Vitamin K2

Vitamin K2 activates the proteins that steer calcium into bone and keep it out of artery walls — the mechanism is well established and the biomarker evidence genuinely strong, but the hard-outcome trials are split: a couple of positive results measuring scans and markers, balanced by an equal weight of trials that found nothing. It is a cheap, very safe pairing with vitamin D3 for bone and vascular health, not the proven life-extender the supplement market implies.

Vitamin K comes in two families. K1 (phylloquinone), from leafy greens, is what the liver uses to make clotting factors, and ordinary diets supply plenty of it. K2 (the menaquinones), from fermented foods and animal fats, is the form that reaches tissues outside the liver — the vasculature, the skeleton — and it is chronically thin in modern Western diets. Almost everything interesting about K2 for longevity sits in that extrahepatic role, and almost none of it is captured by the official intake numbers, which were set for blood clotting alone.

What K2 actually does: the calcium traffic problem

Vitamin K is the required cofactor for one specific chemical step — carboxylation — that switches on a small family of "Gla proteins." Until they are carboxylated, these proteins are inert; they cannot bind calcium. Two of them carry the whole longevity argument:

Osteocalcin, made by bone cells, pulls calcium into the skeletal matrix.
Matrix Gla Protein (MGP), made in artery walls, sweeps calcium out of soft tissue and is the most potent natural brake on vascular calcification known.

When K2 status is low, both proteins circulate in their uncarboxylated, switched-off form. You can absorb calcium perfectly well (vitamin D handles that) and still lack the machinery to route it — the setup behind the so-called "calcium paradox," where the same person loses calcium from bone while depositing it in arteries.^[1] The uncarboxylated form of MGP (dp-ucMGP) is measurable in blood and predicts arterial stiffness and cardiovascular events — which is why K2 has a strong biomarker case even where the hard-outcome trials are missing. (Strong — for the carboxylation mechanism and the biomarker's predictive value; whether moving the biomarker changes outcomes is the open question the trials below wrestle with.)^[2]

This is the same routing logic that makes K2 part of the calcium triad and the bone-density stack — it doesn't add calcium, it decides where calcium goes.

MK-4 versus MK-7: why the form matters

The two K2 forms worth knowing behave very differently in the body.

Form	Half-life	Dosing	Evidence profile
MK-4 (menatetrenone)	1–4 hours	Needs large, multiple daily doses	The only form with fracture-endpoint trial data, at pharmacological doses (~45 mg/day) in Japanese osteoporosis cohorts — but that literature is retraction-contaminated (see below). Nutritional doses don't raise blood levels at all.
MK-7 (menaquinone-7)	~3 days	Once daily, 90–180 µg	Standard nutritional doses produce sustained, measurable blood levels and full protein carboxylation. Strong mechanism and biomarker data; the better-studied surrogate-endpoint trials use this form.

The practical upshot: a small daily MK-7 dose maintains continuous carboxylation in a way that nutritional MK-4 cannot, because MK-4 is cleared within hours and absorbed poorly at food-level doses.^[3] MK-4's fracture data comes from doses roughly 250× higher than a supplement capsule — a Japanese prescription protocol, not a longevity dose — and it does not bear the weight usually placed on it. The meta-analysis that reported large fracture reductions pooled seven Japanese menaquinone trials from a literature in which fourteen menatetrenone papers, twelve of them by one author, have since been retracted for fabricated data; its own senior author later published a caution to that effect.^[4] ^[5] The one large clean trial — 4,378 women, 45 mg/day, three years — missed its primary endpoint in both strata,^[6] and modern pooled estimates are non-significant.^[7] For a healthy adult, MK-7 at 90–180 µg/day is the sensible default.

The cardiovascular evidence: strong mechanism, thinner outcomes

This is where honesty matters most, because the marketing runs well ahead of the trials.

The observational signal is real but observational. The landmark population data is the Rotterdam Study, which followed 4,807 adults for up to a decade: those in the highest third of dietary menaquinone (K2) intake had roughly half the coronary-heart-disease mortality and severe aortic calcification of the lowest third — while K1 intake showed no association at all, neatly fitting the extrahepatic mechanism.^[8] But this is a dietary cohort: people eating more K2 (from cheese and fermented foods) differ in many ways, and no observational study can prove the K2 itself did the work.

Two interventional trials are positive — both on surrogate endpoints. (Moderate.) The first was a three-year, double-blind, placebo-controlled trial that gave 244 healthy postmenopausal women 180 µg/day of MK-7 and found it not only halted the expected age-related arterial stiffening but modestly reversed it, measured by carotid stiffness and pulse-wave velocity.^[9] The second, reported in 2026, is the first to move an actual calcification imaging endpoint: 167 patients with symptomatic coronary artery disease (most already on statins) took MK-7 360 µg/day or placebo for two years, and the treatment arm showed roughly a quarter to a third less progression of coronary artery calcium on a CT scan.^[10] The investigators themselves call the effect "modest" and note it is a surrogate — coronary calcium score, not heart attacks or death — in a small, short trial. Both remain risk markers rather than hard endpoints.

But an equally serious trial is null, and the kidney-disease story is worse. In a larger, longer, higher-dose Danish trial, 365 older men with aortic-valve calcification took MK-7 720 µg/day plus vitamin D for two years and showed no effect on valve calcification progression — nor, in the coronary substudy, on overall coronary calcium — even though the dp-ucMGP biomarker fell as expected.^[11] And in chronic kidney disease — where patients have the worst vascular calcification and the highest dp-ucMGP, making them the obvious test case — the pattern is starkest: multiple trials confirm K2 reliably lowers dp-ucMGP, yet none has reduced actual calcification, arterial stiffness, or cardiovascular events, and systematic reviews conclude there is no clear benefit.^[12] The biomarker moves; the outcome, so far, does not. That gap is the honest centre of the K2 story.

One caveat threads through the favourable vascular trials: the positive MK-7 studies (both Knapen trials and the 2026 coronary trial) used product supplied by the dominant commercial MK-7 manufacturer, so the industry-independent evidence is thinner than the raw count of positive trials suggests.

The longevity signal, honestly localized. (Moderate for the association; unproven as cause.) Beyond the natto and Rotterdam diet cohorts, the cleaner evidence comes from blood levels rather than food-frequency questionnaires. Pooling individual data from three US cohorts, adults with the lowest circulating vitamin K1 had about a fifth higher all-cause mortality than those with ample levels — but, tellingly, no significant excess of cardiovascular events specifically.^[13] The vascular biomarker points the same way: in a 15-year cohort, higher dp-ucMGP predicted higher all-cause and cardiovascular mortality.^[14] Both findings are observational, both localize the signal to overall mortality rather than heart attacks, and both are plausibly confounded by the fact that people with good vitamin K status also eat more vegetables — no trial has shown that raising vitamin K status, or lowering dp-ucMGP, actually lowers mortality.

The bone evidence: maintenance, not dramatic gains

The same Knapen cohort showed that 180 µg/day of MK-7 over three years significantly slowed the loss of bone mineral density (BMD) and content at the lumbar spine and femoral neck versus placebo. (Moderate, for the maintenance effect.)^[15] But a comparable three-year trial in postmenopausal women who already had osteopenia — and who were all taking background calcium and vitamin D — found the opposite: a higher MK-7 dose switched on osteocalcin just as expected, yet bone density fell equally in both arms, with no benefit at any site. (Moderate, for the null.)^[16] The likeliest explanation for the split is that MK-7 helps most in healthy bone with thin vitamin K status, and adds little once calcium and vitamin D are already replete or bone is frankly diseased. Meta-analytic data confirm K2 shifts bone-turnover markers in a favourable direction — chiefly by converting uncarboxylated osteocalcin to its active form.^[17] Combined vitamin D3 + K2 outperforms either alone on density and osteocalcin in pooled trials.^[18]

The realistic framing: K2 is a supporting nutrient for bone, not a primary driver. The dominant osteogenic stimulus is mechanical — heavy resistance training and impact loading do far more for bone than any micronutrient. K2's job is to make sure the calcium you do absorb is directed correctly. See Bone density.

Food sources: why supplementation even comes up

K2 is genuinely scarce in Western diets, which is the real argument for a supplement.

Natto (fermented soybeans) is in a class of its own — roughly 1,000 µg of MK-7 per 100 g, an order of magnitude beyond any other common food. Habitual natto eaters show higher carboxylated osteocalcin and modestly higher bone density.^[19] The mortality signal comes from separate Japanese cohorts (Weak–Moderate — observational, and natto-eaters differ in diet and lifestyle): across ~93,000 adults followed nearly 15 years, fermented soy — but not unfermented soy — was tied to lower all-cause mortality, and natto specifically to lower cardiovascular mortality,^[20] while a 15-year cohort of elderly Japanese men linked habitual natto intake to lower all-cause mortality (about 40% lower in the moderate-intake group).^[21] See Fermented foods.
Aged and fermented cheeses are the main Western source, varying widely by culture and ripening — Dutch and Swiss hard cheeses and French soft cheeses (Munster, Camembert, Brie) carry meaningful amounts; Mediterranean cheeses (mozzarella, feta, parmesan) carry almost none.^[22] This K2 content is part of why fermented dairy reads as cardiovascular-neutral-to-favourable despite its saturated fat. See Dietary fats.
Egg yolk, butter, and dark poultry meat supply MK-4, but in amounts far below what nutritional studies suggest is needed to shift the proteins.

If you eat natto regularly, you don't need a K2 supplement. Almost no one outside Japan does.

Dose, safety, and the one real caveat

Dose. 90–180 µg/day of MK-7 is the band used in the three-year trials and is what the rest of this site recommends when pairing with D3. There is no clear added benefit demonstrated above that for healthy adults; specialist osteoporosis protocols sometimes go higher.

The official intake numbers are lower, and set for a different job. Regulators never established a full recommended dietary allowance for vitamin K — only an Adequate Intake (120 µg/day for men, 90 µg/day for women), and explicitly on the basis of blood clotting alone, because the non-clotting endpoints were too uncertain to set a number.^[23] That level suffices for the liver's clotting factors but leaves a meaningful fraction of osteocalcin and MGP uncarboxylated in ordinary adults; fully switching on those extrahepatic proteins takes more vitamin K than clotting does.^[24] That gap — enough for clotting, short for the bone and vascular proteins — is the core argument for a modest MK-7 supplement, and it is why status is tracked with a dedicated blood marker for each tissue's uncarboxylated protein: PIVKA-II (undercarboxylated clotting factors) for clotting, uncarboxylated osteocalcin for bone, and dp-ucMGP for the vasculature.

Safety is excellent. (Strong.) Natural vitamin K has no established toxicity, and neither the NIH nor the European Food Safety Authority (EFSA) set a tolerable upper limit because there's no signal to base one on.^[25] An industry-independent review put the highest observed safe intake of MK-7 at 375 µg/day with no effect on clotting parameters.^[26] (Synthetic vitamin K3, menadione, is a different molecule — cytotoxic and not used in human supplements.)

The one caveat that matters: warfarin and other vitamin-K-antagonist anticoagulants. These drugs work by blocking vitamin K recycling, so supplemental K2 directly opposes them — doses as low as 10–20 µg/day can destabilise the international normalised ratio (INR), the standard measure of anticoagulation. If you take warfarin (or acenocoumarol, phenprocoumon), do not start K2 without your prescriber and INR monitoring. The newer direct oral anticoagulants (apixaban, rivaroxaban, dabigatran) don't touch the vitamin K cycle, so K2 doesn't interfere with them.^[27]

What is not established

The source literature for K2 is full of dramatic longevity claims that don't survive scrutiny, and it's worth naming them so you can discount the marketing. (Weak / preliminary throughout — observational or animal data only.)

Telomeres. A single cross-sectional analysis found higher dietary vitamin K associated with marginally longer telomeres (about 0.2 base pairs per µg).^[28] This is observational, tiny, and not a basis for any claim that K2 slows cellular aging.
Sirtuins / lifespan extension. K2 extends lifespan in nematode (C. elegans) models via a stress-response pathway.^[29] Worm data is hypothesis-generating, not evidence for humans.
Brain and cognition. MK-4 is the dominant K vitamer in brain tissue, and post-mortem cohorts link higher brain MK-4 to less Alzheimer pathology — but this is observational, and any cognitive benefit in living humans is unproven.^[30]
Insulin sensitivity and metabolic health. Osteocalcin biology hints at a glucose connection, but a meta-analysis of eight randomized trials found no effect of vitamin K on insulin sensitivity, fasting glucose, or fasting insulin.^[31] Treat any metabolic benefit as unestablished.

Treat these as mechanism and curiosity, not as reasons to supplement.

The practical short version

If you take vitamin D3 for bone or already supplement calcium, add MK-7 at 90–180 µg/day. It's cheap, very safe, and biologically coherent — it directs the calcium D helps you absorb. See Vitamin D and Calcium.
If you eat natto regularly, skip the supplement — you're already well above any studied dose.
Take it with a fat-containing meal (K2 is fat-soluble) and pair it with adequate magnesium for the full bone stack. See Magnesium.
Don't expect a longevity miracle. The honest evidence supports K2 as a sensible adjunct for bone density and arterial flexibility, with two positive surrogate-endpoint trials behind it and an equal weight of null ones — not as a proven extender of lifespan.
On warfarin: coordinate with your prescriber first. On a direct oral anticoagulant: no interaction.