Vitamin K2 for Bones, Arteries, and Coronary Calcium

Calcium does two jobs and the second one can kill you. It builds bone, and it hardens arteries. Vitamin K2 sits at the switch that decides which happens. That single sentence is why a vitamin almost nobody discussed a decade ago now sells alongside vitamin D in nearly every bone and heart stack on the shelf.

The mechanism is real and well-described. The marketing built on top of it has run well ahead of what the trials show. The famous fracture numbers come from a 45-milligram drug and a set of trials that were later retracted for fabrication. Most of the human evidence rests on surrogate markers rather than broken bones or heart attacks. A 2026 trial finally moved the needle on coronary calcium, and the authors who ran it call the effect modest. Here is the part that holds up.

01K2 Activates Calcium Routing Proteins

Vitamin K is a cofactor for one specific chemical reaction. An enzyme called gamma-glutamyl carboxylase uses it to add a carboxyl group to certain proteins, a step that lets those proteins bind calcium. Two of those proteins decide where calcium ends up in the body.

Osteocalcin, made by bone-forming cells, binds calcium into the bone mineral matrix once it is carboxylated. Matrix Gla protein, made in artery walls and cartilage, is the body's most powerful local inhibitor of calcification. Carboxylated matrix Gla protein actively prevents calcium from depositing in vascular tissue.¹ Without enough vitamin K, both proteins are produced but left in their inactive, undercarboxylated state. The bone protein cannot do its job and the artery protein cannot do its.

You can measure this directly. Dephospho-uncarboxylated matrix Gla protein, written dp-ucMGP, circulates in proportion to how much inactive protein is sitting around. A high value means poor vitamin K status. In a Dutch general-population cohort, 31 percent of adults had functional vitamin K insufficiency by this marker, and the figure rose toward half in older and sicker groups.¹⁰ The biology here is clean. The leap worth flagging early is that lowering a biomarker is not the same as preventing a fracture or a cardiac event, and most of the human data lives at the biomarker level.

02K1, MK-4, and MK-7 are not interchangeable

Vitamin K comes in two families. Phylloquinone, or K1, is the form in leafy greens. Menaquinones, the K2 family, come from bacterial fermentation and animal tissue, and they are numbered by side-chain length from MK-4 to MK-13. The two that matter for supplements are MK-4 and MK-7, and they behave very differently.

The pharmacokinetic gap is the whole reason MK-7 is the supplement form people land on. In a head-to-head trial, both K1 and MK-7 peaked around four hours after a dose, but K1 cleared within hours while MK-7 accumulated to seven to eight times higher serum levels over prolonged intake and produced more complete osteocalcin carboxylation.² A long half-life means a once-daily dose can maintain more continuous exposure. MK-4, despite being the form your tissues actually make and use, clears too fast to work as a low-dose daily supplement, which is why the Japanese bone studies used pharmacological milligram doses.

Your body does convert some K1 into MK-4 in tissue using an enzyme called UBIAD1, which is why dietary K1 is not useless for extrahepatic functions.¹² That conversion produces MK-4, not MK-7, so eating more spinach does not reproduce the kinetics of an MK-7 supplement.

03K2 Moves Markers More Than Outcomes

This is where the gap between mechanism and proof opens up. The cleaner you are about study design, the smaller the claims get.

The Rotterdam association. The study that launched the cardiovascular interest in K2 followed 4,807 adults over age 55. Those in the highest tertile of dietary menaquinone intake had a relative risk of coronary heart disease mortality of 0.43 and an odds ratio for severe aortic calcification of 0.48 compared with the lowest tertile. Dietary K1 showed no association at all.³ This is a genuinely interesting signal and it is observational. The K2 in that population came largely from cheese and fermented foods, and people who eat differently differ in other ways. It generates the hypothesis. It does not test it.

The Knapen bone and stiffness trials. The best long-term supplement evidence comes from a three-year trial of 180 mcg of MK-7 per day in healthy postmenopausal women. On bone, MK-7 slowed the decline in bone mineral density at the lumbar spine and femoral neck and reduced loss of vertebral height, though the total-hip site did not reach significance.⁴ On arteries, the same cohort saw dp-ucMGP fall by roughly half, and carotid-femoral pulse wave velocity, a measure of arterial stiffness, decreased significantly. The broader stiffness panel improved mainly in the women who started with the stiffest arteries.⁷ These are real, well-run results. They are also a slowed rate of decline in a surrogate measure, not a reduction in fractures or cardiac events.

The coronary calcium question, updated in 2026. For years the supplement-dose cardiovascular endpoint trials came up empty, and the honest summary was that K2 lowered the biomarker without touching imaging outcomes. That changed with VitaK-CAC, published in 2026. In 167 patients with established coronary disease and baseline coronary artery calcium scores between 50 and 400, MK-7 at 360 mcg per day produced a smaller rise in Agatston calcium score over two years than placebo, 184 versus 214 arbitrary units, a difference that reached statistical significance at P equals 0.02.⁸ The trial investigators describe the effect as modest and write that it is uncertain whether a reduction this small translates into fewer cardiovascular events. A separate trial in aortic-valve calcification using 720 mcg of MK-7 with vitamin D found no effect on its primary endpoint.⁹ Put together, K2 can slow a calcium imaging marker under the right conditions, no trial has shown it prevents an actual heart attack or stroke, and apolipoprotein B remains the lipid number that drives atherosclerotic risk.

04Fracture Claims Rely on Drug Doses

If you read a supplement page, you will likely see striking numbers for K2 and broken bones. A 2006 meta-analysis reported that menaquinone reduced vertebral fractures by 60 percent, hip fractures by 77 percent, and other fractures by 81 percent.⁵ Those numbers are the foundation of the K2-for-bones pitch, and there are two reasons they do not transfer to a bottle of MK-7.

First, almost all of that data used menatetrenone, an MK-4 drug prescribed in Japan at 45 mg per day. That is roughly 250 to 500 times the dose in a typical 90 to 180 mcg supplement. Pharmacological MK-4 and nutritional MK-7 are different interventions.

Second, much of the favorable Japanese trial data came from a research group whose work was later retracted for data fabrication. When the meta-analysis was revisited with the compromised trials removed, the fracture benefit did not survive.⁶ A Western trial of 5 mg of K1 in osteopenic women, well above any dietary amount, found no protection of bone density either.¹⁵ The fracture story is the single most overstated claim in this space.

05The vitamin D pairing

The most common way K2 is sold is alongside vitamin D3, with the argument that D drives calcium absorption while K2 directs that calcium into bone and away from arteries. The mechanism is biologically reasonable. Vitamin D increases osteocalcin and matrix Gla protein production, and K2 is what activates them, so on paper the two work in sequence.

The trial evidence for the pairing is thinner than the marketing implies. Most comparisons focus on bone density, bone turnover, or calcification imaging. Meta-analyses suggest possible BMD effects in selected groups, but they do not establish fewer fractures or cardiovascular events when K2 is added to D3.⁹¹⁶ Treat the combination as a plausible idea worth pairing if you already supplement both, not as an established result. The same caution that applies to magnesium as a vitamin D cofactor applies here, with the difference that magnesium's role in activating vitamin D is better supported than K2's role in the pairing.

06Food Sources Rarely Match MK-7 Supplements

Most of the K1 supply comes from leafy greens, which cover the clotting requirement easily. K2 is sparser in a typical diet, and the sources are uneven.¹³

Natto is the one food that delivers an MK-7 dose comparable to a supplement, and most people outside Japan rarely eat it.¹³ Gut bacteria do produce long-chain menaquinones, mostly in the distal colon where absorption is poor, so do not count on your microbiome to cover the gap.¹⁴

A supplement makes the most sense for a narrow set of people. Postmenopausal women managing bone loss, anyone with documented vascular calcification working with a cardiologist, and people who eat almost no fermented food, cheese, egg yolk, or organ meat have the strongest case. A reasonable dose is 90 to 200 mcg of MK-7 per day, taken with a fat-containing meal because vitamin K is fat-soluble. Tie the decision to the same logic as other micronutrient supplements, which is to test or identify a real gap rather than add a capsule by default.

One absolute exception. Anyone taking warfarin or another vitamin K antagonist should not start K2 without their prescriber managing it. The drug works by blocking vitamin K, and even small doses change anticoagulation. As little as 45 mcg of MK-7 per day cut a clotting marker by around 40 percent in a controlled study, which is enough to destabilize an INR.¹¹ That is a feature of how the vitamin works, and it is also the clearest reason the supplement is not automatically safe for everyone.

The honest summary is that vitamin K2 sits on a strong mechanism and a thin clinical record. The biology of where calcium goes is real, a meaningful share of adults are functionally insufficient, and MK-7 reliably activates the relevant proteins. What no one can yet promise is that supplementing prevents a fracture or a cardiac event, and the place to track that question is the next generation of hard-endpoint trials rather than the label on the bottle.