🧬 Aging Wiki

d3-k2-calcium-partitioning

Properties1
aliasescalcium paradox resolution, bone-vascular axis synergy, D3+K2 synergy, D3 K2 calcium partitioning, vitamin D3 vitamin K2 bone vascular

24 min read

D3 + K2 Calcium-Partitioning Hypothesis

The calcium-partitioning hypothesis proposes that combined supplementation with vitamin D3 (cholecalciferol / calcitriol) and vitamin K2 (menaquinone-7, MK-7) synergistically partitions calcium into bone and away from arterial walls — resolving what has been called the “calcium paradox” of aging, in which bone loses mineral while arteries paradoxically gain it. The hypothesis rests on a mechanistically coherent but clinically under-confirmed chain: calcitriol upregulates synthesis of vitamin K-dependent Gla proteins (osteocalcin and matrix-gla-protein) via vitamin D response elements (VDREs); vitamin K2 provides the cofactor required by ggcx to gamma-carboxylate those proteins; carboxylated MGP then inhibits arterial calcification, while carboxylated osteocalcin supports bone mineralization. The net claimed effect is less arterial calcification plus better bone mineral density — an appealing “dual benefit” for the bone-vascular axis paradox documented in bone §6.

Status: contested. Biomarker evidence (dp-ucMGP normalization) is robust and replicated. Functional endpoint evidence (arterial stiffness) rests on a single unreplicated RCT (Knapen 2015). Hard endpoint evidence (aortic valve calcification, coronary calcium, MACE) is largely null or absent. No hard-endpoint RCT of the combination itself exists. The mechanistic “D3 upregulates Gla protein synthesis” claim, while supported by classic molecular biology, has not been tested in a human dose-response framework that isolates this link.

The claim

Dietary vitamin D3, when converted to calcitriol (1,25(OH)₂D₃) by renal 1α-hydroxylase, activates the vitamin D receptor (VDR), which binds VDREs in the promoters of both the MGP gene and the BGLAP (osteocalcin) gene, increasing their transcription in vascular smooth muscle cells and osteoblasts, respectively. This upregulates the pool of uncarboxylated Gla-protein substrates available for ggcx-mediated activation. Vitamin K2 (MK-7), through the vitamin-k-cycle, provides the cofactor (vitamin K hydroquinone) that GGCX requires for gamma-carboxylation. The carboxylated outputs — cMGP in the arterial wall and cOC in bone — then divide calcium partitioning: cMGP blocks hydroxyapatite nucleation in vascular media; cOC incorporates calcium into bone mineral. In this model, D3 raises the carboxylatable substrate ceiling while K2 fills it, making the combined intervention mechanistically super-additive in a bounded, substrate-limited fashion.

Status: contested

The hypothesis is mechanistically motivated (the individual arrows are individually supported) but has not been tested as a unit by a powered hard-endpoint RCT. The contested status reflects:

  1. No combined D3+K2 RCT has tested a hard clinical endpoint (MACE, fracture, all-cause mortality). AVADEC and its substudies have tested calcification surrogates and one composite safety endpoint.
  2. The largest co-administration RCT (AVADEC) used K2+D3 specifically and was null on its primary endpoint (AVC progression, p=0.64); a CAC substudy (Hasific 2023) showed a non-significant trend (p=0.089) with a hypothesis-generating subgroup signal (CAC ≥400 AU; p=0.047).
  3. D3 monotherapy RCTs (VITAL, D-Health) are null on CVD and cancer mortality in the general population.
  4. The “D3 → Gla protein transcription” link, while supported by classic cell and animal studies, has not been shown to be the rate-limiting step in humans supplementing at standard doses.
  5. K2 alone has a plausible but largely surrogate-only evidence base (see vitamin-k2).

Evidence does not falsify the hypothesis; it fails to confirm the prediction at the hard-endpoint level. The hypothesis remains biologically plausible and generates testable predictions (see below).

Key predictions

The hypothesis predicts that, compared with neither supplement, D3+K2 co-administration should:

  1. Reduce circulating dp-ucMGP more than K2 alone (because D3 increases MGP substrate supply, amplifying K2’s carboxylation effect). [Testable; tested only in very small populations (El Borolossy 2022 pediatric CKD, n=60) — not confirmed in adults with primary-prevention populations.]
  2. Slow arterial calcification (CAC score, aortic valve calcification, pulse-wave velocity) more than K2 alone. [Directly tested in AVADEC 2022 — null; tested in El Borolossy 2022 on dp-ucMGP/ucOC surrogates only.]
  3. Improve bone mineral density more than D3 alone (because K2 supplies the cofactor to carboxylate the D3-induced osteocalcin). [Weakly tested in pediatric populations; no powered adult primary-prevention RCT of this specific prediction exists.]
  4. Show greater calcification benefit in populations with lower baseline K2 status (where the carboxylation ceiling is more substrate-limited). [Unconfirmed; AVADEC did not stratify by baseline dp-ucMGP or K2 status for its primary analysis.]

Evidence supporting

Genetic / animal (strong mechanism)

  • Mgp-/- mice develop fatal arterial calcification within 2 months — demonstrating that uncarboxylated or absent MGP permits uncontrolled vascular hydroxyapatite deposition. Keutel syndrome (biallelic MGP loss-of-function in humans) mirrors the mouse phenotype. This establishes MGP carboxylation as non-redundant for vascular calcification prevention. See matrix-gla-protein (verified 2026-05-23).

  • GGCX loss-of-function (VKCFD1) produces an MGP carboxylation deficit resulting in premature arterial calcification alongside skeletal phenotypes, confirming that the GGCX-dependent carboxylation step is load-bearing in humans. See ggcx (verified 2026-06-02).

Molecular biology of VDRE-driven Gla-protein transcription

  • Osteocalcin/BGLAP: 1,25(OH)₂D₃ induces osteocalcin transcription via a well-characterised VDRE in the osteocalcin gene promoter; VDR-DNA binding requires both the hormone and a nuclear accessory factor 1. Demay & Kronenberg 1989 identified the specific promoter regions mediating this effect 2. The rat osteocalcin VDRE has been characterized as a functional enhancer element 3. gap/needs-human-replication — while the molecular biology of VDRE-mediated osteocalcin induction is well established in cell and animal models, the quantitative contribution of D3 supplementation to the circulating or tissue osteocalcin pool in vitamin-D-replete adults is uncertain and has not been tested in a dose-response RCT isolating this link.

  • MGP: The human MGP gene promoter contains putative vitamin D response element sequences 4. Retinoic acid (a related fat-soluble nuclear receptor ligand) induces MGP gene expression ~25-fold in human osteoblasts 5, establishing the principle that nuclear receptor engagement drives MGP transcription. Whether 1,25(OH)₂D₃ at supplemental doses drives a biologically meaningful increase in VSMC MGP output in humans is gap/unsourced — the VDRE-MGP link is biologically plausible but its quantitative relevance to supplementation in humans is not directly demonstrated in primary sources this wiki can point to.

dp-ucMGP biomarker (robust)

  • K2 (MK-7) supplementation consistently and dose-dependently reduces dp-ucMGP across all studied populations — this is the most replicated finding in the K2 literature. See vitamin-k2 § dp-ucMGP normalization.

  • El Borolossy 2022 combined K2+D3 RCT (pediatric CKD, n=60): Combined vitamin K2 (100 µg MK-7) + vitamin D (10 µg) showed “the most significant decrease in dp-ucMGP and ucOC” compared with either supplement alone in pediatric hemodialysis patients 6. This is the only direct test of prediction #1 (D3+K2 > K2 alone on the biomarker). Caveats are severe: pediatric CKD patients have extreme baseline K deficiency and vitamin D dysregulation; n=60; population not generalizable to primary-prevention adults. gap/needs-replication in adults.

Combined vitamin D + vitamin K intervention trials (direct, but none is MK-7 + D3 on a hard endpoint)

Besides El Borolossy 2022, two other RCTs supplemented vitamin D and vitamin K together and reported benefit on a vascular surrogate — but each carries a form/endpoint caveat that keeps it short of confirming the D3 + MK-7 prediction:

  • Braam 2004 (n=181 postmenopausal women, 3 years): a placebo vs minerals+D vs minerals+D+vitamin K1 design; the combined D+K1 arm preserved carotid arterial elasticity (distensibility/compliance) while the placebo and D-alone arms deteriorated 7. An early, direct demonstration that combined D+K outperforms D alone on a functional vascular measure — but the K was phylloquinone (K1), not MK-7, and the endpoint is arterial elasticity, a surrogate. See braam-2004-vitamin-dk-vessel-elasticity.
  • Asemi 2016 (n=66 overweight type-2-diabetic patients with CHD, 12 weeks): D + vitamin K (90 µg) + calcium co-supplementation reduced maximum left carotid intima-media thickness (−0.04 vs +0.04 mm, P=0.02) and improved insulin sensitivity, HDL and hs-CRP 8. Caveats are heavy: short (12 weeks), small, a three-way D+K+Ca combination (calcium confounds the calcium-partitioning interpretation), in a metabolically-compromised population — and the authors themselves note the CIMT effect “could be a chance finding.”

Taken together, the direct combined-arm evidence (El Borolossy, Braam, Asemi) is positive on biomarkers and surrogates in small/short/sick-population trials that used K1 or added calcium, while the one trial that tested MK-7 + D3 on a structural calcification endpoint — AVADEC — was null (below). The combination’s appeal rests on surrogate/biomarker engagement, not on a hard-endpoint or even a clean MK-7+D3 surrogate win.

Arterial stiffness (single unreplicated positive RCT)

  • Knapen 2015 (n=244 healthy postmenopausal women, MK-7 180 µg/day × 3 years): MK-7 significantly decreased cfPWV and Stiffness Index β vs placebo; the subgroup with above-median baseline stiffness showed greater benefit. dp-ucMGP reduced ~50% in the MK-7 group. This is the primary positive functional endpoint supporting the K2 arm. See vascular-calcification (verified 2026-05-23) and vitamin-k2 (verified 2026-05-08) for full numerics. The trial is MK-7 alone, not combined D3+K2.

Observational epidemiology (consistent but confounded)

  • Rotterdam Study (Geleijnse 2004, n=4807): Dietary menaquinone (K2) in the upper tertile vs lower tertile was associated with reduced CHD mortality (RR 0.43, 95% CI 0.24–0.77) and reduced all-cause mortality in adjusted models. Phylloquinone (K1) was not significantly associated. See vitamin-k2. Effect sizes are large relative to RCT data — residual confounding by dietary pattern (K2 from fermented dairy, a diet-quality marker) is likely. This is observational support for the K2 arm of the hypothesis only; D3 is not a variable in this analysis.

Bone benefit (K2 surrogate data)

  • Knapen 2013 (n=244 postmenopausal women, 180 µg/d MK-7 × 3 years): MK-7 significantly reduced age-related decline in lumbar spine and femoral neck BMC/BMD and improved vertebral height. See vitamin-k2 § Bone evidence.

Evidence against

Combined K2+D3 null on hard calcification endpoints (AVADEC 2022)

  • AVADEC (Diederichsen 2022, Circulation, n=365 men, mean age 71): This is the most directly relevant trial — it used both MK-7 (720 µg/day) and vitamin D3 (25 µg/day) together for 24 months in men with established aortic valve calcification (AVC score >300 AU). Primary endpoint was null: AVC progression +275 AU (treatment) vs +292 AU (placebo), p=0.64 9. dp-ucMGP was significantly reduced in the treatment arm, confirming biomarker engagement. All secondary endpoints including cardiovascular events, valve function, and all-cause mortality were null. This directly tests the combined D3+K2 hypothesis on a calcification endpoint and finds no benefit — a critical negative result. See vitamin-k2 (verified 2026-05-08).

  • AVADEC coronary artery calcification substudy (Hasific 2023, JACC Advances, n=304): From the same AVADEC cohort, excluding participants with prior ischemic heart disease. The primary endpoint (mean CAC progression, all participants) was non-significant (Δ203 AU treatment vs Δ254 AU placebo; p=0.089). A pre-specified subgroup with baseline CAC ≥400 AU showed a significant difference (Δ288 vs Δ380 AU; p=0.047). Safety composite events were fewer in the treatment arm (1.9% vs 6.7%; p=0.048). The CAC ≥400 AU subgroup result and safety signal are hypothesis-generating but do not overturn the overall neutral primary endpoint 10.

  • AVADEC epicardial adipose tissue substudy (Hasific 2025, Atherosclerosis, n=388): The same MK-7 720 µg + D3 25 µg dose did not affect epicardial adipose tissue, pericardial adipose tissue, or systemic inflammatory markers despite significant dp-ucMGP reduction 11. Mechanistic biomarker signal does not translate to structural or inflammatory endpoint benefit.

D3 monotherapy null on hard endpoints (VITAL, D-Health)

  • VITAL (Manson 2019, NEJM, n=25,871): Vitamin D3 2000 IU/day for median 5.3 years did not reduce major cardiovascular events (HR 0.97, 95% CI 0.85–1.12, p=0.69) or total cancer incidence (HR 0.96, 95% CI 0.88–1.06, p=0.47) vs placebo 12. A possible cancer mortality reduction emerged in secondary analyses excluding the first 2 years of follow-up, but this was a secondary, post-hoc analysis. For the cardiovascular and bone-vascular axis claims of the D3+K2 hypothesis, VITAL provides the strongest evidence that D3 alone does not shift hard cardiovascular outcomes. The D3 arm of the hypothesis thus carries no hard-endpoint support.

K2 monotherapy largely null on calcification endpoints

  • The Li/Wang 2023 meta-analysis (14 RCTs, n=1,533 enrolled) found a statistically significant but small reduction in CAC progression with vitamin K supplementation in a 4-study sub-analysis (MD −17.37 Agatston units, 95% CI −34.18 to −0.56, p=0.04), but 94.4% of the statistical weight derived from a single community trial (Shea 2009). The result is highly sensitive to exclusion of that trial. See vascular-calcification.

  • Murali 2023 systematic review (49 RCTs, n=9,901, non-CKD adults): Classified vitamin K as “possibly reduces progression” — a guarded category — and concluded that “the vitamin K pathway may not represent a therapeutic target to mitigate this in a non-CKD cohort.” Conflicting conclusions from Li/Wang 2023 vs Murali 2023 are explained by different inclusion criteria and endpoint heterogeneity, not a clean CKD-vs-non-CKD split. See vascular-calcification for full discordance framing.

No hard-endpoint D3+K2 combination trial exists

There is no completed, powered RCT in adults testing the D3+K2 combination against a primary hard endpoint (MACE, fracture, or all-cause mortality) in a primary-prevention population. The hypothesis’s central clinical prediction remains untested at the hard-endpoint level. This is categorically different from being falsified — but the prediction has not been tested, and the available biomarker and surrogate evidence is insufficient to substitute for a hard endpoint.

Evidence type taxonomy

EvidenceTypeTierStatus
MGP-null mice / Keutel syndrome (genetic LOF)Direct mechanismGenetic LoFConfirms carboxylated MGP is essential; does not validate supplementation in replete adults
VDRE-mediated osteocalcin/MGP transcription (molecular biology)MechanisticCell/animalWell-established mechanism; quantitative relevance at supplement doses in replete humans: undemonstrated
dp-ucMGP normalization with K2 (biomarker)Target engagementReplicated RCTRobust — but dp-ucMGP is not a validated surrogate endpoint for MACE or calcification
El Borolossy 2022 (pediatric CKD, n=60) combined K2+D3 on dp-ucMGP/ucOCCombined biomarkerSmall RCT, extreme populationSupports prediction #1 in an extreme-deficiency context; low generalizability
Braam 2004 (n=181, 3 yr) combined D+K1 on arterial elasticityCombined surrogate (functional)RCTCombined D+K1 > D-alone on vessel elasticity; K1 not MK-7, surrogate endpoint
Asemi 2016 (n=66, 12 wk) D+K+Ca on CIMTCombined surrogateSmall/short RCTWeak positive; 3-way (Ca confound), sick population, authors flag possible chance finding
Knapen 2015 (n=244, cfPWV) K2 aloneSurrogate endpointSingle unreplicated RCTPositive for arterial stiffness in postmenopausal women; K2 only (not combined)
Rotterdam Study (Geleijnse 2004) dietary K2 and CHD mortalityObservationaln=4807, prospective cohortFavorable; likely confounded
AVADEC 2022 (n=365, combined K2+D3, AVC progression)Hard-ish calcification endpointRCTNull — the largest combined trial on AVC; directly contradicts the hypothesis
Hasific 2023 JACC Adv (n=304, combined K2+D3, CAC progression)Calcification endpointRCT substudyPrimary endpoint null (p=0.089); CAC ≥400 AU subgroup p=0.047; safety composite favored treatment (p=0.048) — hypothesis-generating only
VITAL 2019 (n=25,871, D3 alone, CVD/cancer)Hard endpointLarge RCTNull for CVD; D3 monotherapy contributes no hard-endpoint support
Murali 2023 non-CKD systematic review, 49 RCTsSynthesized evidenceMeta-analysisNegative/guarded for vitamin K on CAC in non-CKD adults

What would update this hypothesis

Toward confirmation:

  1. A powered (n≥1000), long-duration (≥5 years) RCT in non-CKD adults with early CAC (score 1–400, not established severe calcification) testing D3+K2 vs K2 alone vs D3 alone vs placebo, with MACE or CAC progression as primary endpoint and baseline dp-ucMGP and 25-OH-D status as pre-specified stratifiers. No such trial is currently registered or recruiting as of 2026-06-14.
  2. A mechanistic RCT demonstrating that pharmacologically dosed D3 in vitamin-D-replete adults raises circulating or tissue MGP and osteocalcin protein levels (the “D3 lifts the substrate ceiling” prediction) in isolation from confounders.
  3. Confirmation in an adult primary-prevention population that combined K2+D3 reduces dp-ucMGP more than K2 alone (prediction #1; currently tested only in pediatric CKD, n=60).

Toward falsification:

  1. A powered primary-prevention combined K2+D3 RCT with null result on both calcification and bone endpoints would substantially narrow the hypothesis, particularly if baseline dp-ucMGP was low (confirming target engagement).
  2. A systematic review confirming no dp-ucMGP-reducing effect of D3 in vitamin-D-replete adults (which would undermine the “D3 expands substrate pool” arm).
  3. A Mendelian randomization study using genetic variants in CYP27B1 (the gene encoding 1α-hydroxylase) or VDR as instruments, showing no causal effect on arterial calcification or bone density independent of phosphate homeostasis — which would localize any D3 effect to PTH/phosphate rather than the VDRE-Gla protein axis.
Vitamin D3
  ↓ [converted by 25-hydroxylase + 1α-hydroxylase]
Calcitriol (1,25(OH)₂D₃)
  ↓ [VDR + VDRE binding at BGLAP and MGP promoters]
  → SUPPORTED in cell + animal models; quantitative relevance in replete adults: NOT DEMONSTRATED
Increased MGP and osteocalcin protein synthesis
  ↓
Larger pool of uncarboxylated Gla-protein substrate
  ↓ [gated by vitamin K2 (MK-7) + [[ggcx]]]
  → Requires adequate K2; this is where K2 enters the combined hypothesis
Carboxylated MGP (cMGP) in vascular wall    Carboxylated osteocalcin in bone
  ↓                                             ↓
  Inhibits hydroxyapatite nucleation            Incorporates Ca²⁺ into hydroxyapatite
  → STRONGLY SUPPORTED (genetic LoF + human)   → SUPPORTED (biochemical + RCT surrogate)
Slower vascular calcification progression
  → PARTIALLY SUPPORTED (K2 surrogate; AVADEC = null combined K2+D3)
Improved bone mineral density
  → PARTIALLY SUPPORTED (K2 alone; Knapen 2013 positive)
Reduced cardiovascular events / fracture mortality
  → NOT DEMONSTRATED (no hard-endpoint combined trial; D3 alone null in VITAL)

The “D3 + K2 = synergy” claim is specifically located between links 2 and 3 in this chain. The shared downstream biology (cMGP inhibits calcification; cOC supports bone) is well supported; the unique D3 contribution to substrate supply at human supplemental doses is the weakest link.

  • information-theory-of-aging — adjacent at the level of “biological information loss drives aging pathology”; the calcification-partitioning hypothesis operates at a much more specific mechanistic level
  • The “warfarin = calcification risk” corollary (see vitamin-k-cycle) is the pharmacological negative of this hypothesis: VKOR inhibition by warfarin prevents MGP carboxylation, demonstrating the pathway’s necessity via iatrogenic calcification
  • altered-intercellular-communication — the primary hallmark linkage: vascular calcification drives progressive arterial stiffening, which degrades the pulse-wave information content of arterial mechanosensing and baroreceptor function; the VSMC-to-ECM communication shift from compliant to calcified is the canonical intercellular communication disruption in this axis
  • chronic-inflammation — uncarboxylated Gas6 (a parallel VKDP substrate of ggcx) impairs macrophage efferocytosis, driving unresolved inflammation; also, arterial calcification and the associated SASP from senescent VSMCs contribute to systemic inflammaging
  • deregulated-nutrient-sensing — the FGF23/Klotho/phosphate axis (see bone §5 and vascular-calcification) intersects with this hypothesis: calcitriol synthesis is regulated by FGF23-mediated suppression of 1α-hydroxylase, so aging-associated FGF23 rise reduces the D3→calcitriol conversion efficiency, potentially limiting the D3 arm of the hypothesis in older adults
  • vitamin-k2 — the K2 arm; full evidence base there (verified 2026-05-08)
  • vitamin-k-cycle — enzymatic cycle that GGCX and VKOR1 operate; the mechanistic substrate for the K2 arm (verified 2026-06-02)
  • matrix-gla-protein — primary effector protein in the vascular arm (verified 2026-05-23)
  • osteocalcin — primary effector protein in the bone arm (protein page: stub — see below)
  • vascular-calcification — the downstream process the hypothesis targets (verified 2026-05-23)
  • arterial-stiffening — the functional outcome downstream of vascular calcification
  • bone — the bone-vascular axis paradox context (see §6 of that page, verified 2026-05-23)

Vitamin D seeding note

There is no molecules/compounds/vitamin-d.md page in the wiki as of 2026-06-14. The VITAL trial null result, calcitriol endocrine biology (intestinal Ca absorption, PTH suppression, 1α-hydroxylase regulation by FGF23), and the detailed D3+K2 synergy quantification all require a dedicated vitamin D compound page before the D3 arm of this hypothesis can be fully wiki-grounded. See model/vitamin-d3-k2-intervention-design.md for the Phase 0 content plan. Until that page exists, D3-specific claims on this page rely on external sources cited below and should be considered gap/unsourced at the wiki-internal cross-link level.

Limitations and open questions

  • #gap/needs-human-replication — No combined D3+K2 RCT has tested a hard clinical endpoint (MACE, fracture, all-cause mortality) in a non-CKD primary-prevention adult population. AVADEC tested the combination on a calcification endpoint (AVC score) and was null (p=0.64).
  • #gap/unsourced — The quantitative contribution of calcitriol (at supplemental D3 doses) to MGP and osteocalcin transcription in vitamin-D-replete adults has not been directly demonstrated in a human RCT isolating this link. The molecular biology of VDRE-driven Gla-protein transcription is established in cells and animals; the dose-response in humans at physiological-to-supplemental D3 levels is not.
  • #gap/dose-response-unclear — Optimal D3 + K2 dose combination is unknown. AVADEC used 720 µg MK-7 + 25 µg D3; the Knapen studies used 180 µg MK-7. Whether the AVADEC null reflects the specific calcification endpoint (aortic valve, which may differ mechanistically from medial calcification), the combination dose, or a genuine absence of benefit is unresolved.
  • #gap/contradictory-evidence — Knapen 2015 (K2 alone, 180 µg/d, n=244 postmenopausal) showed positive cfPWV benefit; AVADEC (K2+D3, 720 µg/d, n=365 men, AVC endpoint) showed null. The discordance could reflect endpoint difference (cfPWV vs AVC score), population difference (women vs men; stiffness vs established calcification), dose difference, or genuine heterogeneity of effect. These discordances are not reconciled.
  • Population specificity — Positive K2 signals cluster in populations with extreme vitamin K deficiency (CKD, dialysis, established vascular disease). Whether the hypothesis applies to healthy aging adults with moderate dietary K2 insufficiency is uncertain.
  • D3 confounded by PTH/phosphate axis — FGF23 rise with aging suppresses renal 1α-hydroxylase, reducing calcitriol conversion from D3. Supplemental D3 at standard doses may not overcome FGF23-mediated suppression in older adults, limiting the D3 contribution to the hypothesized substrate-supply effect.
  • AVADEC population caveat — Men with AVC score >300 AU have advanced established calcification. The hypothesis’s mechanism (inhibiting new deposition) may be inadequate to slow progression in already-calcified tissue. A primary-prevention population with early calcification (CAC 0–100) might be a better test.

Notes / synthesis judgment

The D3+K2 calcium-partitioning hypothesis occupies an unusual epistemic position: the individual biological arrows are credible, the endpoint RCT record is largely null, and the hypothesis has never been tested by a hard-endpoint trial in its intended primary-prevention population. This is mechanistically-motivated, biomarker-supported, but hard-endpoint-untested — distinct from both “confirmed” and “falsified.”

The status: contested assignment reflects: (a) AVADEC (the largest combined D3+K2 RCT) was null on its primary AVC endpoint (p=0.64); a CAC substudy (Hasific 2023) shows a non-significant trend (p=0.089) with a hypothesis-generating subgroup signal but an overall neutral primary endpoint; (b) D3 monotherapy is null on CVD in VITAL; (c) K2 surrogate evidence is replicated but the single cfPWV positive RCT (Knapen 2015) is unreplicated; (d) the specific prediction of additive/synergistic D3+K2 benefit over monotherapy is tested only in pediatric CKD (El Borolossy 2022, n=60, extreme population).

The wiki framing at vitamin-k2 correctly positions the evidence: “mechanistic plausibility, weak positive observational data, a single positive arterial stiffness RCT, and null hard-endpoint data.” Adding D3 to this picture does not change the hard-endpoint assessment and introduces additional confounds (AVADEC null; VITAL D3 null).

The cost of D3+K2 supplementation at standard doses is low (~$15–25/month combined) and the safety profile is excellent in patients not on warfarin/coumarin anticoagulants. For aging adults not on VKA anticoagulation, the combination represents a defensible low-risk hedging strategy for vascular health, but this is a very different claim from “demonstrated to slow arterial calcification or reduce cardiovascular mortality.” The wiki’s job is to represent the field honestly: the hypothesis is contested, not confirmed.

Footnotes

Footnotes

  1. PMID 2175914 · Liao J, Ozono K, Sone T, McDonnell DP, Pike JW · Proc Natl Acad Sci USA 1990;87(24):9751–9755 · experimental (cell-free binding assays + transfection) · VDR binds osteocalcin VDRE in a 1,25(OH)₂D₃-dependent manner; binding requires both the hormone and a nuclear accessory protein extracted from mammalian cells · model: cell-free systems + mammalian cell nuclear extracts · foundational VDR-osteocalcin VDRE binding paper ↩

  2. doi:10.1016/s0021-9258(18)94173-6 · Demay MB, Roth DA, Kronenberg HM · J Biol Chem 1989;264(5):2279–2282 · experimental · defined the promoter regions of the rat osteocalcin gene mediating 1,25-dihydroxyvitamin D3 responsiveness; identified the functional VDRE sequence in the osteocalcin promoter · model: in vitro transcription + deletion analysis of rat osteocalcin promoter · canonical molecular biology reference for VDR→osteocalcin transcription ↩

  3. PMID 10321839 · Sneddon WB, Demay MB · J Cell Biochem 1999;73(3):400–407 · experimental · characterized an enhancer element (core GGTTTGG motif at positions −420 to −414; full protein-binding element between −430 and −414) required for 1,25(OH)₂D₃-dependent transactivation of the rat osteocalcin gene; element works cooperatively with the VDRE · model: rat osteocalcin promoter/reporter in cell transfection assays ↩

  4. PMID 2394711 · Cancela L, Hsieh CL, Francke U, Price PA · J Biol Chem 1990;265(25):15040–15048 · molecular biology · determined the structure of the human MGP gene (4 exons, 3 introns, chromosome 12p); identified putative hormone-response elements in the MGP promoter region including sequences homologous to vitamin D and retinoic acid response elements · foundational human MGP gene structure paper ↩

  5. PMID 1727694 · Cancela ML, Price PA · Endocrinology 1992;130(1):102–108 · experimental · retinoic acid induces MGP mRNA ~25-fold in human osteoblasts (EC₅₀ ~0.1 µM); retinoic acid effects in bone/skin/cartilage may operate through MGP synthesis induction · model: human osteoblast-like cells · establishes principle that fat-soluble nuclear receptor ligands (retinoids, by extension vitamin D) can drive MGP transcription ↩

  6. doi:10.1038/s41430-021-01050-w · PMID 34845313 · El Borolossy R, El-Farsy MS · Eur J Clin Nutr 2022;76(6):848–854 · rct · n=60 pediatric hemodialysis patients; 4 arms: K2 100 µg MK-7 + D 10 µg; K2 alone; D alone; placebo · combined K2+D arm showed the most significant decrease in dp-ucMGP and ucOC among arms · primary significance: only direct test of the combined supplementation prediction on dp-ucMGP/ucOC biomarkers; major caveat: pediatric CKD, extreme deficiency context, n=60, low generalizability to primary-prevention adults · model: homo-sapiens (pediatric ESRD) ↩

  7. braam-2004-vitamin-dk-vessel-elasticity · doi:10.1160/TH03-07-0423 · PMID 14961167 · Braam LA, Hoeks AP, Brouns F, Hamulyák K, Gerichhausen MJ, Vermeer C · Thromb Haemost 2004;91(2):373–380 · rct · n=181 postmenopausal women (108 analysed), 3 arms (placebo / minerals+D / minerals+D+K1), 3 years · combined D+K1 arm preserved carotid distensibility & compliance while placebo and D-alone deteriorated; abstract-confirmed between-group contrasts: DC 8.8% / CC 8.6% / pulse-pressure 6.3% (all p<0.05), Young’s modulus 13.2% (p<0.01); IMT no change (NS); exact per-arm absolute values not confirmed (closed-access gap/no-fulltext-access) · K form = phylloquinone (K1), NOT MK-7 · model: homo-sapiens (postmenopausal women) ↩

  8. doi:10.1017/S0007114516001847 · PMID 27198036 · Asemi Z, Raygan F, Bahmani F et al. · Br J Nutr 2016;116(2):286–293 · rct · n=66 overweight T2DM patients with CHD; D 5 µg + vitamin K 90 µg + Ca 500 mg twice daily vs placebo × 12 weeks · max left CIMT −0.04 (sd 0.22) vs +0.04 (sd 0.09) mm, P=0.02; improved insulin/HOMA-IR/HDL/hs-CRP/MDA · CAVEATS: three-way D+K+Ca (Ca confounds), 12 weeks, n=66, metabolically-compromised population; authors state the CIMT effect “could be a chance finding”; K form unspecified (90 µg) · model: homo-sapiens (T2DM+CHD) ↩

  9. doi:10.1161/CIRCULATIONAHA.121.057008 · PMID 35465686 · Diederichsen ACP, Lindholt JS, Möller S et al. (18 authors) · Circulation 2022;145(18):1387–1397 · rct · n=365 men (182 intervention, 183 placebo), mean age 71.0 ±4.4, AVC score >300 AU; combined MK-7 720 µg/day + D3 25 µg/day vs placebo × 24 months · primary endpoint (AVC progression): treatment +275 AU (95% CI 225–326) vs placebo +292 AU (95% CI 246–338), treatment effect −17 AU (95% CI −86 to 53), p=0.64 (null); all secondary endpoints null including aortic valve area, peak jet velocity, cardiovascular events (10 vs 10; p=0.99), all-cause death (1 vs 4; p=0.37); dp-ucMGP significantly reduced in treatment arm (−212 vs +45 pmol/L; p<0.001) · KEY NEGATIVE: only RCT directly testing combined D3+K2 on a calcification endpoint; null despite confirmed target engagement · model: homo-sapiens (men 65–74 with established aortic valve calcification) ↩

  10. doi:10.1016/j.jacadv.2023.100643 · PMID 38938724 · Hasific S, Oevrehus KA, Lindholt JS et al. · JACC Advances 2023;2(9):100643 · rct substudy (AVADEC) · n=304 men (from 389 AVADEC participants; excluded prior ischemic heart disease) · MK-7 720 µg + D3 25 µg vs placebo × 24 months · primary endpoint (mean CAC progression): Δ203 AU treatment vs Δ254 AU placebo, p=0.089 (non-significant trend); pre-specified subgroup CAC ≥400 AU: Δ288 vs Δ380 AU, p=0.047 (significant); safety composite favored treatment (1.9% vs 6.7%, p=0.048); no significant difference in noncalcified plaque volume progression · overall neutral primary endpoint; subgroup and safety signals hypothesis-generating · model: homo-sapiens (men 65–74 with AVC ≥300 AU, no prior MI) ↩

  11. doi:10.1016/j.atherosclerosis.2025.120540 · PMID 41100911 · Hasific S, Ravn EJ, Rasmussen LM et al. · Atherosclerosis 2025;410:120540 · rct substudy (AVADEC) · n=388 men aged 65–74 · MK-7 720 µg + D3 25 µg vs placebo × 24 months · no effect on epicardial adipose tissue, pericardial adipose tissue, or systemic inflammatory markers; dp-ucMGP reduced significantly (target engagement confirmed) · biomarker-positive, structural-endpoint-negative · model: homo-sapiens (men with established aortic valve calcification) ↩

  12. doi:10.1056/NEJMoa1809944 · PMID 30415629 · Manson JE, Cook NR, Lee I-M et al.; VITAL Research Group · N Engl J Med 2019;380(1):33–44 · rct · n=25,871 (men ≥50, women ≥55); D3 2000 IU/day vs placebo × median 5.3 years · primary endpoint: invasive cancer (HR 0.96, 95% CI 0.88–1.06, p=0.47) and major CVD events (HR 0.97, 95% CI 0.85–1.12, p=0.69) — both null · secondary cancer mortality analysis suggested possible benefit (post-hoc, excluding years 1–2) · D3 monotherapy is null on CVD hard endpoints at this dose and follow-up in the general primary-prevention population · model: homo-sapiens (community-dwelling adults) ↩

Graph View

Backlinks