causality separately

Headache, gastrointestinal symptoms and hypersensitivity reactions are reported with ginkgo. Frequency depends on product, population and recording method. Without a specific dataset, exact dyspepsia, headache or skin-reaction percentages should not be assigned to everyone. Intolerance may require discontinuation even when a product meets impurity standards.

Reports of spontaneous intracranial, ocular and postoperative bleeding generated an important safety signal. An individual case with several concomitant medicines does not establish EGb 761 as the cause; product composition and other factors are often poorly documented. The opposite error is declaring risk impossible because usual ginkgolide plasma concentrations lie below laboratory IC₅₀ values. Patient assessment must consider both the clinical signal and limitations of the proposed mechanism.

A controlled warfarin study in 12 healthy volunteers found no significant change in the pharmacokinetic or pharmacodynamic measures studied. This counters inevitable-potentiation claims, but a short single-dose experiment does not reproduce anticoagulation in an older patient with polypharmacy. Prescribed anticoagulants should not be stopped independently to use a herbal product.

Ginkgo use should be disclosed to the surgical and anaesthesia teams before an operation. The European monograph recommends caution and discontinuing the corresponding product 3–4 days before surgery; the final interval depends on product instructions, procedure and team decisions. A universal “7–14 days before any puncture to restore platelets” rule is not established: ginkgolides are not irreversible platelet inhibitors like aspirin.

The corresponding products are contraindicated in pregnancy; breastfeeding use is not recommended because data are insufficient. Adult regimens cannot be transferred to children. In epilepsy, the warning about potentially provoking seizures matters. These are clinical-decision constraints, not evaluations of extract “antioxidant strength”.

CYP and transporters: not all extracts are one substance

In a small study of genotyped volunteers, one ginkgo preparation altered omeprazole metabolism measures related to CYP2C19. Another EGb 761 study using probe substrates found no clinically relevant effects on the major CYP isoenzymes investigated under its regimens. Composition and design differences explain why one study cannot imply mandatory CYP2C19 induction by all ginkgo products.

In-vitro CYP3A4 inhibition also does not establish an interaction of equal magnitude in humans. Drugs with a narrow therapeutic range require assessment of the particular pair; extrapolating omeprazole results to voriconazole without direct data is unreliable. An eight-volunteer digoxin study found no significant pharmacokinetic change, but does not exclude every ginkgo effect on P-glycoprotein. Transport interactions require other evidence too, including the regulatory warning about a potential interaction with dabigatran.

How specialists can apply this knowledge

Practice begins with defining the problem. Subjective memory deterioration may relate to sleep, depression, medicines, hearing impairment or an established neurocognitive disorder. Tinnitus and dizziness do not themselves prove impaired cerebral microcirculation. Extract selection should therefore follow clinical assessment and should not replace it.

For a physician considering symptomatic therapy in an established cognitive disorder, three questions are useful: does the patient match the trial population, does the product match the studied extract, and what result will count as useful? Naming a measurable goal beforehand—such as particular everyday difficulties or neuropsychiatric symptoms—is more valuable than promising “better brain nourishment”. Confirmed contributing causes and established care continue in parallel.

Many EGb 761 trials used 240 mg/day, also specified in the European monograph for the corresponding dry extract. This describes studied regimens, not a prescription to the reader. The monograph’s minimum eight weeks and reassessment if improvement is absent after three months must not be extended to every supplement. A short-term scale response does not justify an automatic 6–12-month course; negative prevention trials do not justify years of intake to prevent dementia.

Some trials used 120–160 mg/day, but the universal division “120 mg for asthenia and vegetative-vascular dystonia, 240 mg for Raynaud and vascular tinnitus” is unsupported. “Vegetative-vascular dystonia” does not identify the cause of symptoms or select an evidence-based indication. There is no general scientifically established ban on intake after 18:00 either; timing follows instructions, the treatment plan and individual tolerability.

A clinical pharmacologist can use mechanisms to assess interaction plausibility but must distinguish laboratory concentrations from patient exposure. Nutrition or sports specialists can clarify product identity, dose and supplement combinations, identify duplication and refer for medical advice. This material does not justify prescribing extract for athletic performance, “blood thinning” or self-treatment of vascular disease.

What to observe, and why

Response assessment should match the original question. MoCA or MMSE can contribute to structured evaluation, but learning, hearing, language and baseline status influence scores. Everyday function requires the patient’s and a close person’s views, descriptions of particular actions and comparable reassessment. Tinnitus assessment concerns symptom severity and life impact; Raynaud assessment concerns attack diaries and evidence of secondary disease. Changes should not automatically be attributed to the product.

A blood-count platelet number does not measure function, and normal PT and aPTT do not exclude impaired primary haemostasis. Doppler imaging of supplying arteries and nailfold capillaroscopy answer their own diagnostic questions; they are not universal ginkgo-response monitoring tools. HAS-BLED, developed for a particular anticoagulation context, cannot become an independent eligibility test for a herbal product.

What remains when hypotheses are separated from results

EGb 761 is a well-characterized pharmacological research subject whose flavonoid glycosides, ginkgolides and bilobalide have different experimental properties. PAFR antagonism and effects on NO signalling, oxidative injury and mitochondrial function create plausible research directions. Their value does not depend on turning every diagram arrow into a clinical promise.

For practice, the decisive conclusion is narrower: a specific extract has evidence of symptomatic effects in some patients with established dementia, whereas large prevention trials did not confirm disease prevention. Tinnitus, Raynaud, glaucoma and vascular-surrogate findings have substantial limits and do not establish ginkgo as a universal microcirculation remedy. Extract identity, diagnosis, a measurable outcome and compatibility with current treatment provide more useful information than the general label “neuroprotectant”.