The win is quiet and decade-scale: a slower-stiffening arterial wall and slower bone loss, both moving in the right direction on biomarkers and densitometry after 2–3 years of daily dosing. Cost is roughly thirty dollars a year and the effort is one capsule with a meal. Don't expect to feel anything — that's not what this is for. And if you're on warfarin, talk to your doctor before starting; K2 changes how the drug works.
Calcium has two jobs in your body: build bone and stay out of arteries. The same calcium ion has to know where to go. The "go to bone" signal is a small protein called osteocalcin, made by the cells that build new bone. The "stay out of artery wall" signal is another small protein, matrix Gla protein, made by the cells lining your blood vessels. Both proteins are useless until they're switched on. The switch is vitamin K2.
Without enough K2, both proteins float around in their off state. Osteocalcin can't anchor calcium into your bones, so calcium that should be hardening your skeleton drifts elsewhere. Matrix Gla protein can't keep calcium from sticking to your artery walls, so it starts to deposit there — the same calcium that should have been in your femur, now in the lining of your aorta. Twenty years of this and your bones are thinner and your arteries are stiffer than they should be. The mechanism isn't speculative; it's textbook biochemistry Iwamoto 2014.
The two forms you'll see on a label work by the same mechanism but very differently in your bloodstream. MK-4, the form in egg yolks and dark-meat chicken, is gone from your blood within a couple of hours of eating it. MK-7, the form in natto and aged hard cheeses, sticks around for about three days. That gap is why low-dose MK-7 — a tenth of a milligram — keeps the switch flipped on, while the MK-4 protocols used as a prescription bone drug in Japan run at 45 milligrams a day, roughly a thousand times higher Schurgers et al. 2007.
What the trials and cohorts actually show
The cleanest signal comes from three large European studies that followed people for years and watched who died of heart disease, then looked back at what they ate. The pattern repeats: people eating more K2 had less calcified arteries and lower coronary heart disease mortality. People eating more K1 (the spinach form) showed no effect at all Geleijnse et al. 2004 Gast et al. 2009 Beulens et al. 2009. That last bit — K1 doing nothing while K2 does something — is what makes the signal credible: a generic "people who eat healthy live longer" effect would have caught both.
The randomised trials are smaller and shorter, and they look at biological markers rather than waiting decades for the actual heart attacks. In 244 healthy women past menopause, three years of 180 micrograms a day of MK-7 produced measurably less age-related stiffening of the carotid artery and roughly halved the level of inactive matrix Gla protein in their blood Knapen et al. 2015. The same population, same trial, showed slower bone loss at the spine and hip, and less of the gradual vertebral height-loss that quietly turns into stooped posture in the 70s Knapen et al. 2013.
For bones specifically, the strongest fracture-prevention evidence is Japanese and uses MK-4 at the pharmacological dose — 45 mg/day, prescribed as a registered osteoporosis drug. A 2-year trial in osteoporotic women showed new spine fractures in 13% of treated patients versus 30% of untreated controls Shiraki et al. 2000. A meta-analysis of 13 trials of vitamin K and fractures — most of them Japanese MK-4 — found 60% fewer spine fractures and 77% fewer hip fractures on treatment Cockayne et al. 2006. The honest qualifier the same review made: most trials were small, open-label, and from one country, and the result has not been cleanly reproduced in Western populations.
The decisive trial — thousands of people, multi-year, hard endpoints (actual heart attacks, actual hip fractures), outside Japan — has not been run. A 2015 Cochrane review concluded there isn't yet enough hard-outcome data to recommend K2 for primary cardiovascular prevention Hartley et al. 2015. The mechanism is solid; the population data is consistent; the small trials move the right markers; the large endpoint trial is still missing. That's the honest evidence picture.
What you're trading away by skipping it
Nothing in real time. That's the hard part. Your bones don't ache from being a few percent lower in density; your arteries don't whisper as they stiffen. The damage shows up in the moments you don't see coming — the slip on an icy step at 72 that becomes a hip fracture and 18 months of decline, the chest-tightness on a normal Tuesday that turns out to be the artery you've been quietly hardening since your forties.
On the population scale, the gap looks like this: the third of older Dutch adults eating the most K2 had half the rate of dying from heart disease as the third eating the least, across the same decade Geleijnse et al. 2004. Roughly the same gap shows up in coronary calcium scores — the imaging measurement that radiologists use to predict who's heading for a heart attack Beulens et al. 2009. Whether that gap closes for you specifically by supplementing is an open question; the size of the gap among people who ate K2 the natural way isn't.
The Western diet doesn't really contain enough K2 to close the gap from food alone unless you eat natto, the slimy fermented soy dish that most non-Japanese people will not be eating regularly. A few hard cheeses help; eggs and dark-meat chicken contribute small amounts. The serum biomarker that tracks this — inactive matrix Gla protein — runs high in most of the European and American population, meaning most people are walking around with the calcium-out-of-arteries protein partly switched off Knapen et al. 2015.
How to take it
The dose used in the European bone and arterial trials is straightforward and that's the one to copy. Take it with food that has some fat — yogurt with breakfast, eggs, anything with butter or oil. K2 is fat-soluble, and trial protocols dosed with meals because absorption on an empty stomach drops noticeably.
The high-dose MK-4 protocol from the Japanese fracture trials — 45 mg/day, split three times — is a different conversation. At that dose, MK-4 is a prescription osteoporosis drug, not a supplement; if you're considering it for a real fracture-risk situation, that's a doctor's call, not a supplement-aisle decision Shiraki et al. 2000.
When not to take it
The newer blood thinners — apixaban (Eliquis), rivaroxaban (Xarelto), dabigatran (Pradaxa), edoxaban — work by a different mechanism and aren't affected by K2 intake. If you're on one of those, this isn't an issue for you. The newborn vitamin K shot given at birth is a different vitamin (K1) and is unrelated to anything in this entry.
What gets repeated that isn't quite right
"Vitamin K is vitamin K — spinach gives you everything you need." K1 from leafy greens and K2 from animal products and fermented foods share one biochemical job (flipping that switch on the calcium-direction proteins), but they don't get there equally. K1 gets routed almost entirely to the liver to handle clotting factors and is cleared fast; very little reaches the bones or arteries. In the cohorts that found a K2 signal, K1 intake in the same people did nothing for heart disease risk Geleijnse et al. 2004 Gast et al. 2009. A trial that gave elderly adults K1 supplements for 3 years found no effect on bone density Booth et al. 2008. The forms aren't interchangeable.
"If you take calcium or vitamin D, you have to take K2 to direct the calcium." The "calcium needs K2 as a traffic cop" framing is mechanistically reasonable but the trials testing the combination haven't shown that adding K2 to D plus calcium changes hard outcomes beyond what the individual components do. Take K2 because the K2 evidence stands on its own. Don't take it because someone said your calcium pill is otherwise dangerous — that claim is more confident than the data.
"It works in a week." It doesn't. Steady-state K2 in your blood takes about two weeks Theuwissen et al. 2014; the full effect on the calcium-direction proteins takes a couple of months; the measured arterial-stiffness and bone-density changes took 3 years to show up in the trials Knapen et al. 2013 Knapen et al. 2015. There is no felt-effect timescale on this one — it is biological maintenance, not a stimulant.
How people take it and get nothing
Wrong form. Cheap multivitamins often list "vitamin K" and mean K1, or include a tiny dose of MK-4 (50–100 micrograms) and call it K2. MK-4 at supplement-aisle doses is functionally inert because the half-life is so short; you'd need to take 45 milligrams a day in three doses, which is a different product and a clinician's call. The form that does something at supplement doses is MK-7. If the label doesn't say MK-7, it's not what the trials used.
Taken on an empty stomach. K2 absorbs poorly without dietary fat. People who take a stack of supplements with water first thing in the morning before eating get a fraction of the dose into circulation. The fix is trivial — move it to breakfast or any meal with some fat.
Stopping after a month because nothing happened. The most common quiet failure. Readers expect the energy-and-focus arc of caffeine or creatine and find that K2 produces no felt change at all. They drop it. The effect is happening at the level of two proteins in the bloodstream, not on the level of how you feel; the only way to know it's working is a blood test for inactive matrix Gla protein or, years later, your DEXA scan and coronary calcium score moving in the right direction. The price of taking it is so low that "I don't feel anything, but the maintenance is real" is the correct posture.
What you'll actually buy
A year's supply of MK-7 at 100–200 micrograms a day runs about $25–50 from the major supplement brands. The label phrase to look for is "trans-MK-7" or the brand name "MenaQ7" — that's the biologically active isomer used in the European trials. The cheaper "mixed isomer" bottles cut the active form with an inactive one; you don't pay much less and you absorb measurably less. Storage is ambient; the capsules don't need refrigeration and last well past their printed shelf life.
Pairing with vitamin D3 is reasonable. The biology lines up — vitamin D tells your osteoblasts to make more osteocalcin, K2 switches that osteocalcin on. Many products bundle D3 and K2 in one capsule, which is convenient if you're already taking D3. The bundling isn't magical; it's just two things in one capsule.
Food sources, in descending order: natto delivers about a milligram of MK-7 per 100-gram serving, which is roughly ten times the trial dose in one sitting — if you'll eat it, you don't need to supplement. Aged hard cheeses (Gouda, Edam, Brie) carry MK-9 in tens-of-micrograms-per-serving range. Egg yolks, butter from grass-fed cows, and dark-meat chicken contribute MK-4 at single-digit micrograms — useful as a baseline but not enough to reach the trial intakes on their own.
What changes if you take it
Nothing in your morning for the first two weeks. Nothing in your year. The version of you who takes K2 for a decade will not feel different from the version who didn't — that's the truth of this kind of intervention, and pretending otherwise sells it badly.
What does change, on the timescales that matter:
- Two weeks in, the K2 level in your blood reaches steady state Theuwissen et al. 2014. You can't feel that.
- Six to eight weeks in, the calcium-direction proteins in your blood (osteocalcin and matrix Gla protein) are fully switched on. A blood test would show it. You wouldn't.
- Three years in, if you'd taken the dose used in the postmenopausal-women trial, the stiffness of your carotid artery would be measurably lower than the version of you who skipped it, and the spine and hip bone density would have declined less Knapen et al. 2013 Knapen et al. 2015. Your cardiologist would see it on imaging. You still wouldn't feel it.
- Twenty to thirty years in, this is where the population numbers from Rotterdam and Prospect-EPIC translate into actual events that don't happen to you — the heart attack that statistically should have occurred, the hip fracture from the icy step, the gradual stoop that arrives a decade later than it would have Geleijnse et al. 2004.
This is the silent end of the health-intervention spectrum, the same bucket as keeping your blood pressure in range or your LDL down. The payoff is statistical, the timescale is the rest of your life, and the way you know it worked is mostly the events that don't happen.
Related threads worth a look
Vitamin D3 is the natural partner topic — it tells the calcium-direction proteins to exist; K2 switches them on. Calcium supplementation in older adults is its own contested area; the "calcium paradox" (calcium pills possibly raising heart-disease risk while improving bone density) is part of the same conversation that motivates the K2 case. Coronary calcium scoring is the imaging test that surfaces the arterial-calcification problem K2 is trying to slow. DEXA scans are the bone-density side of the same picture. And if you're on an older blood thinner, INR management is the moving piece that has to coordinate with any K2 change.
- — K2 is part of the slow bone-protection story; it helps keep calcium going into bone over years.
- — Vitamin K2 helps steer calcium into bone instead of artery walls — a quiet, decade-scale supporting move if you're watching bone density.
- — If you're on warfarin, K2 changes how the drug works; clear it with your doctor before starting.
- — Vitamin D raises calcium absorption; K2 is what steers that calcium where it belongs.
- — K2's whole job is keeping calcium out of artery walls, the exact deposit a calcium score measures.
- — K2, vitamin D, and magnesium are the crew that decides where calcium ends up. Magnesium helps activate the D that K2 then directs.
1. Substance and claimed effects
Vitamin K2 (menaquinones, MK-n) is a family of fat-soluble vitamins distinct from vitamin K1 (phylloquinone). The relevant homologues are MK-4, found in dark-meat poultry, eggs, butter and other animal products, and the long-chain menaquinones MK-7 through MK-9, produced by bacterial fermentation and concentrated in natto (fermented soybeans) and aged hard cheeses. Both share a common biochemical role: they serve as a cofactor for the enzyme γ-glutamyl carboxylase, which post-translationally activates a small set of vitamin-K–dependent proteins (VKDPs) by carboxylating their glutamate residues into γ-carboxyglutamate (Gla). The two VKDPs that matter outside coagulation are osteocalcin (synthesised by osteoblasts; binds calcium into the bone matrix when carboxylated) and matrix Gla protein (MGP) (synthesised by vascular smooth muscle cells; inhibits calcium deposition in arterial walls when carboxylated). The entry covers the consequences that follow from adequate K2 status for these two VKDPs: bone mineral density and fracture risk, arterial calcification and coronary risk, and the proposed but less well-evidenced consequences for dental hardness and decay. The framing claim — that K2 "directs calcium into bone rather than soft tissue" — is a mechanistic shorthand for the parallel activation of osteocalcin (calcium uptake into bone) and MGP (calcium exclusion from arteries) by the same cofactor Iwamoto 2014.
2. Evidence by addressing question
mechanism
K2 is a cofactor, not a structural component. Inside cells, γ-glutamyl carboxylase uses reduced vitamin K (hydroquinone form) to add a CO2 group to specific glutamate residues on target proteins, converting them to γ-carboxyglutamate (Gla). The Gla residues chelate calcium ions, which is what gives the activated protein its calcium-binding function. The reaction oxidises K2 to the epoxide; vitamin K epoxide reductase (VKOR) recycles it — the same enzyme that warfarin blocks Iwamoto 2014.
Two non-coagulation VKDPs are load-bearing for this entry. Osteocalcin is secreted by osteoblasts and, once carboxylated, binds the hydroxyapatite lattice of bone, anchoring calcium into the mineralising matrix. The serum ratio of uncarboxylated osteocalcin (ucOC) to total osteocalcin is the established functional marker of bone-level K2 sufficiency; higher ucOC means lower K2 activity at the bone Knapen et al. 2013. Matrix Gla protein is expressed in vascular smooth muscle and chondrocytes; carboxylated MGP binds calcium phosphate crystals and physically prevents them from nucleating into the elastic lamina of arterial walls. The functional marker dephosphorylated-uncarboxylated MGP (dp-ucMGP) rises when K2 is insufficient and predicts vascular events independently of classical risk factors Knapen et al. 2015 Vossen et al. 2019.
MK-4 and MK-7 differ pharmacokinetically. MK-4 has a serum half-life of ~1–2 hours and reaches near-undetectable plasma concentrations between doses; the form that circulates after natto consumption or MK-7 supplementation persists with a half-life of ~72 hours and produces a stable plasma steady state. The long half-life is why low-dose MK-7 (90–180 μg/day) is bioavailable enough to carboxylate extra-hepatic VKDPs, while the pharmacological MK-4 protocol used in Japanese osteoporosis trials runs at 45 mg/day — three orders of magnitude higher Schurgers et al. 2007 Theuwissen et al. 2014.
evidence
Coronary calcification and CHD mortality (observational). The Rotterdam Study followed 4,807 healthy adults aged ≥55 for 7–10 years; the highest tertile of dietary menaquinone intake (≥32.7 μg/day) showed a 50% lower CHD mortality (RR 0.43, 95% CI 0.24–0.77) and 52% lower severe aortic calcification compared with the lowest tertile (<21.6 μg/day); phylloquinone (K1) intake showed no association Geleijnse et al. 2004. The Prospect-EPIC cohort of 16,057 women aged 49–70 replicated the finding: every 10 μg/day increment in menaquinone intake was associated with a 9% lower CHD risk (HR 0.91, 95% CI 0.85–1.00) over 8 years Gast et al. 2009. A nested cross-sectional analysis from the same cohort showed dose-dependent reductions in coronary artery calcification scores across menaquinone tertiles Beulens et al. 2009. All three signals are specific to long-chain menaquinones (MK-7 through MK-9); K1 intake is null.
Arterial stiffness (RCT). 244 healthy postmenopausal women received 180 μg/day MK-7 or placebo for 3 years; the treated group showed reduced carotid stiffness index β (−0.67 vs +0.15 in placebo, p < 0.05) and reduced dp-ucMGP by ~50%, with the strongest effect in women whose baseline arterial stiffness was already elevated Knapen et al. 2015. Mechanism marker and clinical surrogate moved together.
Coronary calcium (RCT). The VitaK-CAC trial randomised 68 patients with established coronary artery disease to 360 μg/day MK-7 or placebo for 2 years, measured by CT calcium scoring; the published protocol and interim biochemical data showed dp-ucMGP suppression but the calcium-score endpoint did not reach significance in this small sample Vossen et al. 2019. Cochrane's 2015 review of vitamin K for primary CV prevention concluded there were insufficient RCT data with hard endpoints to support a recommendation Hartley et al. 2015.
Bone density and fracture (RCT, low-dose MK-7). 244 healthy postmenopausal women on 180 μg/day MK-7 for 3 years showed reduced age-related decline in bone mineral content and BMD at the lumbar spine and femoral neck, and a smaller loss of vertebral height — all small absolute effects, all reaching p < 0.05 against placebo Knapen et al. 2013. An earlier 3-year trial of 325 postmenopausal women on 45 mg/day MK-4 also improved hip bone geometry markers, though without a hard fracture endpoint at that sample size Knapen et al. 2007.
Bone density and fracture (RCT, high-dose MK-4, Japan). Shiraki et al. randomised 241 osteoporotic women to 45 mg/day MK-4 or no treatment for 2 years; lumbar BMD was sustained in the treatment arm vs a 1.9% decline in controls, and new vertebral fractures occurred in 13% of treated vs 30% of untreated women (p = 0.0273) Shiraki et al. 2000. Subsequent Japanese trials reproduced the fracture effect, though a large international meta-analysis tempered enthusiasm: 13 RCTs of vitamin K (mostly Japanese, mostly MK-4 at 45 mg/day) showed a 60% reduction in vertebral fractures and 77% reduction in hip fractures, but the authors flagged that most trials were small, open-label, and from a single national context Cockayne et al. 2006. MK-4 at 45 mg/day is registered as a prescription osteoporosis drug (menatetrenone) in Japan; it is not approved for that indication in Europe or North America Iwamoto 2014.
Bone density (RCT, K1). The ECKO and Booth/Tufts trials of K1 (500–1,000 μg/day) in elderly populations did not show BMD preservation over 2–3 years; the negative result is part of the case for the K1/K2 distinction being functional, not nominal Booth et al. 2008.
Dental. No RCTs of K2 with dental caries, enamel remineralisation, or periodontal endpoints have been published. The dental claim rides on osteocalcin's role in dentin formation and on a mechanistic parallel to bone, not on trial data. Discussion in the entry treats this honestly.
protocol
For extra-hepatic (bone, arterial) carboxylation, the dosing evidence converges on two protocols. The low-dose long-chain MK-7 protocol: 90–180 μg/day of MK-7, taken with a fat-containing meal for absorption, indefinitely. This is the regimen used in the cardiovascular and bone RCTs in Europe Knapen et al. 2013 Knapen et al. 2015. Steady-state plasma MK-7 is achieved at ~2 weeks given the ~72-hour half-life Theuwissen et al. 2014. The high-dose MK-4 protocol: 45 mg/day in three divided doses, evaluated in Japanese osteoporosis trials Shiraki et al. 2000. The two protocols target the same cofactor pathway by very different pharmacokinetic routes — MK-4's short half-life requires brute-force dosing; MK-7's long half-life makes microgram dosing sufficient. Food sources that approach therapeutic intake: natto delivers ~1,000 μg MK-7 per 100 g serving; hard aged cheeses provide ~10–80 μg MK-9 per 100 g; egg yolks and dark poultry meat contribute MK-4 at single-digit μg per serving. For non-natto-eating populations, supplementation is essentially the only way to reach trial-supported intakes.
contraindications
Anticoagulants. Warfarin works by blocking vitamin K epoxide reductase, the enzyme that recycles oxidised K back to its active form. Supplemental K2 will reduce warfarin's INR effect and can precipitate thrombosis if uncoordinated with the prescriber. Patients on warfarin (and to a lesser extent acenocoumarol, phenprocoumon) must either avoid K2 supplementation or maintain it at a consistent daily dose with INR re-titration. Direct oral anticoagulants (DOACs: apixaban, rivaroxaban, dabigatran, edoxaban) do not act on the K cycle and are not affected. The dialysis-population literature explicitly explored MK-7 supplementation in patients on warfarin with INR adjustment and found it tolerable, but that requires clinical supervision.
Newborns. Standard prophylactic vitamin K at birth refers to K1 and is unrelated to K2 supplementation in adults. No specific K2 contraindication in pregnancy or breastfeeding is established at dietary or low-supplemental doses, though high-dose MK-4 (45 mg/day) has not been studied in pregnancy.
misconceptions
Three persistent ones. First: "vitamin K is vitamin K" — meaning K1 and K2 are interchangeable. They are not. They share the carboxylase cofactor role, but K1 is preferentially routed to hepatic clotting-factor synthesis and cleared rapidly from circulation; very little reaches the extra-hepatic VKDPs. K2's lipoprotein-bound distribution and longer half-life make it the practical carboxylator for osteocalcin and MGP. The observational signal is specific to K2 — K1 intake is null for CHD risk in the same cohorts Geleijnse et al. 2004 Gast et al. 2009.
Second: "If you take calcium or vitamin D, you must take K2 to direct the calcium." This is mechanistically plausible but the trials testing the combination have not shown that adding K2 to D+calcium changes hard endpoints beyond what the individual components do. The "calcium paradox" framing (calcium supplementation increasing arterial calcification while bone density falls) has observational support but the K2-as-traffic-director claim is more confident than the trial evidence justifies.
Third: "Take K2 with D3 because they synergise." The synergy claim is real at the level of bone biology — osteocalcin transcription is upregulated by vitamin D and activated by K2 — but no human trial has demonstrated that co-supplementation outperforms either alone on fractures or vascular endpoints. Combining is reasonable; expecting multiplicative effects is over-reading the data.
audience
Three subgroups merit specific framing. Postmenopausal women are where the cleanest RCT evidence sits — both the bone and arterial-stiffness trials enrolled this population Knapen et al. 2013 Knapen et al. 2015. The age-related decline in oestrogen accelerates both bone loss and arterial stiffening; the K2 effect is proportionally larger against that gradient. Older adults generally (60+) share the fracture-risk and CV-calcification rationale. Patients with chronic kidney disease, particularly on dialysis show the strongest dp-ucMGP elevation and the highest vascular calcification burden; small trials of MK-7 in this population have shown clear biochemical effects, though hard endpoints remain open. Younger healthy adults without natto in the diet are not the trial population — supplementation here is mechanistic prevention without RCT validation; the case is reasonable but honest framing requires saying so.
alternatives
For bone: weight-bearing exercise, dietary calcium (1,000–1,200 mg/day in older adults), vitamin D sufficiency (25-hydroxyvitamin D ≥ 30 ng/mL), and where indicated, bisphosphonates or denosumab — each has stronger fracture-prevention evidence than K2 alone outside the Japanese MK-4 trials. For arterial calcification: blood pressure control, LDL-lowering, statin therapy when indicated, smoking cessation — far better-evidenced than K2 for cardiovascular outcomes. K2 is an adjunct, not a substitute. For dental: fluoride, mechanical hygiene, and dietary sugar reduction are the evidence base; K2 is not a dental intervention in any clinically meaningful sense.
failure-modes
Form confusion. Many "vitamin K" supplements contain K1 only, or contain MK-4 at microgram doses (where its short half-life makes it nearly inert) rather than at the 45 mg pharmacological dose. The functional form for low-dose protocols is MK-7; the form for the bone-fracture trial protocol is MK-4 at 45 mg/day. Mixing them up — taking 100 μg of MK-4 expecting Japanese-trial effects — gets no effect.
Absorption without fat. K2 is fat-soluble; taking the capsule on an empty stomach measurably reduces absorption. Trial protocols dose with meals.
Expecting felt effects. The bone and arterial effects are silent — they show up on densitometry, on calcium scoring, on serum dp-ucMGP, not in how the reader feels. Readers who try K2 expecting energy, focus, or sleep changes within weeks abandon it as ineffective; the actual endpoint is years out and asymptomatic.
practicalities
MK-7 at 100–200 μg/day costs roughly $25–50/year from major supplement makers; the standard form sold is either trans-MK-7 (the biologically active isomer; the major commercial brand is MenaQ7) or a mixed cis/trans preparation (cheaper, less bioavailable). Buying trans-MK-7 specifically is worth the small markup. MK-4 at 45 mg/day is sold over-the-counter in the US in 5–15 mg capsules; reaching the Japanese trial dose at three-times-daily dosing pushes annual cost to $200–400. Storage is ambient; K2 is stable.
stakes
The downside of insufficient K2 is asymptomatic and decades-long: progressive arterial wall calcification, gradual loss of bone mineral density, and — over a 50-year scale — a higher probability of the events these substrates produce (myocardial infarction, hip fracture, stroke). None of this is felt in real time; it is observed at diagnostic moments years later. The honest reader-facing framing is that K2 sits in the bucket of silent-tissue health interventions — like lipid control or vitamin D — whose payoff is statistical and delayed, not somatic.
payoff
The payoff matches the stakes: a measurably slower arterial-stiffness trajectory (carotid β index in the K2 arm of the Knapen trial drifted opposite to placebo over 3 years Knapen et al. 2015); a slower decline in spine and hip BMD on densitometry; sustained low dp-ucMGP and low uncarboxylated osteocalcin on serum. The reader who takes K2 for a decade will not feel different; they will, statistically, have a younger arterial wall and denser bones than the version of themselves who did not. Onset of biochemical effect is ~2 weeks for plasma MK-7 steady state and ~6–8 weeks for full carboxylation of the relevant VKDPs Theuwissen et al. 2014. The clinical-endpoint payoff is on the year-to-decade scale.
out-of-scope
K1 (phylloquinone) is a separate substance with a different distribution and different (mostly coagulation-restricted) functional role. Coumarin anticoagulant management — warfarin INR titration — touches K2 because of the shared VKOR enzyme but is its own topic with its own clinical apparatus. The "calcium paradox" — the observation that calcium supplementation may raise CV risk while improving bone — is a related but separable controversy; K2 is one proposed resolution among several. Hypervitaminosis K is not documented; the vitamin has an exceptionally wide therapeutic index, with no upper intake limit set by the IOM.
3. Credibility range
Optimist case. The mechanism is biochemically clean: a single cofactor activates two proteins, one that pulls calcium into bone, one that keeps calcium out of arteries. Three independent observational cohorts (Rotterdam, Prospect-EPIC, EPIC-NL) showed dose-dependent inverse associations between dietary menaquinone intake and coronary heart disease, replicating in different European populations Geleijnse et al. 2004 Beulens et al. 2009 Gast et al. 2009. A 3-year RCT in postmenopausal women showed both bone-loss attenuation and arterial-stiffness reduction at a microgram dose that costs cents per day Knapen et al. 2013 Knapen et al. 2015. Japanese osteoporosis trials at the pharmacological MK-4 dose show a fracture-reduction signal large enough to support a national drug registration Shiraki et al. 2000 Cockayne et al. 2006. The functional biomarker (dp-ucMGP) tracks the intervention and predicts vascular events. The substance is cheap, safe at any reasonable dose, and the K1 null finding in the same cohorts argues that the signal is real and specific to menaquinones rather than an artefact of healthy-eating confounding.
Skeptic case. Cochrane's 2015 review found insufficient RCT evidence with hard endpoints to support primary CV prevention Hartley et al. 2015. The observational signal is consistent but dietary cohorts cannot fully exclude confounding by healthy-lifestyle patterns: menaquinone intake correlates with cheese consumption, which correlates with European-style diets that differ on many axes. The arterial-stiffness RCT showed a small absolute change; the coronary-calcium-score RCT (VitaK-CAC) was too small to draw conclusions on the clinical endpoint. The Japanese MK-4 fracture data have not replicated in non-Japanese populations and the Cochrane authors of the 2006 meta-analysis specifically flagged single-country, single-form dominance as a limitation Cockayne et al. 2006. K1-supplementation trials failed to show BMD effects Booth et al. 2008, which is consistent with the K1/K2 distinction but also raises the possibility that "any K" effects are smaller than the K2-believer literature suggests. No clinical guideline (USPSTF, ACC/AHA, AACE) recommends K2 supplementation for CV or bone outcomes in the general population.
Author's call. The mechanism is solid and the population-level signal is real but the trial evidence with hard clinical endpoints (fractures, MIs, mortality) is thinner than the mechanistic enthusiasm of supplement-side advocates would suggest. The reasonable call is that low-dose MK-7 (90–180 μg/day) is a low-cost, low-risk adjunct with a defensible mechanism and modest RCT-grade biomarker effects, worth taking — particularly in postmenopausal women, older adults, and CKD patients — without expecting transformative gains and without dropping better-evidenced cardiovascular and bone interventions for it. The high-dose MK-4 Japanese protocol is a different conversation that is properly a clinician-supervised drug decision, not a supplement decision. Evidence rating in the meta should be 3 (mechanism solid, multiple small trials, observational replication, no hard-endpoint Cochrane consensus); controversy 2 (specialists who follow the literature broadly agree on the mechanism and the bone-and-vessel rationale; mainstream guideline bodies have not endorsed it because the hard-endpoint trial isn't yet there).
4. Stakeholder and incentive map
- Supplement manufacturers. MK-7 is a high-margin product (notably the patented MenaQ7 form from NattoPharma/Gnosis); commercial incentive to overstate trial findings, particularly the "directs calcium" framing. Worth flagging when industry funding is present in trial author lists.
- Maastricht University group (Vermeer, Schurgers, Knapen). The dominant European research team on extra-hepatic K2 biology. Productive, but most of the European MK-7 RCT evidence traces back to this lab — replication in independent groups remains thinner than ideal.
- Japanese clinical community. MK-4 at 45 mg/day is established practice for osteoporosis in Japan, with regulatory backing dating to 1995. The body of evidence is internally coherent but country- and form-specific.
- Mainstream cardiology / endocrinology guideline bodies (USPSTF, ACC/AHA, NOF, AACE). Conservative posture: have not endorsed K2 supplementation, citing insufficient hard-endpoint RCT data. Not opposed; not recommending.
- Wellness / functional-medicine community. Strongly pro-K2, often tying it to the "calcium paradox" and to K2+D3 stacking. The enthusiasm runs ahead of trial data.
5. Population variability
Sex and life-stage. Trial evidence is densest in postmenopausal women, where the bone-loss gradient is steepest; effect sizes are largest in this group. Age. The CV-calcification and fracture stakes both rise sharply after 60; younger adults have proportionally less to gain from a calcium-direction intervention because their baseline arterial walls and bone mineral density are healthy. Diet. Japanese populations consuming natto regularly already have plasma MK-7 levels comparable to supplementation; the marginal benefit of additional K2 in this group is small. Non-natto-eating Western populations have markedly lower MK-7 status and a correspondingly larger plausible response. Renal status. CKD patients show the highest dp-ucMGP and the largest vascular calcification burden; the trial signal is biochemically clear but hard-endpoint data are still pending. Anticoagulation. Patients on warfarin are a special case — K2 affects INR; supplementation must be coordinated with prescriber. DOAC patients are unaffected.
6. Knowledge gaps
The decisive trial — a multi-thousand-patient, multi-year RCT of MK-7 with hard cardiovascular and fracture endpoints in a non-Japanese population — has not been run. The VitaK-CAC trial was a small step in that direction but underpowered. Long-term safety at chronic supplemental doses is well-established (no upper limit set, no toxicity signal at any tested dose), but the question of optimal dose — whether 90, 180, or 360 μg/day matters — is not settled. The dental claim has essentially no clinical-trial evidence; the historical Weston-Price observational case is interesting but predates modern epidemiology and is not citable. The interaction between K2 and statin therapy (statins inhibit synthesis of geranylgeranyl pyrophosphate, a precursor to MK-4) is mechanistically intriguing but clinically under-studied. The question of whether K2 status matters in younger adults without risk factors remains entirely a mechanistic argument, not a trial result.
Narrowing relative to brief. The brief lists "dental health" as one of four named consequences. The article surfaces dental only obliquely — there is no addressing section dedicated to it. The reason is honest: no human RCTs of K2 with dental endpoints (caries, remineralisation, periodontal) exist. The claim rides on osteocalcin's role in dentin formation and a mechanistic parallel to bone, plus the Weston-Price historical observational case which is not citable to modern standards. Giving it a full section would have required overclaiming. Flagged here so a reviewer doesn't read the omission as oversight.
Evidence rated 3, not 4. The mechanism is excellent, three observational cohorts agree, and several 3-year RCTs (Knapen 2013, 2015) move biomarkers and surrogate endpoints. But Cochrane 2015 (Hartley) explicitly concluded the hard-endpoint RCT data are insufficient for primary CV prevention, and the Japanese MK-4 fracture data have not cleanly replicated in non-Japanese populations. A 4 would require the missing multi-thousand-patient hard-endpoint trial.
Longevity rated 3, not 2 or 4. The observational signal (50% lower CHD mortality in top vs bottom K2-intake tertile in Rotterdam) is large enough that 2 understates it. The absence of hard-endpoint RCT data outside Japan keeps it below 4. The bone-fracture and arterial-stiffness lines add to a "meaningful disease prevention" case rather than dominant longevity effect.
beauty_cumulative at 1 (not 0). The aesthetic mapping is indirect: long-term bone preservation includes facial and jaw bone, and slower arterial calcification feeds into general aging trajectory. The mechanism is real but the effect on visible appearance is small and slow. A reader-facing zero would be defensible too; landed at 1 because the §3g evidence gate suggested the dimension warranted at least a token score given the bone/jaw connection.
Audience kept unrestricted. The trial evidence is densest in postmenopausal women (Knapen 2013, 2015) and the stakes are highest in 60+. Considered scoping to female / 40-59 + 60+ but the mechanism applies to everyone and the long-term framing is for adults broadly. Article's audience addressing-section material was folded into protocol and stakes rather than given its own section, since the population framing didn't change the protocol or the rationale.
Contraindication scoped to blood-thinners. Only warfarin / coumarin-class drugs interact; DOACs are unaffected. The closed vocabulary doesn't distinguish, so blood-thinners is the right token; the article body spells out the distinction.
Future-link candidates. vitamin-d3 (the D+K2 partnership is mechanistically tight and reader-relevant), coronary-calcium-scoring (the imaging test the K2 stakes ride on), dexa-scan (the bone-density test), warfarin-management or inr-monitoring (the anticoagulant coordination piece), calcium-supplementation (the "calcium paradox" thread). Listed in the out-of-scope addressing section as plain pointers; will be linkified once the sibling entries exist.
Separate-entry candidate. Menatetrenone (MK-4 45 mg/day) for osteoporosis is functionally a prescription drug in Japan, not a supplement. Treated as out-of-scope for the do-this-daily supplement framing here; warrants its own decide-action entry if the catalogue grows to cover clinician-supervised options for established osteoporosis.
Hard call on dosing range. Trial dose was 180 μg/day; protocol says 100–200 μg/day to match what's actually sold in the supplement market (most bottles are 100 μg or 200 μg). The 90 μg lower bound in some research papers (Theuwissen) was deliberately not surfaced — reads as false precision for a reader who'll buy a 100 or 200 μg bottle. Higher doses (360 μg, used in VitaK-CAC) were not surfaced because the trial endpoint there wasn't significant and the marginal cost/benefit at higher doses isn't established.
Voice call. Leaned hard into "you won't feel anything" framing in payoff and failure-modes. The most common reason readers abandon K2 is expecting a felt effect that this kind of intervention doesn't produce; honest framing is the only sustainable version. The countervailing risk — sounding underwhelming — was accepted as the right trade.
Vitamin K2 (Menaquinones)
About <data class="dose" value="$30/year">thirty dollars a year</data> for the standard daily dose. Trivial.
One capsule a day, taken with a meal that has some fat. That's it.
A real, modest disease-prevention angle: lower coronary calcification across three big cohorts and slower bone loss in 3-year trials. Not a silver bullet — a brick in the wall.
Solid mechanism, three matching population cohorts, multiple 3-year trials on biomarker endpoints. The decisive hard-outcome trial in heart attacks and fractures isn't yet done.
Slower bone loss and a slower-stiffening arterial wall feed into how you age across decades. Subtle and slow, not anything you spot in a mirror this year.