🧬 Aging Wiki

Vitamin K supplementation and progression of coronary artery calcium in older men and women

7 min read

Vitamin K supplementation and progression of coronary artery calcium in older men and women

Shea MK, O’Donnell CJ, Hoffmann U, Dallal GE, Dawson-Hughes B, Ordovas JM, Price PA, Williamson MK, Booth SL · American Journal of Clinical Nutrition 89(6):1799–1807 · 2009 · DOI: 10.3945/ajcn.2008.27338

Open access via PMC2682995.

TL;DR

A 3-year RCT in 388 healthy older adults testing 500 µg/d phylloquinone (vitamin K1) plus multivitamin versus multivitamin alone found no significant difference in coronary artery calcium (CAC) progression in the intention-to-treat analysis (mean Agatston change 27±6 vs 37±7; p=0.26). A pre-specified per-protocol analysis in participants with ≥85% adherence found significantly less CAC progression in the vitamin K1 arm (mean change 17 vs 37 Agatston units; p=0.03). An exploratory subgroup with pre-existing CAC (baseline Agatston >10) showed the most striking between-arm difference (25 vs 59 Agatston units; p=0.03). The trial is critically important to read carefully because it is the sole RCT dominating the Li 2023 CAC meta-analysis (~94% weight), the headline “vitamin K slows calcification” conclusion rests entirely on the per-protocol subgroup, and the compound tested was K1 (phylloquinone), not K2 (menaquinone).

Background and rationale

Vascular calcification is an active, cell-mediated process — not passive mineral deposition — and is an independent predictor of cardiovascular events and all-cause mortality. Matrix Gla protein (MGP) is a potent local inhibitor of vascular calcification whose activity requires vitamin K-dependent gamma-carboxylation; undercarboxylated MGP (ucMGP) is associated with arterial calcification in epidemiological studies. This trial was designed to test whether pharmacological vitamin K1 supplementation could slow CAC progression, using the Framingham Heart Study infrastructure and dual-energy X-ray absorptiometry–calibrated CT for CAC scoring.

Study design

FeatureDetail
DesignRandomised, double-blind, placebo-controlled
PopulationAmbulatory men and postmenopausal women aged 60–80 years (Framingham Heart Study offspring and spouse cohorts); known coronary heart disease, atrial fibrillation, pacemaker, or warfarin use were exclusion criteria
n (randomised)388 (200 treatment; 188 control)
Intervention500 µg/d phylloquinone (vitamin K1) + multivitamin + calcium 600 mg/d + vitamin D 400 IU/d (cholecalciferol)
ComparatorMultivitamin + calcium 600 mg/d + vitamin D 400 IU/d (no phylloquinone)
Duration3 years
Primary endpointChange in coronary artery calcium (Agatston score) on CT
Secondary endpointsSerum matrix Gla protein (MGP); carboxylated vs undercarboxylated MGP fractions
Adherence threshold for per-protocol analysis≥85% of dispensed capsules taken

Results

Primary outcome — intention-to-treat (ITT)

GroupnMean CAC change (Agatston ± SEM)P-value
Vitamin K1 (500 µg/d)20027 ± 6—
Control (multivitamin only)18837 ± 7—
Between-group difference——0.26 (NS)

The primary analysis failed to show a significant treatment effect on CAC progression 1.

Per-protocol subgroup (≥85% adherence)

GroupnMean CAC change (Agatston)P-value
Vitamin K1 (≥85% adherent)14917—
Control (≥85% adherent)14637—
Between-group difference——0.03

Among the 295 participants meeting the adherence threshold, CAC progression was significantly lower in the vitamin K1 arm 2. This is the result most cited in downstream reviews. Readers should note: the ≥85% adherence threshold was predefined (stated as a secondary analysis in the methods), but it was not the pre-registered primary endpoint. The total number of participants meeting the ≥85% adherence criterion was 367; the analytical n of 295 (149+146) reflects those with all covariate data available for the secondary analysis.

Pre-existing CAC subgroup (baseline Agatston >10)

GroupnMean CAC change (Agatston)P-value
Vitamin K18125—
Control8959—
Between-group difference——0.03

The treatment effect was concentrated in participants with pre-existing calcification at baseline — the sub-population where clinically meaningful attenuation would matter most 3.

Matrix Gla protein (MGP) secondary endpoints

Treatment groups showed divergent MGP carboxylation responses (p ≤ 0.03 across analyses); however, controlling statistically for MGP changes did not attenuate the observed CAC progression effect. The authors conclude that vitamin K’s influence on CAC operated independently of its measurable effect on total MGP concentrations — suggesting either that MGP fraction assays were imprecise, that the mechanistic route involves other vitamin K-dependent proteins (e.g., Gas6, protein-S), or that MGP concentration is a poor surrogate for functional MGP activity. gap/no-mechanism

Critical caveats for wiki cross-referencing

1. Vitamin K1 (phylloquinone), NOT K2

This trial tested phylloquinone (K1), the dietary form found in leafy greens. The wiki’s primary hypothesis for D3 + K2 cardiovascular synergy concerns menaquinone-7 (MK-7, vitamin K2), which has longer half-life, better extrahepatic (vascular) tissue distribution, and distinct carboxylation kinetics. Results from this K1 trial cannot be directly extrapolated to K2/MK-7 supplementation without explicit bridging evidence. See vitamin-k2-mk7 and vascular-calcification.

2. Headline result is a pre-specified secondary / adherent-subgroup finding

The ITT primary analysis was null (p=0.26). The “vitamin K slows CAC progression” claim rests on the ≥85%-adherent subgroup (n=295; adherence threshold was predefined per the methods) and the pre-existing-CAC exploratory subgroup (n=170). The per-protocol subgroup was a pre-specified secondary analysis (not post-hoc), but it was not the pre-registered primary endpoint. The pre-existing-CAC subgroup was a hypothesis-generating exploratory analysis. This hierarchy matters when weighting the evidence.

3. This trial dominates the Li 2023 CAC meta-analysis

This single trial contributes approximately 94% of the weight in the Li 2023 meta-analysis of vitamin K supplementation and CAC progression — meaning the meta-analytic result is essentially a restatement of this trial’s subgroup finding, not independent replication. The vascular-calcification page discusses this concentration of evidence. Genuine independent replication in a K2/MK-7 trial or a larger K1 trial is needed before treating the meta-analysis as convergent evidence. gap/needs-replication

DimensionStatusNotes
Pathway conserved in humans?yesHuman RCT; MGP carboxylation pathway directly tested
Phenotype (CAC) conserved?yesHard imaging endpoint in humans
Replicated in humans?noThis trial dominates the only meta-analysis; no independent K1 or K2 RCT for CAC with comparable power

Context and limitations

Strengths:

  • Double-blind RCT in well-characterised cohort (Framingham offspring/spouse); 3-year duration; validated CT CAC endpoint
  • Multiple subgroup analyses pre-specified or clearly labelled
  • Open-access full text (PMC2682995) permits independent verification

Limitations:

  • ITT primary endpoint was null; secondary/subgroup findings require replication
  • K1 form limits direct relevance to K2/MK-7-based intervention literature
  • 500 µg/d K1 is a pharmacological dose (~4–5Ă— dietary reference intake); effects may not extrapolate to dietary K1 variation
  • Healthy community-dwelling participants; limited generalisability to populations with high vascular disease burden
  • MGP mechanistic interpretation is ambiguous — effect independent of measured MGP changes
  • Relatively modest sample (n=388); the pre-existing CAC subgroup (n=170) is underpowered for definitive conclusions

Significance and wiki role

This is the primary human RCT anchor for claims linking vitamin K supplementation to reduced vascular calcification. Its methodological details — K1 not K2; ITT null with per-protocol subgroup positive; near-total domination of the only CAC meta-analysis — must be explicitly preserved whenever it is cited on downstream wiki pages. The per-protocol subgroup result is hypothesis-generating; a dedicated K2/MK-7 RCT with pre-existing CAC enrichment and ITT as primary endpoint would resolve the evidence gap.

Cross-references

PageRelationship
vascular-calcificationProcess page discussing CAC biology and the Li 2023 meta-analysis this trial dominates
vitamin-k1Compound page for phylloquinone — form tested in this trial
vitamin-k2-mk7MK-7 compound page — distinct form; this trial’s results do not directly support K2 claims
matrix-gla-proteinProposed mechanistic mediator; secondary endpoint in this trial
deregulated-nutrient-sensingVitamin K-dependent carboxylation intersects with nutrient-sensing hallmark

Gaps and open questions

gap/needs-replication — This single trial dominates the entire meta-analytic CAC literature for vitamin K. An independent RCT — ideally using MK-7 (K2), enriched for pre-existing CAC, with ITT CAC progression as the pre-registered primary endpoint — is needed before treating “vitamin K slows vascular calcification” as a settled claim.

gap/no-mechanism — The vitamin K1 effect on CAC progression was independent of measurable MGP concentration changes. The mechanism by which K1 attenuates calcification (if the effect is real) remains unclear: alternative K-dependent vascular proteins, local carboxylation kinetics, or unmeasured MGP fractions are candidate explanations.

gap/dose-response-unclear — The trial used a single pharmacological dose (500 µg/d K1). Whether lower dietary doses, different forms (K2/MK-7), or different delivery schedules produce equivalent effects is unknown.

Footnotes

Footnotes

  1. doi:10.3945/ajcn.2008.27338 (this page) · n=388 (200 treatment, 188 control) · rct · p=0.26 (NS) · model: healthy older men and postmenopausal women, community-dwelling; Framingham offspring/spouse cohort · ITT primary analysis; full text confirmed via PMC2682995 ↩

  2. doi:10.3945/ajcn.2008.27338 (this page) · n=295 (149 treatment, 146 control; ≥85% adherent) · rct (per-protocol subgroup) · p=0.03 · model: same cohort, adherent participants only; secondary analysis ↩

  3. doi:10.3945/ajcn.2008.27338 (this page) · n=170 (81 treatment, 89 control; baseline Agatston >10) · rct (exploratory subgroup) · p=0.03 · model: pre-existing CAC enriched; mean Agatston change 25 vs 59 units; exploratory post-hoc analysis ↩

Graph View

Backlinks