I thought this was worth chatGPT5.6max paid
Worked for 6m 47s
Overall assessment
The DANCODE trial provides fairly strong evidence that 720 μg/day of MK-7 plus 25 μg/day of vitamin D3 slows the rise in coronary calcium in older people who already have severe calcification.
It does not demonstrate removal of existing calcium, regression of atherosclerosis, fewer cardiovascular events, or benefit from vitamin K2 alone.
Summary
DANCODE was a three-centre Danish, randomized, double-blind, placebo-controlled trial:
398 participants with CAC scores ≥400 Agatston units.
Median age 71; 30% women.
Median baseline CAC 903; 44% had CAC ≥1000.
Approximately 87% were taking statins.
Intervention: MK-7 720 μg plus vitamin D3 25 μg—equivalent to 1,000 IU—daily for 24 months.
Noncontrast CT was performed at baseline, 12 and 24 months.
Twenty-four-month scans were available for 356 participants: 90% of the intervention group and 89% of placebo.
Principal results
Outcome over 24 months K2+D3 Placebo Between-group result
CAC-score increase 196 AU 248 AU −52 AU (95% CI −79 to −24), P<0.001
Calcified plaque-volume increase* 15.0 mm³ 22.8 mm³ −7.81 mm³, P=0.007
Total plaque-volume increase* 21.0 mm³ 30.5 mm³ −9.49 mm³, P=0.34
Noncalcified plaque-volume increase* 5.9 mm³ 7.7 mm³ −1.75 mm³, P=0.85
*Selected CT-angiography subset of 143 participants.
The 52-AU difference represents 21% less progression relative to placebo progression, not a 21% reduction in the participants’ total calcium burden. Calcium increased substantially in both groups. Despite the trial’s “decalcification” name, it did not demonstrate decalcification.
The effect was directionally similar:
Men: −47 AU.
Women: −61 AU.
Baseline CAC 400–999: −39 AU.
Baseline CAC ≥1000: −70 AU.
The interaction tests were nonsignificant, although the subgroups were not sufficiently powered to establish that the effect is genuinely identical across sex or disease severity.
Mechanistic and safety findings
Inactive matrix Gla protein, dp-ucMGP, changed from approximately 558 to 383 pmol/L with supplementation, versus 560 to 591 pmol/L with placebo. Vitamin D increased from approximately 65 to 90 nmol/L. These findings confirm substantial biological exposure and are consistent with enhanced vitamin-K-dependent MGP carboxylation.
Only five major cardiovascular events occurred—two versus three—so the trial provides essentially no information about heart attacks, strokes, revascularization or mortality. No important safety signal emerged, but 398 participants followed for two years cannot exclude uncommon or longer-term adverse effects.
What is genuinely novel?
- Prospective confirmation in severe CAC
The earlier AVADEC analysis found a signal only in a prespecified subgroup of men with CAC ≥400. DANCODE deliberately recruited this high-progression population and confirmed the signal in a trial designed around CAC as its primary outcome.
- Extension to women
The severe-CAC AVADEC evidence was restricted to men. DANCODE included women and found no evidence of a sex interaction, although only 120 participants were women.
- Convergence with an independent K2-only trial
DANCODE is not the first positive randomized trial. The 2026 VitaK-CAC trial found that 360 μg/day MK-7 alone slowed CAC progression in 167 people with milder CAC scores of 50–400. Together, VitaK-CAC and DANCODE provide more persuasive evidence for a vitamin-K-related effect than either trial alone.
- Plaque-composition evidence
The CT-angiography substudy attempts to address an important concern: that reducing macroscopic calcification might leave more lipid-rich, potentially vulnerable plaque. It found less calcified-plaque progression without a detected increase in noncalcified plaque. This is useful mechanistic evidence, but less conclusive than the authors imply.
Strengths
Proper randomized, double-blind, placebo-controlled design.
Good allocation concealment and blinded central CT assessment.
Balanced baseline characteristics and almost identical attrition between groups.
Repeated measurements at 12 and 24 months.
A prespecified primary endpoint with a clear, statistically strong result.
Results reportedly remained similar after log-transforming CAC, adjusting for baseline CAC and adjusting for statin use.
Biochemical evidence of target engagement.
The findings are coherent with both AVADEC and VitaK-CAC.
These features make confounding or simple measurement bias an unlikely explanation for the primary CAC result.
Important criticisms
- CAC progression is not a validated treatment surrogate
High CAC and naturally occurring CAC progression predict cardiovascular risk, but this does not prove that deliberately slowing CAC progression improves outcomes.
Statins illustrate the problem: they reduce cardiovascular events while sometimes increasing calcium density as plaques become more fibrotic and stable. Therefore, the direction of a treatment-induced CAC change cannot automatically be translated into benefit.
There is also no validated “minimum clinically important difference” for CAC progression. The confidence interval indicates 24–79 AU less progression, but its meaning for myocardial infarction risk is unknown.
- The trial cannot identify the active ingredient
DANCODE compared:
MK-7 + D3 versus placebo
It did not contain K2-only, D3-only or factorial arms. Consequently it cannot determine:
Whether K2 produced the effect.
Whether vitamin D contributed.
Whether the combination is synergistic.
Whether D3 reduced or enhanced the K2 effect.
VitaK-CAC makes K2 the more likely active component, but that is an inference across trials, not a DANCODE finding.
- The plaque substudy is considerably weaker than the primary result
Only 143 of 398 participants entered the plaque analysis, restricted to CAC <1000 and acceptable scans at both time points. Randomization was no longer well protected in this selected subset: baseline total plaque volume was 237 mm³ in placebo versus 193 mm³ in the intervention group.
Although baseline-adjusted sensitivity analyses reportedly produced similar results, residual selection and regression-to-the-mean concerns remain.
More importantly, “no increase in noncalcified plaque” should be phrased as no statistically detected between-group difference. Its 95% confidence interval ranged from approximately 19.5 mm³ less to 16.0 mm³ more noncalcified plaque with treatment. That interval does not establish equivalence or exclude a clinically relevant increase.
- There are prespecification and reporting discrepancies
The public statistical analysis plan raises several questions:
The paper states that the SAP was finalized and uploaded before the first participant was enrolled. However, the registry lists the actual study start as 8 February 2023, whereas SAP version 1.0 is dated 24 February 2023. An earlier plan or a distinction between study start and first randomization may explain this, but the apparent discrepancy should be clarified.
The SAP required a 100-imputation sensitivity analysis if more than 5% of 24-month observations were missing. Forty-two of 398 participants—10.6%—lacked a 24-month scan, yet this prespecified analysis is not reported among the paper’s listed sensitivity analyses.
The SAP specified a random-intercept model; the article added random slopes. This may be statistically reasonable, but it is a model change requiring explanation.
The SAP planned to report aortic-valve calcification, quality of life and low-attenuation noncalcified plaque in the primary publication. These are absent.
The planned cardiovascular composite comprised myocardial infarction, revascularization and cardiac death; the article instead describes myocardial infarction, stroke and revascularization.
None of these observations readily explains away the strong primary result, particularly given balanced attrition. They do, however, weaken the paper’s assertion that no important prespecified changes occurred.
- Generalizability is narrow
The results apply most directly to:
Older Danish patients.
Severe CAC ≥400.
Patients investigated because of possible coronary symptoms.
Predominantly male participants.
A population already receiving substantial preventive therapy.
People not taking vitamin-K antagonists and without previous coronary revascularization.
They should not be extrapolated directly to younger people, CAC-zero populations, general prevention, other K2 forms or doses, or people taking warfarin-type anticoagulants.
- Funding and replication
The trial was investigator initiated, but Kappa Bioscience/Balchem provided financial support, and Kappa and Orkla donated the tablets. The companies are reported to have had no role in design, analysis or publication. The design and blinding reduce concern, but fully independent replication would still be valuable.
Bottom line
My assessment is:
High confidence: the tested K2+D3 regimen modestly slows CT-measured CAC progression in this severe-CAC population.
Moderate confidence: the effect involves vitamin-K-dependent MGP biology.
Low or no confidence: that the effect represents plaque regression, improves plaque stability, prevents cardiovascular events or extends life.
Scientifically, this is an important and fairly convincing proof that the coronary calcification pathway is pharmacologically modifiable. Clinically, it is not yet practice-changing. The decisive next study would be a larger, longer, preferably 2×2 factorial K2/D3 trial powered for cardiovascular events and including rigorous plaque-phenotype imaging and adherence measurement.
