An enzyme from fermented soybeans that digests fibrin, the protein mesh that holds a blood clot together. It is the one supplement here where the dose decides whether anything happens at all.
Mechanisms, studies and evidence grades for each claim above.
Nattokinase is a serine protease produced by Bacillus subtilis during the fermentation of soybeans into natto, the sticky Japanese breakfast food. It is a single enzyme of about 28 kDa, isolated from the fermented mass and sold in capsules.
Two things are commonly confused. Natto the food is rich in vitamin K2 as MK-7, which supports clotting and is the reason people on warfarin are told to avoid it. Nattokinase the supplement is the enzyme, usually with the K2 stripped out, and it pushes the opposite way — toward breaking clots down. Taking them to be interchangeable gets the direction of the effect backwards. The vitamin is covered separately under vitamin K2.
Dose is quoted in fibrinolytic units (FU), a measure of enzyme activity, not weight. Milligram figures on a label say nothing useful without the FU rating, and FU is not interchangeable with IU.
↑ back to summaryFibrin is the protein mesh that turns a platelet plug into a stable clot. Nattokinase attacks it from several directions at once, which is unusual for a dietary enzyme.
It cleaves fibrin directly, degrading all three chains of fibrinogen without needing any activation step. Separately it inactivates PAI-1, the inhibitor that normally restrains the body's own clot-dissolving machinery — removing that brake lets more plasmin form. It also promotes release of tissue plasminogen activator and can convert prourokinase to urokinase. The net effect is a push toward fibrinolysis from both the enzymatic and regulatory side.
In people, this shows up as measurable falls in plasma fibrinogen, factor VII and factor VIII — roughly 7–19% over two months at 4,000 FU/day, and to a similar extent in healthy subjects, people with cardiovascular risk factors, and dialysis patients alike.9 Those are real human measurements rather than cell-culture inference, which is why this section carries a filled dot while the downstream claims do not. Two caveats on that trial: it was open-label and self-controlled rather than placebo-controlled, and it found blood lipids unaffected.
A cleaner design tested a single dose. Twelve healthy men, double-blind, placebo-controlled crossover, one 2,000 FU capsule: fibrin and fibrinogen degradation products rose at 4 hours, D-dimer at 6 and 8 hours, factor VIII activity fell, antithrombin rose, and activated partial thromboplastin time lengthened — several fibrinolytic and anticoagulant pathways moving at once from one ordinary dose.8 The authors add the qualifier that gets dropped in every popular retelling: all of those changes stayed within the normal reference range. The enzyme does something measurable within hours; it does not produce a pharmacological anticoagulant state.
One honest limit: nattokinase is a protein, and proteins are what the stomach is built to destroy. How much intact enzyme reaches the bloodstream is genuinely unsettled. Proposed routes — limited transcytosis, paracellular passage of active fragments, indirect signalling by peptides — are hypotheses rather than established pathways, and the clotting-factor changes are currently better evidence that something crosses than any pharmacokinetic study is.
↑ back to summaryThis is the claim that sells the supplement, and the evidence splits cleanly along dose.
The positive side. A trial of 1,062 participants taking 10,800 FU/day for twelve months reported significant reductions in carotid intima-media thickness and plaque size, with improvement rates between 66.5% and 95.4%. The same study tested 3,600 FU/day and found it ineffective for both lipids and plaque. An earlier and much smaller trial, 76 completers over 26 weeks, compared 6,000 FU/day against simvastatin 20 mg and reported greater reduction in carotid thickness and plaque size on the enzyme, despite the statin lowering LDL further. In that trial the change in plaque did not track the change in lipids — whatever moved the plaque, it was not cholesterol lowering.
The null side. The best-designed trial is the Nattokinase Atherothrombotic Prevention Study: double-blind, 265 participants, median age 65, randomised to 2,000 FU or placebo and followed with serial carotid ultrasound every six months for a median of three years. It found no difference in the rate of change of carotid thickness or arterial stiffness, and no effect on blood pressure or any laboratory measure. It was run by an academic atherosclerosis-imaging group with no product interest.
Those two results are usually presented as a contradiction. They are more likely a dose-response. The null trial used 2,000 FU; the trial that worked used more than five times that and separately showed 3,600 FU doing nothing. The null trial also enrolled healthy people at low cardiovascular risk, who have less plaque to lose than the patients in the positive trials.
Two cautions on the positive side, which matter because they point the same way. Five of the ten authors of the 1,062-participant study were employed by a nattokinase manufacturer, and that paper carries a published corrigendum. The smaller comparison trial was single-centre. Neither is disqualifying; both argue for treating the effect size as a ceiling rather than an estimate.
↑ back to summaryThe blood-pressure literature on nattokinase is a set of small trials reporting modest reductions, and the three-year trial above found none.1 Chasing a few points of blood pressure is the wrong reason to take this. The more interesting claim is about how easily blood moves, and it starts with fibrinogen.
Why fibrinogen makes blood sticky. Fibrinogen is the raw material for the fibrin mesh in a clot, but it also circulates at high levels all the time, about 2–4 g per litre of plasma. It is a long, rod-shaped molecule, and it is the main protein that sets how thick plasma is. It also has a second effect. Each end can loosely attach to a red blood cell, so one fibrinogen molecule can bridge two cells. When blood is moving slowly, those bridges pull red cells face to face into stacks like piles of coins, called rouleaux. A stack is much harder to push through a vessel than single cells. In fast-flowing arteries the force of the flow breaks the stacks apart, so the effect is small there. In venules and the slow parts of the microcirculation, where flow is gentle, the stacks hold together, so blood is thickest exactly where tissue gets its oxygen. More fibrinogen means more bridges, bigger stacks and thicker blood at slow flow.
Why that would matter. In a large pooled analysis of 154,211 people across 31 prospective studies, each extra 1 g/L of fibrinogen went with about 2.4 times the risk of coronary heart disease and 2.1 times the risk of stroke. Those ratios fell to about 1.8 after adjusting for other risk factors.10 Higher fibrinogen means a clot forms more easily if a plaque ruptures and blood flows less freely through small vessels. Both effects point the same way. The same analysis also linked fibrinogen to deaths that had nothing to do with blood vessels. That suggests part of the association is fibrinogen marking inflammation, because the liver makes more of it when inflammation is high. The authors said causation is still unproven.
What nattokinase actually does to that chain. The first step is shown in people: two months at 4,000 FU/day lowered fibrinogen by roughly 7–19%.9 The step after that, to thinner blood and better flow in small vessels, is inferred from what fibrinogen does in general. It hasn't been shown in a nattokinase trial. The one long trial that measured blood flow properties directly (blood rheology) is the three-year study at 2,000 FU/day, and it found no change in them, or in any other lab value.1 That fits the dose pattern elsewhere on this page. 2,000 FU is enough to move clotting markers for a few hours, but not enough to change the blood's resting state. Whether 4,000 FU and above lowers viscosity in practice hasn't been tested.
↑ back to summaryThe high-dose trials report improved lipid profiles alongside the plaque changes, with the effect larger in smokers, drinkers and people with higher BMI. The three-year independent trial found no change in any laboratory value, including lipids.
One frequently cited trial tested nattokinase combined with monascus — red yeast rice — and found lipid improvements. That combination cannot speak to the enzyme, because red yeast rice contains monacolin K, which is chemically lovastatin. A lipid effect in that trial is the expected result of giving people a statin.
A third trial points the same way: two months at 4,000 FU/day moved all three coagulation factors and left blood lipids unchanged.9 Whatever the enzyme is doing, it is doing it to the clotting system rather than to cholesterol.
On present evidence, lipid lowering is not a reason to take nattokinase. If plaque does move at high doses, the trial that measured both suggests it moves for reasons other than cholesterol.
↑ back to summaryThis is the part of the page that matters most, and the risk is not theoretical.
Bleeding. Nattokinase is a fibrinolytic agent. Case reports include acute cerebellar haemorrhage in a patient with prior ischaemic stroke, intracranial haemorrhage in a patient who added it to aspirin, and a fatal bleed in an elderly woman who was taking no other blood thinner. The trials report it as well tolerated, but trials enrol screened participants and exclude the people most at risk.
Do not combine it with anticoagulants or antiplatelets — warfarin, DOACs, aspirin, clopidogrel — without a prescriber managing it. Note that one of the high-dose trials reported aspirin co-administration as a benefit; read that as a bleeding-risk flag rather than a recommendation.
Warfarin is a special case, because the two effects oppose each other. The enzyme pushes toward bleeding while any residual vitamin K2 pushes the INR down, and the combination is unpredictable in a way that is difficult to dose around. Substituting nattokinase for a prescribed anticoagulant is the most dangerous version of this: a published case describes a patient who stopped warfarin after mechanical valve replacement, clotted the valve, and required a second replacement.
Avoid with a history of deep vein thrombosis, haemorrhagic or ischaemic stroke, bleeding disorders, or active peptic ulcer. Stop before surgery or dental procedures and ask the surgical team how long beforehand.
Soy allergy is a further consideration, since the enzyme is produced on a soybean substrate.
↑ back to summary